Demencia NICE

National Institute for Health and Care
Excellence
Final
Dementia
Assessment, management and support for
people living with dementia and their carers
NICE Guideline 97
Methods, evidence and recommendations
June 2018
Final
Commissioned by the National Institute for

Health and Care Excellence
Dementia - assessment, management and support
Contents
Disclaimer
Healthcare professionals are expected to take NICE clinical guidelines fully into account
when exercising their clinical judgement. However, the guidance does not override the
responsibility of healthcare professionals to make decisions appropriate to the circumstances
of each patient, in consultation with the patient and/or their guardian or carer.
Copyright
© NICE 2018. All rights reserved. See Notice of rights.
ISBN: 978-1-4731-2978-8

Contents
Contents
Context............................................................................................................................... 10
1 Guideline committee membership and NICE technical team .................................. 13
1.1 Guideline committee............................................................................................ 13
1.2 Guideline social care subgroup ........................................................................... 15
1.3 NICE Centre for Guidelines Team ....................................................................... 15
1.3.1 Peer Review ............................................................................................ 17
2 Strength of recommendation ..................................................................................... 18
3 Methods ...................................................................................................................... 19
3.1 Evidence synthesis and meta-analyses ............................................................... 19
3.2 Evidence of effectiveness of interventions ........................................................... 19
3.2.1 Quality assessment .................................................................................. 19
3.2.2 Methods for combining intervention evidence .......................................... 19
3.2.3 Minimal clinically important differences (MIDs) ......................................... 20
3.2.4 GRADE for pairwise meta-analyses of interventional evidence ................ 21
3.3 Methods for combining direct and indirect evidence (network meta-analysis)
for interventions ................................................................................................... 22
3.3.1 Synthesis ................................................................................................. 22
3.3.2 Modified GRADE for network meta-analyses ........................................... 22
3.4 Diagnostic test accuracy evidence ...................................................................... 23
3.4.2 Methods for combining diagnostic test accuracy evidence ....................... 24
3.4.3 Modified GRADE for diagnostic test accuracy evidence ........................... 25
3.5 Qualitative evidence ............................................................................................ 26
3.5.2 Methods for combining qualitative evidence ............................................. 26
3.5.3 CERQual for qualitative studies ............................................................... 26
3.6 Health economics ................................................................................................ 27
4 Summary of recommendations ................................................................................. 29
4.1 Recommendations summary ............................................................................... 29
4.2 Research recommendations summary ................................................................ 43
5 Dementia diagnosis.................................................................................................... 45
5.1 Dementia diagnosis ............................................................................................. 46
5.1.1 Introduction .............................................................................................. 46
5.1.2 Evidence review ....................................................................................... 47
5.1.3 Health economic evidence ....................................................................... 57
5.1.4 Evidence statements ................................................................................ 64
5.1.5 Evidence to recommendations ................................................................. 99
5.1.6 Recommendations ................................................................................. 110

4
Contents
5.1.7 Research recommendations .................................................................. 113

5.2 Distinguishing dementia from delirium or delirium with dementia ....................... 114
5.2.1 Introduction ............................................................................................ 114
5.2.2 Evidence review ..................................................................................... 114
5.2.3 Health economic evidence ..................................................................... 116
5.2.4 Evidence statements .............................................................................. 116
5.2.5 Evidence to recommendations ............................................................... 121
5.3 Case finding for people at high risk of dementia ................................................ 124
5.3.1 Introduction ............................................................................................ 124
5.3.2 Evidence review ..................................................................................... 124
6 Involving people living with dementia in decisions about care ............................ 128
6.1 Barriers and facilitators to involvement in decision making for people living
with dementia .................................................................................................... 129
6.1.1 Introduction ............................................................................................ 129
6.1.2 Evidence review ..................................................................................... 129
7 Care planning, review and co-ordination ................................................................ 138
7.1 Health and social care co-ordination.................................................................. 139
7.1.1 Introduction ............................................................................................ 139
7.1.2 Evidence review ..................................................................................... 140
7.2 Post diagnosis review for people living with dementia ....................................... 159
7.2.1 Introduction ............................................................................................ 159
7.2.2 Evidence review ..................................................................................... 160

5
Contents

8 Inpatient care ............................................................................................................ 166
8.1 Caring for people living with dementia who are admitted to hospital .................. 167
8.1.1 Introduction ............................................................................................ 167
8.1.2 Evidence review ..................................................................................... 167
9 Care setting transitions............................................................................................ 174
9.1 Managing the transition between different settings for people living with
dementia ........................................................................................................... 175
9.1.1 Introduction ............................................................................................ 175
9.1.2 Evidence review ..................................................................................... 176
10 Modifying risk factors for dementia progression ................................................... 179
10.1 Risk factors for dementia progression ............................................................... 180
10.1.1 Introduction ............................................................................................ 180
10.1.2 Evidence review ..................................................................................... 180
11 Cholinesterase inhibitors and memantine for dementia........................................ 190
11.1 Cholinesterase inhibitors and memantine for people living with Alzheimer’s
disease .............................................................................................................. 191
11.1.1 Introduction ............................................................................................ 191
11.1.2 Evidence review ..................................................................................... 191
11.2 Co-prescription and withdrawal of cholinesterase inhibitors and memantine in
Alzheimer’s disease .......................................................................................... 200
11.2.1 Introduction ............................................................................................ 200
11.2.2 Evidence review ..................................................................................... 201

6
Contents

11.3 Pharmacological management of dementia with Lewy bodies ........................... 216
11.3.1 Introduction ............................................................................................ 216
11.3.2 Evidence review ..................................................................................... 217
11.4 Cholinesterase inhibitors and memantine for types of dementia other than
typical Alzheimer’s disease ............................................................................... 234
11.4.1 Introduction ............................................................................................ 234
11.4.2 Evidence review ..................................................................................... 235
12 Drugs that may worsen cognitive decline .............................................................. 248
12.1 Drugs that may cause cognitive decline ............................................................ 249
12.1.1 Introduction ............................................................................................ 249
12.1.2 Evidence review ..................................................................................... 249
13 Non-pharmacological interventions for people living with dementia ................... 254
13.1 Pre-, peri- and post-diagnostic counselling and support for people living with
dementia and their families ................................................................................ 255
13.1.1 Introduction ............................................................................................ 255
13.1.2 Evidence review ..................................................................................... 255
13.2 Interventions to promote cognition, independence and wellbeing ...................... 264
13.2.1 Introduction ............................................................................................ 264
13.2.2 Evidence review ..................................................................................... 264

7
Contents

14 Managing non-cognitive symptoms ........................................................................ 298
14.1 Interventions for treating illness emergent non-cognitive symptoms in people
living with dementia ........................................................................................... 299
14.1.1 Introduction ............................................................................................ 299
14.1.2 Evidence review ..................................................................................... 300
15 Supporting informal carers ...................................................................................... 332
15.1 Supporting informal carers of people living with dementia ................................. 333
15.1.1 Introduction ............................................................................................ 333
15.1.2 Evidence review ..................................................................................... 334
16 Staff training ............................................................................................................. 351
16.1 Staff training ...................................................................................................... 352
16.1.1 Introduction ............................................................................................ 352
16.1.2 Evidence review ..................................................................................... 353
16.1.5 Evidence to recommendation ................................................................. 363
17 Needs of younger people living with dementia ...................................................... 368
17.1 The specific needs of younger people living with dementia ............................... 369
17.1.1 Introduction ............................................................................................ 369
17.1.2 Evidence review ..................................................................................... 369

8
Contents

18 Assessing and managing comorbidities ................................................................ 374
18.1 Assessing and treating intercurrent illness in people living with dementia ......... 375
18.1.1 Introduction ............................................................................................ 375
18.1.2 Evidence review ..................................................................................... 376
18.2 Management strategies for people living with dementia and co-existing
physical long term conditions............................................................................. 392
18.2.1 Introduction ............................................................................................ 392
18.2.2 Evidence review ..................................................................................... 393
18.3 Managing mental health conditions alongside dementia .................................... 401
18.3.1 Introduction ............................................................................................ 401
18.3.2 Evidence review ..................................................................................... 402
19 Palliative care: care towards and at the end of life ................................................ 404
19.1 Palliative care .................................................................................................... 405
19.1.1 Introduction ............................................................................................ 405
19.1.2 Evidence review ..................................................................................... 406
20 Glossary ................................................................................................................... 417

9
Guideline committee membership and NICE technical team
Context
Dementia is a term used to describe a range of cognitive and behavioural symptoms that can
include memory loss, problems with reasoning and communication and change in
personality, and a reduction in a person's ability to carry out daily activities, such as
shopping, washing, dressing and cooking. The most common types of dementia are:
Alzheimer's disease, vascular dementia, mixed dementia, dementia with Lewy bodies and
frontotemporal dementia. Dementia is a progressive condition, which means that the
symptoms will gradually get worse. This progression will vary from person to person and
each will experience dementia in a different way – people may often have some of the same
general symptoms, but the degree to which these affect each person will vary (Dementia
Gateway, Social Care Institute for Excellence).
A report published by the Alzheimer’s Society found that in 2013 there were approximately
815,000 people living with dementia in the UK. If current trends continue, this number is
expected to increase to 1,143,000 by 2025. In England, the National Dementia and
Antipsychotic Prescribing Audit found that approximately 31,000 people were newly
diagnosed with dementia in 2011. This is an increase of 8% between 2006 and 2011. Finally,
in December 2017, there were 456,739 people on GP registers with a formal diagnosis of
dementia, up from approximately 290,000 people in 2009/10, with the majority of this
difference accounted for by an increase in diagnosis rates.
The Alzheimer’s Society report found that in 2013 the total cost of dementia in the UK was
estimated to be £26.3 billion. Of this, approximately £4.3 billion consists of health care, and
approximately £10.3 billion consists of social care. The remaining £11.6 billion accounts for
estimated unpaid care contributions.
Why is it needed?
Providing care and support is very complex, because of the number of people living with
dementia and the variation in the symptoms each person faces. This has led to considerable
variation in practice. Areas that pose particular challenges for services and practitioners may
include:
 coordinating care and support between different services
 what support carers need, and how this should be provided
 staff training.
This guideline makes evidence-based recommendations aiming to support these areas of

practice.
Dementia also has significant costs for health and social care services. Because of this, it is
important to ensure that people living with dementia can get the care and support they need,
and that services provide this in an efficient and cost-effective way.
In addition, new methods for diagnosing and assessing dementia have been developed.
Amyloid imaging techniques have been licensed for use in the UK, and new evidence is
available for cerebrospinal fluid examination. There is also evidence on different approaches
to assess and diagnose dementia subtypes. The guideline makes new recommendations on
dementia diagnosis, based on a review of the latest evidence.

10
What does it cover?

This guideline addresses how dementia should be assessed and diagnosed. It covers
person-centred care and support, tailored to the specific needs of each person living with
dementia. As part of this, it can help professionals involve people living with dementia and
their carers in decision-making, so they can get the care and support they need. It also
addresses care coordination and staff training, and how dementia may impact on the care
offered for other conditions.
The guideline does not cover every aspect of dementia care or support, or areas where
recommendations would be the same for people with or without dementia. It focuses on
areas where:
 there is variation in practice, and enough evidence is available to identify what works best
 people living with dementia need different care and support to people in the same
situation who do not have dementia.
How has it been developed?

This guideline has been developed by a multidisciplinary guideline committee, using an
extensive review of research evidence. To ensure that the committee had the necessary
social care expertise, a subgroup of social care practitioners was recruited to develop
recommendations in this area.
Given the costs of dementia and the financial pressures facing health and social care
services, the committee focused on making recommendations in areas where there is good
evidence available. This will help services make the most of limited resources. For areas with
a lack of evidence, the committee has made recommendations for future research (on health
and social care topics) to address gaps in the evidence base. Future updates of the guideline
will look at any relevant new research that has been published.
Some recommendations are made with more certainty than others. We word our
recommendations to reflect this. In the sections on interventions we use 'offer' to reflect a
strong recommendation, usually where there is clear evidence of benefit. We use 'consider'
to reflect a recommendation for which the evidence of benefit is less certain. For more
information see making decisions using NICE guidelines.
How does it relate to statutory and non-statutory

guidance?
The guideline complements existing legislation and guidance. It describes how services and
professionals can provide high-quality care and support.
The Prime Minister’s Challenge on Dementia 2020 sets out the UK Government's strategy for
transforming dementia care within the UK. The aims of the strategy include:
 improving diagnosis, assessment and care for people living with dementia
 ensuring that all people living with dementia have equal access to diagnosis
 providing all NHS staff with training on dementia appropriate to their role
 ensuring that every person diagnosed with dementia receives meaningful care.

11
Since the 2006 NICE guideline on dementia was developed, key new legislation has been
implemented. The Care Act 2014 created a new legislative framework for adult social care,
and also gives carers a legal right to assessment and support.
Relevant legislation and statutory guidance

 NHS England (2015) Accessible Information Standard
 Care Act 2014
 Health and Social Care Act 2008 (Regulated Activities) Regulations 2014
 Department of Health (2014) Care Act 2014: Statutory Guidance for Implementation
 Department of Health (2014) Positive and Proactive Care: Reducing the need for
restrictive interventions
 Health and Social Care Act 2012
 Equality Act 2010
 Mental Capacity Act 2005
 Human Rights Act 1998
Relevant policies and non-statutory guidance

 Information Commissioner’s Office (2017) Guide to the General Data Protection
Regulation
 NHS England (2017) Dementia: Good Care Planning
 NHS England (2015) Implementation guide and resource pack for dementia care
 Skills for Health, Health Education England and Skills for Care (2015) Dementia Core
Skills Education and Training Framework. This framework was commissioned and funded
by the Department of Health and developed in collaboration by Skills for Health and
Health Education England in partnership with Skills for Care
 Department of Health (2014) NHS Outcomes Framework 2015 to 2016
 Department of Health (2014) Adult Social Care Outcomes Framework 2015 to 2016

12
1 Guideline committee membership and
NICE technical team
1.1 Guideline committee
Damien Longson (committee chair)
Consultant Liaison Psychiatrist, Greater Manchester Mental Health Foundation Trust
Louise Allan
Professor of Geriatric Medicine, University of Exeter
Joanne Brady (Co-opted member)
Consultant in Palliative Medicine, Royal Free London NHS Foundation Trust and North
London Hospice
Linda Clare
Professor of Clinical Psychology of Ageing and Dementia, University of Exeter
Richard Clibbens
Nurse Consultant, Wakefield Memory Service, South West Yorkshire Partnership NHS
Foundation
Carol Duff
Consultant Occupational Therapist, Older Adults Specialist Mental Health Services,

Lincolnshire Partnership NHS Foundation Trust
Paul Dunnery (Resigned November 2016)
Lay member – organisation (Alzheimer’s Society)
Sandra Evans
Consultant Psychiatrist and Lecturer in Psychiatry, St Bartholomew’s Hospital, London
Wayne Goddard (From March 2017)
Head of Strategy and Delivery Integrated Dementia Lead, Doncaster CCG
Kim Grosvenor (From March 2017)
Senior Manager – Dementia Transformation, HWLH CCG
13
Karen Harrison Dening
Head of Research & Publications, Admiral Nurse, Dementia UK
Jeremy Isaacs (Co-opted member)
Consultant Neurologist and Dementia Clinical Lead, St George’s University Hospitals NHS
Foundation Trust
Hannah Luff
Lead Speech & Language Therapist, South London and Maudsley NHS Foundation Trust
Kate Mitchell (Co-opted member) (Resigned November 2016)
Commissioner and Programme Lead for Long Term Conditions, NHS Kernow
Kevin Minier
Lay member – carer for a person living with dementia
John O’Brien
Professor of Old Age Psychiatry, University of Cambridge
Ruth O’Dea
Care Home Manager, Somerset Redstone Trust
Catherine Pascoe (Co-opted member) (Resigned October 2016)
Commissioning Manager Adult Services, Hampshire County Council
Sarah Partington (Co-opted member)
Community matron, ANP complex care, Bradford District Care NHS Foundation Trust
Chris Roberts
Lay member – person living with dementia
Louise Robinson
GP and Professor of Primary Care and Ageing, Newcastle University
Tracy Wright
Directorate Manager – (Social Worker) Directorate Manager HMR Integrated Teams,

Pennine Acute
14
1.2 Guideline social care subgroup
Tracey Wright (Subgroup chair) (Guideline committee member)
(Social Worker) Directorate Manager HMR Integrated Teams, Pennine Acute
Belinda Black
Chief Executive, Sheffcare
Sally English
Practice Manager, East Sussex County Council
Maggie Murdoch
Lay member – carer for a person living with dementia
Pauline Shaw
Director of Care & Service Development, the Royal Star & Garter Homes, West Midlands
Ruth O’Dea (Guideline committee member)
Care Home Manager, Somerset Redstone Trust
Ben Williams
Practice Excellence Advisor, Genesis Housing Association
For a full list of guideline development group and declarations of interest, see Appendix A.
1.3 NICE Centre for Guidelines Team

Harry Allen (Until January 2017)
Consultant Clinical Adviser
Sohaib Ashraf (From November 2016)
Health Economist
Elizabeth Barrett
Information Specialist
Jean Bennie (From June 2017)
Technical Analyst
15
Daniel Davies (From April 2016 until March 2017)
Project Manager
Sue Ellerby (From December 2015 until March 2017)
Consultant Clinical Adviser
Jamie Elvidge (From July until November 2016)
Health Economist
Victoria Gillis-Elliott
Technical Analyst
James Hall
Editor
Marie Harrisingh (From November 2016)
Technical Analyst
Holly Irwin (Until January 2016)
Project Manager
Yolanda Martinez (From February to May 2017)
Technical Analyst
Toby Mercer (From November 2016)
Technical Analyst
Michael Mellors (From November 2016 until March 2017)
Social Care Adviser
Kate McAllister (From September until November 2016)
Technical Analyst
Hugh McGuire (Until December 2015)
Technical Adviser
Sarah Mills (From February 2016 until April 2016)
Project Manager
16
Vonda Murray (From February 2016 until April 2016) & (From March 2017)
Project Manager
Angela Parkin
Medicines Adviser
Joshua Pink (From February 2016)
Technical Adviser
Gabriel Rogers
Technical Adviser, Health Economics
Susan Spiers
Associate Director
Jeffrey Tabiri-Essuman (From April to May 2017)
Technical Analyst
Steven Ward (Until April 2016)
Health Economist
Verena Wolfram (From July 2017)
Technical Analyst
1.3.1 Peer Review
Jane Silvester
Associate Director - Guidance and Quality Assurance management, Implementation
For social care recommendations
17
2 Strength of recommendation
Some recommendations can be made with more certainty than others. The Guideline
committee makes a recommendation based on the trade-off between the benefits and harms
of an intervention, taking into account the quality of the underpinning evidence. For some
interventions, the Guideline committee is confident that, given the information it has looked
at, most patients would choose the intervention. The wording used in the recommendations
in this guideline denotes the certainty with which the recommendation is made (the strength
of the recommendation).
For all recommendations, NICE expects that there is discussion with the patient about the
risks and benefits of the interventions, and their values and preferences. This discussion
aims to help them to reach a fully informed decision (see also ‘Patient-centred care’).
Interventions that must (or must not) be used

We usually use ‘must’ or ‘must not’ only if there is a legal duty to apply the recommendation.
Occasionally we use ‘must’ (or ‘must not’) if the consequences of not following the
recommendation could be extremely serious or potentially life threatening.
Interventions that should (or should not) be used – a

‘strong’ recommendation
We use ‘offer’ (and similar words such as ‘refer’ or ‘advise’) when we are confident that, for
the vast majority of patients, an intervention will do more good than harm, and be cost
effective. We use similar forms of words (for example, ‘Do not offer…’) when we are
confident that an intervention will not be of benefit for most patients.
Interventions that could be used

We use ‘consider’ when we are confident that an intervention will do more good than harm
for most patients, and be cost effective, but other options may be similarly cost effective. The
choice of intervention, and whether or not to have the intervention at all, is more likely to
depend on the patient’s values and preferences than for a strong recommendation, and so
the healthcare professional should spend more time considering and discussing the options
with the patient.
18
3 Methods
This guideline was developed in accordance with the process set out in ‘Developing NICE
guidelines: the manual (2014)’. There is more information about how NICE clinical guidelines
are developed on the NICE website. A booklet, ‘How NICE clinical guidelines are developed:
an overview for stakeholders, the public and the NHS’ is available. In instances where the
guidelines manual does not provide advice, additional methods are used as described below,
organised by study type.
3.1 Evidence synthesis and meta-analyses

Where possible, meta-analyses were conducted to combine the results of quantitative
studies for each outcome. For continuous outcomes analysed as mean differences, where
change from baseline data were reported in the trials and were accompanied by a measure
of spread (for example standard deviation), these were extracted and used in the meta-
analysis. Where measures of spread for change from baseline values were not reported, the
corresponding values at study end were used and were combined with change from baseline
values to produce summary estimates of effect. These studies were assessed to ensure that
baseline values were balanced across the treatment groups; if there were significant
differences at baseline these studies were not included in any meta-analysis and were
reported separately. For continuous outcomes analysed as standardised mean differences,
where only baseline and final time point values were available, change from baseline
standard deviations were estimated, assuming a correlation coefficient of 0.5.
3.2 Evidence of effectiveness of interventions

3.2.1 Quality assessment
GRADE was used to assess the quality of evidence for the selected outcomes as specified in
‘The guidelines manual (2014)’. Where RCTs are available, these are initially rated as high
quality and the quality of the evidence for each outcome was downgraded or not from this
initial point. If non-RCT evidence was included for intervention-type systematic reviews then
these are initially rated as low quality and the quality of the evidence for each outcome was
downgraded or not from this point.
Individual RCTs, cohort studies and case-control studies were quality assessed using the
CASP RCT, cohort study and case-control checklists, respectively. Each individual study was
classified as being either at low, moderate or high risk of bias based on that assessment.
3.2.2 Methods for combining intervention evidence
Meta-analysis of interventional data was conducted with reference to the Cochrane

Handbook for Systematic Reviews of Interventions (Higgins et al. 2011).
Where different studies presented continuous data measuring the same outcome but using
different numerical scales (e.g. a 0-10 and a 0-100 visual analogue scale), these outcomes
were all converted to the same scale before meta-analysis was conducted on the mean
differences. Where outcomes measured the same underlying construct but used different
instruments/metrics, data were analysed using standardised mean differences (Hedges’ g).
A pooled relative risk was calculated for dichotomous outcomes (using the Mantel–Haenszel
method).
19
Fixed- and random-effects models (der Simonian and Laird) were fitted for all syntheses, with
the presented analysis dependent on the degree of heterogeneity in the assembled
evidence. Fixed-effects models were the preferred choice to report, but in situations where
the assumption of a shared mean for fixed-effects model were clearly not met, even after
appropriate pre-specified subgroup analyses were conducted, random-effects results are
presented. Fixed-effects models were deemed to be inappropriate if one or both of the
following conditions was met:
 Significant between study heterogeneity in methodology, population, intervention or
comparator was identified by the reviewer in advance of data analysis. This decision was
made and recorded before any data analysis was undertaken.
 The presence of significant statistical heterogeneity in the meta-analysis, defined as
I2≥40%.
Meta-analyses were performed in Cochrane Review Manager v5.3.
3.2.3 Minimal clinically important differences (MIDs)
The Core Outcome Measures in Effectiveness Trials (COMET) database was searched to
identify published minimal clinically important difference thresholds relevant to this guideline,
and this list was supplemented by any additional MIDs found through studies included in the
guideline. Identified MIDs were assessed to ensure they had been developed and validated
in a methodologically rigorous way, and were applicable to the populations, interventions and
outcomes specified in this guideline. In addition, the Guideline committee were asked to
prospectively specify any outcomes where they felt a consensus MID could be defined from
their experience. In particular, any questions looking to evaluate non-inferiority (that one
treatment is not meaningfully worse than another) required an MID to be defined to act as a
non-inferiority margin.
MIDs found through this process are given in Table 1. For other continuous outcomes not
specified in the table below, no MID was defined.
Table 1: Identified MIDs

Outcome MID Source
BADLS 3.5 Howard R, Phillips P, Johnson T, et al. Determining the
minimum clinically important differences for outcomes in the
DOMINO trial. Int J Geriatr Psychiatry 2011; 26: 812–817.
IDDD 5 Meeuwsen EJ, Melis RJF, van der Aa GCHM. Effectiveness of
dementia follow-up care by memory clinics or general
practitioners: randomised controlled trial. BMJ 2012;344:e3086.
MMSE 1.4 Howard R, Phillips P, Johnson T, et al. Determining the
NPI total score 8 Howard R, Phillips P, Johnson T, et al. Determining the
QoL-15D 0.03 Koivisto AM, Hallikainen I, Vlimki T, et al. Early psychosocial
intervention does not delay institutionalization in persons with
mild Alzheimer disease and has impact on neither disease
progression nor caregivers’ well-being: ALSOVA 3-year follow-
up. Int J Geriatr Psychiatry 2016; 31: 273–283.
QoL-AD 3 Meeuwsen EJ, Melis RJF, van der Aa GCHM, et al.
Effectiveness of dementia follow-up care by memory clinics or
general practitioners: randomised controlled trial. BMJ
2012;344:e3086.
20
Outcome MID Source
BADLS: Bristol Activities of Daily Living Scale
IDDD: Interview for deterioration in daily living activities in dementia
MMSE: Mini-mental State Examination
NPI: Neuropsychiatric Inventory
QoL-15D: 15D health related quality of life instrument
QoL-AD: Quality of life in Alzheimer’s disease instrument
For standardised mean differences where no other MID was available, an MID of 0.2 was
specified by the committee, corresponding to the threshold for a small effect size initially
suggested by Cohen et al. (1988). For dichotomous outcome measures, the committee
agreed that any changes in mortality, entry to long stay care and the proportions of people
achieving a clinically meaningful improvement would themselves be clinically meaningful,
whilst for other measures an MID interval of 0.8 to 1.25 was used.
The committee noted that the MIDs identified for specific outcome scales were all based on
the level of short-term change needed to make a meaningful difference to an individual with
an established diagnosis of dementia, and this made interpretation difficult when applied to
mean differences between groups, particularly because dementia is a highly heterogeneous
condition and therefore it would be expected there would be considerable variability between
individuals in the level of response to an intervention. Additionally, it would be likely that
smaller MIDs would be found for these outcomes in cases where the effects of interventions
persist in the longer-term. The committee also noted that since these MIDs were derived
from RCT data with clearly defined populations, it was unclear whether they would be
applicable for people with different subtypes of dementia, or at different stages of severity.
Therefore, it was agreed the above MIDs would not be used to downgrade for imprecision,
with the line of no effect being used instead, but they would be taken in to account by the
committee as parts of their discussions as to whether the findings of a review were clinically
meaningful.
When decisions were made in situations where MIDs were not available, the ‘Evidence to
Recommendations’ section of that review should make explicit the committee’s view of the
expected clinical importance and relevance of the findings. In particular, this includes
consideration of whether the whole effect of a treatment (which may be felt across multiple
independent outcome domains) would be likely to be clinically meaningful, rather than simply
whether each individual sub outcome might be meaningful in isolation.
3.2.4 GRADE for pairwise meta-analyses of interventional evidence
The quality of the evidence for each outcome was downgraded where appropriate for the
reasons outlined in Table 2
Table 2: Rationale for downgrading evidence for intervention studies

GRADE criteria Reasons for downgrading quality
Risk of bias The quality of the evidence was downgraded if there were concerns about the
design or execution of the study, including concealment of allocation, masking,
loss to follow up using intervention checklists in the NICE guidelines manual
(2014)
Inconsistency The quality of the evidence was downgraded if, after appropriate pre-specified
sensitivity analyses were conducted, there were remaining concerns about
inconsistency of effects across studies: occurring when there is variability in
the treatment effect demonstrated across studies (heterogeneity).
This was downgraded either if important differences were found between
populations, interventions and/or comparators across studies included in a
meta-analysis, or if there was significant unexplained statistical heterogeneity,
21
assessed using the I2 statistic, where I2 ≥ 40% was categorised as serious
inconsistency.
Indirectness The quality of the evidence was downgraded if there were concerns about the
population, interventions and outcomes in the included studies and how
directly these variables could address the specific review question.
Imprecision If MIDs other than the line of no effect (1 corresponding to meaningful benefit;
1 corresponding to meaningful harm) were defined for the outcome, the
outcome was downgraded once if the 95% confidence interval for the effect
size crossed 1 MID, and twice if it crossed both the upper and lower MIDs.
If an MID was not defined for the outcome, or the line of no effect was specified
as an MID, it was downgraded once if the 95% confidence interval for the
effect size crossed the line of no effect (i.e. the outcome was not statistically
significant), and twice if additionally the sample size of the study was
sufficiently small that it is not plausible any realistic effect size could have been
detected.
In situations where data was included, but only p values were available and not
confidence intervals, the data were downgraded once for imprecision if the
sample size of the study was less than 100.
3.3 Methods for combining direct and indirect evidence

(network meta-analysis) for interventions
Conventional pairwise meta-analysis involves the statistical combination of direct evidence
about pairs of interventions that originate from 2 or more separate studies (for example,
where there are two or more studies comparing A vs B).
In situations where there are more than 2 interventions, pairwise meta-analysis of the direct
evidence alone is of limited use. This is because multiple pairwise comparisons need to be
performed to analyse each pair of interventions in the evidence, and these results can be
difficult to interpret. Furthermore, direct evidence about interventions of interest may not be
available. For example studies may compare A vs B and B vs C, but there may be no direct
evidence comparing A vs C. Network meta-analysis (NMA) overcomes these problems by
combining all evidence into a single, internally consistent model, synthesising data from
direct and indirect comparisons, and providing estimates of relative effectiveness for all
comparators and the ranking of different interventions.
3.3.1 Synthesis
Frequentist NMAs were undertaken using the netmeta package in R v3.4.1. This uses a
graph-theoretical method which is mathematically equivalent to frequentist network meta-
analysis (Rücker 2012). Inconsistency was assessed using the overall I2 value for the whole
network, which is a weighted average of the I2 value for all comparisons where there are
multiple trials (both direct and indirect), and random-effects models were used if the I2 value
was above 50% (this was interpreted as showing the assumption of consistent, shared
underlying means was not met, and therefore a fixed-effects model was inappropriate).
3.3.2 Modified GRADE for network meta-analyses
A modified version of the standard GRADE approach for pairwise interventions was used to
assess the quality of evidence across the network meta-analyses undertaken. While most
criteria for pairwise meta-analyses still apply, it is important to adapt some of the criteria to
take into consideration additional factors, such as how each 'link' or pairwise comparison
within the network applies to the others. As a result, the following was used when modifying
22
the GRADE framework to a network meta-analysis. It is designed to provide a single overall
quality rating for an NMA, which can then be combined with pairwise quality ratings for
individual comparisons (if appropriate), to judge the overall strength of evidence for each
comparison.
Table 3: Rationale for downgrading quality of evidence for intervention studies

Risk of bias Not serious: If fewer than 33.3% of the studies in the network meta-analysis
were at moderate or high risk of bias, the overall network was not downgraded.
Serious: If greater than 33.3% of the studies in the network meta-analysis were
at moderate or high risk of bias, the network was downgraded one level.
Very serious: If greater than 33.3% of the studies in the network meta-analysis
were at high risk of bias, the network was downgraded two levels.
Indirectness Not serious: If fewer than 33.3% of the studies in the network meta-analysis
were partially indirect or indirect, the overall network was not downgraded.
Serious: If greater than 33.3% of the studies in the network meta-analysis were
partially indirect or indirect, the network was downgraded one level.
Very serious: If greater than 33.3% of the studies in the network meta-analysis
were indirect, the network was downgraded two levels.
Inconsistency N/A: Inconsistency was marked as not applicable if there were no links in the
network where data from multiple studies (either direct or indirect) were
synthesised.
For network meta-analyses conducted under a frequentist framework, the
network was downgraded one level if the I2 was greater than 50%.
In addition, the direct and indirect treatment estimates were compared as a
check on the consistency of the network.
Imprecision The overall network was downgraded for imprecision if it was not possible to
differentiate between any meaningfully distinct treatments options in the
network (based on 95% confidence/credible intervals). Whether two options
were meaningfully distinct was judged using the MIDs defined above for
pairwise meta-analysis of the outcomes, if available; or the line of no effect if
MIDs were not available.
3.4 Diagnostic test accuracy evidence

In this guideline, diagnostic test accuracy (DTA) data are classified as any data in which a
feature – be it a symptom, a risk factor, a test result or the output of some algorithm that
combines many such features – is observed in some people who have the condition of
interest at the time of the test and some people who do not. Such data either explicitly
provide, or can be manipulated to generate, a 2x2 classification of true positives and false
negatives (in people who, according to the reference standard, truly have the condition) and
false positives and true negatives (in people who, according to the reference standard, do
not).
The ‘raw’ 2x2 data can be summarised in a variety of ways. Those that were used for
decision making in this guideline are as follows:
 Positive likelihood ratios describe how many times more likely positive features are in
people with the condition compared with people without the condition. Values greater than
1 indicate that a positive result makes the condition more likely.
o LR+ = (TP/[TP+FN])/(FP/[FP+TN])
 Negative likelihood ratios describe how many times less likely negative features are in
people with the condition compared with people without the condition. Values less than 1
indicate that a negative result makes the condition less likely.
o LR- = (FN/[TP+FN])/(TN/[FP+TN])
23
 Sensitivity is the probability that the feature will be positive in a person with the condition.
o sensitivity = TP/(TP+FN)
 Specificity is the probability that the feature will be negative in a person without the
condition.
o specificity = TN/(FP+TN)
The following schema, adapted from the suggestions of Jaeschke et al. (1994), was used to
interpret the likelihood ratio findings from diagnostic test accuracy reviews.
Table 4: Interpretation of likelihood ratios

Value of likelihood ratio Interpretation
LR ≤ 0.1 Very large decrease in probability of disease
0.1 < LR ≤ 0.2 Large decrease in probability of disease
0.2 < LR ≤ 0.5 Moderate decrease in probability of disease
0.5 < LR ≤ 1.0 Slight decrease in probability of disease
1.0 < LR < 2.0 Slight increase in probability of disease
2.0 ≤ LR < 5.0 Moderate increase in probability of disease
5.0 ≤ LR < 10.0 Large increase in probability of disease
LR ≥ 10.0 Very large increase in probability of disease
The schema above has the effect of setting a minimal important difference for positive
likelihoods ratio at 2, and a corresponding minimal important difference for negative
likelihood ratios at 0.5. Likelihood ratios (whether positive or negative) falling between these
thresholds were judged to indicate no meaningful change in the probability of disease.
Individual studies were quality assessed using the QUADAS-2 tool, which contains four
domains: patient selection, index test, reference standard, and flow and timing. Each
individual study was classified as being either at low, moderate or high risk of bias based on
that assessment.
3.4.2 Methods for combining diagnostic test accuracy evidence
Meta-analysis of diagnostic test accuracy data was conducted with reference to the
Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy (Deeks et al.
2010).
Where applicable, diagnostic syntheses were stratified by:

 Presenting symptomatology (features shared by all participants in the study, but not all
people who could be considered for a diagnosis in clinical practice).
 The reference standard used for true diagnosis.
Where five or more studies were available for all included strata, a bivariate model was fitted
using the mada package in R v3.4.1, which accounts for the correlations between positive
and negative likelihood ratios, and between sensitivities and specificities. This model
requires five parameters to be fitted and is therefore not appropriate when only a small
number of studies are available (Reitsma et al. 2005). Where sufficient data were not
available (2-4 studies), separate independent pooling was performed for positive likelihood
ratios, negative likelihood ratios, sensitivity and specificity, using Microsoft Excel. This
approach is conservative as it is likely to somewhat underestimate test accuracy, due to
24
failing to account for the correlation and trade-off between sensitivity and specificity (see
Deeks 2010), but these errors in the majority of cases will not be large enough to
systematically affect decision making, and therefore an analysis was not marked down for
risk of bias solely due to being based on a univariate model.
Random-effects models (der Simonian and Laird) were fitted for all syntheses, as
recommended in the Cochrane Handbook for Systematic Reviews of Diagnostic Test
Accuracy (Deeks et al. 2010).
3.4.3 Modified GRADE for diagnostic test accuracy evidence
GRADE has not been developed for use with diagnostic studies; therefore a modified
approach was applied using the GRADE framework. GRADE assessments were only
undertaken for positive and negative likelihood ratios, as the MIDs used to assess
imprecision were based on these outcomes, but results for sensitivity and specificity are also
presented alongside those data.
Cross-sectional and cohort studies were initially rated as high-quality evidence if well
conducted, and then downgraded according to the standard GRADE criteria (risk of bias,
inconsistency, imprecision and indirectness) as detailed in Table 5 below.
Table 5: Rationale for downgrading quality of evidence for diagnostic questions

Risk of bias Not serious: If less than 33.3% of the weight in a meta-analysis came from
studies at moderate or high risk of bias, the overall outcome was not
downgraded.
Serious: If greater than 33.3% of the weight in a meta-analysis came from
studies at moderate or high risk of bias, the outcome was downgraded one
level.
Very serious: If greater than 33.3% of the weight in a meta-analysis came from
studies at high risk of bias, the outcome was downgraded two levels.
Outcomes meeting the criteria for downgrading above were not downgraded if
there was evidence the effect size was not meaningfully different between
studies at high and low risk of bias.
Indirectness Not serious: If less than 33.3% of the weight in a meta-analysis came from
partially indirect or indirect studies, the overall outcome was not downgraded.
Serious: If greater than 33.3% of the weight in a meta-analysis came from
partially indirect or indirect studies, the outcome was downgraded one level.
Very serious: If greater than 33.3% of the weight in a meta-analysis came from
indirect studies, the outcome was downgraded two levels.
direct and indirect studies.
Inconsistency Concerns about inconsistency of effects across studies, occurring when there
is unexplained variability in the test accuracy demonstrated across studies
(heterogeneity), after appropriate pre-specified subgroup analyses have been
conducted. This was assessed using the I2 statistic.
N/A: Inconsistency was marked as not applicable if data on the outcome was
only available from one study.
Not serious: If the I2 was less than 33.3%, the outcome was not downgraded.
Serious: If the I2 was between 33.3% and 66.7%, the outcome was
downgraded one level.
Very serious: If the I2 was greater than 66.7%, the outcome was downgraded
two levels.
25
studies with the smallest and largest effect sizes.
Imprecision If the 95% confidence interval for a positive likelihood ratio spanned 2, the
outcome was downgraded one level, as the data were deemed to be
consistent with a meaningful increase in risk and no meaningful predictive
value. Similarly, negative likelihood ratios that spanned 0.5 led to downgrading
for serious imprecision. Any likelihood ratios that spanned both 0.5 and 2 were
downgraded twice, as suffering from very serious imprecision.
the confidence interval was sufficiently narrow that the upper and lower bounds
would correspond to clinically equivalent scenarios.
3.5 Qualitative evidence

Individual qualitative studies were quality assessed using the CASP qualitative checklist.
Each individual study was classified as being either at low, moderate or high risk of bias
based on that assessment.
3.5.2 Methods for combining qualitative evidence
Where multiple qualitative studies were identified for a single question, information from the
studies was combined using a thematic synthesis. By examining the findings of each
included study, descriptive themes were independently identified and coded. Once all of the
included studies had been examined and coded, the resulting themes and sub-themes were
evaluated to examine their relevance to the review question, the importance given to each
theme, and the extent to which each theme recurred across the different studies. The
qualitative synthesis then proceeded by using these ‘descriptive themes’ to develop
‘analytical themes’, which were interpreted by the reviewer in light of the overarching review
questions.
3.5.3 CERQual for qualitative studies
CERQual was used to assess the confidence we have in the summary findings of each of the
identified themes. Evidence from all qualitative study designs (interviews, focus groups etc.)
was initially rated as high confidence and the confidence in the evidence for each theme was
then downgraded from this initial point as detailed in Table 6 below.
Table 6: Rationale for downgrading confidence in evidence for qualitative questions

CERQual criteria Reasons for downgrading confidence
Methodological Not serious: If the theme was identified in studies at low risk of bias, the
limitations outcome was not downgraded
Serious: If the theme was identified only in studies at moderate or high risk of
bias, the outcome was downgraded one level.
Very serious: If the theme was identified only in studies at high risk of bias, the
outcome was downgraded two levels.
Relevance High: If the theme was identified in highly relevant studies, the outcome was
not downgraded
Moderate: If the theme was identified only in relevant and partially relevant
studies, the outcome was downgraded one level.
26
CERQual criteria Reasons for downgrading confidence
Low: If the theme was identified only in partially relevant studies, the outcome
was downgraded two levels.
Coherence Coherence was addressed based on two factors:
 Between study – does the theme consistently emerge from all relevant
studies
 Theoretical – does the theme provide a convincing theoretical explanation for
the patterns found in the data
The outcome was downgraded once if there were concerns about one of these
elements of coherence, and twice if there were concerns about both elements.
Adequacy of data The outcome was downgraded if there was insufficient data to develop an
understanding of the phenomenon of interest, either due to insufficient studies,
participants or observations.
3.6 Health economics

Literature reviews seeking to identify published cost–utility analyses of relevance to the
issues under consideration were conducted for all questions. In each case, the search
undertaken for the clinical review was modified, retaining population and intervention
descriptors, but removing any study-design filter and adding a filter designed to identify
relevant health economic analyses. In assessing studies for inclusion, population,
intervention and comparator, criteria were always identical to those used in the parallel
clinical search; only cost–utility analyses were included. Economic evidence profiles,
including critical appraisal according to the Guidelines manual, were completed for included
studies.
Economic studies identified through a systematic search of the literature are appraised using
a methodology checklist designed for economic evaluations (NICE guidelines manual; 2014).
This checklist is not intended to judge the quality of a study per se, but to determine whether
an existing economic evaluation is useful to inform the decision-making of the committee for
a specific topic within the guideline.
There are 2 parts of the appraisal process. The first step is to assess applicability (that is, the
relevance of the study to the specific guideline topic and the NICE reference case);
evaluations are categorised according to the criteria in Table 7.
Table 7: Applicability criteria

Level Explanation
Directly applicable The study meets all applicability criteria, or fails to meet one or
more applicability criteria but this is unlikely to change the
conclusions about cost effectiveness
Partially applicable The study fails to meet one or more applicability criteria, and
this could change the conclusions about cost effectiveness
Not applicable The study fails to meet one or more applicability criteria, and
this is likely to change the conclusions about cost
effectiveness. These studies are excluded from further
consideration
In the second step, only those studies deemed directly or partially applicable are further
assessed for limitations (that is, methodological quality); see categorisation criteria in Table
8.
27
Table 8: Methodological criteria
Level Explanation
Minor limitations Meets all quality criteria, or fails to meet one or more quality
criteria but this is unlikely to change the conclusions about cost
effectiveness
Potentially serious Fails to meet one or more quality criteria and this could change
limitations the conclusions about cost effectiveness
Very serious limitations Fails to meet one or more quality criteria and this is highly likely
to change the conclusions about cost effectiveness. Such
studies should usually be excluded from further consideration
Where relevant, a summary of the main findings from the systematic search, review and
appraisal of economic evidence is presented in an economic evidence profile alongside the
clinical evidence.
28
4 Summary of recommendations
4.1 Recommendations summary
Initial assessment in non-specialist settings

1. At the initial assessment take a history (including cognitive, behavioural
and psychological symptoms, and the impact symptoms have on their
daily life):
 from the person with suspected dementia and
 if possible, from someone who knows the person well (such as a
family member).
2. If dementia is still suspected after initial assessment:
 conduct a physical examination and
 undertake appropriate blood and urine tests to exclude reversible
causes of cognitive decline and
 use cognitive testing.
3. When using cognitive testing, use a validated brief structured cognitive
instrument such as:
 the 10-point cognitive screener (10-CS)
 the 6-item cognitive impairment test (6CIT)
 the 6-item screener
 the Memory Impairment Screen (MIS)
 the Mini-Cog
 Test Your Memory (TYM).
4. Do not rule out dementia solely because the person has a normal score
on a cognitive instrument.
5. When taking a history from someone who knows the person with
suspected dementia, consider supplementing this with a structured
instrument such as the Informant Questionnaire on Cognitive Decline in
the Elderly (IQCODE) or the Functional Activities Questionnaire (FAQ)
6. Refer the person to a specialist dementia diagnostic service (such as a
memory clinic or community old age psychiatry service) if:
 reversible causes of cognitive decline (including delirium,
depression, sensory impairment [such as sight or hearing loss]
or cognitive impairment from medicines associated with
increased anticholinergic burden) have been investigated and
 dementia is still suspected.
7. If the person has suspected rapidly-progressive dementia, refer them to a
neurological service with access to tests (including cerebrospinal fluid
examination) for Creutzfeldt-Jakob disease and similar conditions.
8. For more guidance on assessing for dementia in people with learning
disabilities, see the NICE guideline on mental health problems in people
with learning disabilities.
Dementia diagnosis in specialist settings
29
9. Diagnose a dementia subtype (if possible) if initial specialist assessment
(including an appropriate neurological examination and cognitive testing)
confirms cognitive decline and reversible causes have been ruled out.
10. If Alzheimer’s disease is suspected, include a test of verbal episodic
memory in the assessment.
11. Consider neuropsychological testing if it is unclear:
 whether or not the person has cognitive impairment or
 whether or not their cognitive impairment is caused by dementia
or
 what the correct subtype diagnosis is.
12. Use validated criteria to guide clinical judgement when diagnosing
dementia subtypes, such as:
 International consensus criteria for dementia with Lewy bodies
 International FTD criteria for frontotemporal dementia (primary
non-fluent aphasia and semantic dementia)
 International Frontotemporal Dementia Consortium criteria for
behavioural variant frontotemporal dementia
 NINDS-AIREN criteria (National Institute of Neurological
Disorders and Stroke and Association Internationale pour la
Recherché et l'Enseignement en Neurosciences) for vascular
dementia
 NIA criteria (National Institute on Aging) for Alzheimer’s disease
 Movement disorders Society criteria for Parkinson’s disease
dementia
 International criteria for Creutzfeldt-Jakob disease.
13. Offer structural imaging to rule out reversible causes of cognitive decline
and to assist with subtype diagnosis, unless dementia is well established
and the subtype diagnosis is clear.
14. Only consider further diagnostic tests (recommendations 15-28) if:
 it would help to diagnose a dementia subtype and
 knowing more about the dementia subtype would change
management.
Further tests for Alzheimer’s disease
15. If the diagnosis is uncertain (see recommendation 14) and Alzheimer’s
disease is suspected, consider either:
 FDG-PET (fluorodeoxyglucose-positron emission
tomography-CT), or perfusion SPECT (single-photon emission
CT) if FDG-PET is unavailable.
or
 examining cerebrospinal fluid for:
 either total tau or total tau and phosphorylated-tau 181 and
 either amyloid beta 1–42 or amyloid beta 1–42 and amyloid beta
1–40
If a diagnosis cannot be made after one of these tests, consider using the
other one.
30
16. Be aware that the older a person is, more likely they are to get a false
positive with cerebrospinal fluid examination.
17. Do not rule out Alzheimer’s disease based solely on the results of CT or
MRI scans.
18. Do not use Apolipoprotein E genotyping or electroencephalography to
diagnose Alzheimer’s disease.
19. Be aware that young-onset Alzheimer’s disease has a genetic cause in
some people.
Further tests for dementia with Lewy bodies
20. If the diagnosis is uncertain (see recommendation 14) and dementia with
Lewy bodies is suspected, use 123I-FP-CIT SPECT.
21. If 123I-FP-CIT SPECT is unavailable, consider 123I-MIBG cardiac
scintigraphy.
22. Do not rule out dementia with Lewy bodies based solely on normal results
on 123I-FP-CIT SPECT or 123I-MIBG cardiac scintigraphy.
Further tests for frontotemporal dementia
23. If the diagnosis is uncertain (see recommendation 14) and frontotemporal
dementia is suspected, use either:
 FDG-PET or
 perfusion SPECT.
24. Do not rule out frontotemporal dementia based solely on the results of
structural, perfusion or metabolic imaging tests.
25. Be aware that frontotemporal dementia has a genetic cause in some
people.
Further tests for vascular dementia
26. If the dementia subtype is uncertain (see recommendation 14) and
vascular dementia is suspected, use MRI. If MRI is unavailable or
contraindicated, use CT.
27. Do not diagnose vascular dementia based solely on vascular lesion
burden.
28. Be aware that young-onset vascular dementia has a genetic cause in
some people.
Telling the difference between delirium and dementia in people without a
diagnosis of either
29. For people who are in hospital and have cognitive impairment with an
unknown cause, consider using one of the following to find out whether
they have delirium or delirium superimposed on dementia, compared with
dementia alone:
 the long confusion assessment method (CAM)
 the Observational Scale of Level of Arousal (OSLA).
30. Do not use standardised instruments (including cognitive instruments)
alone to distinguish delirium from delirium superimposed on dementia.
31. If it is not possible to tell whether a person has delirium, dementia, or
delirium superimposed on dementia, treat for delirium first. For guidance
on treating delirium, see treating delirium in the NICE guideline on
delirium.
31
32. Only conduct case finding for suspected dementia as part of a clinical trial
that also provides an intervention to people diagnosed with dementia.
Information provision
33. Provide people living with dementia and their family members or carers
(as appropriate) with information that is relevant to their circumstances
and the stage of their condition.
34. Be aware of the obligation to provide accessible information as detailed in
the NHS Accessible Information Standard. For more guidance on
providing information and discussing people’s preferences with them, see
the NICE guideline on patient experience in adult NHS services and
people’s experience in adult social care services.
35. At diagnosis, offer the person and their family members or carers (as
appropriate) oral and written information that explains:
 what their dementia subtype is and the changes to expect as the
condition progresses
 which healthcare professionals and social care teams will be
involved in their care and how to contact them
 if appropriate, how dementia affects driving, and that they need
to tell the Driver and Vehicle Licensing Agency (DVLA) and their
car insurer about their dementia diagnosis
 their legal rights and responsibilities
 their right to reasonable adjustments (in line with the Equality Act
2010) if they are working or looking for work
 how the following groups can help and how to contact them:
 local support groups, online forums and national charities
 financial and legal advice services
 advocacy services.
36. If it has not been documented earlier, ask the person at diagnosis:
 for their consent for services to share information
 which people they would like services to share information with
(for example family members or carers)
 what information they would like services to share.
Document these decisions in the person’s records.
37. After diagnosis, direct people and their family members or carers (as
appropriate) to relevant services for information and support (see
recommendations 47 and 48 on care coordination).
38. For people who do not want follow-up appointments and who are not
using other services, ask if they would like to be contacted again at a
specified future date.
39. Ensure that people living with dementia and their carers know how to get
more information and who from if their needs change.
40. Tell people living with dementia (at all stages of the condition) about
research studies they could participate in.
Advance care planning
41. Offer early and ongoing opportunities for people living with dementia and
people involved in their care (see recommendation 36) to discuss:
32
 the benefits of planning ahead
 lasting power of attorney (for health and welfare decisions and
property and financial affairs decisions)
 an advance statement about their wishes, preferences, beliefs
and values regarding their future care
 advance decisions to refuse treatment
 their preferences for place of care and place of death.
Explain that they will be given chances to review and change any advance
statements and decisions they have made.
42. At each care review, offer people the chance to review and change any
advance statements and decisions they have made.
Involving people in decision-making
43. Encourage and enable people living with dementia to give their own views
and opinions about their care.
44. If needed, use additional or modified ways of communicating (for example
visual aids or simplified text).
Staff training
45. Ensure that all health and social care staff are aware of:
 The extent of their responsibility to protect confidentiality under
data protection legislation and
 any rights that family members, carers and others have to
information about the person’s care (see recommendation 48 on
information sharing between different care settings).
46. Health and social care professionals advising people living with dementia
(including professionals involved in diagnosis) should be trained in
starting and holding difficult and emotionally challenging conversations.
Care coordination
47. Provide people living with dementia with a single named health or social
care professional who is responsible for coordinating their care.
48. Named professionals should:
 arrange an initial assessment of the person’s needs, which
should be face to face if possible.
 provide information about available services and how to access
them.
 involve the person’s family members or carers (as appropriate) in
support and decision-making.
 give special consideration to the views of people who do not
have capacity to make decisions about their care, in line with the
principles of the Mental Capacity Act 2005
 ensure that people are aware of their rights to and the availability
of local advocacy services, and if appropriate to the immediate
situation an independent mental capacity advocate
 develop a care and support plan, and:
 agree and review it with the involvement of the person, their
family members or carers (as appropriate) and relevant
professionals
33
 specify in the plan when and how often it will be reviewed
 evaluate and record progress towards the objectives at each
review
 ensure it covers the management of any comorbidities
 provide a copy of the plan to the person and their family
members or carers (as appropriate).
49. When developing care and support plans and advance care and support
plans, request consent to transfer these to different care settings as
needed.
50. Service providers should ensure that information (such as care and
support plans and advance care and support plans) can be easily
transferred between different care settings (for example home, inpatient,
community and residential care).
51. Staff delivering care and support should maximise continuity and
consistency of care. Ensure that relevant information is shared and
recorded in the person’s care and support plan.
52. Service providers should design services to be accessible to as many
people living with dementia as possible, including:
 people who do not have a carer or whose carer cannot support
them on their own
 people who do not have access to affordable transport, or find
transport difficult to use
 people who have responsibilities (such as work, children or being
a carer themselves)
 people with learning disabilities, sensory impairment (such as
sight or hearing loss) or physical disabilities
 people who may be less likely to access health and social care
services, such as people from black, Asian and minority ethnic
groups.
53. After a person is diagnosed with dementia, ensure they and their family
members or carers (as appropriate) have access to a memory service or
equivalent hospital- or primary-care-based multidisciplinary dementia
service.
54. Memory services and equivalent hospital- and primary-care-based
multidisciplinary dementia services should offer a choice of flexible
access or prescheduled monitoring appointments.
55. When people living with dementia or their carers have a primary care
appointment, assess for any emerging dementia-related needs and ask
them if they need any more support.
56. Be aware of the increased risk of delirium in people living with dementia
who are admitted to hospital. See the NICE guideline on delirium for
interventions to prevent and treat delirium.
57. For guidance on managing transition between care settings for people
living with dementia, see:
 the NICE guideline on transition between inpatient hospital
settings and community or care home settings for adults with
social care needs
34
 the NICE guideline on transition between inpatient mental health
settings and community or care home settings
 section 1.2 of the NICE guideline on medicines optimisation.
Follow the principles in these guidelines for transitions between other settings
(for example from home to a care home or respite care).
58. Review the person's needs and wishes (including any care and support
plans and advance care and support plans) after every transition.
59. Do not offer the following specifically to slow the progress of Alzheimer's
disease, except as part of a randomised controlled trial:
 diabetes medicines
 hypertension medicines
 statins
 non-steroidal anti-inflammatory drugs (NSAIDs), including
aspirin.
60. The three acetylcholinesterase (AChE) inhibitors donepezil, galantamine
and rivastigmine as monotherapies are recommended as options for
managing mild to moderate Alzheimer's disease under all of the
conditions specified in 62 and 63.
61. Memantine monotherapy is recommended as an option for managing
Alzheimer's disease for people with:
 moderate Alzheimer's disease who are intolerant of or have a
contraindication to AChE inhibitors or
 severe Alzheimer's disease.
Treatment should be under the conditions specified in recommendation 6.
62. Treatment should be under the following conditions:
 For people who are not taking an AChE inhibitor or memantine,
prescribers should only start treatment with these on the advice
of a clinician who has the necessary knowledge and skills. This
could include:
 secondary care medical specialists such as psychiatrists,
geriatricians and neurologists
 other healthcare professionals (such as GPs, nurse consultants
and advanced nurse practitioners), if they have specialist
expertise in diagnosing and treating Alzheimer’s disease.
 Once a decision has been made to start cholinesterase inhibitors
or memantine, the first prescription may be made in primary
care.
 Ensure that local arrangements for prescribing, supply and
treatment review follow the NICE guideline on medicines
optimisation.
63. If prescribing an AChE inhibitor (donepezil, galantamine or rivastigmine),
treatment should normally be started with the drug with the lowest
acquisition cost (taking into account required daily dose and the price per
dose once shared care has started). However, an alternative AChE
inhibitor could be prescribed if it is considered appropriate when taking
into account adverse event profile, expectations about adherence,
medical comorbidity, possibility of drug interactions and dosing profiles.
35
64. When using assessment scales to determine the severity of Alzheimer's
disease, healthcare professionals should take into account any physical,
sensory or learning disabilities, or communication difficulties that could
affect the results and make any adjustments they consider appropriate.
Healthcare professionals should also be mindful of the need to secure
equality of access to treatment for patients from different ethnic groups, in
particular those from different cultural backgrounds.
65. When assessing the severity of Alzheimer's disease and the need for
treatment, healthcare professionals should not rely solely on cognition
scores in circumstances in which it would be inappropriate to do so.
These include:
 if the cognition score is not, or is not by itself, a clinically
appropriate tool for assessing the severity of that patient's
dementia because of the patient's learning difficulties or other
disabilities (for example, sensory impairments), linguistic or other
communication difficulties or level of education or
 if it is not possible to apply the tool in a language in which the
patient is sufficiently fluent for it to be appropriate for assessing
the severity of dementia or
 if there are other similar reasons why using a cognition score, or
the score alone, would be inappropriate for assessing the
severity of dementia.
In such cases healthcare professionals should determine the need for initiation
or continuation of treatment by using another appropriate method of
assessment.
66. For guidance on managing medicines (including covert administration),
see the NICE guidelines on managing medicines for adults receiving
social care in the community and managing medicines in care homes.
67. Do not stop AChE inhibitors in people with Alzheimer’s disease because
of disease severity alone.
68. For people with an established diagnosis of Alzheimer’s disease who are
already taking an AChE inhibitor:
 consider memantine in addition to an AChE inhibitor if they have
moderate disease
 offer memantine in addition to an AChE inhibitor if they have
severe disease.
69. For people with an established diagnosis of Alzheimer’s disease who are
already taking an AChE inhibitor, primary care prescribers may start
treatment with memantine (see recommendation 68) without taking
advice from a specialist clinician.
70. Offer donepezil or rivastigmine to people with mild to moderate dementia
with Lewy bodies.
71. Only consider galantamine for people with mild to moderate dementia with
Lewy bodies if donepezil and rivastigmine are not tolerated.
72. Consider donepezil or rivastigmine for people with severe dementia with
Lewy bodies.
73. Consider memantine for people with dementia with Lewy bodies if AChE
inhibitors are not tolerated or are contraindicated.
36
74. For guidance on pharmacological management of Parkinson’s disease
dementia, see Parkinson’s disease dementia in the NICE guideline on
Parkinson’s disease.
75. Only consider AChE inhibitors or memantine to people with vascular
dementia if they have suspected comorbid Alzheimer’s disease,
Parkinson’s disease dementia or dementia with Lewy bodies.
76. Do not offer AChE inhibitors or memantine for people with frontotemporal
dementia.
77. Do not offer AChE inhibitors or memantine to people with cognitive
impairment caused by multiple sclerosis.
78. Be aware that some commonly prescribed medicines are associated with
increased anticholinergic burden, and therefore cognitive impairment.
79. Consider minimising the use of medicines associated with increased
anticholinergic burden, and if possible look for alternatives:
 when assessing whether to refer a person with suspected
dementia for diagnosis
 during medication reviews with people living with dementia.
80. Be aware that there are validated tools for assessing anticholinergic
burden (for example, the Anticholinergic Cognitive Burden Scale), but
there is insufficient evidence to recommend one over the others.
81. For guidance on carrying out medication reviews, see medication review
in the NICE guideline on medicines optimisation.
82. Offer a range of activities to promote wellbeing that are tailored to the
person’s individual preferences.
83. Offer group cognitive stimulation therapy to people living with mild to
moderate dementia.
84. Consider group reminiscence therapy for people living with mild to
moderate dementia.
85. Consider cognitive rehabilitation or occupational therapy to support
functional ability in people living with mild to moderate dementia.
86. Do not offer acupuncture to treat dementia.
87. Do not offer ginseng, vitamin E supplements or herbal formulations to
treat dementia.
88. Do not offer cognitive training to treat mild to moderate Alzheimer’s
disease.
89. Do not offer interpersonal therapy to treat the cognitive symptoms of mild
to moderate Alzheimer’s disease.
90. Do not offer non-invasive brain stimulation (including transcranial
magnetic stimulation) to treat mild to moderate Alzheimer’s disease,
except as part of a randomised controlled trial.
Agitation, aggression, distress and psychosis
91. Before starting non-pharmacological or pharmacological treatment for
distress in people living with dementia, conduct a structured assessment
to:
 explore possible reasons for the person’s distress and
 check for and address clinical or environmental causes (for
example pain, delirium or inappropriate care).
37
92. As initial and ongoing management, offer psychosocial and environmental
interventions to reduce distress in people living with dementia.
93. Only offer antipsychotics, for people living with dementia who are either:
 at risk of harming themselves or others or
 experiencing agitation, hallucinations or delusions that are
causing them severe distress.
94. Be aware that for people with dementia with Lewy bodies or Parkinson’s
disease dementia, antipsychotics can worsen the motor features of the
condition, and in some cases cause severe antipsychotic sensitivity
reactions. For more guidance, see the advice on managing delusions and
hallucinations in the NICE guideline on Parkinson’s disease. Be aware
that interventions may need to be modified for people living with dementia
95. Before starting antipsychotics, discuss the benefits and harms with the
person and their family members or carers (as appropriate). Consider
using a decision aid to support this discussion.
96. When using antipsychotics:
 use the lowest effective dose and use them for the shortest
possible time
 reassess the person at least every 6 weeks, to check whether
they still need medication.
97. Stop treatment with antipsychotics:
 the person is not getting a clear ongoing benefit from taking them
and
 after discussion with the person taking them and their family
members or carers (as appropriate).
98. Ensure that people living with dementia can continue to access
psychosocial and environmental interventions for distress while they are
taking antipsychotics and after they have stopped taking them.
99. For people living with dementia who experience agitation or aggression,
offer personalised activities to promote engagement, pleasure and
interest.
100. Do not offer valproate to manage agitation or aggression in people living
with dementia, unless it is indicated for another condition.
Depression and anxiety
101. For people living with mild to moderate dementia who have mild to
moderate depression and/or anxiety, consider psychological treatments.
102. Do not routinely offer antidepressants to manage mild to moderate
depression in people living with mild to moderate dementia, unless they
are indicated for a pre-existing severe mental health condition.
Sleep problems
103. Do not offer melatonin to manage insomnia in people living with
Alzheimer’s disease.
104. For people living with dementia who have sleep problems, consider a
personalised multicomponent sleep management approach that includes
sleep hygiene education, exposure to daylight, exercise and personalised
activities.
Parkinson’s disease dementia and dementia with Lewy bodies
38
105. For guidance on the management of Parkinson’s disease symptoms in
people with Parkinson’s disease dementia and Dementia with Lewy
bodies, see the NICE guideline on Parkinson’s disease. Be aware these
interventions may need to be modified for people living with dementia.
106. Offer carers of people living with dementia a psychoeducation and skills
training intervention that includes:
 education about dementia, its symptoms and the changes to
expect as the condition progresses
 developing personalised strategies and building carer skills
 training to help them provide care, including how to understand
and respond to changes in behaviour
 training to help them adapt their communication styles to improve
interactions with the person living with dementia
 advice on how to look after their own physical and mental health,
and their emotional and spiritual wellbeing
 advice on planning enjoyable and meaningful activities to do with
the person they care for
 information about relevant services (including support services
and psychological therapies for carers) and how to access them
 advice on planning for the future.
107. Ensure that the support offered to carers is:
 tailored to their needs and preferences and to what they want it
to achieve (for example, providing information on carer’s
employment rights for carers who work or want to work)
 designed to help them support people living with dementia
 available at a location they can get to easily
 provided in a format suitable for them (for example individual or
group sessions, or online training and support)
 available from diagnosis and as needed after this.
108. Be aware that carer interventions are likely to be most effective when
provided as group sessions.
109. Advise carers about their right to the following and how to get them:
 a formal assessment of their own needs (known as a 'Carer's
Assessment'), including their physical and mental health
 an assessment of their need for short breaks and other respite
care .
110. Be aware that carers of people living with dementia are at an increased
risk of depression. For guidance on identifying and managing depression,
see the NICE guideline on depression in adults.
111. Care and support providers should provide all staff with training in person-
centred and outcome-focused care for people living with dementia, which
should include:
 understanding the signs and symptoms of dementia, and the
changes to expect as the condition progresses
 understanding the person as an individual, and their life story
39
 respecting the person's individual identity, sexuality and culture
 understanding the needs of the person and their family members
or carers
 the principles of the Mental Capacity Act 2005 and the Care Act
2014.
112. Care providers should provide additional face-to-face training and
mentoring to staff who deliver care and support to people living with
dementia. This should include:
 understanding the organisation’s model of dementia care and
how it provides care
 how to monitor and respond to the lived experience of people
living with dementia, including adapting communication styles
 initial training on understanding, reacting to and helping people
living with dementia who experience agitation, aggression, pain,
or other behaviours indicating distress
 follow-up sessions where staff can receive additional feedback
and discuss particular situations
 advice on interventions that reduce the need for antipsychotics
and allow doses to be safely reduced
 promoting freedom of movement and minimising the use of
restraint
 if relevant to staff, the specific needs of younger people living
with dementia and people who are working or looking for work.
113. Consider giving carers and/or family members the opportunity to attend
and take part in staff dementia training sessions.
114. Consider training staff to provide multi-sensory stimulation for people with
moderate to severe dementia and communication difficulties.
Pain
115. Consider using a structured observational pain assessment tool:
 alongside self-reported pain and standard clinical assessment for
people living with moderate to severe dementia
 alongside standard clinical assessment for people living with
dementia who are unable to self-report pain.
116. For people living with dementia who are in pain, consider using a
stepwise treatment protocol that balances pain management and
potential adverse events.
117. Repeat pain assessments for people living with dementia:
 who seem to be in pain
 who show signs of behavioural changes that may be caused by
pain
 after any pain management intervention.
Falls
118. For guidance on managing the risk of falling for people living with
dementia (in community and inpatient settings), see the NICE guideline
on falls in older people. When using this guideline:
40
 take account of the additional support people living with dementia
may need to participate effectively
 be aware that multifactorial falls interventions may not be suitable
for a person living with severe dementia.
119. Ensure that people living with dementia have equivalent access to
diagnosis, treatment and care services for comorbidities to people who do
not have dementia. For more guidance on assessing and managing
multimorbidity, see the NICE guidelines on multimorbidity and older
people with social care needs and multiple long-term conditions.
120. For more guidance on providing support for older adults with learning
disabilities, see the NICE guideline on care and support of people
growing older with learning disabilities.
121. For guidance on setting HbA1c targets for people living with severe
dementia who have type 2 diabetes, see recommendation 1.6.9 in the
NICE guideline on type 2 diabetes in adults.
122. For guidance on pharmacological treatment of overactive bladder, see the
NICE technology appraisal on mirabegron for treating symptoms of
overactive bladder.
123. For guidance on treating faecal incontinence, see recommendations 1.7.2
and 1.7.8 in the NICE guideline on faecal incontinence.
Sensory impairment (such as sight loss, hearing loss, or both)
124. For guidance on hearing assessments for people with suspected or
diagnosed dementia, see adults with suspected dementia in the NICE
guideline on hearing loss
125. Encourage people living with dementia to have eye tests every 2 years.
Consider referring people who cannot organise appointments themselves.
126. From diagnosis, offer people living with dementia flexible, needs-based
palliative care that takes into account how unpredictable dementia
progression can be.
127. For people living with dementia who are approaching the end of life, use
an anticipatory healthcare planning process (see recommendation 41 on
advance care planning). Involve the person and their family members or
carers (as appropriate) as far as possible, and use the principles of best-
interest decision-making if the person cannot make decisions about their
own care.
128. For standards and measures on palliative care, see the NICE quality
standard on end of life care for adults.
129. For guidance on care for people in the last days of life, see the NICE
guideline on care of dying adults.
130. For guidance, on best interests decision-making, see the NICE guideline
on decision-making and mental capacity.
131. Encourage and support people living with dementia to eat and drink,
taking into account their nutritional needs.
132. Consider involving a speech and language therapist if there are concerns
about a person’s safety when eating and drinking.
133. Do not routinely use enteral feeding in people living with severe dementia,
unless indicated for a potentially reversible comorbidity.
41
134. When thinking about admission to hospital for a person living with severe
dementia, carry out an assessment that balances their current medical
needs with the additional harms they may face in hospital, for example:
 disorientation
 a longer length of stay
 increased mortality
 increased morbidity on discharge
 delirium
 the effects of being in an impersonal or institutional environment.
135. When thinking about admission to hospital for a person living with
dementia, take into account:
 any advance care and support plans
 the value of keeping them in a familiar environment.
136. Consider using a structured tool to assess the likes and dislikes, routines
and personal history of a person living with dementia..
42
4.2 Research recommendations summary
1. Does amyloid PET imaging provide additional diagnostic value, and is it

cost effective, for the diagnosis of Alzheimer’s disease and other
dementias when compared with standard diagnostic procedures and
other imaging or biomarker tests?
2. In people with treated delirium who no longer meet the DSM-5 criteria for
delirium, but who have persistent cognitive deficits, when is the most
appropriate time to carry out an assessment for dementia?
3. What is the effectiveness of structured case finding (including a
subsequent intervention for people identified as having dementia) in
people at high risk of dementia, following up both people identified as
having or not having dementia?
4. What is the effectiveness and cost effectiveness of high-intensity case
management compared with usual care on quality of life (for the person
living with dementia and for their carer) and the timing of entry to long-
term care?
5. What are the most effective methods of care planning for people in
residential care settings?
6. What are the most effective methods of care planning for people who do
not have regular contact with an informal carer?
7. What is the effectiveness of structured transfer plans to ease the
transition between different environments for people living with dementia
and their carers?
8. What is the effectiveness of combination treatment with a cholinesterase
inhibitor and memantine for people with dementia with Lewy bodies if
treatment with a cholinesterase inhibitor alone is not effective or no longer
effective?
9. Does actively reducing anticholinergic burden in people living with
dementia improve cognitive outcomes compared with usual care?
10. What are the most effective psychosocial interventions for improving
cognition, independence, activities of daily living and wellbeing in people
living with dementia?
11. What is the effectiveness of unstructured community activities on
wellbeing for people living with dementia?
12. What is the effectiveness and cost-effectiveness of self-management
training for people living with dementia and their carers?
13. What are the most effective psychological treatments for managing
depression or anxiety in people living with dementia at each stage of the
condition?
14. What is the effectiveness and cost-effectiveness of dextromethorphan-
quinidine for managing agitation in people living with dementia?
15. What is the effectiveness and cost-effectiveness of choline alphoscerate
for managing apathy in people living with dementia?
16. What is the effectiveness of pharmacological treatments for sleep
problems in people who have not responded to non-pharmacological
management?
43
17. What is the effectiveness and cost-effectiveness of group-based cognitive
behavioural therapy for carers of people living with dementia who are at
high risk of developing depression?
18. What is the cost effectiveness of using a dementia-specific addition to the
Care Certificate for community staff, including dementia-specific elements
on managing anxiety, communication, nutritional status and personal
care?
19. What is the effectiveness of training acute hospital staff in managing
behaviours that challenge in people living with dementia on improving
outcomes for people and their carers?
20. What are the most clinically and cost-effective non-pharmacological
interventions for helping the long-term recovery of people with delirium
superimposed on dementia?
21. What is the effectiveness of interventions to improve faecal and urinary
continence in people living with dementia?
22. What is the impact on cognition, quality of life and mortality of withdrawing
treatments for the primary and secondary prevention of vascular
outcomes in people with severe dementia?
intensive treatments for diabetic control in people with severe dementia?
24. What are the optimal management strategies for people with enduring
mental health problems (including schizophrenia) who subsequently
develop dementia?
25. What are the most effective models of general and specialist palliative
care support to meet the needs of people with advanced dementia?
26. What are the most effective interventions to support staff to recognise
advanced dementia and develop appropriate escalation/end of life plans
to facilitate care to remain at home?
44
5 Dementia diagnosis
In order to access treatment and support, people living with dementia must first receive a
diagnosis. Diagnosis rates in England have been rising in recent years, but current estimates
suggest that 32% of people living with dementia still do not have a formal diagnosis
(http://digital.nhs.uk/catalogue/PUB30051). The provision of a dementia diagnosis should be
timely, personalised and accurate; certain interventions are only suitable for specific
dementia subtypes, and the implications of a diagnosis e.g. in terms of prognosis or the risk
of having a genetic form of the condition, can vary significantly between subtypes. It is also
vitally important to rule out reversible causes of cognitive decline, and to distinguish
dementia from delirium.
Population screening for dementia is outside the scope of this guideline and is not currently
recommended in the UK (https://www.gov.uk/government/news/recommendation-against-
national-dementia-screening). The starting point for dementia assessment is usually the
presentation of an individual to a primary care practitioner with memory or other cognitive
concerns. An important part of an initial assessment is an informant history, for which
structured tools are available. A range of brief instruments exist to help practitioners
determine the severity of cognitive decline; however cut-off scores that define "normal"
versus "impaired" cognition might not be valid in individual patients, for a variety of cultural,
educational and other reasons. Common dementia mimics in primary care include
depression, side-effects of medicines and sensory impairments such as hearing loss. In
specialist settings, there is an increasing range of diagnostic tests to determine the
underlying cause of the dementia syndrome, with a recent focus on biomarker-based tests
for the presence of Alzheimer's disease neuropathology. However, in many cases the
diagnosis of dementia and the identification of the subtype can be made on the basis of
clinical assessment, with brain imaging used simply to exclude mimics such as brain tumours
or hydrocephalus.
Expert consensus suggests that in the UK approximately 62% of dementia is due to

Alzheimer's disease, 17% to cerebrovascular disease, 10% to mixed aetiologies, 4%
to dementia with Lewy bodies, 2% to Parkinson's disease dementia, 2% to frontotemporal
dementia and 3% to other causes (Dementia UK 2nd edition). However, among people aged
under 65 these proportions are different, with a lower contribution from vascular dementia
and a greater relative incidence of frontotemporal dementia (Mercy 2008), while in people
aged 90 and over, mixed dementias are a larger proportion of the total (James 2012).
People living with dementia are at significantly increased risk of delirium, and many older
people with delirium have undiagnosed dementia. However, some older people with delirium
make an excellent cognitive recovery. Therefore distinguishing delirium alone, dementia
alone and delirium superimposed on dementia is important, particularly in acute hospital
settings where it might influence decisions about medicines, discharge planning and follow-
up arrangements.
45
5.1 Dementia diagnosis
Review questions
 What are the most effective methods of primary assessment to decide whether a person
with suspected dementia should be referred to a dementia service?
 What are the most effective methods of diagnosing dementia and dementia subtypes in
specialist dementia diagnostic services?
5.1.1 Introduction
This review has two aims:

 To identify which tools and tests are the most accurate for determining which people
suspected of having dementia are likely to have dementia and should be referred to a
specialist dementia diagnostic service for further investigation.
 To identify which tools and tests are the most accurate for making/confirming a diagnosis
of dementia and for diagnosing dementia subtypes in specialist dementia diagnostic
services.
The review focused on identifying studies that fulfilled the conditions specified in Table 9 and
Table 10. For full details of the review protocols, see Appendix C.
Table 9: Review summary: primary care assessment

Population People (aged 40 years and over) with a suspected diagnosis of
dementia
Diagnostic variables Potential diagnostic variables include:
 Clinical history
 Clinical cognitive assessment (e.g. Mini-Mental State Examination,
(MMSE))
 Neuropsychological testing
 Physical examination
 Medication review
Outcomes  Incidence of accurately identified dementia
 Diagnostic accuracy measures
 Resource use and costs
Table 10: Review summary: specialist care diagnosis

dementia
Diagnostic variables Potential diagnostic variables include:
 Specified diagnostic criteria
 Structural imaging (Magnetic Resonance Imaging (MRI) and
Computed Tomography (CT))
 Single-photon emission computed tomography (SPECT) (e.g. blood
flow, dopamine)
 Positron emission tomography (PET) (e.g. fluorodeoxyglucose
(FDG), amyloid)
 Cerebrospinal fluid (CSF) examination
 Electroencephalography (EEG)
 Brain biopsy
 Neuropsychological assessment
 Functional assessment
46
 Genetic testing
 Neurological examination
5.1.2 Evidence review
The search strategy for this review question consisted of several separate searches for
different types of evidence that were combined to identify relevant primary dementia DTA
(diagnostic test accuracy) studies. They are summarised in Appendix D.
Systematic searches were initially carried out for dementia DTA systematic reviews (SR) and
these reviews were mined for primary studies that matched our review protocol. The search
identified 3,698 references; and 114 of these matched the review protocol at title and
abstract level. These were screened as full texts and if they still met the review protocol then
the individual studies included in the SRs were also screened as title and abstracts to identify
potentially relevant primary DTA studies. If multiple SRs were found on the same topic then
the latest and highest quality reviews were prioritised as sources of papers. In addition, we
also identified 2 Cochrane reviews that were published after the search date, and 2 at the
time unpublished Cochrane reviews were shared with us by the Cochrane Dementia and
Cognitive Improvement Group (Seitz 2017; Chan, 2017).
Of the 118 SR references screened at full text, 37 SRs were included as sources of primary
DTA studies, plus 1 primary study which was incidentally identified through this search.
Systematic reviews were excluded if they did not match the review protocol, or if no primary
studies were extracted from them.
Using the 37 included SRs, we identified 156 primary studies that matched our review
protocol at title and abstract screening. This made 157 primary studies in total, including the
additional primary study identified directly by the SR search. These potentially relevant
references were ordered for full text review and 68 were included for data extraction based
on their relevance to the review protocol and the presentation of data in a format suitable for
analysis. We excluded studies that were not written in English or that were conference
abstracts, unless data for these studies could be obtained from a Cochrane review (see
analyses section below for details).
A second systematic literature search was carried out to cover the time between the
searches in the SRs and the current date. In the cases where SRs were not identified for
specific tests, settings (e.g. primary care) or dementia subtypes that were considered
important by the committee, additional searches were carried out to bridge the evidence
gaps. This search identified 8,047 references, of which 216 were screened as full texts and
56 additional primary studies were identified.
Seventy-six references from the original dementia guideline were also screened, with 7 being
included for full text screening. Several of these references were already included from other
searches and were excluded as duplicates on this basis (leaving 5 references to be
included). One additional reference was identified from an included primary study and
another 2 were provided by committee members.
In total, there were 380 primary studies included after title and abstract screening, with 124
meeting the review protocol as full texts and being used for data extraction. Prior to
consultation, the searches were re-run and an additional 1,524 references were identified.
Following de-duplication to remove references already identified by the previous search,
47
1,048 remained for title and abstract screening. Of these references, 10 were screened at full
text with 5 additional studies added to the evidence review.
The included primary studies are summarised below (Section 1.4) with full references in
Appendix I. The excluded studies are listed, with reasons for their exclusion, in Appendix F.
Evidence tables for the included studies are presented in Appendix P.
5.1.2.1 Analyses
Calculations of diagnostic test sensitivity, specificity, positive likelihood and negative

likelihood ratios (LR) were carried out and are presented in the GRADE tables in Appendix
G. The 2x2 tables for each individual test and the results of the QUADAS-2 assessments of
risk of bias and applicability are presented in the evidence tables in Appendix P.
Data extraction and analysis was carried out using standard methods (see section 3.4) with
the additional following decision rules:
1. The SRs were used as a source of primary studies rather than data itself with the
exception of the Cochrane reviews stated above. If the Cochrane review used
unpublished data or data from studies published in languages other than English the data
was extracted from the review directly as the Cochrane reviews were judged to be of
sufficiently high quality to be a reliable source of evidence. If the data was available in the
original paper in an accessible format then this was used instead of the Cochrane review.
2. Studies involving screening people with Parkinson’s disease (PD) or stroke for dementia
were excluded unless the participants had suspected dementia at baseline, as specified
in the review protocols above.
3. Studies that used the index test as part of the recruitment criteria for the trial were
excluded or that index was excluded if several index tests were reported.
4. Risk of bias and indirectness/applicability was assessed at the study and index test level
so that a single study could be at low risk of bias for one index test and high for another
depending on how the tests were carried out and analysed.
5. Studies that were judged to be at high risk of reporting bias due to the selective reporting
of only the most accurate test outcomes were excluded from the analysis or, if they
contained multiple groups of tests such as neuroimaging, biomarkers and
neuropsychological tests, the group of tests at risk of reporting bias was excluded but the
other tests were analysed.
6. Study settings were divided into primary care, secondary care (general) and secondary
care (specialist dementia) to facilitate meta-analysis.
7. The reference standard was divided into 3 categories of increasing accuracy: clinical
criteria alone (applied by researchers without clinician involvement), clinician diagnosis
(whether using or not using clinical criteria) and neuropathology (including autopsy and/or
biopsy results).
8. Studies using unspecified criteria for diagnosis were downgraded for risk of bias as we
were unable to determine whether the reference standard could accurately diagnose
dementia/subtype of dementia, but studies that did not use criteria (but just listed tests)
were not downgraded automatically.
9. For the evaluation of clinical criteria, autopsy was considered the most accurate
reference standard, although delayed diagnosis until further symptoms emerge and
diagnosis is confirmed was considered to be acceptable.
10. Diagnostic comparisons examining subgroups of participants (e.g. Alzheimer’s disease
(AD) versus frontotemporal dementia, [FTD]) that excluded > 10% study population were
downgraded for risk of bias. This applied to analyses performed by the study authors and
by NICE.
11. Data for all possible diagnostic comparisons was extracted and grouped into the following
categories to simplify analysis:
48
a. Dementia versus no dementia. Subjective memory complaints (SMC), mild
cognitive impairment (MCI) and other non-dementia diagnoses are included in no
dementia group.
b. Dementia subtype versus non-dementia subtype (e.g. AD versus non-AD). Non-
dementia subtype group includes SMC, MCI, other dementias and other non-
dementia diagnoses.
c. Dementia subtype versus non-dementia subtype plus unclassifiable cases (as b.
but with unclassifiable cases included with the non-dementia grouping for
comparison)
d. Dementia subtype versus no dementia (e.g. AD versus no dementia). The no
dementia subtype group includes SMC, MCI and other non-dementia diagnoses.
Other dementias are excluded.
e. Dementia subtype versus other dementias (e.g. AD versus other dementias). The
other dementias group includes all other dementias diagnosed, and excludes
SMC, MCI and other non-dementia diagnoses.
f. Dementia subtype versus another specific dementia subtype (e.g. AD versus
FTD). All other groups are excluded.
12. If study participants were diagnosed with MCI they were analysed with the no dementia
group where possible. If this was not possible based on the original data provided or they
were excluded from analysis by the study authors then this fact was noted and the study
downgraded if >10% study population was excluded.
13. Where possible during analysis probable Alzheimer’s disease was separated from
possible Alzheimer’s disease for comparison against non-Alzheimer’s disease groups.
14. Where possible, all relevant subgroup comparisons were carried out by the NICE
reviewers. In the case of most tests AD versus FTD is equivalent to FTD versus AD so
both options are not presented. However, with certain neuroimaging tests a particular
pattern may indicate a particular dementia subtype and thus AD versus FTD (using the
AD image pattern as an index test positive outcome) is not equivalent to FTD versus AD
(using the FTD image pattern as an index test positive outcome) and both comparisons
are included.
15. Single-photon emission computed tomography (SPECT) studies were analysed in
subgroups based on their camera types (single- or multiple-headed) as the two were not
considered to be comparable by the committee. If the camera type could not be
determined from information in the study then a cut-off date of 2010 was applied and all
studies with data collected after this date were deemed to have used a multiple-headed
camera.
16. SPECT studies using single-headed cameras were not downgraded for indirectness even
though modern SPECT machines use multiple-headed cameras.
17. Pittsburgh compound B (PIB) positron emission tomography (PET) studies were
excluded based on committee comments as this ligand is only used in research.
18. If a study presented multiple test cut-offs then the standard/index paper cut-offs were
extracted along with the best 3 based on sensitivity and specificity. If the standard cut-off
was not used/unclear then the 4 best results were extracted.
19. Studies using optimised cut-offs or presenting multiple cut-offs were downgraded for a
risk of bias for that test and cut-off, although any standard cut-off result was not
downgraded.
20. If the researchers carrying out the reference test were not blind/blinding was unclear to
the results of the index test then this was considered a high risk of bias only if the index
test was likely to influence the reference test outcome. For example, knowledge of Mini-
Mental State Examination (MMSE) results was considered unlikely to alter a final
reference diagnosis of dementia, but knowledge of the results of a SPECT test could
influence the diagnosis of dementia and dementia subtype.
21. A study was not downgraded for risk of bias and applicability/indirectness for the same
issue (e.g. exclusions at recruitment stage). In these instances the study was
downgraded for risk of bias only.
49
22. If the index test was part of the reference test this was not considered an additional
source of bias. The study was not downgraded for risk of bias as long as the researchers
were blind to the other test result.
23. Studies examining diagnostic criteria were only included if they referred to the current
version of the criteria (to the best of our knowledge) at the time of the evidence review.
Although a new version of the DLB criteria was published during this review (McKeith et
al 2017), Skogseth et al 2017 was included in the evidence review as it related to the
current criteria at the start of the review.
5.1.2.2 Description of included studies
A total of 124 cohort studies containing relevant diagnostic tests (Table 11, Table 12 and
Table 13), and clinical criteria (Table 14) were identified. These included: cognitive screening
and neuropsychological tests; informant questionnaires; clinician rating scales (Table 11);
structural and other imaging tests (Table 12); and biomarker and other related tests (Table
13). Some studies looked at multiple tests using the same cohort of people and many studies
presented data for several test cut-offs.
Table 11: Summary of cognitive screening and neuropsychological tests, informant

questionnaires and clinician rating scales
Number
of References
Test Studies (short title) Diagnosis category
Cognitive screening tests
10- point Cognitive screener 1 Apolinario 2015 Dementia versus no dementia
(10-CS)
Total weighted, free and total 1 Mormont 2012 Dementia versus no dementia
recall scores of the 5-word AD versus no dementia
test
6-item screener 1 Callahan 2002 Dementia versus no dementia
6-item Cognitive Impairment 1 Abdel-Aziz 2015 Dementia versus no dementia
Test (6-CIT)
7 minute screen 1 Skjerve 2008 Dementia versus no dementia
Abbreviated Mental Test 1 Flicker 1997 Dementia versus no dementia
(AMT)
Addenbrooke’s Cognitive 2 Larner 2007 Dementia versus no dementia
Exam (ACE) Mathuranath 2000
Addenbrooke’s Cognitive 3 Bastide 2012 Dementia versus no dementia
Exam-Revised (ACE-R) Hancock 2011
Terpening 2011
Addenbrooke’s Cognitive 1 Jubb 2015 Dementia versus no dementia
Exam-III (ACE- III)
Clock drawing test (CDT) 4 Beinhoff 2005 Dementia versus no dementia
Berger 2008
Milian 2012
Sager 2006
HIV Dementia Scale (HDS) 1 Skinner 2009 HAND (HIV-associated
neurocognitive disorder) versus
other neurological disorders in
HIV+ people
International HIV Dementia 1 Skinner 2009 HAND (HIV-associated
Scale (IHDS) neurocognitive disorder) versus
50
Number
of References
other neurological disorders in
HIV+ people
Letter sorting Test (LST) 1 Beinhoff 2005 Dementia versus no dementia
Memory impairment screen 2 Carnero-Pardo 2011 Dementia versus no dementia
(MIS) Beinhoff 2005
Mini-Addenbrooke’s 1 Larner 2017 Dementia versus no dementia
Cognitive Exam (Mini-ACE)
Mini-Cog 2 Carnero-Pardo 2013 Dementia versus no dementia
Milian 2012
Mini-Mental State 18 Abdel-Aziz 2015 Dementia versus no dementia
Examination (MMSE) Bastide 2012 AD versus no dementia
Callahan 2002
Carnero-Pardo 2013
Cruz-Orduna 2012
Flicker 1997
Goncalves 2011
Hancock 2011
Knaefelc 2003,
Kukull 1994
Larner 2015
Mathuranath 2000
Milian 2012
Mormont 2012
Nielsen 2013
Postel-Vinay 2014
Sager 2006
Yeung 2014
Montreal Cognitive 4 Chen 2011 Dementia versus no dementia
Assessment (MoCA) Goldstein 2014
Larner 2017
Yeung 2014
Orientation (OR) 1 Beinhoff 2005 Dementia versus no dementia
Phototest 1 Carnero-Pardo 2011 Dementia versus no dementia
Rowland Universal Dementia 2 Goncalves 2011 Dementia versus no dementia
Assessment Scale (RUDAS) Nielsen 2013
Short Portable Mental Status 1 Malhotra 2013 Dementia versus no dementia
Questionnaire (SPMSQ)
Syndrom Kurztest 1 Skjerve 2008 Dementia versus no dementia
Test your Memory (TYM) 2 Hancock 2011 Dementia versus no dementia
Postel-Vinay 2014
Clinician rating scales
Alzheimer’s disease (AD) 1 Gustafson 2010 AD versus other dementias
scale
Frontotemporal dementia 1 Gustafson 2010 FTD versus other dementias
(FTD) scale
Hachinski Ischemic Scale 2 Bachetta 2007 VaD versus AD and mixed
(HIS) Gustafson 2010 dementias (AD with VaD)
VaD versus other dementias
51
Number
of References
Lewy Body Composite Risk 1 Galvin 2015 DLB versus AD
score (LBCR) DLB versus other dementias
Neurological tests
Applause sign 1 Bonello 2016 Dementia versus no dementia
Palmo Mental Reflex 1 Streit 2015 Dementia versus no dementia
Olfactory Test 1 Christensen 2017 AD versus non-AD
Short smell test 1 Streit 2015 Dementia versus no dementia
Neuropsychological tests
Boston Naming Test (BNT) 1 Beinhoff 2005 Dementia versus no dementia
Brief Neuropsychological 1 Coutinho 2013 Dementia versus no dementia
Test Battery
Consortium to Establish a 1 Hentschel 2005 Dementia versus no dementia
Registry for Alzheimer’s
Disease (CERAD) battery
Verbal Category Fluency 1 Beinhoff 2005, Sager Dementia versus no dementia
(Animal Naming) 2006
Questionnaires completed by informant
AD8 1 Larner 2015 Dementia versus no dementia
Functional Activities 1 Cruz-Orduna 2012 Dementia versus no dementia
Questionnaire (FAQ)
Informant Questionnaire on 7 Cruz-Orduna 2012 Dementia versus no dementia
Cognitive Decline (IQCODE, Flicker 1997 AD versus no dementia
16 and 26 item versions) Hancock 2009
Garcia 2002
Goncalves 2003
Knaefelc 2003
Sikkes 2010
The diagnosis category refers to the comparisons where data was available. This does not
necessarily mean that the test would be used for this diagnosis in practice. Abbreviations for
dementia subtypes are as follows: Alzheimer’s disease (AD), Cerebral Autosomal Dominant
Arteriopathy with Subcortical Infarcts and Leukoencephalopathy (CADASIL), corticobasal
degeneration (CBD), Creutzfeldt-Jakob disease (CJD), dementia with Lewy bodies (DLB),
frontotemporal dementia (FTD), HIV-associated neurocognitive disorder (HAND), Parkinson’s
disease dementia (PDD), primary progressive aphasia (PPA), vascular dementia (VaD).
Table 12: Imaging and other related tests.

Number
of References
123 I- 7 Estorch 2008 DLB versus non-DLB
metaiodobenzylguanidine Hanyu 2006 DLB versus other dementias
cardiac scintigraphy Manabe 2017 PDD and DLB versus other
(123I MIBG cardiac Treglia 2012 dementias
scintigraphy)
Sakamoto 2014
Sakamoto 2017
Slaets 2015
Dopaminergic 6 Kemp 2011 DLB versus non-DLB
iodine-123-radiolabelled Treglia 2012 DLB versus other dementias
2β- carbomethoxy-3β-(4-i O’Brien 2009
52
Number
of References
odophenyl)-N-(3-fluoropro Walker 2007
pyl) nortropane single- Walker 2009
photon emission computed Thomas 2017
tomography (123I-FP-CIT
SPECT)
N-isopropyl-p- 1 Sakamoto 2014 DLB versus non-DLB
[123I]iodoamphetamine
single-photon emission
computed tomography
(123I-IMP SPECT)
123I-IMP SPECT and 123I- 1 Sakamoto 2014 DLB versus non-DLB
MIBG cardiac scintigraphy
Technetium-99m ethyl 2 Kaneta 2016 AD versus non-AD
cysteinate dimer single- Tripathi 2010 FTD versus non-FTD
photon emission computed
tomography
(99mTc ECD SPECT)
Technetium-99m 11 Bergman 1997 Dementia versus no dementia
hexamethylpropyleneamine Boutoleu- Bretonniere AD versus non-AD dementia
oxime single-photon 2012 plus unclassifiable
emission computed Dobert 2005 AD versus non-AD
tomography (99mTc
Holman 1992 AD versus other dementias
HMPAO SPECT)
Launes 1991 AD versus FTD
Masterman 1997 AD versus VaD
McMurdo 1994 FTD versus non- FTD dementia
Read 1995 plus unclassifiable
Rollin- Sillaire 2012 FTD versus non- FTD
Talbot 1998 FTD versus AD
Velakoulis 1007 FTD versus other dementias
FTD versus VaD
VaD versus AD
VaD versus FTD
VaD versus non-VaD
Computed tomography 1 O’Brien 2000 Dementia versus no dementia
(CT) AD versus other dementias
AD versus VaD
Electroencephalogram 3 Engedal 2015 AD versus non-AD
(EEG) Tagliapietra 2013 CJD versus non-CJD
Tschampa 2005 DLB versus non-DLB
(2-[18F]fluoro-2-deoxy-d- 11 Arslan 2015 Dementia versus no dementia
glucose positron emission Dobert 2005 AD versus no dementia
tomography (FDG-PET) Frisoni 2009 AD versus non-AD
Hoffman 2000 AD versus other dementias
Jagust 2007 AD versus DLB
Kerklaan 2014 AD versus FTD
Motara 2017 DLB versus non-DLB
Ossenkoppele 2013 DLB versus other dementias
Panegyres 2009 bv-FTD versus non-bv-FTD
Silverman 2001 bv-FTD/fd+ versus non-bv-
Yakushev 2010 FTD/fd+
FTD versus DLB
53
Number
of References
FTD versus non- FTD
FTD versus other dementias
[18F]flutemetamol PET 1 Zwan 2017 AD versus non-AD
MRI (Magnetic resonance 10 Boutoleau- Dementia versus no dementia
imaging) Bretonniere 2012 AD versus non-AD dementia
Frisoni 2009 plus unclassifiable
Hentschel 2005 AD versus non-AD
Koikkalainen 2016 AD versus other dementias
Schroter 2000 AD versus DLB
Suppa 2015 AD versus FTD
Tagliapietra 2013 AD versus VaD
Tschampa 2005
bv-FTD versus non-bv-FTD
Van Everbroeck 2004
CJD versus non-CJD
Vijverberg 2016b
DLB versus AD
DLB versus FTD
DLB versus non-DLB
DLB versus other dementias
DLB versus VaD
FTD versus AD
FTD versus DLB
FTD versus non- FTD dementia
plus unclassifiable
FTD versus other dementias
FTD versus VaD
VaD versus AD
VaD versus DLB
VaD versus FTD
VaD versus non-VaD plus
unclassifiable
VaD versus non-VaD
VaD versus other dementias
frontotemporal dementia (FTD), behavioural variant FTD (bv-FTD), behavioural variant FTD with
functional decline (bv-FTD/fd+), HIV-associated neurocognitive disorder (HAND), Parkinson’s
Table 13: Biomarkers and other related tests.

Number
of References
14-3-3 (ELISA, 17 Bahl 2008 CJD versus non-CJD
immunoblotting, Beudry 1998
Automated Capillary Burkhard 2001
Western Assay )
Chohan 2011
Cuadro-Corrales 2006
Fourier 2017
Foutz 2017
54
Number
of References
Hamlin 2012
Kenney 2000
Lattanzio 2017
Lemstra 2000
Leitao 2016
Rohan 2015
Tagliapetra 2013
Tschampa 2005
Van Everbroeck 2003
Zerr 2000
14-3-3 and Amyloid 1 Van Everbroeck 2003 CJD versus non-CJD
Beta
14-3-3 and S100B 1 Chohan 2010 CJD versus non-CJD
14-3-3 and total tau 1 Chohan 2010 CJD versus non-CJD
14-3-3, total tau and 1 Chohan 2010 CJD versus non-CJD
S100B
Amyloid Beta 1-42 11 Andreasen 2001 AD versus other dementias
Boutoleu-Bretonniere AD versus no dementia
2012 AD versus non-AD
Brandt 2008 AD versus other dementias
Duits 2014 AD versus DLB
Dumurgier 2015 AD versus VaD
Gabelle 2012 CJD versus non-CJD
Ibach 2006
Knapskgog 2016
Maddalena 2003
Mulder 2010
Van Everbroeck 2003
Amyloid Beta 1-42 and 3 Frisoni 2009 Dementia versus no dementia
Total Tau Toledo 2012 AD versus non-AD
Van Everbroeck 2003 AD versus FTD
CJD versus not CJD
Amyloid Beta 1- 1 Gabelle 2012 AD versus non-AD

42/phosphorylated tau
181 (p-tau 181)
Amyloid Beta 1-42/Total 1 Gabelle 2012 AD versus non-AD
tau
Amyloid Beta 1-42/1-40 Dumurgier 2015 AD versus non-AD
Apolipoprotein E (Apo 1 Mayeux 1998 AD versus non-AD
E)
Combinations of 5 Boutoleau-Bretonniere AD versus non-AD
Amyloid Beta 1-42, total 2012
tau and p-tau 181 Duits 2014
abnormal Brandt 2008
Dumurgier 2015
Jahn 2011
Mass Spectrometry 1 Jahn 2011 AD versus non-AD
Neuron-specific enolase 2 Bahl 2008, CJD versus non-CJD
55
Number
of References
Beudry 1998
Polymerase chain 1 Dumaresq 2013 Neurosyphilis versus not
Reaction (PCR) for T. neurosyphilis
pallidum genes polA,
Tpp47, and bmp
Real-time quaking- 2 Foutz 2017 CJD versus non-CJD
induced prion Lattanzio 2017
conversion (RT-QuIC)
S100B 3 Chohan 2010 CJD versus non-CJD
Beudry 1998
Coulthart 2011
Skin biopsy 1 Ampuero 2009 CADASIL versus CADASIL-like
syndromes
Phosphorylated -tau 181 10 Boutoleau-Bretonniere AD versus no dementia
(p-tau 181) 2012 AD versus non-AD
Brandt 2008 AD versus other dementias
Duits 2014 AD versus FTD
Dumurgier 2015
Gabelle 2012
Knapskgog 2010
Ibach 2006
Maddalena 2003
Mulder 2010
Toledo 2012
p-tau 181/Amyloid Beta 3 Maddalena 2003 AD versus no dementia
1-42 Duits 2014 AD versus non-AD
Dumurgier 2015
p-tau 181/ total tau 1 Bahl 2008, Leitao 2016 CJD versus non-CJD
p-tau 181 and Amyloid 1 Dumurgier 2015 AD versus non-AD
Beta 1-42/1-40
Total tau 18 Bahl 2008 AD versus no dementia
Brandt 2008 CJD versus non-CJD
Chohan 2010
Coulthart 2011
Duits 2014
Dumurgier 2015
Foutz 2017
Gabelle 2012
Hamlin 2010
Knapskgog 2016,
Lattanzio 2017
Leitao 2016
Mulder 2010
Rohan 2015
Tagliapietra 2013
Van Everbroeck 2003
and 2004
Yakushev 2010
Total Tau and S100B 1 Chohan 2010 CJD versus non-CJD
56
Number
of References
Total tau/Amyloid Beta 1 Duits 2014 AD versus non-AD
1-42
Urinary AD7c-NTP 1 Goodman 2007 AD versus non-AD
frontotemporal dementia (FTD), HIV-associated neurocognitive disorder (HAND), Parkinson’s
Table 14: Clinical criteria

Number
of References
ADDTC 1 Gold 2002 VaD versus AD and mixed
dementia (AD with VaD)
CBD consensus criteria 1 Alexander 2014 CBD versus non-CBD
DLB consensus criteria 1 Skogseth 2017 DLB versus other dementias
FTD consensus criteria 1 Mendez 2007 FTD versus non-FTD
Criteria for CJD: 3 Brandel 2000 CJD versus non-CJD
WHO CJD criteria Heath 2010
French and European Zerr 2009
criteria for CJD
Master’s criteria for CJD
FTDC criteria for bv- 1 Harris 2013 bv-FTD versus non-bv-FTD
FTD
Movement disorders 1 Kiesman 2013 PDD versus non-PDD
criteria for PDD
NINDS-AIREN 2 Gold 2002 VaD versus AD and mixed
Bachetta 2002 dementias (AD with VaD)
necessarily mean that the criteria would be used for this diagnosis in practice. Abbreviations for
frontotemporal dementia (FTD), behavioural variant FTD (bv-FTD), HIV-associated neurocognitive
disorder (HAND), Parkinson’s disease dementia (PDD), primary progressive aphasia (PPA),
vascular dementia (VaD).
5.1.3 Health economic evidence
A systematic literature search was undertaken to identify existing cost–utility analyses

(CUAs) evaluating diagnostic strategies that have been published since the literature reviews
in CG42. In total, 2,347 articles were returned, of which 6 were selected as potentially
relevant and retrieved for full text review. Additionally, 1 study included in CG42 was deemed
to be suitable for full text review against the current protocol. Of the 7 potentially suitable
publications, 6 were judged to be at least partially applicable to the review question and were
therefore included.
Details of the literature search are provided in Appendix D.
57
5.1.3.1 GP-administered diagnostics
Tong et al. (2016) conducted a model-based cost–utility analysis, comparing 3 diagnostic

strategies to diagnose Alzheimer’s disease, that could be administered by a GP (MMSE,
6CIT, and GPCOG) compared with unassisted GP judgement. The patient-level model
simulated a population aged over 65 years, who are assessed for cognitive impairment by
their GPs in England. The primary outcome measures were costs and QALYs over the
patient’s life time horizon. For further details, please see the economic evidence profile in
Appendix M.
The authors’ base case adopted a NHS and PPS perspective that is consistent with the
NICE reference case. An additional analysis was presented that took broader perspective
that valued private social care costs (both in the community and patients who were in full-
time care) along with informal care costs.
Diagnostic outcomes for each strategy were estimated from a range of observational
literature. Transition probabilities were calculated from five pooled studies from the Ward et
al. (2012) systematic review. Sources for resource-use and cost inputs included a NICE QOF
cost impact statement and estimates in an Alzheimer’s Society report. The price year used
was 2016 and costs were expressed in UK pounds. Health utilities for patients were
calculated using an equation reported in a cost–utility analysis of drug treatment for
Alzheimer’s disease; carer utility was not included in any analysis.
Results presented by the authors allowed the incremental analysis of each treatment option
and the removal of the cost of the MMSE diagnostic test (in case a version of the MMSE
diagnostic test is available as a royalty free for use by general practitioners in the UK).
Base-case results (Table 15) suggested that compared with GPCOP, both GP unassisted
judgement and the MMSE are dominated strategies. The 6CIT produces more QALYs than
GPCOG, but also costs an additional £186.54 per patient, resulting in an ICER of
£58,689/QALY.
Table 15: Cost–utility results from Tong et al. (2016) adjusted to show per-patient
incremental cost and effects, along with the removal of the cost of MMSE
licence fee
Absolute Incremental
Diagnostic test Cost Effect Cost Effect ICER
GPCOG NR NR
GP unassisted judgement NR NR £185.85 -0.0003 Dominated
QALYs
MMSE NR NR £119.13 -0.0002 Dominated
QALYs
6CIT NR NR £186.54 0.0032 £58,689
QALYs /QALY
In probabilistic sensitivity analysis compared with, the probability of the GPCOG being the
best option at a threshold of £30,000 per QALY was 75%. The probability of the 6CIT being
the best option became higher than the GPCOG’s when the threshold was above £50,000
per QALY.
The authors concluded that using any of the 3 cognitive screening tests was more cost-
effective than the GP unassisted judgement. Among the 3 cognitive tests, the GPCOG was
considered the most cost-effective option for the NHS given the referenced threshold of
58
£30,000 per QALY. The authors also noted that the results are sensitive to assumptions
about the effectiveness of dementia medicines, and that the model results should be treated
with caution.
Wolfs et al. (2009) conducted a cost–utility analysis alongside a 12-month cluster RCT
(n=230) in the Netherlands. 33 GP practices were randomised to a multidisciplinary
diagnostic strategy (DOC-PG) whilst 37 were randomised to usual care. The primary
outcome measures were QALYs and costs over 12 months (no extrapolation was undertaken
beyond the RCT results). For further details, please see the economic evidence profile in
Appendix M.
DOC-PG consisted of a home visit by the community mental health team (CMHT) and 2
visits to university hospital departments of geriatric medicine and geriatric psychiatry. In
addition, a CT scan and various blood tests were performed. The results were then
discussed at a weekly interdisciplinary meeting in which a definitive diagnosis was made and
a treatment plan was formulated. Usual care meant that either the diagnosis was made by
the GP, or the GP referred the patient to one of the existing separate regional services.
The authors’ base case adopted a broad societal perspective, including an attempt to value
informal care costs; however, disaggregated results are reported, enabling the recalculation
of results with a perspective that is consistent with the NICE reference case (that is, NHS
and PSS costs only). Information about resources used was derived from a case report form
provided by the carer. All cost prices were adopted from a standard Dutch source. The price
year used was 2005 and costs were expressed in Euros.
Utilities were measured only for the patient, using the EQ-5D, administered by the patients’
proxy at baseline, 6 and 12 months, with weights derived from a UK population.
Base-case results with costs not consistent with the NICE reference case removed (Table
16) suggest that DOC-GP is both more effective and more expensive than usual care, with
an ICER of €11,510 per QALY gained.
Table 16: Adjusted analysis from Wolfs et al. (2009), where costs not relevant to the
NICE reference case were removed
Treatment Cost Effect (95% CI) Cost Effect (95% CI) ICER
Usual care €26,171 0.452 QALYs (0.432 to – – –
0.472)
DOC-PG €26,758 0.503 QALYs (0.487 to €587 0.051 QALYs (−0.01 to €11,510
0.519) 0.13) /QALY
The authors’ analysis, which included productivity loss and informal care in the costs,
resulted in a smaller additional cost of €65 for DOC-GP compared with usual care, resulting
in an ICER of €1,267 per additional QALY produced.
It is not possible to remove costs excluded from the NICE reference case from the authors’
probabilistic analysis. The incremental costs in the bootstrap simulation ranged over a 95%
confidence interval of −€7,435 to €6,750; the equivalent range for incremental effectiveness
was from −0.01 to 0.13. The probability that the DOC-PG is cost effective was 72% when
QALYs are valued at €45,000 each.
The authors concluded that an integrated approach to dementia diagnosis is not

demonstrably more expensive and has a high probability of being more effective than usual
care in terms of QALYs.
59
5.1.3.2 Imaging diagnostics
Biasttu et al. (2012) conducted a model-based cost–utility analysis, comparing 3 diagnostic

strategies (standard diagnosis, standard MRI, and MRI + contrastophore-linker-
pharmacophore [MRI+CLP]) for a cohort of 70 year-olds consulting for the first time following
mild cognitive impairment symptoms in a French context. The diagnostic target was early
Alzheimer’s disease, and where this was detected, the effects of a hypothetical treatment
efficient in early Alzheimer’s disease were tested. The primary outcome measures were
costs and QALYs over 3 years. For further details, please see the economic evidence profile
in Appendix M.
The authors’ base case adopted a societal perspective, including several indirect costs. It
was not possible to disaggregate these costs to conduct an analysis that is consistent with
the NICE reference case (that is, NHS and PSS costs only). Information about accuracy of
the diagnostic strategies was derived from a range of separate observational studies.
Information about resource use and costs was derived from a published economic evaluation
of drugs for Alzheimer’s disease. The cost of MRI was obtained from the ‘‘Classification
Commune des Actes Médicaux’’, a fixed-costs scale of medical procedures based on
practitioners’ fees, fixed costs for the medical procedures themselves, and fixed costs for
operating the equipment. All prices were converted to the year 2009 and expressed in Euros.
The authors estimated population mean quality-of-life weights people without Alzheimer’s
disease and published utilities for people with Alzheimer’s disease at each disease stage and
care setting (institution or community).
In the primary analysis, standard diagnosis compared with standard MRI costs more and
produced fewer QALYs and was therefore a dominated treatment strategy. The MRI+CLP
treatment strategy compared with Standard MRI cost more but also produced additional
QALYS, resulting in an ICER of €88,439/QALY.
The “Screen and treat analysis”, which looked at targeted screening of individuals carrying
the e4 allele of the apolipoprotein E gene (ApoE4), found that the Standard Diagnosis
compared with Standard MRI costs more and produced fewer QALYs and was therefore a
dominated treatment strategy. The MRI+CLP treatment strategy compared with Standard
MRI cost more but also produced additional QALYS, resulting in an ICER of €641,326/QALY.
The mean costs, effects and ICERs are presented in Table 17.
Table 17: Analysis from Biasttu et al. (2012)

Strategy Cost Effect Cost Effect ICER
MRI €36,161 1.7710 QALYs – – –
-0.00470
Standard €36,294 1.7663 QALYs €133 QALYs Dominated
0.00210
MRI+CLP €36,313 1.7731 QALYs €152 QALYs €88,439 /QALYs
Standard diagnosis was dominated by standard MRI compared within all scenarios.
MRI+CLP was found to produce a small increase in QALYs, but was also associated with
additional costs, leading to an ICER compared with standard MRI of €88,439 per QALY
gained.
60
In a scenario involving a hypothetical new treatment, which would decrease progression from
mild to moderate stage AD, the ICER for MRI+CLP compared with standard MRI decreased,
but only to €60,923/QALY.
In probabilistic sensitivity analyses, the probability of MRI+CLP being cost-effective

compared with standard MRI remained lower than 4% when QALYs were valued at €200,000
each.
Homberger et al. (2015) conducted a decision-tree analysis, comparing Florbetapir-PET

with standard clinical examination alone for the diagnosis of Alzheimer’s disease. In the base
case, the target population were 70-year-old patients with an MMSE score of 20, who were
undergoing initial assessment for cognitive impairment in Spain. The primary outcome
measures were costs and QALYs over a 10-year time horizon. The authors’ base case
adopted a Spanish societal perspective that is broadly consistent with the NHS and PPS
perspective. For further details, please see the economic evidence profile in Appendix M.
Test characteristics for the comparators were derived from disparate sources, including a
cohort study for Florbetapir-PET and a review of registry data. Healthcare costs included
diagnostic testing costs, medicine costs, carer time and residence in a public nursing home.
All costs were adjusted to 2013 and were expressed in Euros. Health utility scores were
taken from observational sources.
In the base case (Table 18), Florbetapir-PET was associated with small additional costs and
QALY gains, compared with standard examination, resulting in an ICER of €4,769 per QALY.
In a scenario analysis, in which initial assessment was assumed to take place at an MMSE
score of 22 compared with Florbetapir-PET produced a saving of €1,534 and produced an
additional 0.019 QALYs, compared with standard examination, making it a dominant
strategy.
Table 18: Analysis from Homberger et al. (2015)

3.022
Standard examination €155,686 QALYs
3.030 0.008 €4,769
Florbetapir-PET €155,722 QALYs €36 QALYs /QALYs
Over 82% of the PSA simulations showed Florbetapir-PET to be associated with an ICER of
€30,000 per QALY or better. One-way sensitivity analysis showed that the model was most
sensitive to the hazard ratio of institutionalisation per unit increase in MMSE.
The authors concluded that the addition for Florbetapir-PET to standard clinical examination
could facilitate the diagnostic decision-making, thereby improving the treatment of patients
under evaluation for cognitive impairment.
In Hornberger et al. (2017), the same authors updated their analysis to assess Amyloid-β
PET (Aβ-PET) imaging as an adjunct to standard diagnostic assessment with or without CSF
testing for the diagnosis of Alzheimer’s disease in France. The base case assumes an
MMSE score of 22 at the time of evaluation and treatment initiation. The primary outcome
measures were costs and QALYs over a 10-year time horizon.
Test characteristics for Aβ-PET and standard assessment were derived from separate
studies. All costs were from French sources; resource use estimates were extracted from
multiple sources, including government websites. Currency was standardised to 2016 Euros
61
(€). Both the base-case scenario and the alternative scenario included caregiver costs that
were likely to be informal caregiver costs; the analyses presented here remove these costs
where possible.
In the base case (Table 19), the addition of CSF to standard assessment alone made a
negligible difference to costs and QALYs. Aβ-PET was associated with additional costs but
also conferred greater benefits, with an ICER of 43,000/QALY.
Table 19: Analysis from Hornberger et al. (2017).

3.175
Standard assessment + CSF €89,408 QALYs
3.175 0.000
Standard assessment alone €89,445 QALYs €37 QALYs Dominated
3.197 0.022
Standard assessment + Aβ-PET €90,354 QALYs €946 QALYs €43,000/QALY
The authors also conducted 2 additional scenario analyses, in which earlier testing and
treatment initiation was assumed (at an MMSE score of 25) and additional diagnostic tests
were simulated. Both scenarios suggested improved cost effectiveness for Aβ-PET;
however, it is not possible to disaggregate costs that are inconsistent with the NICE
reference case from these analyses.
Probabilistic sensitivity analysis, including costs that are inconsistent with the NICE reference
case, suggested that there was a 95% probability that the ICER for Aβ-PET compared with
standard assessment was €40,000 per QALY or better.
McMahon et al. (2000) conducted a model-based cost–utility analysis, comparing 4

diagnostic strategies (standard examination, visual SPECT, computed SPECT and contrast-
enhanced MRI) for patients who present to an Alzheimer’s disease centre in the United
States. The model classified patients by disease severity and healthcare setting (community
or nursing home). The primary outcome measures were costs and QALYs over an 18-month
time horizon. For further details, please see the economic evidence profile in Appendix M.
The authors’ base case included costs that are not consistent to the NICE reference case
(patient time and self-funded travel costs). However, an additional analysis was conducted
by the authors that excluded these.
The diagnostic accuracy of the tests, were derived from a single observational study, and the
accuracy of standard examination was based on authors’ assumption.
Resource use for the initial diagnostic work-up was based on published literature and
assessment of resource use at Massachusetts General Hospital. Costs were mostly based
on Medicare reimbursement rates. All costs were adjusted to the price year 1998 and were
expressed in US dollars ($).
Quality of life weights for patients without Alzheimer’s disease were based on a large general
population cohort; utilities for people with Alzheimer’s disease came from commonly cited
sources. Carer utility was not included in any analysis.
In the base case analysis (Table 20), compared with standard examination, both visual and
computed SPECT cost more money and produced fewer QALYs, and were therefore
62
considered dominated strategies. Compared with contrast-enhanced MRI produced a small
QALY benefit, but the additional costs that were also associated with the approach led to an
ICER of almost $500,000 per QALY gained.
Table 20: Original analysis from McMahon et al. (2000).

0.9889
Standard examination $54,762 QALYs
0.9581 -0.0308
Visual SPECT $55,362 QALYs $600 QALYs Dominated
0.9888 -0.0001
Computed SPECT $55,549 QALYs $787 QALYs Dominated
0.9910 0.0021 $479,500
Contrast-enhanced MRI $55,769 QALYs $1,007 QALYs /QALY
The sensitivity analysis conducted by the author, where patient time and travel costs (neither
of which are relevant to the NICE reference case) were removed, shows a similar pattern to
the authors’ base case in that Visual SPECT and Computed SPECT both remained
dominated treatment strategies, whilst contrast-enhanced MRI had an ICER of $328,830 per
QALY.
The authors concluded that the base-case analysis suggest that it is not cost-effective to add
functional imaging to the standard diagnostic work-up for Alzheimer disease.
The same authors produced an updated analysis (McMahon et al., 2003), comparing
standard examination, contrast-enhanced MRI, FDG PET and computed SPECT.
The authors’ base case adopted a societal perspective, incorporating costs ‘regardless of
who incurred them’. This is likely to include items that are not consistent with the NICE
reference case; however, details are not specified.
Diagnostic accuracy parameters were updated to use a wider range of observational data.
No information about resource use was provided, and is therefore assumed to be the same
as McMahon (2000). All costs were adjusted to the price year 1999 by using the medical
component of the consumer price index and were expressed in US dollars ($). Health related
quality-of-life weights were updated to Health Utilities Index Mark 3 (HUI3) values.
In the base case (Table 21), MRI was once again associated with an ICER in the order of
£0.5m/QALY. Compared with MRI, both SPECT and PET were dominated.
Table 21: Analysis from McMahon et al. (2003).

0.7092
Standard examination $56,859 QALYs
0.7109 0.0017 $598,824
Contrast-enhanced MRI $57,877 QALYs $1,018 QALYs /QALY
0.7063 -0.0046
FDG PET $58,590 QALYs $713 QALYs Dominated
0.7093 -0.0016
Computed SPECT $58,872 QALYs $995 QALYs Dominated
63
The authors concluded that the results of this analysis suggest that a combined structural
and functional examination, such as dynamic susceptibility weighted contrast-enhanced MR
imaging, may be preferable to PET for the diagnosis of AD. With improvements in therapies
or with negative consequences of inappropriate treatment, the incremental cost-effectiveness
ratio of dynamic susceptibility weighted contrast-enhanced MR imaging becomes more
favourable. Improved non-pharmacologic strategies for AD management could also make
functional imaging more useful.
5.1.4 Evidence statements
The evidence statements in this review for diagnosing dementia are written with reference to
the size of the likelihood ratios in the GRADE tables in appendix P, using the interpretation
detailed in the methods section on diagnostic test accuracy (Table 4) for both point estimates
and confidence intervals. Positive likelihood ratios, and their associated 95% confidence
intervals, were used to determine which tests increase the probability of diagnosing dementia
and negative likelihood ratios, and their associated 95% confidence intervals, were used to
determine which tests decrease the probability of diagnosing dementia.
5.1.4.1 Dementia versus no dementia
5.1.4.1.1 Results which increase the probability of diagnosing dementia
The following test results increase the probability of diagnosing dementia in a person with
suspected dementia to a degree that is likely to be very large:
 10-CS (≤5) (low quality, 95% confidence interval ranged from large to very large)
 6 item screener (≥4) (moderate quality, 95% confidence interval ranged from large to very
large)
 6 item screener (≥5) (moderate quality, 95% confidence interval ranged from very large to
very large)
 6 item screener (≥6) (moderate quality, 95% confidence interval ranged from very large to
very large)
 ACE-III (<81) (low quality, 95% confidence interval ranged from moderate to very large)
 ACE-R (<74) (moderate quality, 95% confidence interval ranged from large to very large)
 CERAD battery positive (low quality, 95% confidence interval ranged from large to very
large)
 CDT, Shulman scoring method (>2) (moderate quality, 95% confidence interval ranged
from large to very large)
 LST (<1) (moderate quality, 95% confidence interval ranged from moderate to very large)
 Mini-Cog (Scanlan and Borson algorithm positive) (moderate quality, 95% confidence
interval ranged from large to very large)
 OR (<7) (moderate quality, 95% confidence interval ranged from large to very large)
 Total Weighted Score of the 5 word Test (≤15) (low quality, 95% confidence interval
ranged from moderate to very large)
suspected dementia to a degree that is likely to be large:
 10-CS (≤6) (low quality, 95% confidence interval ranged from moderate to large)
 6 item screener (≥3) (moderate quality, 95% confidence interval ranged from large to very
large)
 BNT (<13) (moderate quality, 95% confidence interval ranged from moderate to very
large)
64
 Brief Neuropsychological Test Battery (high quality, 95% confidence interval ranged from
moderate to very large)
 CDT, Shulman scoring method (>1) (moderate quality, 95% confidence interval ranged
from moderate to large)
 Free recall score of the 5 word test (≤6) (low quality, 95% confidence interval ranged from
 LST (<2) (moderate quality, 95% confidence interval ranged from moderate to very large)
 MMSE (<17) (moderate quality, 95% confidence interval ranged from large to very large)
 MMSE (<18 or <22 or <23 or <24) (very low to low quality, 95% confidence interval
 MMSE (<19) (moderate quality, 95% confidence interval ranged from moderate to large)
 MoCA (<19) (very low quality, 95% confidence interval ranged from slight to very large)
 OR (<8) (moderate quality, 95% confidence interval ranged from moderate to very large)
 Phototest (<27) (high quality, 95% confidence interval ranged from moderate to very
large)
 RUDAS (<21 or <22) (low to moderate quality, 95% confidence interval ranged from
 TYM (≤30) (moderate quality, 95% confidence interval ranged from moderate to large)
 Total Recall Score of 5 word test (≤9) (low quality, 95% confidence interval ranged from
suspected dementia to a degree that is likely to be moderate:
 10-CS (≤7) (very low quality, 95% confidence interval ranged from slight to moderate)
 6 item screener (≥1) (low quality, 95% confidence interval ranged from slight to moderate)
 6 item screener (≥1) (low quality, 95% confidence interval ranged from slight to moderate)
 6 item screener (≥2) (moderate quality, 95% confidence interval ranged from moderate to
large)
 6 CIT (>9) (high quality, 95% confidence interval ranged from moderate to large)
 ACE (<75) (high quality, 95% confidence interval ranged from moderate to large)
 ACE (<88) (very low quality, 95% confidence interval ranged from slight to moderate)
 ACE-III (<82) (low quality, 95% confidence interval ranged from slight to moderate)
 ACE-III (<84) (very low quality, 95% confidence interval ranged from slight to moderate)
 ACE-R (<83) (moderate quality, 95% confidence interval ranged from moderate to
moderate)
 ACE-R (<85) (moderate quality, 95% confidence interval ranged from slight to moderate)
 ACE-R (<89) (low quality, 95% confidence interval ranged from slight to moderate)
 AMT (<7) (low quality, 95% confidence interval ranged from moderate to large)
 AMT (<8) (very low quality, 95% confidence interval ranged from slight to moderate)
 Applause sign (<3) (high quality, 95% confidence interval ranged from moderate to large)
 BNT (<14) (moderate quality, 95% confidence interval ranged from moderate to large)
 CDT, Shulman scoring method (>3) (low quality, 95% confidence interval ranged from
slight to moderate)
 CDT, Watson scoring method (>4) (low quality, 95% confidence interval ranged from slight
to moderate)
 CDT, Wolf-Klein scoring method (<7) (moderate quality, 95% confidence interval ranged
from moderate to moderate)
65
 CDT, scoring method unclear (<8) (high quality, 95% confidence interval ranged from
moderate to large)
 CDT, Manos and Wu scoring method (<8) (low quality, 95% confidence interval ranged
from slight to moderate)
 CT positive (moderate quality, 95% confidence interval ranged from slight to large)
 FAQ (<9) moderate quality, 95% confidence interval ranged from moderate to large)
 FDG-PET positive (high quality, 95% confidence interval ranged from moderate to large)
 IQCODE 16 item (>4.1) (low quality, 95% confidence interval ranged from slight to
moderate)
 IQCODE 26 item (>3.6 or > 3.7 or >3.8 or >3.9) (very low quality, 95% confidence interval
ranged from slight decrease to moderate)
 IQCODE 26 item (>4.0) (very low quality, 95% confidence interval ranged from moderate
to moderate)
 IQCODE 26 item (>4.1) (low quality, 95% confidence interval ranged from moderate to
large)
 LST (<3) (low quality, 95% confidence interval ranged from slight to moderate)
 MIS (<4) (high quality, 95% confidence interval ranged from moderate to large)
 MIS (<5 or <6) (moderate quality, 95% confidence interval ranged from moderate to
moderate)
 MMSE (<20 or <25) (very low to moderate quality, 95% confidence interval ranged from
moderate to large)
 MMSE (<21 or <27 or < 28) (low to moderate quality, 95% confidence interval ranged
 MMSE (<26) (very low quality, 95% confidence interval ranged from slight to large)
 Palmo-Mental Reflex positive (low quality, 95% confidence interval ranged from slight to
moderate)
 Palmo-Mental Reflex and short smell test (both positive) (low quality, 95% confidence
interval ranged from slight to very large)
 RUDAS (<23 or <24) (low quality, 95% confidence interval ranged from moderate to large)
 RUDAS (<25 or <26) (very low quality, 95% confidence interval ranged from slight to
moderate)
 7 minute screen (P>0.6 or P>0.7) (low to moderate quality, 95% confidence interval
ranged from slight to moderate)
 7 minute screen (P>0.8) (low quality, 95% confidence interval ranged from slight to large)
 Short Smell Test positive (very low quality, 95% confidence interval ranged from slight to
moderate)
 SPMSQ (≥4 or ≥ 5) (very low quality, 95% confidence interval ranged from slight to large)
 Test your memory, TYM (≤39) (moderate quality, 95% confidence interval ranged from
moderate to moderate)
 Verbal category fluency (animal naming) (<14) (moderate quality, 95% confidence interval
 Verbal category fluency (animal naming) (<19 or <20) (low quality, 95% confidence
interval ranged from slight to moderate)
The following results did not provide any meaningful diagnostic value that could be
differentiated from random chance:
 10-CS (≤8) (low quality, 95% confidence interval ranged from slight increase in probability
to slight increase)
66
 6 item screener (≥0) (moderate quality, 95% confidence interval ranged from slight
decrease in probability to slight increase)
 99mTc-HMPAO SPECT positive (low quality, 95% confidence interval ranged from
moderate decrease in probability to moderate increase)
 ACE (<83) (very low quality, 95% confidence interval ranged from slight decrease in
probability to very large increase)
 AD8 (≥2) (high quality, 95% confidence interval ranged from slight increase in probability
to slight increase)
 Amyloid Beta and Total Tau (low quality, 95% confidence interval ranged from slight
decrease in probability to moderate increase)
 ACE-III (<88) (low quality, 95% confidence interval ranged from slight increase in
probability to moderate increase)
 AMT (<10) (low quality, 95% confidence interval ranged from slight increase in probability
to slight increase)
 AMT (<9) (very low quality, 95% confidence interval ranged from slight increase in
 BNT (<15) (low quality, 95% confidence interval ranged from slight increase in probability
to moderate increase)
 CDT, Shulman scoring method (>0) (low quality, 95% confidence interval ranged from
slight increase in probability to moderate increase)
 CDT, Manos and Wu scoring method (<9) (low quality, 95% confidence interval ranged
from slight increase in probability to slight increase)
 CDT, Lin scoring method (<3) (low quality, 95% confidence interval ranged from slight
increase in probability to moderate increase)
 IQCODE 16 item (>3.5) (very low quality, 95% confidence interval ranged from slight
decrease in probability to large increase)
 IQCODE 26 item (>3.5) (very low quality, 95% confidence interval ranged from slight
 Mini-ACE (<26) (high quality, 95% confidence interval ranged from slight increase in
probability to slight increase)
 Mini-Cog (≤2) (moderate quality, 95% confidence interval ranged from slight increase in
 MIS (<7) (low quality, 95% confidence interval ranged from slight increase in probability to
moderate increase)
 MIS (<8) (moderate quality, 95% confidence interval ranged from slight increase in
 MoCA (<22 or <24 or <25 or <26) (moderate to high quality, 95% confidence interval
ranged from slight increase in probability to slight increase)
 MRI positive (very low quality, 95% confidence interval ranged from slight increase in
 Palmo-Mental Reflex and short smell test (one positive) (low quality, 95% confidence
interval ranged from slight increase in probability to moderate increase)
 SPMSQ (≥ 6) (very low quality, 95% confidence interval ranged from slight decrease in
 Syndrom Kurztest (low to moderate quality, 95% confidence interval ranged from slight
 TYM (≤42) (moderate quality, 95% confidence interval ranged from slight increase in
67
 Verbal category fluency (animal naming) (<21 or <22) (low quality, 95% confidence
interval ranged from slight increase in probability to moderate increase)
 Verbal category fluency (animal naming) (<23 or <24) (moderate quality, 95% confidence
interval ranged from slight increase in probability to slight increase)
5.1.4.1.2 Results which decrease the probability of diagnosing dementia
The following test results decrease the probability of diagnosing dementia in a person with
 10-CS (>7) (low quality, 95% confidence interval ranged from moderate to very large)
 10-CS (>8) (low quality, 95% confidence interval ranged from moderate to very large)
 6 item screener (<1) (moderate quality, 95% confidence interval ranged from large to very
large)
 ACE (≥88) (low quality, 95% confidence interval ranged from moderate to very large)
 ACE-III (≥88) (low quality, 95% confidence interval ranged from slight to very large)
 AMT (≥10) (low quality, 95% confidence interval ranged from moderate to very large)
 Mini-ACE (≥26) (high quality, 95% confidence interval ranged moderate to very large)
 Mini-Cog (>2 ) (moderate quality, 95% confidence interval ranged moderate to very large)
 MIS (≥4) (high quality, 95% confidence interval ranged moderate to very large)
 MIS (≥8) (moderate quality, 95% confidence interval ranged moderate to very large)
 MMSE (≥28) (very low quality, 95% confidence interval ranged from large to very large)
 MOCA (≥19 or ≥22) (low to moderate quality, 95% confidence interval ranged from large
to very large)
 MOCA (≥26) (moderate quality, 95% confidence interval ranged from moderate to very
large)
 TYM (>39 or >42) (moderate to high quality, 95% confidence interval ranged from
 Verbal category fluency (animal naming) (≥20 or ≥ 22 or ≥23 or ≥24 ) (moderate quality,
95% confidence interval ranged from moderate to very large)
 6 item screener (<2) (moderate quality, 95% confidence interval ranged from large to very
large)
 6CIT (≥9) (high quality, 95% confidence interval ranged from moderate to large)
 ACE (≥75) (high quality, 95% confidence interval ranged from moderate to large)
 ACE (≥83) (low quality, 95% confidence interval ranged from moderate to very large)
 ACE-III (≥81) (low quality, 95% confidence interval ranged from moderate to very large)
 ACE-III (≥84) (low quality, 95% confidence interval ranged from moderate to very large)
 ACE-R (≥74) (moderate quality, 95% confidence interval ranged from moderate to very
large)
 ACE-R (≥83) (low quality, 95% confidence interval ranged from moderate to very large)
large)
large)
 Brief Neuropsychological Test Battery (high quality, 95% confidence interval ranged from
moderate to large)
68
 CDT, Shulman scoring method (≤3) (moderate quality, 95% confidence interval ranged
 CDT, Manos and Wu scoring method (≥9) (low quality, 95% confidence interval ranged
 FAQ (≥9) (low quality, 95% confidence interval ranged from slight to very large)
 FDG-PET negative (very low quality, 95% confidence interval ranged from slight to very
large)
 IQCODE 16 item (≤3.5) (moderate quality, 95% confidence interval ranged from large to
very large)
 Mini-Cog (Scanlan and Borson algorithm negative) (moderate quality, 95% confidence
interval ranged from large to large)
 MIS (≥5) (very low quality, 95% confidence interval ranged slight to very large)
 MIS (≥6 or ≥7) (moderate quality, 95% confidence interval ranged moderate to very large)
 MMSE (≥25 or ≥26 ) (very low quality, 95% confidence interval ranged from moderate to
large)
 MMSE (≥27) (low quality, 95% confidence interval ranged from moderate to very large)
 MOCA (≥24) (low quality, 95% confidence interval ranged from slight to very large)
 TYM (>30) (moderate quality, 95% confidence interval ranged from large to large)
 Verbal category fluency (animal naming) (≥21 ) (moderate quality, 95% confidence
interval ranged from moderate to very large)
 10-CS (>5) (low quality, 95% confidence interval ranged from moderate to moderate)
 10-CS (>6) (low quality, 95% confidence interval ranged from moderate to large).
 6 item screener (<3) (moderate quality, 95% confidence interval ranged from moderate to
large)
 6 item screener (<4) (moderate quality, 95% confidence interval ranged from moderate to
moderate)
 ACE-III (≥82) (low quality, 95% confidence interval ranged from slight to large)
 AMT (≥7) (very low quality, 95% confidence interval ranged from slight to moderate)
 AMT (≥8) (low quality, 95% confidence interval ranged from slight to moderate)
 AMT (≥9) (low quality, 95% confidence interval ranged from moderate to large)
 BNT (≥15) (low quality, 95% confidence interval ranged from slight to moderate)
 CDT, Watson scoring method (≤4) (low quality, 95% confidence interval ranged from slight
to moderate)
 CDT, scoring method unclear (≥8) (high quality, 95% confidence interval ranged from
 CDT, Manos and Wu scoring method (≥8) (moderate quality, 95% confidence interval
ranged from moderate to moderate)
69
 CDT, Lin scoring method (≥3) (moderate quality, 95% confidence interval ranged from
moderate to large)
 CERAD Battery (low quality, 95% confidence interval ranged moderate to large)
 Free recall score of the 5 word test (>6) (low quality, 95% confidence interval ranged from
moderate to large)
 IQCODE 16 item (≤4.1) (moderate quality, 95% confidence interval ranged from slight to
moderate)
 IQCODE 26 item (≤3.5) (low quality, 95% confidence interval ranged from moderate to
large)
 IQCODE 26 item (≤3.6 or ≤3.7 or ≤3.8 or ≤3.9) (very low to low quality, 95% confidence
interval ranged from moderate to moderate)
 IQCODE 26 item (≤4.0 or ≤4.1 ) (very low quality, 95% confidence interval ranged from
slight to moderate)
 LST (≥3) (moderate quality, 95% confidence interval ranged from moderate to large)
 MMSE (≥17 or ≥22 or ≥24) (low to moderate quality, 95% confidence interval ranged from
 MMSE (≥18 or ≥19 or ≥23) (very low quality, 95% confidence interval ranged from slight to
large)
 OR (≤7) (low quality, 95% confidence interval ranged from slight to moderate)
 Palmo-Mental Reflex and short smell test (both negative) (moderate quality, 95%
confidence interval ranged from slight to moderate)
 Phototest (≥27) (high quality, 95% confidence interval ranged from moderate to large)
 RUDAS (≥21) (moderate quality, 95% confidence interval ranged from moderate to
moderate)
 RUDAS (≥22 or ≥23 or≥24 or ≥25) (very low quality, 95% confidence interval ranged from
slight to moderate)
 RUDAS (≥26) (low quality, 95% confidence interval ranged from moderate to large)
 7 minute screen (P≤0.6 or ≤0.7 or ≤0.8) (low quality, 95% confidence interval ranged from
slight to moderate)
 SPMSQ (<4 or <5) (low quality, 95% confidence interval ranged from moderate to large)
 Total Recall Score of 5 word test (>9) (low quality, 95% confidence interval ranged from
moderate to large)
 Total Weighted Score of the 5 word Test (>15) (low quality, 95% confidence interval
ranged from moderate to large)
 Verbal category fluency (animal naming) (≥14) (high quality, 95% confidence interval
 Verbal category fluency (animal naming) (≥19 ) (moderate quality, 95% confidence
interval ranged from moderate to large)
 6 item screener (<0) (very low quality, 95% confidence interval ranged from very large
decrease in probability to very large increase)
 99mTc-HMPAO SPECT negative (low quality, 95% confidence interval ranged from very
large decrease in probability to slight increase)
 AD8 (<2) (moderate quality, 95% confidence interval ranged from very large decrease in
70
 6 item screener (<5) (low quality, 95% confidence interval ranged from slight decrease in
probability to moderate decrease)
 6 item screener (<6) (low quality, 95% confidence interval ranged from slight decrease in
probability to slight decrease)
 Applause sign (≥3) (moderate quality, 95% confidence interval ranged from slight
decrease in probability to moderate decrease)
 BNT (≥13) (moderate quality, 95% confidence interval ranged from slight decrease in
 BNT (≥14) (low quality, 95% confidence interval ranged from slight decrease in probability
to moderate decrease)
 CDT, Shulman scoring method (≤2) (very low quality, 95% confidence interval ranged
from large decrease in probability to slight increase)
 CDT, Wolf-Klein scoring method (≥7) (low quality, 95% confidence interval ranged from
slight decrease in probability to moderate decrease)
 CT negative (moderate quality, 95% confidence interval ranged from slight decrease in
 LST (≥1 or ≥2) (low to moderate quality, 95% confidence interval ranged from slight
decrease in probability to slight decrease)
 MMSE (≥20 or ≥21) (very low quality, 95% confidence interval ranged from very large
 MOCA (≥25) (low quality, 95% confidence interval ranged from very large decrease in
 MRI negative (very low quality, 95% confidence interval ranged from very large decrease
in probability to slight increase)
 OR (≤7) (moderate quality, 95% confidence interval ranged from slight decrease in
 Palmo-Mental Reflex negative (low quality, 95% confidence interval ranged from
moderate decrease in probability to slight increase)
 Palmo-Mental Reflex and short smell test (one negative) (low quality, 95% confidence
interval ranged from slight decrease in probability to slight increase)
 RUDAS (≥22) (very low quality, 95% confidence interval ranged from slight decrease in
 Short Smell Test negative (very low quality, 95% confidence interval ranged from
 SPMSQ (<6) (low quality, 95% confidence interval ranged from moderate decrease in
 Syndrom Kurztest (low to moderate quality, 95% confidence interval ranged from slight
5.1.4.2 AD versus DLB
 Amyloid Beta 1-42 positive (moderate quality, 95% confidence interval ranged from slight
 FDG-PET positive (very low quality, 95% confidence interval ranged from moderate
71
 MRI positive (moderate quality, 95% confidence interval ranged from slight decrease in
 Amyloid Beta 1-42 negative (moderate quality, 95% confidence interval ranged from slight
 FDG-PET negative (very low quality, 95% confidence interval ranged from moderate
5.1.4.3 AD versus FTD
 Amyloid Beta 1-42 and total tau positive (moderate quality, 95% confidence interval
 p-tau 181 positive (moderate quality, 95% confidence interval ranged from moderate to
very large)
 99mTc-HMPAO SPECT positive (very low quality, 95% confidence interval ranged from
slight to large)
 FDG-PET positive (very low quality, 95% confidence interval ranged from slight to very
large)
 p-tau 181 negative (moderate quality, 95% confidence interval ranged from large to very
large)
 Amyloid Beta 1-42 and total tau negative (moderate quality, 95% confidence interval
72
 99mTc-HMPAO SPECT negative (very low quality, 95% confidence interval ranged from
slight to moderate)
 FDG-PET negative (very low quality, 95% confidence interval ranged from slight decrease
in probability to moderate decrease)
 MRI negative (moderate quality, 95% confidence interval ranged from slight decrease in
5.1.4.4 AD versus no dementia
 Free recall score of the five word test (≤5) (low quality, 95% confidence interval ranged
 Total tau positive (very low quality, 95% confidence interval ranged from slight to very
large)
 Total weighted score of 5-word test (≤15) (low quality, 95% confidence interval ranged
 Amyloid Beta 1-42 positive (low quality, 95% confidence interval ranged from slight to very
large)
 FDG-PET positive (moderate quality, 95% confidence interval ranged from moderate to
very large)
 p-tau 181/Amyloid beta 1-42 positive (moderate quality, 95% confidence interval ranged
from moderate to very large)
 Total recall score of 5-word test (≤9) (low quality, 95% confidence interval ranged from
 IQCODE (16 item, >3.4) (very low quality, 95% confidence interval ranged from slight to
moderate)
 IQCODE (16 item, >3.5 or >3.6) (low quality, 95% confidence interval ranged from
 MMSE (16 item, <28) (low quality, 95% confidence interval ranged from moderate to
large)
 IQCODE (16 item, >3.2) (low quality, 95% confidence interval ranged from slight increase
 IQCODE (16 item, >3.3) (very low quality, 95% confidence interval ranged from slight
 p-tau 181 positive (low quality, 95% confidence interval ranged from slight decrease in
73
 IQCODE (16 item, ≤3.2 or ≤3.3) (low quality, 95% confidence interval ranged from large to
very large)
 MMSE (16 item, ≥28) (low quality, 95% confidence interval ranged from large to very
large)
 Total recall score of 5-word test (>9) (low quality, 95% confidence interval ranged from
 Amyloid Beta 1-42 negative (moderate quality, 95% confidence interval ranged from
moderate to large)
 Free recall score of the five word test (>5) (low quality, 95% confidence interval ranged
 IQCODE (16 item, ≤3.4 or ≤3.5) (low quality, 95% confidence interval ranged from
 IQCODE (16 item, ≤3.6) (low quality, 95% confidence interval ranged from moderate to
large)
 Total weighted score of 5-word test (>15) (low quality, 95% confidence interval ranged
 FDG-PET positive (low quality, 95% confidence interval ranged from slight to large)
 p-tau 181 negative (low quality, 95% confidence interval ranged from slight to moderate)
 p-tau 181/Amyloid beta 1-42 negative (moderate quality, 95% confidence interval ranged
 Total tau negative (very low quality, 95% confidence interval ranged from slight decrease
5.1.4.5 AD versus non-AD dementia plus unclassifiable
 99mTc-HMPAO SPECT positive (low quality, 95% confidence interval ranged from slight
to moderate)
 MRI positive (very low quality, 95% confidence interval ranged from slight decrease in
74
 99mTc-HMPAO SPECT negative (low quality, 95% confidence interval ranged from slight
to large)
5.1.4.6 AD versus non-AD
 FDG-PET/CT positive (low quality, 95% confidence interval ranged from large to very
large)
 p-tau 181 and Amyloid Beta 1-42 positive (discrepancies resolved by Amyloid Beta 1-
42/1-40) (low quality, 95% confidence interval ranged from large to very large)
 Amyloid Beta 1-42 and total tau positive (moderate quality, 95% confidence interval
 Amyloid Beta 1-42 and total tau/p-tau positive (high quality, 95% confidence interval
 Formula Mattson (biomarkers) positive (high quality, 95% confidence interval ranged from
moderate to large)
 p-tau 181 and Amyloid Beta 1-42/1-40 positive (low quality, 95% confidence interval
ranged from large to very large)
 p-tau 181/Amyloid Beta 1-42 positive (very low quality, 95% confidence interval ranged
 Mass spectrometry positive (moderate quality, 95% confidence interval ranged from
 Amyloid Beta 1-42 and total tau and p-tau (≥2 positive) (high quality, 95% confidence
interval ranged from moderate to moderate)
 Amyloid Beta 1-42 and total tau and p-tau (2 positive) (high quality, 95% confidence
 99mTc-ECD SPECT, all information method positive (low quality, 95% confidence interval
 99mTc-ECD SPECT, automated method positive (moderate quality, 95% confidence
to moderate)
75
 Amyloid Beta 1-42 positive (low quality, 95% confidence interval ranged from moderate to
moderate)
 Amyloid Beta 1-42/p-tau positive (moderate quality, 95% confidence interval ranged from
moderate to large)
 Amyloid Beta 1-42/total tau positive (moderate quality, 95% confidence interval ranged
 Amyloid Beta 1-42/1-40 positive (low quality, 95% confidence interval ranged from
 FDG-PET positive (very low quality, 95% confidence interval ranged from slight to large)
 Flutemetamol PET positive (moderate quality, 95% confidence interval ranged from slight
to moderate)
 Formula Hulstaert (biomarkers) positive (high quality, 95% confidence interval ranged
 Formula Mulder (biomarkers) positive (high quality, 95% confidence interval ranged from
 Formula Schoonenboom (biomarkers) positive (high quality, 95% confidence interval
 p-tau 181 positive (very low quality, 95% confidence interval ranged from moderate to
large)
 Total tau positive (low quality, 95% confidence interval ranged from moderate to
moderate)
 Total tau/Amyloid Beta 1-42 positive (high quality, 95% confidence interval ranged from
moderate to large)
 Urinary AD7c-NTP (22 micrograms/ml) positive (moderate quality, 95% confidence
 Amyloid Beta 1-42 and total tau and p-tau abnormal (3 positive) (very low quality, 95%
confidence interval ranged from slight decrease in probability to very large increase)
 99mTc-ECD SPECT, visual assessment alone method positive (very low quality, 95%
confidence interval ranged from moderate decrease in probability to very large increase)
 EEG positive (high quality, 95% confidence interval ranged from slight increase in
 MRI positive (moderate quality, 95% confidence interval ranged from slight increase in
 Olfactory Test ≥ 3 errors (moderate quality, 95% confidence interval ranged from large
 Olfactory Test ≥ 4 errors (moderate quality, 95% confidence interval ranged from slight
 Olfactory Test ≥ 5 errors (low quality, 95% confidence interval ranged from moderate
 Formula Hulstaert (biomarkers) negative (high quality, 95% confidence interval ranged
76
 Amyloid Beta 1-42 and total tau and p-tau (<2 positive) (high quality, 95% confidence
interval ranged from large to large)
 Amyloid Beta 1-42/total tau positive (moderate quality, 95% confidence interval ranged
 FDG-PET/CT negative (low quality, 95% confidence interval ranged from moderate to
very large)
 Formula Mulder (biomarkers) negative (high quality, 95% confidence interval ranged from
large to very large)
 Formula Schoonenboom (biomarkers) negative (high quality, 95% confidence interval
ranged from large to very large)
 Mass spectrometry negative (moderate quality, 95% confidence interval ranged from
 p-tau 181 and Amyloid Beta 1-42 negative (discrepancies resolved by Amyloid Beta 1-
42/1-40) (low quality, 95% confidence interval ranged from large to very large)
 p-tau 181 and Amyloid Beta 1-42/1-40 positive (low quality, 95% confidence interval
 Total tau/Amyloid Beta 1-42 negative (high quality, 95% confidence interval ranged from
 p-tau 181/Amyloid Beta 1-42 negative (very low quality, 95% confidence interval ranged
 Amyloid Beta 1-42 negative (low quality, 95% confidence interval ranged from moderate
to moderate)
 Amyloid Beta 1-42 and total tau negative (low quality, 95% confidence interval ranged
from slight to large)
 Amyloid Beta 1-42 and total tau/p-tau negative (high quality, 95% confidence interval
 Amyloid Beta 1-42/p-tau negative (low quality, 95% confidence interval ranged from
moderate to large)
 Amyloid Beta 1-42/1-40 negative (very low quality, 95% confidence interval ranged from
slight to very large)
 99mTc-ECD SPECT, all information method negative (low quality, 95% confidence
 FDG-PET negative (very low quality, 95% confidence interval ranged from slight to large)
 Flutemetamol PET negative (moderate quality, 95% confidence interval ranged from slight
to moderate
 Formula Mattson (biomarkers) negative (high quality, 95% confidence interval ranged
 MRI positive (low quality, 95% confidence interval ranged from large decrease to slight
increase)
 p-tau 181 negative (very low quality, 95% confidence interval ranged from moderate to
moderate)
 Total tau negative (very low quality, 95% confidence interval ranged from moderate to
moderate)
77
 Amyloid Beta 1-42 and total tau and p-tau (not 2 positive) (high quality, 95% confidence
interval ranged from slight decrease in probability to slight decrease)
 Amyloid Beta 1-42 and total tau and p-tau (<3 positive) (very low quality, 95% confidence
interval ranged from large decrease in probability to slight increase)
 99mTc-ECD SPECT, automated method negative (high quality, 95% confidence interval
ranged from slight decrease in probability to slight decrease)
 99mTc-ECD SPECT, visual assessment alone method negative (very low quality, 95%
confidence interval ranged from very large decrease in probability to moderate increase)
 99mTc-HMPAO SPECT negative (moderate quality, 95% confidence interval ranged from
 EEG negative (high quality, 95% confidence interval ranged from slight decrease in
 Olfactory Test <3 errors (moderate quality, 95% confidence interval ranged from large
 Olfactory Test <4 errors (moderate quality, 95% confidence interval ranged from moderate
 Olfactory Test <5 errors (high quality, 95% confidence interval ranged from slight
 Urinary AD7c-NTP (22 micrograms/ml) negative (moderate quality, 95% confidence
interval ranged from slight decrease in probability to moderate decrease)
5.1.4.7 AD versus other dementias
 Amyloid Beta 1-42 and total tau positive (low quality, 95% confidence interval ranged from
 AD scale (≥6) (high quality, 95% confidence interval ranged from moderate to very large)
slight to large)
 14-3-3, total tau and p-tau positive (very low quality, 95% confidence interval ranged from
slight to large)
 FDG-PET positive (low quality, 95% confidence interval ranged from slight to moderate)
 p-tau 181 positive (very low quality, 95% confidence interval ranged from moderate to
large)
 p-tau 181/Amyloid Beta 1-42 positive (low quality, 95% confidence interval ranged from
 Total tau positive (very low quality, 95% confidence interval ranged from slight to large)
 Total tau/Amyloid Beta 1-42 positive (very low quality, 95% confidence interval ranged
78
 Amyloid Beta 1-42 positive (very low quality, 95% confidence interval ranged from slight
 Apo E (≥1 allele) (moderate quality, 95% confidence interval ranged from slight increase
in probability to moderate increase)
 CT positive (moderate quality, 95% confidence interval ranged from slight decrease in
 MRI positive (very low quality, 95% confidence interval ranged from slight increase in
 14-3-3, total tau and p-tau negative (low quality, 95% confidence interval ranged from
 99mTc-HMPAO SPECT negative (low quality, 95% confidence interval ranged from large
to very large)
 AD scale (<6) (high quality, 95% confidence interval ranged from moderate to large)
 Amyloid Beta 1-42 negative (very low quality, 95% confidence interval ranged from slight
to moderate)
 Amyloid Beta 1-42 and total tau negative (low quality, 95% confidence interval ranged
 Apo E (0 alleles) (moderate quality, 95% confidence interval ranged from slight to
moderate)
 FDG-PET negative (low quality, 95% confidence interval ranged from slight to moderate)
 p-tau 181 negative (very low quality, 95% confidence interval ranged from slight to large)
 p-tau 181/Amyloid Beta 1-42 negative (low quality, 95% confidence interval ranged from
moderate to large)
 Total tau negative (very low quality, 95% confidence interval ranged from slight to large)
 Total tau/Amyloid Beta 1-42 negative (very low quality, 95% confidence interval ranged
 CT negative (low quality, 95% confidence interval ranged from slight decrease in
 MRI negative (very low quality, 95% confidence interval ranged from large decrease in
79
5.1.4.8 AD versus VaD
slight to large)
 Amyloid Beta 1-42 positive (high quality, 95% confidence interval ranged from slight
decrease to slight increase)
 CT positive (moderate quality, 95% confidence interval ranged from slight decrease to
slight increase)
 MRI positive (low quality, 95% confidence interval ranged from slight decrease to large
increase)
to moderate)
 Amyloid Beta 1-42 negative (moderate quality, 95% confidence interval ranged from
 CT negative (low quality, 95% confidence interval ranged from slight decrease in
5.1.4.9 bvFTD versus non-bvFTD
 FTDC criteria positive (moderate quality, 95% confidence interval ranged from large to
very large)
 MRI positive (moderate quality, 95% confidence interval ranged from moderate to very
large)
 FDG-PET positive (low quality, 95% confidence interval ranged from slight to moderate)
80
 FDG-PET and MRI positive (moderate quality, 95% confidence interval ranged from
moderate to large)
 FTDC criteria (probable) positive (moderate quality, 95% confidence interval ranged from
moderate to large)
 FTDC criteria (possible) positive (moderate quality, 95% confidence interval ranged from
slight decrease in probability to slight increase)
 FDG-PET and MRI positive (moderate quality, 95% confidence interval ranged from
 FDG-PET negative (moderate quality, 95% confidence interval ranged from moderate to
very large)
 FTDC criteria (probable) negative (moderate quality, 95% confidence interval ranged from
 FTDC criteria negative (moderate quality, 95% confidence interval ranged from moderate
to large)
 MRI positive (low quality, 95% confidence interval ranged from slight to large)
 FTDC criteria (possible) negative (low quality, 95% confidence interval ranged from
5.1.4.10 bvFTD/FD+ versus non-bvFTD/FD+
 FDG-PET positive (moderate quality, 95% confidence interval ranged from slight to very
large)
 FDG-PET negative (moderate quality, 95% confidence interval ranged from slight
81
5.1.4.11 CADASIL versus CADASIL-like syndromes
 Skin biopsy positive (high quality, 95% confidence interval ranged from moderate to
moderate)
 Skin biopsy negative (high quality, 95% confidence interval ranged from moderate to very
large)
5.1.4.12 CBD versus non-CBD
 CBD consensus criteria positive (high quality, 95% confidence interval ranged from slight
 CBD consensus criteria negative (low quality, 95% confidence interval ranged from large
5.1.4.13 CJD versus non-CJD
 Amyloid Beta 1-42 and total tau positive (high quality, 95% confidence interval ranged
from very large to very large)
 14-3-3 and S100B (>1.0ng/ml) positive (moderate quality, 95% confidence interval ranged
 14-3-3 and Amyloid Beta 1-42 positive (high quality, 95% confidence interval ranged from
very large to very large)
 14-3-3 (ELISA) positive (moderate quality, 95% confidence interval ranged from very large
to very large)
 14-3-3 Automated Capillary Western Assay positive (high quality, 95% confidence interval
ranged from very large to very large)
 14-3-3, total tau and S100B positive (moderate quality, 95% confidence interval ranged
 MRI (DWI) positive (high quality, 95% confidence interval ranged from moderate to very
large)
82
 RT-QuIC positive (high quality, 95% confidence interval ranged from very large to very
large)
 S100B (>4.2ng/ml) positive (high quality, 95% confidence interval ranged from very large
to very large)
 Total tau and S100B positive (moderate quality, 95% confidence interval ranged from
 14-3-3 and total tau positive (moderate quality, 95% confidence interval ranged from
 14-3-3 (immunoblotting) positive (low quality, 95% confidence interval ranged from
moderate to large)
large)
 Neuron-specific enolase positive (moderate quality, 95% confidence interval ranged from
 p-tau 181/total tau positive (low quality, 95% confidence interval ranged from large to very
large)
 S100B (>1.0ng/ml) positive (moderate quality, 95% confidence interval ranged from
 S100B (>2.5ng/ml) positive (moderate quality, 95% confidence interval ranged from large
to large)
 Total tau positive (low quality, 95% confidence interval ranged from moderate to very
large)
 New criteria for sporadic CJD positive (very low quality, 95% confidence interval ranged
 WHO CJD criteria positive (very low quality, 95% confidence interval ranged from
 14-3-3 (multiple methods) positive (moderate quality, 95% confidence interval ranged from
 EEG positive (very low quality, 95% confidence interval ranged from moderate decrease
in probability to large increase)
 European criteria for CJD positive (high quality, 95% confidence interval ranged from
 French criteria for CJD positive (moderate quality, 95% confidence interval ranged from
 Masters criteria for CJD positive (high quality, 95% confidence interval ranged from slight
83
 14-3-3 and Amyloid Beta 1-42 negative (high quality, 95% confidence interval ranged from
 14-3-3 (ELISA) negative (low quality, 95% confidence interval ranged from moderate to
very large)
 14-3-3 Automated Capillary Western Assay negative (high quality, 95% confidence
interval ranged from large to very large)
 New criteria for sporadic CJD negative (low quality, 95% confidence interval ranged from
 p-tau 181/total tau negative (very low quality, 95% confidence interval ranged from
 Amyloid Beta 1-42 and total tau negative (high quality, 95% confidence interval ranged
 14-3-3 (multiple methods) negative (high quality, 95% confidence interval ranged from
moderate to large)
 14-3-3 (immunoblotting) positive (moderate quality, 95% confidence interval ranged from
large to large)
 Masters criteria for CJD negative (moderate quality, 95% confidence interval ranged from
 RT-QuIC negative (moderate quality, 95% confidence interval ranged from moderate to
very large)
 S100B (>2.5ng/ml) positive (moderate quality, 95% confidence interval ranged from
moderate to large)
 Total tau negative (low quality, 95% confidence interval ranged from large to large)
 WHO CJD criteria negative (very low quality, 95% confidence interval ranged from large to
very large)
 14-3-3 and S100B (>1.0ng/ml) negative (moderate quality, 95% confidence interval
 14-3-3 and total tau negative (moderate quality, 95% confidence interval ranged from
 14-3-3, total tau and S100B positive (low quality, 95% confidence interval ranged from
slight to moderate)
 European criteria for CJD negative (moderate quality, 95% confidence interval ranged
 French criteria for CJD negative (high quality, 95% confidence interval ranged from
moderate to large)
 MRI (DWI) negative (moderate quality, 95% confidence interval ranged from slight to
large)
 Neuron-specific enolase negative (moderate quality, 95% confidence interval ranged from
 S100B (>1.0ng/ml) negative (moderate quality, 95% confidence interval ranged from
 Total tau and S100B positive (low quality, 95% confidence interval ranged from slight to
moderate)
84
 EEG negative (high quality, 95% confidence interval ranged from slight decrease in
 MRI negative (low quality, 95% confidence interval ranged from slight decrease in
 S100B (>4.2ng/ml) negative (moderate quality, 95% confidence interval ranged from slight
5.1.4.14 DLB versus AD
 LBCRS (≥3) (moderate quality, 95% confidence interval ranged from moderate to large)
 MRI positive (low quality, 95% confidence interval ranged from slight increase in
 LBCRS (<3) (moderate quality, 95% confidence interval ranged from moderate to very
large)
5.1.4.15 DLB versus FTD
 MRI positive (low quality, 95% confidence interval ranged from slight to large)
85
5.1.4.16 DLB versus non-DLB
 123I-FP-CIT SPECT positive (moderate quality, 95% confidence interval ranged from
 123I-MIBG scintigraphy positive (low quality, 95% confidence interval ranged from
 FDG-PET positive (very low quality, 95% confidence interval ranged from slight to very
large)
 123I-IMP SPECT and 123I-MIBG scintigraphy positive (moderate quality, 95% confidence
interval ranged from moderate to very large)
 EEG positive (high quality, 95% confidence interval ranged from large to large)
 2 or more of visual hallucinations, Parkinsonism, and RBD (high quality, 95% confidence
 123I-IMP SPECT positive (low quality, 95% confidence interval ranged from slight to
moderate)
 RBD or 2 or more of visual hallucinations, Parkinsonism, and fluctuating
attention/concentration (high quality, 95% confidence interval ranged from moderate to
moderate)
 2 or more of visual hallucinations, Parkinsonism, and fluctuating attention/concentration
(high quality, 95% confidence interval ranged from moderate to moderate)
 2 or more of visual hallucinations, Parkinsonism, and fluctuating attention/concentration,
or RBD (high quality, 95% confidence interval ranged from moderate to moderate)
 MRI positive (moderate quality, 95% confidence interval ranged from slight increase in
 123I-IMP SPECT and 123I-MIBG scintigraphy negative (moderate quality, 95%
confidence interval ranged from moderate to very large)
 123I-MIBG scintigraphy negative (low quality, 95% confidence interval ranged from
 EEG negative (moderate quality, 95% confidence interval ranged from moderate to very
large)
 No RBD or less than 2 of visual hallucinations, Parkinsonism, and fluctuating
attention/concentration (high quality, 95% confidence interval ranged from moderate to
very large)
86
 Less than 2 of visual hallucinations, Parkinsonism, and fluctuating attention/concentration,
or RBD (high quality, 95% confidence interval ranged from moderate to large)
 123I-FP-CIT SPECT negative (high quality, 95% confidence interval ranged from
moderate to large)
 Less than 2 of visual hallucinations, Parkinsonism, and fluctuating attention/concentration
(high quality, 95% confidence interval ranged from moderate to large)
 Less than 2 of visual hallucinations, Parkinsonism, and RBD (high quality, 95%
confidence interval ranged from moderate to large)
 123I-IMP SPECT negative (low quality, 95% confidence interval ranged from slight
 FDG-PET negative (very low quality, 95% confidence interval ranged from large decrease
 MRI negative (high quality, 95% confidence interval ranged from slight decrease in
5.1.4.17 DLB versus other dementias
 123I-FP-CIT SPECT positive (very low quality, 95% confidence interval ranged from slight
to very large)
 123I-MIBG scintigraphy positive (moderate quality, 95% confidence interval ranged from
 DLB consensus criteria positive (moderate quality, 95% confidence interval ranged from
 LBCRS (≥3) positive (moderate quality, 95% confidence interval ranged from moderate to
very large)
 FDG-PET positive (very low quality, 95% confidence interval ranged from slight decrease
in probability to very large increase)
 LBCRS (≥3) negative (moderate quality, 95% confidence interval ranged from large to
very large)
87
 123I-FP-CIT SPECT negative (high quality, 95% confidence interval ranged from
 123I-MIBG scintigraphy negative (high quality, 95% confidence interval ranged from
 DLB consensus criteria negative (low quality, 95% confidence interval ranged from slight
to very large)
5.1.4.18 DLB versus VaD
 MRI positive (low quality, 95% confidence interval ranged from slight to very large)
FTD versus AD
 99mTc-HMPAO SPECT positive (low quality, 95% confidence interval ranged from large
to very large)
88
slight to moderate)
5.1.4.19 FTD versus DLB
large)
 FDG-PET positive (very low quality, 95% confidence interval ranged from moderate
 FDG-PET negative (very low quality, 95% confidence interval ranged from moderate
5.1.4.20 FTD versus non-FTD dementia versus unclassifiable
to large)
 MRI positive (low quality, 95% confidence interval ranged from large decrease in
to large)
89
5.1.4.21 FTD versus non-FTD
 99mTc-ECT SPECT positive (moderate quality, 95% confidence interval ranged from very
 FTD consensus criteria positive (high quality, 95% confidence interval ranged from
 99mTc-HMPAO SPECT positive (low quality, 95% confidence interval ranged from
 FDG-PET positive (very low quality, 95% confidence interval ranged from moderate to
very large)
 MRI positive (low quality, 95% confidence interval ranged from slight to moderate)
 SPECT/PET positive (high quality, 95% confidence interval ranged from moderate to
large)
 99mTc-ECT SPECT negative (moderate quality, 95% confidence interval ranged from
 SPECT/PET negative (high quality, 95% confidence interval ranged from moderate to very
large)
slight to moderate)
90
 FTD consensus criteria negative (high quality, 95% confidence interval ranged from slight
5.1.4.22 FTD versus other dementias
 FTD scale (≥6) (high quality, 95% confidence interval ranged from large to very large)
 MRI positive (low quality, 95% confidence interval ranged from moderate to moderate)
 FDG-PET positive (very low quality, 95% confidence interval ranged from slight increase
 FTD scale (<6) (high quality, 95% confidence interval ranged from moderate to very large)
slight to moderate)
5.1.4.23 FTD versus VaD
 MRI positive (low quality, 95% confidence interval ranged from slight to very large
increase)
91
slight to moderate increase)
5.1.4.24 HAND versus other neurological disorders in HIV+ people
 HIV dementia scale (<11) (low quality, 95% confidence interval ranged from slight to
large)
 International HIV dementia scale (<10) (moderate quality, 95% confidence interval ranged
 Modified HIV dementia scale (<7.5) (very low quality, 95% confidence interval ranged from
slight to moderate)
 HIV dementia scale (<10) (moderate quality, 95% confidence interval ranged from slight
 Grooved pegboard test positive (low quality, 95% confidence interval ranged from slight
increase in probability to slight increase)
 Modified HIV dementia scale and grooved pegboard test, one positive (low quality, 95%
confidence interval ranged from slight increase in probability to slight increase))
 HIV dementia scale (≥11) (low quality, 95% confidence interval ranged from slight to
moderate)
 Modified HIV dementia scale (≥7.5) (very low quality, 95% confidence interval ranged from
slight to moderate)
 HIV dementia scale (≥10) (moderate quality, 95% confidence interval ranged from slight
92
 International HIV dementia scale (≥10) (moderate quality, 95% confidence interval ranged
from large decrease in probability to slight increase)
 Modified HIV dementia scale and grooved pegboard test, both negative (very low quality,
95% confidence interval ranged from slight decrease in probability to slight decrease)
 Grooved pegboard test negative (very low quality, 95% confidence interval ranged from
5.1.4.25 Neurosyphilis versus not neurosyphilis
 CSF EIA positive (moderate quality, 95% confidence interval ranged from slight increase
in probability to moderate increase )
 FTA-ABS positive (moderate quality, 95% confidence interval ranged from slight decrease
 INNO-LIA positive (moderate quality, 95% confidence interval ranged from slight decrease
 PCR for T. pallidum genes positive (low quality, 95% confidence interval ranged from
 TPPA positive (moderate quality, 95% confidence interval ranged from slight decrease in
 CSF EIA negative (moderate quality, 95% confidence interval ranged from slight to very
large)
 FTA-ABS negative (very low quality, 95% confidence interval ranged from very large
 INNO-LIA negative (very low quality, 95% confidence interval ranged from very large
 PCR for T. pallidum genes negative (moderate quality, 95% confidence interval ranged
from slight decrease in probability to slight increase)
 TPPA negative (low quality, 95% confidence interval ranged from moderate decrease in
5.1.4.26 PDD/DLB versus other dementias
 123I-MIBG scintigraphy positive (moderate quality, 95% confidence interval ranged from
93
 123I-MIBG scintigraphy negative (moderate quality, 95% confidence interval ranged from
5.1.4.27 PDD versus non-PDD
 Movement disorders criteria for PDD (≤120) (low quality, 95% confidence interval ranged
from slight to very large)
 FCSRT-IR 3-FR (≤22) (low quality, 95% confidence interval ranged from slight to very
large)
 ROCF (≤22) (low quality, 95% confidence interval ranged from slight to very large)
from slight increase in probability to moderate increase)
 Movement disorders criteria for PDD (>123) (moderate quality, 95% confidence interval
 Movement disorders criteria for PDD (>132) (low quality, 95% confidence interval ranged
 ROCF (>22) (moderate quality, 95% confidence interval ranged from moderate to very
large)
 FCSRT-IR 3-FR (>22) (moderate quality, 95% confidence interval ranged from moderate
to very large)
 Movement disorders criteria for PDD (>120) (low quality, 95% confidence interval ranged
94
5.1.4.28 PPA versus non-PPA
 FDG-PET positive (moderate quality, 95% confidence interval ranged from large to very
large)
 FDG-PET negative (low quality, 95% confidence interval ranged from moderate decrease
5.1.4.29 VaD and mixed dementias versus AD
 HIS (≥5) (low quality, 95% confidence interval ranged from moderate to moderate)
 HIS (<5) (low quality, 95% confidence interval ranged from moderate to large)
5.1.4.30 VaD versus AD and mixed dementia
 ADDTC (possible) positive (moderate quality, 95% confidence interval ranged from slight
to large)
 ADDTC (possible and probable) positive (low quality, 95% confidence interval ranged
 NINDS-AIREN (possible) positive (moderate quality, 95% confidence interval ranged from
slight to large)
 NINDS-AIREN (possible and probable) positive (low quality, 95% confidence interval
 ADDTC (probable) positive (moderate quality, 95% confidence interval ranged from slight
 HIS (≥7) (low quality, 95% confidence interval ranged from slight increase in probability to
moderate increase)
95
 NINDS-AIREN (probable) positive (moderate quality, 95% confidence interval ranged from
slight decrease in probability to large increase)
 ADDTC (possible) negative (moderate quality, 95% confidence interval ranged from slight
to large)
 ADDTC (probable) negative (high quality, 95% confidence interval ranged from slight
 ADDTC (possible and probable) negative (low quality, 95% confidence interval ranged
from slight decrease in probability to moderate decrease)
 HIS (<7) (low quality, 95% confidence interval ranged from moderate decrease in
 NINDS-AIREN (possible) negative (moderate quality, 95% confidence interval ranged
from slight decrease in probability to moderate decrease)
 NINDS-AIREN (probable) negative (high quality, 95% confidence interval ranged from
 NINDS-AIREN (possible and probable) negative (low quality, 95% confidence interval
ranged from slight decrease in probability to moderate decrease)
5.1.4.31 VaD versus AD
 MRI positive (moderate quality, 95% confidence interval ranged from very large to very
large)
to large)
 99mTc-HMPAO SPECT negative (low quality, 95% confidence interval ranged from
moderate to large)
 MRI negative (low quality, 95% confidence interval ranged from slight to large)
96
5.1.4.32 VaD versus DLB
large)
5.1.4.33 VaD versus FTD
 MRI positive (moderate quality, 95% confidence interval ranged from large to very large)
 99mTc-HMPAO SPECT negative (low quality, 95% confidence interval ranged from large
5.1.4.34 VaD versus non-VaD dementia plus unclassifiable
 MRI positive (moderate quality, 95% confidence interval ranged from moderate to large)
 MRI negative (low quality, 95% confidence interval ranged from very large decrease in
97
5.1.4.35 VaD versus non-VaD
 MRI positive (high quality, 95% confidence interval ranged from very large to very large)
to moderate)
 99mTc-HMPAO SPECT negative (moderate quality, 95% confidence interval ranged from
slight to moderate)
 MRI negative (moderate quality, 95% confidence interval ranged from slight to large)
5.1.4.36 VaD versus other dementias
 MRI positive (moderate quality, 95% confidence interval ranged from very large to very
large)
 HIS (≥7) (high quality, 95% confidence interval ranged from moderate to very large)
 HIS (<7) (moderate quality, 95% confidence interval ranged from slight to moderate)
5.1.4.37 Health economics
5.1.4.37.1 GP administered diagnostics
One directly applicable UK cost–utility model with potentially serious limitations explored 3
GP-administered diagnostic strategies for over 65 year-olds, concluding that MMSE, 6CIT,
and GPCOG are all more cost effective than a GP’s unassisted judgement. Among the 3
tests, the GPCOG was considered the most cost-effective option for the NHS, unless QALYs
are valued at £60,000 each, in which case the small additional QALY gain associated with
6CIT would be worth its additional costs. Assuming a threshold of £30,000 per QALY, the
probability of the GPCOG being the best option was 75%.
98
One partially applicable Dutch trial-based cost–utility analysis with potentially serious
limitations compared multidisciplinary diagnosis with usual care in suspected dementia. Once
costs outside the NICE reference case were removed, the multidisciplinary approach cost an
additional €587, and produced an additional 0.051 QALYs, resulting in an ICER of €11,510
per QALY compared with usual care. Probabilistic sensitivity analysis (from which non-
reference-case costs could not be disaggregated) suggested that the probability that
multidisciplinary diagnosis is cost effective was 72% when QALYs are valued at €45,000
each.
5.1.4.37.2 Imaging diagnostics
One partially applicable French cost–utility model with potentially serious limitations
compared MRI, contrast-enhanced MRI and usual diagnosis. Standard diagnosis was
dominated by standard MRI in all scenarios. Contrast enhancement was found to produce a
small increase in QALYs, but was also associated with additional costs. The probability of
contrast-enhanced MRI being optimal remained lower than 4% at all thresholds up to
€200,000/QALY.
One partially applicable Spanish cost–utility model with potentially serious limitations
compared Florbetapir-PET with standard clinical examination, finding additional costs of €36
and benefits of 0.008 QALYs, resulting in an ICER of €4,769/QALY. In a scenario where
patients underwent initial assessment with an MMSE of 22, Florbetapir-PET was found to be
dominant. Over 82% of the PSA simulations showed Florbetapir-PET to be cost effective if
QALYs are valued at €30,000 each. A closely related, partially applicable model with
potentially serious limitations compared Aβ-PET compared with standard diagnostic
assessment in France. It found costs and QALYs were both higher with Aβ-PET, with an
ICER of €43,000/per QALY. Probabilistic sensitivity analysis suggested that ICERs were
below €40,000 per QALY in more than 95% of simulations; however, it was not possible to
remove costs that are inconsistent with the NICE reference case from this analysis.
One partially applicable cost–utility model with potentially serious limitations found that
SPECT (visual and computed) cost more money and produced fewer QALYs than standard
examination. Contrast-enhanced MRI produced a small QALY benefit, but the additional
costs that were also associated with the approach led to an ICER of over $300,000 per
QALY gained. In an updated version of the same model (this time with limitations assessed
as minor) MRI was associated with an ICER in the order of £0.5m/QALY. Both SPECT and
PET were dominated by MRI.
5.1.5 Evidence to recommendations
5.1.5.1 Primary assessment

Relative value of different The committee noted that the outcomes of interest varied during the
outcomes diagnosis process. During the initial assessment in primary care, the
committee looked for cognitive tests with high positive likelihood
ratios/high sensitivity to ensure that the majority of people with
possible dementia would be referred for further assessment. The
negative likelihood ratios/specificity were thought to be less important
at this stage, but there was a balance as the committee chose not to
recommend tests with specificity below 70% to prevent too many
false positives being referred to specialist care. In addition, there
were sufficient tests with higher positive likelihood ratios/sensitivities
and negative likelihood ratios/specificities to render the use of less
specific tests unnecessary. The committee also noted that the four
metrics reported (positive and negative likelihood ratios, sensitivity
and specificity), despite being based on the same underlying data,
covered slightly different aspects of diagnostic accuracy, and
99
therefore it was important to consider all of them as part of decision
making.
Trade-off between General issues
benefits and harms Being diagnosed with dementia can be a stressful and traumatic
experience. The committee recognised this and noted the importance
of guiding people with suspected dementia through the assessment
process carefully. In particular, it is essential that information is
provided throughout the diagnostic process and continued informed
consent is obtained for investigations that may be particularly
stressful (e.g. imaging and lumbar puncture for cerebrospinal fluid
(CSF) biomarker tests).
The committee commented that it was also important to consider
situations where a patient may not want a referral for an
assessment/diagnosis, and the potential disadvantageous outcomes
that might result from this choice (e.g. problems with obtaining
support, care and treatment).
Decision making in primary care
The committee discussed the importance of the initial assessment of
a person presenting with suspected dementia. In particular, they
stressed the value of taking a good history from the person, focusing
not just on cognitive, behavioural and psychological symptoms, but
on the impact of these symptoms on a person’s daily life. They noted
that appointment times in primary care are short compared with those
in secondary care and that, as a result, physicians need to use a brief
validated cognitive test with a high sensitivity and specificity to rapidly
determine whether there is a possibility of dementia. They also noted
that an initial assessment should include a physical examination and
tests (such as blood and urine tests) to exclude reversible causes of
cognitive decline.
The committee reviewed the evidence and noted that a number of
cognitive tests met the high sensitivity and specificity criteria
(including ACE, ACE-III, ACE-R, MoCA, MMSE, CERAD battery), but
that these tests were too long or complex for routine use in primary
care.
The committee noted that although the MoCA had a high sensitivity
and specificity at certain thresholds, in practice the test was not liked
by memory clinic staff due to the large number of false positives
seen. This was reflected in the evidence by the wide confidence
interval for specificity at the optimal test threshold of <19 and by the
much lower specificity point estimates at other test cut-offs. In
addition, based on their experience, the test was not well tolerated by
people with suspected dementia. Taking these issues into account,
the committee decided not to recommend MoCA for the initial
assessment of people with suspected dementia in primary or
secondary care settings.
The committee recommended a choice of brief cognitive tests that
they considered suitable for a primary care setting that had high
sensitivity (≥ 80%) and good specificity (≥ 70%). The committee
agreed that the evidence presented did not favour a single test,
leading them to recommend a selection of possible tests. It was also
noted that selecting a choice of tests using a different method (e.g.
picking those with the largest +ve and –ve likelihood ratios) would
lead to a slightly different list of favoured tests. However, because
there were a considerable number of tests found to have very similar
levels of diagnostic accuracy, the committee were confident the
choice of metric used for analysis would not lead to a
recommendation that was more or less clinically appropriate.
Certain tests (such as the 5 word recall test and verbal fluency tests)
are weighted towards verbal recall memory, which is an important
domain affected in Alzheimer’s disease. The committee agreed that
100
other forms of dementia, such as dementia with Lewy bodies and
vascular dementia, often do not show such pronounced memory
defects at initial presentation and may not be detected using these
tests. The committee noted that, although in the UK about two thirds
of the population living with dementia have Alzheimer’s disease, it is
very important for the physician to select the test appropriately based
on whether the person with suspected dementia presented with
memory impairment.
The committee also made an accompanying ‘do not’
recommendation to stress the importance of the physician not basing
the decision not to refer a person for further assessment solely on the
basis of a normal cognitive test result.
The committee agreed that the person considering referral should
carry out routine investigations and tests to rule out reversible causes
of cognitive impairment. No data was found to support these
investigations or the usefulness of history taking in the diagnosis of
dementia, but the committee considered this to be general good
practice based on their experience and made a recommendation to
reflect this, and agreed that failing to rule out reversible causes of
cognitive decline may lead to over referrals, both wasting resources
and causing unnecessary stress to individuals.
The committee discussed the importance of taking an informant
history. The committee recommended a structured tool, such as the
as the Informant Questionnaire on Cognitive Decline in the Elderly
(IQCODE) or the Functional Activities Questionnaire (FAQ), to help
with this process, noting that this also demonstrated a good balance
between sensitivity and specificity. They stressed that this should be
in addition to and not instead of a conversation with the informant.
The committee also noted that the IQCODE could be completed by
an informant alone and at a later date as needed.
Rapidly progressive dementia
The committee noted there was clear evidence that CSF tests had
both high sensitivity and specificity for detecting CJD. They also
noted that in the UK, standard practice for all people with suspected
rapidly progressive dementia would be to refer them to a specialist
centre for both testing and management, and it is in this specialist
setting that CSF testing would be conducted. The committee
therefore agreed it was appropriate to make a recommendation
reflecting this practice, which was justified by the strong evidence for
the appropriateness of CSF testing in this population.
Consideration of health Although there was a substantial amount of economic evidence for
benefits and resource these review questions, of the 6 economic studies, only 1 study
use (Wolfs et al. 2009) was conducted alongside a clinical trial. This
meant that the remaining 5 studies were based on modelling, and
incorporated a substantial amount of uncertainty in the large number
of model input parameters. Furthermore, the benefits of early
diagnosis were understood to be speculative as the committee
accepts that there are no disease modifying treatments available for
dementia. Many of the included studies modelled hypothetical
treatments that may be efficient in early dementia.
The committee considered the study by Biasttu et al. (2012) but did
not consider the paper to present reliable estimates of cost
effectiveness of the diagnostic strategies considered, as costs not
relevant to the NICE reference case could not be removed from the
analysis. Furthermore, Biasttu et al. (2012) lacked any sensitivity
analyses.
The committee considered the study by Tong et al. (2016) made a
strong case for 6CIT as it was free and had the highest sensitivity for
dementia and mild cognitive impairment from all tests examined that
are used in a UK setting. However, this study made use of old data
101
for the unassisted GP strategy, and data on the GPCOG was from a
screening study rather than a study in people with suspected
dementia, and this reduced the committee’s confidence in the
evidence.
The committee considered the study by Wolfs et al. (2009) which
compared DOC-PG with usual care and were uncertain of the study’s
findings applicability to a UK setting, as it was not possible to remove
costs not relevant to the NICE reference case.
The committee considered two studies by McMahon (McMahon et al.
2000 and McMahon et al. 2003) which examined standard
examination, visual SPECT, computed SPECT and contrast-
enhanced MRI and considered them to be unreliable to form
recommendations as ICERs for all examined treatments were either
dominant or excessive to commonly accepted cost-effectiveness
thresholds, no information was provided as to what was required
before people with dementia were able to present to the specialist
diagnostic centres and the authors had competing interests that the
committee considered of potential importance.
The committee considered two studies by Hornberger (Hornberger et
al. 2015 and Hornberger et al. 2017) and found both studies, which
examined the use of florbetapir positron emission tomography, to be
of limited value to form a positive recommendation, as there was a
very small non-significant increase in QALYs gained and the authors
had competing interests that the committee considered of potential
importance.
The committee agreed that, since the MMSE is under copyright and
attracts a fee, it should not be used where another similar test with
comparable sensitivity and specificity is available for free (e.g. 10-CS,
MIS, TYM for initial assessment in primary care). ACE and ACE-R
also include an MMSE score and should be avoided for the same
reason. Although the MMSE has been used extensively in the past
and is relatively inexpensive, the money saved by using a
comparable, free test can be spent elsewhere where there is greater
need. The committee also noted that more complex (and therefore
time-consuming) tests did not appear to be more effective at
detecting dementia than shorter and simpler tests, and it was
therefore a more efficient use of resources to use these briefer tests
within a time-constrained primary care setting.
The committee commented that the IQCODE (Informant
Questionnaire on Cognitive Decline in the Elderly) and FAQ had the
advantage of being self-administered tests. It could therefore be
completed in the waiting room by the informant. This could free up
time in the appointment for the GP to listen to the patient and
informant, potentially reducing the number of appointments needed to
reach an initial diagnosis and saving money for the NHS. The TYM
(Test Your Memory) test has a similar advantage.
Quality of evidence Positive and negative likelihood ratios were used to determine the
diagnostic accuracy of the tests and the level of confidence/degree of
uncertainty surrounding this estimate. Large positive likelihood ratios
were associated with large increases in the probability that a person
with a positive test had dementia, while large negative likelihood
ratios were associated with large decreases in the probability of
having dementia given a negative test result. The confidence
intervals for these measures, and the minimally important differences
specified for them in section 3.4, were used to determine the level of
precision in the evidence, which together with consideration of the
risk of bias, heterogeneity and applicability made up the overall
quality rating for the studies.
The committee noted that the quality of the evidence was low in the
case of many tests due to serious or very serious risk of bias issues.
102
This was largely due to poor reporting of blinding procedures and that
it was likely that index and reference tests were assessed without
knowledge of each other, but that this could not be assumed in the
absence of a statement to such effect. Other common issues that led
to downgrading for risk of bias included the use of optimised test
thresholds and subgroup analyses that excluded large segments of
the study population (for example, people with mild cognitive
impairment). The committee agreed that it was appropriate to
downgrade for risk of bias if > 10% study population was excluded
from an analysis. However, they also noted that these subgroup
analyses were useful in providing data on test performance in
diagnosing between dementia subtypes, especially as case-control
data was excluded from the evidence review.
The committee noted that a large body of case-control evidence was
omitted from this review on the basis that the comparison of patient
groups of interest to control groups could inflate test sensitivity and
specificity. However, case-control studies often detailed the
development and validation of the diagnostic tests and the committee
was able to use its knowledge of the literature to provide additional
support for its decisions. This was particularly important where the
evidence from cohort studies was of low or very low quality. However,
they acknowledged that the use of cohort studies examining the
diagnosis of people with suspected dementia provided evidence for
test accuracy under circumstances closer to real-life.
There was a shortage of diagnostic test accuracy studies involving
people with suspected dementia in a primary care setting. As a result
the committee were forced to extrapolate from the results obtained
using appropriate tests in a specialist secondary care setting. Based
on the committee’s experience, certain tests (including MRI, CT,
SPECT, PET and biomarkers) were not considered to be useful to
determine whether to refer a person with suspected dementia for
further investigation. These tests would not be available in a primary
care setting and their value lies later in the diagnostic pathway to help
with dementia subtype classification.
It was noticeable that some tests were missing from the evidence
base. The committee commented that there was a lack of data on the
GPCOG test, which is routinely used in GP surgeries in the UK.
However, this test is used to screen for dementia and no published
evidence was found in a population with suspected dementia (the
available studies were all on dementia screening, which is not within
the scope of this guideline). A similar problem was found with the
CAMCOG test, as the studies identified were excluded for using the
test to screen for dementia and/or being case-control studies.
Primary studies using Cantab mobile and Cantab insight were also
missing. In addition, there was a shortage of evidence for diagnosing
dementia in people with Parkinson’s disease or stroke as the
evidence found was focused on screening for dementia in these
people and was inadmissible as a result.
Other considerations The committee agreed that physicians should be aware of the
additional challenges of diagnosing dementia in certain vulnerable
groups, such as people with learning difficulties and Down’s
syndrome, and those people with language and sensory impairment,
lower educational levels and a low standard of literacy. Whilst the
evidence base did not allow them to make specific recommendations
for how the diagnostic pathway should be different for these groups
of individuals, they agreed that it was important that people from
these more difficult to diagnose groups should be assessed by a
clinician with specialist skills in those areas, who would be familiar
with the difficulties and able to make appropriate adaptations to the
process used. A cross-reference to the NICE guideline on mental
health problems in people with learning disabilities was also added,
103
which provides advice on some specific instruments to use when
assessing for dementia in people with learning disabilities.
The committee agreed that some tests (e.g. MoCA) are less robust in
certain population groups due to cultural differences (educational
levels, language issues), and this can skew the resulting diagnosis of
dementia/ continued suspicion of dementia. They stressed the need
to take these issues into account when choosing cognitive tests and
to select test cut-offs appropriately based on the characteristic of the
person with suspected dementia. For this reason, the committee
agreed it would not be appropriate to specify cut-offs that should be
used for tests, as these would need to be adjusted for particular
individuals.
The committee noted that neuropsychological tests are not designed
to be used to diagnose dementia. Rather they either assess
performance in one single aspect of cognition (e.g. episodic memory
or naming ability) or may be grouped into batteries that either cover
performance in a range of aspects (e.g. RBANS) or give a detailed
profile of performance across the elements of one broad construct
such as memory (e.g. Wechsler Memory Scale). By providing this
detailed information about cognitive functioning a thorough
neuropsychological assessment can contribute useful information to
the process of reaching a dementia diagnosis and distinguishing
between dementia sub-types.
The committee commented that primary care physicians may need
support and training to develop the skills to make appropriate
referrals for dementia diagnosis and that the use of the IQCODE
could help provide some structure to their discussions with
informants.
5.1.5.2 Diagnosing dementia in specialist services

Relative value of different The committee agreed that, once a person was being assessed for
outcomes dementia in specialist diagnostic services, it was important that any
test (or battery of tests) used should have a balance between positive
likelihood ratios/sensitivity and negative likelihood ratios/specificity.
They also noted that people who are difficult to diagnose may go
through multiple sequential tests before arriving at a diagnosis, and
therefore it was important to consider all test that might provide
relevant diagnostic value, and not only what is the single best test for
a given diagnosis.
The committee also noted that the four metrics reported (positive and
negative likelihood ratios, sensitivity and specificity), despite being
based on the same underlying data, covered slightly different aspects
of diagnostic accuracy, and therefore it was important to consider all
of them as part of decision making.
Trade-off between General issues
benefits and harms The committee discussed the potentially stressful and unpleasant
diagnostic tests that could be used in a specialist setting. These
include lumbar puncture to obtain cerebrospinal fluid (CSF) for
biomarker tests, MRI and other imaging tests. These tests may not
be well tolerated by all patients, particularly those with claustrophobia
(MRI) or people with more severe dementia. The committee noted
that it was important to use these tests only if they are required to
reduce diagnostic uncertainty, if the person with suspected
dementia/with dementia requiring subtype diagnosis agrees and if
they can comply with test requirements. The committee agreed that
to avoid unnecessary tests being undertaken, it was important to
include a specific recommendation stating these tests only be
undertaken if they would reduce diagnostic uncertainty and reducing
that uncertainty would change management.
104
In addition, the committee discussed the advantage to the person
with an uncertain diagnosis of dementia, and for the NHS (see
below), of carrying out certain tests (e.g. SPECT, PET imaging and
biomarker tests) in series rather than parallel. Carrying out these
tests in parallel may result in a faster diagnosis, but this may also
mean that unnecessary tests are performed and patients are
exposed to additional stressful situations. As a result the committee
recommended that these tests be performed sequentially, in either
order, but that the remaining test should be ‘considered’ if diagnosis
remained unclear.
Initial diagnosis in a specialist care setting
Following referral to specialist care, the committee recommended the
use of validated criteria to guide the clinician during the diagnostic
process. The committee noted that there was limited evidence for the
diagnostic accuracy of the criteria. The quality of the data was
variable with high sensitivity and specificity in some cases (DLB
consensus criteria (high quality), FTDC criteria for bv-FTD (moderate
quality), Movement Disorders Society for PDD (low to moderate
quality) and lower sensitivity with good specificity in others (NINDS-
AIREN (low to moderate quality), FTDC criteria for frontotemporal
dementia (high quality). However, as clinical criteria had formed the
reference standard for the assessment of diagnostic test accuracy for
the majority of tests, the committee agreed it was appropriate to
accept them as the current gold standard approach for diagnosis of
dementia subtypes. The committee recognised that the most
accurate reference standard (neuropathology) was usually
unavailable for use before death.
The committee agreed it was appropriate only to consider studies
that looked at the current versions of clinical criteria, although this
was not always easy to determine (for example in the case of
Creutzfeldt-Jakob disease). No studies were found that evaluated the
current version of the NINCDS-ADRDA criteria. In addition, data for
the diagnostic accuracy of the revised FTD diagnostic criteria
(Rascovsky 2011) were not included in this evidence review as it was
presented in such a way that specificity could not be calculated. (All
of the study participants had a diagnosis of FTD at baseline and the
study did not include the final diagnostic results for people who were
not diagnosed with FTD initially). The committee noted that a new
version of the DLB consensus criteria was due to be published, but
decided to examine the evidence obtained for the existing criteria in
the meantime. The committee agreed that if a new version of clinical
criteria had yet to be validated, but had been developed using a
robust methodology and based on the original version, it was
reasonable to assume that it would be at least as accurate as the
earlier version of the criteria.
The committee reviewed the evidence for the diagnosis of dementia
subtypes. The committee also noted that in the UK about two thirds
of people with dementia are diagnosed with Alzheimer’s disease. The
committee noted that people with Alzheimer’s disease often have
deficits in verbal recall and memory (see also discussion in table
above) and as a result it recommended that a test of verbal episodic
memory should be used where Alzheimer’s disease is suspected.
The committee agreed there was also a subset of people with an
unclear diagnosis in whom a referral for formal neuropsychological
testing would be appropriate, either if it is unclear if a person has
cognitive decline, or if the cause of cognitive decline or correct
subtype diagnosis is unclear. The committee agreed it was
appropriate to make a ‘consider’ recommendation around
neuropsychological testing to cover these issues. They noted that
there was evidence for a range of individual test instruments, but that
105
selecting the appropriate ones for an individual and correctly
interpreting the results often needed input from a neuropsychologist.
The committee discussed the importance of making an accurate
diagnosis based on the fewest tests possible to limit the number of
tests a person with suspected dementia is subjected to and to reduce
costs to the NHS (see above for more details). The committee
recommended that only after an initial assessment has been
completed should structural imaging (e.g. CT or MRI) be offered to
people who still have suspected dementia or require dementia
subtype diagnosis. The committee agreed that although MRI had
good specificity for most dementia subtypes the sensitivity was low.
However, the evidence showed that it had good sensitivity, specificity
and moderate to large positive LRs for vascular dementia and
behavioural variant frontotemporal dementia based on low to
moderate quality evidence. The committee also noted there are a
proportion of people for which dementia is well established and the
subtype diagnosis is clear who would not need structural imaging,
and agreed this was important to reflect in the recommendation.
The committee noted that there can be problems with the
interpretation of imaging data by non-specialists and commented that
where scans are requested by primary care physicians/non-
specialists, specialist input should be obtained to help them interpret
scan data. This will facilitate faster, more accurate diagnosis of
dementia and reduce unnecessary tests and referrals.
The committee discussed which tests to perform to diagnose
dementia subtypes if diagnostic uncertainty remained at this point.
Diagnosing Alzheimer’s disease
In keeping with the committee’s earlier discussion about minimising
test burden for patients, they decided to recommend that the
physician ‘consider’ the use of biomarkers or more specialised
imaging (FDG-PET, SPECT) for people with suspected Alzheimer’s
disease where diagnostic uncertainty remains. The committee noted
the role of imaging to help exclude other pathologies. Based on the
strength of the biomarker evidence (ranging from low to high quality
for different tests and combinations of tests) they used a ‘consider’
recommendation.
Biomarkers can be useful for the diagnosis of certain dementia
subtypes, but there are a number of issues that need to be
considered, including difficulty in obtaining samples from patients
(see above) and the lack of reliability of some of these tests in older
people. Specifically, the committee noted that it was very important to
consider the age of a patient before referring them for biomarker
tests. Biomarker levels are altered in certain types of dementia (such
as Alzheimer’s disease and Creutzfeldt-Jakob disease), but their
levels change with age as well. This means that the tests will give
more false positives in older people as the levels of biomarkers are
also altered in cognitively normal people of this age. The committee
made a ‘be aware’ recommendation to accompany the use of
biomarkers for the diagnosis of Alzheimer’s disease to highlight this
issue.
The committee noted the importance of knowing the mean age and
age range for the interpretation of diagnostic test accuracy evidence,
in particular for biomarker data. In the case of Toledo (2012), they
commented that the sensitivities and specificities of p-tau 181, and
amyloid beta and total tau combined were particularly high compared
with other studies (e.g. Frisoni, 2009; Dumurgier, 2015). This may be
linked to the mean age (69 years) of the participants, which is about
10 years lower than the average memory clinic population in the UK.
Where there was clinical suspicion of Alzheimer’s disease, the
committee recommended that the use of FDG-PET be ‘considered’
106
by the physician based on the variable quality of evidence from very
low to moderate. Good sensitivities, specificities and LRs were seen
when the test was used to distinguish AD from other dementias, no
dementia and non-AD, but the test was less diagnostically informative
for distinguishing AD from FTD and DLB.
If FDG-PET was unavailable, the committee recommended that
perfusion SPECT be ‘considered’. This category included HMPAO
SPECT and ECD SPECT. The committee agreed that there was
insufficient difference between the test accuracy and likelihood ratios
for these tests to warrant recommendation of either one over the
other. In both cases, evidence quality was variable (very low to
moderate), sensitivity and specificity were good (around 70 -80%),
with mostly moderate positive and negative LRs. They noted
however, that HMPAO SPECT was the type routinely used in the UK,
and this was unlikely to change.
Based on the committee’s earlier discussion about the benefits to the
person with suspected dementia of carrying out certain tests in
series, they recommended that the other test should be ‘considered’
should diagnostic uncertainty remain.
The committee decided to write a ‘do not use’ recommendation for
apolipoprotein E (ApoE) genetic testing as ApoE status is associated
with an increased risk of dementia rather than being a biomarker for
the occurrence of dementia. The committee also included EEG in the
‘do not use’ recommendation as the test provided no meaningful
diagnostic information for diagnosing AD based on the positive and
negative likelihood ratios from high-quality evidence. The committee
noted that while there were other tests that also did not provide
meaningful diagnostic information, these were not commonly used in
the UK, and they agreed it was important to retain the
recommendation on EEG from the old guideline so as to not give the
impression that the situation has changed.
The committee noted that a proportion of people with early-onset
Alzheimer’s disease will have a genetic cause for their dementia, and
agreed it was appropriate to include a ‘be aware’ recommendation on
this issue.
Diagnosing dementia with Lewy bodies
The committee noted that DLB resulted in distinctive features on
SPECT that could be used for diagnosis. The committee
recommended the use of 123I-FP-CIT SPECT for the diagnosis of
DLB based on low to high quality evidence from 5 studies comparing
DLB to non-DLB or other dementias. The positive and negative
likelihood ratios were large to very large with high sensitivity and
specificity.
The committee chose not to recommend IMP-SPECT as it had lower
sensitivity and specificity than 123I-FP-CIT SPECT, with a moderate
positive likelihood ratio and a negative likelihood ratio that showed
the test gave no meaningful diagnostic information and was based on
low quality evidence.
The committee discussed the evidence for the use of MIBG cardiac
scintigraphy for the diagnosis of DLB. The evidence from 5 studies
looking at DLB versus non-DLB was of low quality, but had good
sensitivity, specificity and large to very large likelihood ratios. Another
single study looking at DLB versus other dementias provided
moderate to high quality evidence with similar diagnostic accuracy.
Taken together, this evidence led the committee to make a ‘consider’
recommendation. Importantly, they observed that this test was not
useful for diagnosing PDD, as MIBG imaging results were already
abnormal in PD.
The committee discussed the importance of not ruling out a diagnosis
of DLB solely based on these imaging tests, as they are not
107
completely accurate, and it made a recommendation to reflect this
issue.
The evidence presented suggested that EEG could be useful for the
diagnosis of DLB (moderate-high quality evidence, large positive and
negative likelihood ratios and high sensitivity and specificity).
However, the committee commented that the data was from 1 study
that used complex processing of results (and a convoluted algorithm)
and that this would be beyond a standard clinical neurophysiology
department in the UK. In practice, EEG is used qualitatively in the
NHS, and no evidence was available using this method. They did not
recommend EEG here as a result.
The committee also noted that the Lewy body composite risk score
(LBCRS) is not actually a diagnostic test for dementia, but rather
asks about symptoms to facilitate diagnosis.
People with PD do not only develop PDD, but may be diagnosed with
PD then AD. The committee commented that the Rey-Osterrieth
complex figure test (ROCF) and cued recall tests are not suitable to
distinguish PDD from PD with AD.
Diagnosing frontotemporal dementia
The committee discussed the evidence regarding FDG-PET and
perfusion SPECT (ECD-SPECT and HMPAO-SPECT) for the
diagnosis of FTD. They agreed that the evidence for FDG-PET was
of very low to moderate quality, with low sensitivity, high specificity,
moderate to large positive likelihood ratios and not diagnostically
meaningful to large negative likelihood ratios. In the case of
diagnosing PPA versus non-PPA the evidence quality was higher
(low to moderate) and the positive likelihood ratio was very large. As
a result, the committee decided to recommend that physicians use
this test when diagnostic uncertainty remains, in particular because of
the very high specificities seen.
There were multiple studies looking at HMPAO SPECT for the
diagnosis of FTD versus other types of dementia. In all cases, the
evidence for multiple-headed cameras alone gave higher sensitivity
(around 0.7 versus 0.5) and similar specificity (around 0.9) than when
this was combined with data for single-headed cameras, which are
no longer in use. Data quality was very low to moderate, with
moderate to very large positive and negative likelihood ratios. In
contrast, the evidence for ECD-SPECT was confined to 1 moderate
quality study which found that this test was very sensitive and
specific, with very large positive and negative likelihood ratios for the
diagnosis of FTD. Based on the evidence and their experience that
HMPAO SPECT imaging was a more established technique in the
UK, the committee decided not to specify which type of SPECT be
carried out, but to recommend perfusion SPECT be considered as an
alternative to FDG-PET.
The committee commented that FTD may be easy to diagnose if the
scan demonstrates a clear pattern of unilateral or bilateral frontal or
temporal lobe atrophy, but a person can have a normal scan and still
have FTD. As a result, the committee made a ‘do not’
recommendation to prevent FTD being ruled out solely based on the
above recommended tests.
The committee noted that a proportion of people with frontotemporal
dementia will have a genetic cause for their dementia, and agreed it
was appropriate to include a ‘be aware’ recommendation on this
issue.
Diagnosing vascular dementia
MRI was recommended by the committee for cases where diagnostic
uncertainty remains, but there is suspicion of vascular dementia. This
recommendation was based on moderate to high quality evidence
from 2 studies associated with moderate to large positive likelihood
108
ratios and good sensitivity (0.7-0.8) and specificity. Low to moderate
quality evidence from the same studies was associated with
moderate negative likelihood ratios.
In cases where MRI was unavailable or contraindicated, the
committee recommended that computed tomography (CT) be used
instead. Whilst no specific evidence was identified looking at the
diagnostic accuracy of CT for vascular dementia, they agreed that the
positive results found for MRI implied that other forms of structural
imaging were also likely to provide diagnostic value, as similar
features would be detected on CT and MRI. They also agreed that
MRI should be the first choice of test, as this was the test with clear
and robust evidence to support its use.
The committee commented that it was unclear how much vascular
damage on a scan was sufficient to impair patient functioning. They
noted that it was relatively straightforward to interpret effects on
function from low or high levels of vascular disease, but the clinical
relevance of moderate levels of vascular disease on imaging was
often less clear. In these cases, the committee suggested that
physicians refer to clinical criteria and seek expert advice where
needed.
The committee noted that a proportion of people with early-onset
vascular dementia will have a genetic cause for their dementia, and
agreed it was appropriate to include a ‘be aware’ recommendation on
this issue.
Consideration of health The committee agreed that in cases of diagnostic uncertainty,
benefits and resource carrying out biomarker and imaging tests in series, rather than in
use parallel, may result in a clear diagnosis with reduced patient
exposure to potentially traumatic tests and at a reduced cost to the
NHS.
The committee noted that, whilst there were costs associated with
many of the tests recommended (in particular, imaging and
biomarkers) these tests were already routinely in use in the NHS and
therefore there should not be a substantial change in resource use
from the recommendations made. They also noted that, for imaging
tests, the largest cost was the initial purchase of the machinery itself,
and afterwards the incremental cost of each test begin carried out
was much lower.
Quality of evidence Positive and negative likelihood ratios were used to determine the
diagnostic accuracy of the tests and the level of confidence/degree of
uncertainty surrounding this estimate. Large positive likelihood ratios
were associated with large increases in the probability that a person
with a positive test had dementia, while large negative likelihood
ratios were associated with large decreases in the probability of
having dementia given a negative test result. The confidence
intervals for these measures, and the minimally important differences
specified for them in section 3.4, were used to determine the level of
precision in the evidence, which together with consideration of the
risk of bias, heterogeneity and applicability made up the overall
quality rating for the studies.
The committee discussed the importance of ensuring that a study
population was relevant for UK diagnostic services and agreed it was
appropriate to downgrade a study for indirectness if the participants
had less education than the general UK population, or if the age of
the people recruited was lower than was considered to be
representative of the UK memory clinic population. It noted that the
variability in sensitivity and specificity observed between studies
using the same tests may be due to differences in the populations
tested. The committee agreed that certain tests are more vulnerable
to this issue than others (e.g. certain cognitive tests, discussed in
primary assessment table above).
109
The committee agreed that the evidence on rarer dementia subtypes
(HAND, neurosyphilis, CADASIL, and corticobasal degeneration) was
not sufficiently strong for any recommendations to be made. They
also agreed that these people were likely to be in highly specialist
care, and therefore an inability to make recommendations in the
guideline was unlikely to have an impact on their care.
Other considerations Please see the primary care table above for a discussion of the
issues surrounding the diagnosis of dementia in vulnerable
populations.
The committee noted there was only limited evidence available
around the accuracy and cost-effectiveness of amyloid imaging
despite there now being licensed products available for this. They
agreed the evidence was not sufficient to make recommendations on
its use, but that it was appropriate to make a research
recommendation on this topic. They noted that amyloid imaging was
likely to be used towards the end of the diagnostic pathway in
individuals difficult to diagnose with more standard tests, and
therefore agreed the research recommendation should focus on the
additional value provided by amyloid imaging over and above
standard diagnostic assessment, and the comparative value with
other imaging or biomarker tests that could be used.
5.1.6 Recommendations
Initial assessment in non-specialist settings
1. At the initial assessment take a history (including cognitive, behavioural and

psychological symptoms, and the impact symptoms have on their daily life):
 from the person with suspected dementia and
 if possible, from someone who knows the person well (such as a family
member).
2. If dementia is still suspected after initial assessment:

 conduct a physical examination and
 undertake appropriate blood and urine tests to exclude reversible
causes of cognitive decline and
 use cognitive testing.
3. When using cognitive testing, use a validated brief structured cognitive

instrument such as:
 the 10-point cognitive screener (10-CS)
 the 6-item cognitive impairment test (6CIT)
 the 6-item screener
 the Memory Impairment Screen (MIS)
 the Mini-Cog
 Test Your Memory (TYM).
4. Do not rule out dementia solely because the person has a normal score on a
cognitive instrument.
5. When taking a history from someone who knows the person with suspected
dementia, consider supplementing this with a structured instrument such as the
110
Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) or the
Functional Activities Questionnaire (FAQ)
6. Refer the person to a specialist dementia diagnostic service (such as a memory

clinic or community old age psychiatry service) if:
 reversible causes of cognitive decline (including delirium, depression,
sensory impairment [such as sight or hearing loss] or cognitive
impairment from medicines associated with increased anticholinergic
burden) have been investigated and
 dementia is still suspected.
7. If the person has suspected rapidly-progressive dementia, refer them to a

neurological service with access to tests (including cerebrospinal fluid
examination) for Creutzfeldt-Jakob disease and similar conditions.
8. For more guidance on assessing for dementia in people with learning disabilities,
see the NICE guideline on mental health problems in people with learning
disabilities.
Dementia diagnosis in specialist settings
9. Diagnose a dementia subtype (if possible) if initial specialist assessment

(including an appropriate neurological examination and cognitive testing)
confirms cognitive decline and reversible causes have been ruled out.
10. If Alzheimer’s disease is suspected, include a test of verbal episodic memory in

the assessment.
11. Consider neuropsychological testing if it is unclear:

 whether or not the person has cognitive impairment or
 whether or not their cognitive impairment is caused by dementia or
 what the correct subtype diagnosis is.
12. Use validated criteria to guide clinical judgement when diagnosing dementia
subtypes, such as:
 International consensus criteria for dementia with Lewy bodies
 International FTD criteria for frontotemporal dementia (primary non-
fluent aphasia and semantic dementia)
 International Frontotemporal Dementia Consortium criteria for
behavioural variant frontotemporal dementia
 NINDS-AIREN criteria (National Institute of Neurological Disorders and
Stroke and Association Internationale pour la Recherché et
l'Enseignement en Neurosciences) for vascular dementia
 NIA criteria (National Institute on Aging) for Alzheimer’s disease
 Movement disorders Society criteria for Parkinson’s disease dementia
 International criteria for Creutzfeldt-Jakob disease.
13. Offer structural imaging to rule out reversible causes of cognitive decline and to
assist with subtype diagnosis, unless dementia is well established and the
subtype diagnosis is clear.
111
14. Only consider further diagnostic tests (recommendations 15-28) if:
 it would help to diagnose a dementia subtype and
 knowing more about the dementia subtype would change management.
Further tests for Alzheimer’s disease
15. If the diagnosis is uncertain (see recommendation 14) and Alzheimer’s disease is
suspected, consider either:
 FDG-PET (fluorodeoxyglucose-positron emission tomography-CT), or
perfusion SPECT (single-photon emission CT) if FDG-PET is
unavailable.
or
 examining cerebrospinal fluid for:
 either total tau or total tau and phosphorylated-tau 181 and
 either amyloid beta 1–42 or amyloid beta 1–42 and amyloid beta 1–
40
If a diagnosis cannot be made after one of these tests, consider using the other
one.
16. Be aware that the older a person is, more likely they are to get a false positive with
cerebrospinal fluid examination.
17. Do not rule out Alzheimer’s disease based solely on the results of CT or MRI
scans.
18. Do not use Apolipoprotein E genotyping or electroencephalography to diagnose

19. Be aware that young-onset Alzheimer’s disease has a genetic cause in some
people.
Further tests for dementia with Lewy bodies
20. If the diagnosis is uncertain (see recommendation 14) and dementia with Lewy
bodies is suspected, use 123I-FP-CIT SPECT.
21. If 123I-FP-CIT SPECT is unavailable, consider 123I-MIBG cardiac scintigraphy.
22. Do not rule out dementia with Lewy bodies based solely on normal results on 123I-
FP-CIT SPECT or 123I-MIBG cardiac scintigraphy.
Further tests for frontotemporal dementia
23. If the diagnosis is uncertain (see recommendation 14) and frontotemporal

dementia is suspected, use either:
 FDG-PET or
 perfusion SPECT.
24. Do not rule out frontotemporal dementia based solely on the results of structural,
perfusion or metabolic imaging tests.
112
25. Be aware that frontotemporal dementia has a genetic cause in some people.
Further tests for vascular dementia
26. If the dementia subtype is uncertain (see recommendation 14) and vascular
dementia is suspected, use MRI. If MRI is unavailable or contraindicated, use CT.
27. Do not diagnose vascular dementia based solely on vascular lesion burden.
28. Be aware that young-onset vascular dementia has a genetic cause in some
people.
5.1.7 Research recommendations
1. Does amyloid PET imaging provide additional diagnostic value, and is it cost
effective, for the diagnosis of Alzheimer’s disease and other dementias when
compared with standard diagnostic procedures and other imaging or biomarker
tests?
For more details on the research recommendation made, and the rationale behind it, see
appendix L.
113
5.2 Distinguishing dementia from delirium or delirium with
dementia
Review question
 What are the most effective methods of differentiating dementia or dementia with delirium
from delirium alone?
5.2.1 Introduction
The aim of this review question was to determine the effectiveness of different diagnostic
tests in differentiating between groups of patients with dementia, delirium with dementia or
delirium alone, in people without a current diagnosis of either dementia or delirium. The
review identified studies that fulfilled the conditions specified in Table 22. The full review
protocol is available in Appendix C.
Table 22: Review summary: differential diagnosis of dementia and delirium

Population People (aged 40 years and over) with cognitive impairment and no
current diagnosis of dementia or delirium
Diagnostic variables Relevant diagnostic variables may include:
 History data
 Duration of delirium
 IQCODE
 Inappropriate discharge rates
 Inadequate care planning rates
A systematic literature search was carried out to identify cohort studies, cross-sectional
studies or systematic reviews of diagnostic accuracy studies. Two thousand four hundred
and fifty eight references were screened at the title and abstract level, with 40 potentially
relevant references being ordered for full text review. Of these references, 5 were selected
for inclusion based on their relevance to the review protocol and the presentation of data
which was in a useful format for analysis. One additional paper was screened at full text and
included from re-run searches conducted at the end of the guideline. The excluded studies
are listed, with reasons for their exclusion, in Appendix F. Evidence tables for the included
studies are presented in Appendix E.
5.2.2.1 Analyses
Calculations of diagnostic test sensitivity, specificity, positive likelihood and negative

likelihood ratios (LR) were carried out and are presented in the GRADE tables in Appendix
G. Initial analyses examine the ability of the test to distinguish between delirium or delirium
with dementia patients from dementia alone patients. If the 95% confidence interval for the
likelihood ratios in this analysis did not cross 1 (i.e. the result was statistically significant)
then the data for separating delirium alone versus delirium with dementia patients was also
analysed.
114
The characteristics of the included studies are summarised in Table 23. References for the
included studies are given in appendix I.
Table 23: Summary of included studies

Study Diagnostic
Study details population test(s) Reference test Outcome(s)
Cole (2002) Patients ≥ 65  Confusion Diagnostic and Comparison between
years admitted to Assessment Statistical Manual the number of errors
a medical Method (CAM) of Mental and prevalence of
department from  Delirium Index Disorders III symptoms on the
the emergency (DI) (DSM-III-R) for CAM and DI and
department delirium and diagnosis using
Informant DSM-III-R
Questionnaire on
Cognitive Decline
in the Elderly
(IQCODE) for
dementia
Erkinjuntti Patients ≥ 65  Short Portable Dementia Scale Comparison between
(1987) years admitted to Mental Status the number of
a medical Questionnaire SPMSQ errors and
department (SPMSQ) diagnosis using the
dementia scale
Leonard Patients ≥ 60  Delirium Diagnostic and Comparison between
(2016) years with Rating Scale Statistical Manual the DRS-R98 mean
altered mental Revised 98 of Mental and item severity
state referred to (DRS-R98) Disorders (DSM) scores, CTD or NPI
a psychiatry for  Cognitive Test IV for delirium or scores and the
later life for Delirium dementia and reference diagnosis
consultation- (CTD) (IQCODE-Short
liaison service form) for dementia
 Neuropsychiat
and cognitive
ric Inventory
difficulties
(NPI)
Meagher Patients with  Delirium Diagnostic and Comparison between
(2010) altered mental Rating Scale Statistical Manual the DRS-R98 mean
state identified Revised 98 of Mental and item severity
on daily rounds (DRS-R98) Disorders (DSM) scores or CTD scores
 Cognitive Test IV for delirium or and the reference
for Delirium dementia diagnosis
(CTD)
Richardson Patients > 70  Attention test DSM-5 for delirium The diagnostic
(2017) years old who  Observational with IQCODE and accuracy of the index
were admitted to Scale of Level MMSE for tests compared with
5 acute or of Arousal dementia the reference
rehabilitation (OSLA) diagnosis.
hospitals.
Trzepacz Patients with  Delirium Diagnostic and Comparison between
(2001) dementia or Rating Scale Statistical Manual the DRS-R98 mean
delirium, Revised 98 of Mental scores (severity and
Schizophrenia, (DRS-R98) Disorders (DSM) total) or CTD scores
depression or  Cognitive Test IV for delirium or and the reference
other psychiatric for Delirium dementia diagnosis
disorders from a (CTD)
range of medical
115
Study Diagnostic
Study details population test(s) Reference test Outcome(s)
and nursing
home settings
Standard health economic filters were applied to the clinical search for this question, and a
total of 262 citations was returned. Following review of titles and abstracts, no full text studies
were retrieved for detailed consideration. Therefore, no relevant cost–utility analyses were
identified for this question.
The evidence statements in this review for distinguishing delirium from dementia are written
with reference to the size of the likelihood ratios in the GRADE tables in appendix G, section
G1.2, using the interpretation detailed in the methods section on diagnostic test accuracy
(Table 4) for both point estimates and confidence intervals. Positive likelihood ratios, and
their associated 95% confidence intervals, were used to determine which tests increase the
probability of having delirium or delirium with dementia and negative likelihood ratios, and
their associated 95% confidence intervals, were used to determine which tests decrease the
probability of having delirium or delirium with dementia. This rationale also applies to the
evidence statements for the sections on distinguishing delirium from delirium with dementia,
and delirium with dementia from dementia.
5.2.4.1 Distinguishing delirium and delirium with dementia from dementia
5.2.4.1.1 Results that increase the probability of having delirium or delirium with dementia
The following test results increase the probability a person has delirium or delirium with
dementia to a degree that is likely to be very large:
 DRS-R98 Total score, cut off >21.50 (very low quality, 95% CI goes from slight to very
large)
 DRS-R98 Total score, cut off >22.50 (very low quality, 95% CI goes from slight to very
large)
 DRS-R98 Severity score, cut off >17.00 (very low quality, 95% CI goes from slight to very
large)
 Combination of OSLA and Attention Test, cut off >9 (low quality, 95% confidence interval
goes from large to very large)
dementia to a degree that is likely to be large:
 <4 errors using the SPMSQ (low quality, 95% CI goes from slight to very large)
 DRS-R98 Item Severities- Temporal onset of symptoms, scoring ≥ 2 (moderate quality,
95% CI goes from moderate to very large)
 DRS-R98 Total score, cut off 17.75 (very low quality, 95% CI goes from slight to very
large)
dementia to a degree that is likely to be moderate:
 > 5 symptoms using the CAM (low quality, 95% CI goes from slight to moderate)
116
 > 6 symptoms using the CAM (moderate quality, 95% CI goes from moderate to
moderate)
 > 4 symptoms using the DI (moderate quality, 95% CI goes from moderate to moderate)
 <5 errors using the SPMSQ (low quality, 95% CI goes from slight to large)
 DRS-R98 Item Severities- Sleep-wake cycle disturbance, scoring ≥ 2 (low quality, 95% CI
goes from slight to large)
 DRS-R98 Item Severities- Perceptual disturbances and hallucinations, scoring ≥ 2 (low
quality, 95% CI goes from slight to very large)
 DRS-R98 Item Severities- Lability of affect, scoring ≥ 2 (low quality, 95% CI goes from
 DRS-R98 Item Severities- Language, scoring ≥ 2 (low quality, 95% CI goes from slight to
large)
 DRS-R98 Item Severities- Thought process abnormalities, scoring ≥ 2 (low quality, 95%
CI goes from slight to moderate)
 DRS-R98 Item Severities- Motor agitation, scoring ≥ 2 (low quality, 95% CI goes from
slight to large)
 DRS-R98 Item Severities- Orientation, scoring ≥ 2 (low quality, 95% CI goes from slight to
moderate)
 DRS-R98 Item Severities- Attention, scoring ≥ 2 (low quality, 95% CI goes from slight to
moderate)
 DRS-R98 Item Severities- Physical disorder, scoring ≥ 2 (low quality, 95% CI goes from
slight to moderate)
 DRS-R98 Severity score, cut off 15.25 (very low quality, 95% CI goes from moderate to
large)
 <4 points on the SSF item of CTD (low quality, 95% CI goes from slight to very large)
 OSLA, cut off >4 (low quality, 95% confidence interval goes from moderate to large)
dementia to a degree that is likely to be slight:
 > 2 symptoms using the DI (moderate quality, 95% CI goes from slight to moderate)
 > 3 symptoms using the DI (low quality, 95% CI goes from slight to moderate)
The following results were not significantly different from random chance:
 <3 errors using the SPMSQ (low quality)
 DRS-R98 Item Severities- Delusions, scoring ≥ 2 (low quality)
 DRS-R98 Item Severities- Motor retardation, scoring ≥ 2 (low quality)
 DRS-R98 Item Severities- Short-term memory, scoring ≥ 2 (low quality)
 DRS-R98 Item Severities- Long-term memory, scoring ≥ 2 (low quality)
 DRS-R98 Item Severities- Visuospatial processing, scoring ≥ 2 (low quality)
 DRS-R98 Item Severities- Fluctuations in symptom severity, scoring ≥ 2 (low quality)
5.2.4.1.2 Results that decrease the probability of having delirium or delirium with dementia
The following results decrease the probability a person has delirium or delirium with
dementia to a degree that is likely to be very large:
 ≤ 5 symptoms using the CAM (moderate quality, 95% CI goes from very large to very
large)
117
 ≤ 6 symptoms using the CAM (moderate quality, 95% CI goes from very large to very
large)
 DRS-R98 Total score, cut off <17.75 (low quality, 95% CI goes from very large to very
large)
 DRS-R98 Total score, cut off <21.50 (low quality, 95% CI goes from very large to
moderate)
 DRS-R98 Severity score, cut off <15.25 (low quality, 95% CI goes from very large to
moderate)
dementia to a degree that is likely to be large:
 DRS-R98 Item Severities- Physical disorder, scoring < 2 (moderate quality, 95% CI goes
from large to moderate)
 DRS-R98 Total score, cut off <22.50 (low quality, 95% CI goes from very large to
moderate)
 DRS-R98 Severity score, cut off <17.00 (low quality, 95% CI goes from very large to
moderate)
 Combination of OSLA and Attention Test, cut off ≤9 (low quality, 95% confidence interval
goes from very large to moderate)
dementia to a degree that is likely to be moderate:
 ≤2 symptoms using the DI (low quality, 95% CI goes from moderate to slight)
 ≥4 errors using the SPMSQ (low quality, 95% CI goes from moderate to slight)
 ≥5 errors using the SPMSQ (low quality, 95% CI goes from large to slight)
 DRS-R98 Item Severities- Sleep-wake cycle disturbance, scoring <2 (low quality, 95% CI
goes from moderate to slight)
 DRS-R98 Item Severities- Attention, scoring <2 (low quality, 95% CI goes from moderate
to slight)
 DRS-R98 Item Severities- Temporal onset of symptoms, scoring <2 (low quality, 95% CI
 ≥4 points on the SSF item of CTD (low quality, 95% CI goes from moderate to slight)
 OSLA, cut off ≤4 (low quality, 95% CI goes from very large to moderate)
dementia to a degree that is likely to be slight:
 ≤4 symptoms using the DI (low quality, 95% CI goes from slight to slight)
 ≥3 errors using the SPMSQ (moderate quality, 95% CI goes from slight to slight)
 DRS-R98 Item Severities- Perceptual disturbances and hallucinations, scoring <2
(moderate quality, 95% CI goes from slight to slight)
 DRS-R98 Item Severities- Lability of affect, scoring <2 (moderate quality, 95% CI goes
from slight to slight)
 DRS-R98 Item Severities- Language, scoring <2 (moderate quality, 95% CI goes from
slight to slight)
 DRS-R98 Item Severities- Thought process abnormalities, scoring <2 (moderate quality,
95% CI goes from slight to slight)
 DRS-R98 Item Severities- Motor agitation, scoring <2 (moderate quality, 95% CI goes
118
 DRS-R98 Item Severities- Motor retardation, scoring <2 (moderate quality, 95% CI goes
 DRS-R98 Item Severities- Orientation, scoring <2 (moderate quality, 95% CI goes from
slight to slight)
 DRS-R98 Item Severities- Delusions, scoring <2 (moderate quality)
 DRS-R98 Item Severities- Short-term memory, scoring <2 (moderate quality)
 DRS-R98 Item Severities- Long-term memory, scoring <2 (moderate quality)
 DRS-R98 Item Severities- Visuospatial processing, scoring <2 (moderate quality)
 DRS-R98 Item Severities- Fluctuations in symptom severity, scoring <2 (moderate quality)
5.2.4.2 Distinguishing delirium from delirium with dementia
5.2.4.2.1 Results that increase the probability of having delirium alone
The following results increase the probability a person has delirium alone to a degree that is
likely to be slight:
 ≤3 symptoms using the DI (low quality, 95% CI goes from slight to moderate)
 DRS-R98 Item Severities- Sleep-wake cycle disturbances, scoring ≥ 2 (moderate quality,
 DRS-R98 Item Severities- Thought process abnormalities, scoring ≥ 2 (low quality, 95%
CI goes from slight to moderate)
 DRS-R98 Item Severities- Temporal onset of symptoms, scoring ≥ 2 (moderate quality,
 >5 symptoms using the CAM (moderate quality)
 >6 symptoms using the CAM (moderate quality)
 >2 symptoms using the DI (moderate quality)
 >3 symptoms using the DI (moderate quality)
 >4 symptoms using the DI (low quality)
 DRS-R98 Item Severities- Perceptual disturbances and hallucinations, scoring ≥ 2 (low
quality)
 DRS-R98 Item Severities- Lability of affect, scoring ≥ 2 (low quality)
 DRS-R98 Item Severities- Language, scoring ≥ 2 (low quality)
 DRS-R98 Item Severities- Motor agitation, scoring ≥ 2 (low quality)
 DRS-R98 Item Severities- Orientation, scoring ≥ 2 (low quality)
 DRS-R98 Item Severities- Attention, scoring ≥ 2 (moderate quality)
 DRS-R98 Item Severities- Physical disorder, scoring ≥ 2 (moderate quality)
 <4 points on the SSF item of CTD (moderate quality)
119
5.2.4.2.2 Results that decrease the probability of having delirium alone
The following results decrease the probability a person has delirium alone to a degree that is
likely to be moderate:
 ≥5 errors using the SPMSQ (low quality, 95% CI goes from large to slight)
 DRS-R98 Item Severities- Thought process abnormalities, scoring <2 (low quality, 95% CI
 DRS-R98 Item Severities- Temporal onset of symptoms, scoring <2 (low quality, 95% CI
The following results decrease the probability a person has delirium alone to a degree that is
likely to be slight:
 DRS-R98 Item Severities- Sleep-wake cycle disturbances, scoring <2 (low quality, 95% CI
 ≤2 symptoms using the DI (low quality)
 ≤5 symptoms using the CAM (moderate quality)
 ≤6 symptoms using the CAM (moderate quality)
 ≥3 errors using the SPMSQ (moderate quality)
 ≥4 errors using the SPMSQ (low quality)
 DRS-R98 Item Severities- Perceptual disturbances and hallucinations, scoring <2
(moderate quality)
 DRS-R98 Item Severities- Lability of affect, scoring <2 (moderate quality)
 DRS-R98 Item Severities- Language, scoring <2 (moderate quality)
 DRS-R98 Item Severities- Motor agitation, scoring <2 (moderate quality)
 DRS-R98 Item Severities- Orientation, scoring <2 (moderate quality)
 DRS-R98 Item Severities- Attention, scoring <2 (low quality)
 DRS-R98 Item Severities- Physical disorder, scoring <2 (low quality)
 ≥4 points on the SSF item of CTD (moderate quality)
5.2.4.3 Distinguishing delirium with dementia from dementia
5.2.4.3.1 Results that increase the probability of having delirium with dementia
The following test results increase the probability a person has delirium with dementia to a
degree that is likely to be very large:
 OSLA, cut off >4 (low quality, 95% confidence interval goes from moderate to very large)
 Combination of OSLA and Attention Test, cut off >9 (low quality, 95% confidence interval
goes from moderate to very large)
5.2.4.3.2 Results that decrease the probability of having delirium with dementia
The following test results decrease the probability a person has delirium with dementia to a
degree that is likely to be very large:
 Combination of OSLA and Attention Test, cut off ≤9 (low quality, 95% confidence interval
goes from very large to moderate)
120
The following test results decrease the probability a person has delirium with dementia to a
degree that is likely to be moderate:
 OSLA, cut off ≤4 (low quality, 95% confidence interval goes from large to moderate)
5.2.4.4 Health economic evidence
No health economic evidence was identified for this review question.

Relative value of different The committee noted that the tests for delirium examined in this review
outcomes consist of multiple domains some of which are impaired in both people
with dementia and those with delirium, while other domains such as
attention and concentration are important issues in delirium, but not
necessarily impaired in dementia. This meant that some tests were
inherently better suited to differentiate between delirium and dementia
than others.
The committee agreed that both likelihood ratios and sensitivity and
specificity were relevant outcomes for the review, and the decision as to
which was most important in a given situation would depend on whether
the test was being used for ruling out or ruling in potential cases.
Trade-off between The committee agreed that the length of the test for delirium was very
benefits and harms important. The DRS-R98 was considered too long to be usefully applied
in a non-specialist NHS setting. The CTD was also considered
problematic due to its length and the requirement for props, but most
particularly as it included a number of non-specific items to detect
cognitive problems that were likely to be positive in people with
dementia as well as those with delirium.
The committee agreed that the high sensitivity, reasonable specificity
(particularly using a cut-off of >6 symptoms) and positive likelihood ratio
of the CAM test meant that it was a suitable choice to distinguish
between delirium and delirium superimposed on dementia from
dementia alone and therefore decided to recommend the use of this test
for this purpose with a ‘consider’ level of strength due to the low-
moderate quality of the evidence reviewed. The committee agreed the
evidence was equally strong for the use of the Observational Scale of
Level of Arousal (OSLA), and this was also added to the same
recommendation. Whilst the sensitivity and specificity were somewhat
different between these two measures, the committee agreed this was
likely to be a result of the particular cut-offs chosen to use on the tests
in the studies, and the overall diagnostic accuracy of the tests was
similar between the studies. The committee noted there were other tests
(such as the Short Portable Mental State Questionnaire) which also had
good positive likelihood ratios, but decided not to recommend them as
the negative likelihood ratios were less good than for either the long
CAM or the OSLA.
The committee commented that the CAM test existed in both long (10
questions) and short (the first 4 questions) forms. The committee noted
that the short CAM was already recommended for use in the delirium
guideline section 1.5.1, but as the data provided in Cole 2002, referred
to the long CAM it was only able to recommend the use of the long CAM
to distinguish dementia from delirium in this guideline. They also agreed
that it made logical sense that the additional information provided by the
long CAM may be helpful in situations where differential diagnosis is
more complex, whilst the short CAM remains appropriate in less
complex cases.
The committee agreed that the tests presented lacked sufficient
specificity to allow differentiation between people with delirium
superimposed on dementia from delirium alone. Therefore it decided to
121
issue a ‘do not use’ recommendation to highlight the importance of the
clinician not relying on standardised tests alone to distinguish groups
here. The committee agreed that if any uncertainty regarding diagnosis
remained then the person exhibiting an altered mental state should be
treated for delirium first. It decided to make a similar recommendation to
that in the delirium guideline at the end of part 1.5.1 where this was
clearly stated.
The committee agreed that when people with an altered mental state
screened negative for delirium it was important to determine whether
they had dementia instead and this should follow the standard advice
given in the section of the guideline on diagnosing dementia and
dementia subtypes, which already includes reference to ruling out
reversible causes of cognitive decline.
The committee discussed the importance of assessing people who were
successfully treated for delirium for the presence of underlying
dementia, but noted the uncertainty surrounding the best time interval
for this assessment. The committee agreed that this issue would make
a suitable research recommendation, and would best be evaluated
using a cohort study that regularly assessed peoples cognitive
performance following delirium (for example, at 1, 2, 3, 6, 9 and 12
months), and identified the point at which cognition scores stabilise,
which would be the appropriate time to assess for dementia.
Trade-off between net No economic evidence was identified for this review question and
health benefits and economic evidence was not prioritised. The committee agreed that
resource use inaccurate diagnoses of either delirium or dementia would be both
expensive and harmful for the person involved, and therefore the
potential additional time for using the long version of the CAM would be
justified by the costs saved from not providing inappropriate treatments
based on incorrect diagnoses.
Quality of evidence The committee noted that although the DRS-R98 total and severity
scores reported high sensitivity and specificity in detecting delirium
versus dementia, the very low quality of the data in this study (Trzepacz,
2001) and the length of the test meant that they were unable to
recommend it. The committee was unable to recommend the use of the
short CAM in preference to the long CAM to diagnose delirium in the
absence of evidence examining the short CAM in the population of
interest (people with an altered mental state who could have dementia,
delirium or delirium superimposed on dementia).
Other considerations The committee noted that people with cognitive impairment who were
diagnosed with delirium had specific social needs that remained to be
determined, but considered that this fell into the category of a delirium
guideline research recommendation rather than one that could be
included here.
The committee noted that people who had previously experienced
delirium were at increased risk of developing dementia. It commented
that older people usually took longer to recover from delirium and that
delirium was rarely diagnosed in people <35 years old.
Telling the difference between delirium and dementia in people without a diagnosis of
either
29. For people who are in hospital and have cognitive impairment with an unknown
cause, consider using one of the following to find out whether they have delirium
or delirium superimposed on dementia, compared with dementia alone:
 the long confusion assessment method (CAM)
122
 the Observational Scale of Level of Arousal (OSLA).
30. Do not use standardised instruments (including cognitive instruments) alone to

distinguish delirium from delirium superimposed on dementia.
31. If it is not possible to tell whether a person has delirium, dementia, or delirium
superimposed on dementia, treat for delirium first. For guidance on treating
delirium, see treating delirium in the NICE guideline on delirium.
2. In people with treated delirium who no longer meet the DSM-5 criteria for delirium,
but who have persistent cognitive deficits, when is the most appropriate time to
carry out an assessment for dementia?
appendix L.
123
5.3 Case finding for people at high risk of dementia
Review question
 What are the most effective methods of case finding for people at high risk of dementia?
5.3.1 Introduction
The aims of this review were to establish whether case finding alters the incidence of correct
dementia diagnoses in people at high risk of dementia and what effect this has on the
outcomes for these people. Studies were not considered to be relevant if they only reported
the incidence of dementia identified; studies also needed to contain an intervention
component for individuals identified as living with dementia. The review focused on
identifying studies that fulfilled the conditions specified in Table 24. For full details of the
review protocol, see Appendix C.
Table 24: Review summary: case finding for people at high risk of dementia
Population People (aged 40 years and over) who are at high risk of dementia in:
 Primary care
 Acute hospitals
 Care homes
 People over 60 at high vascular risk (prior stroke)
 People with learning disabilities
 People with other neurological disorders (MS)
Intervention Standard cognitive tests
Outcomes  Incidence of dementia (and other conditions) correctly identified in
people classified as at risk
 Delay to diagnosis
 Sensitivity, specificity, NPV, PPV
 Health related quality of life
 Overtreatment
 Resource use and cost
A systematic literature search was carried out to identify prospective cohort studies and
RCTs of case finding approaches. Two thousand, two hundred and seventy eight references
were screened at the title and abstract level, with 6 potentially relevant references being
ordered for full text review, including a systematic review for the US Preventative Services
Task Force (Lin, 2013). One extra study (Borson, 2007) was identified during screening for
another related review. Of these references, 1 (van den Dungen, 2016) was selected for
inclusion based on its relevance to the review protocol. The systematic review (Lin, 2013)
was unable to find any trials that directly assessed whether case finding affected decision
making and outcomes for patients, carers or society. This is in agreement with our findings
as the single identified RCT was published after the systematic review. ClinicalTrials.gov was
also checked to identify additional clinical trials relevant to this question. One ongoing
dementia diagnosis RCT was found (Fowler, 2014), but not included. The excluded studies
are listed, with reasons for their exclusion, in Appendix F. Evidence tables for the included
studies are presented in Appendix E, and GRADE profiles in appendix G.
124
The characteristics of the included study are summarised in Table 25. References for the
Table 25: Included study

Study Study Study Intervention & Relevant
reference design population comparator outcomes Comments
Van den Cluster In stage 1: 647 Intervention Stage 1: Primary Location:
Dungen RCT people with training of family outcome: new Netherlands
(2016) possible physicians to diagnose diagnoses of Follow up:
cognitive dementia MCI and 12 months
impairment Intervention Stage 2: dementia
across 15 assessment of Secondary
primary care cognition and outcome:
practices. functioning by study mental health
In stage 2: 145 two practice nurses. of patients and
of the patients Comparator for both their relatives
from stage 1. stages: no additional
training and usual care
total of 684 citations was returned. Following review of titles and abstracts, no full text studies
were retrieved for detailed consideration. Therefore, no relevant cost–utility analyses were
identified for this question.
5.3.4.1 Primary outcomes
Very low-quality evidence from a single cluster RCT containing 647 participants across 15
primary care practices could not detect a difference in the number of new diagnoses of
dementia and MCI between control practices and those taking part in a dementia education
programme for family physicians over a 12 month period.
Very low-quality evidence from a single cluster RCT containing 145 participants across 15
primary care practices could not detect a difference in the number of new diagnoses of
dementia and MCI between control practices and those taking part in a dementia education
programme for family physicians with practice nurse assessment for dementia over a 12
month period.
5.3.4.2 Secondary outcomes
Moderate-quality evidence from a single cluster RCT containing 145 participants across 15
primary care practices could not detect a difference in the mental health or quality of life of
trial participants or their relatives between people attending the control and intervention
practices over a 12 month period, as measured by MH5, EQ5D and QoL-AD (outcomes for
participants) and EQ-5D, GHQ12, SSCQ (outcomes for close relatives) questionnaires.
125
Relative value of different The committee agreed that while case finding required the targeted
outcomes screening of people at high risk of dementia, it was important that this
was coupled with an intervention to provide care to people after
diagnosis. An intervention that merely identified people but then did
not alter anything about their future care was unlikely to be beneficial,
and could indeed increase anxiety in people without providing any
equivalent advantages.
The committee noted that the secondary outcomes used to measure
participant and carer outcomes in the van den Dungen trial were
appropriate and likely to detect any benefit and harm associated with
case finding and subsequent case management.
The committee commented that the use of the CAMCOG in
preliminary studies by van den Dungen (2012) could have
overestimated the likely yield of dementia diagnoses if normative data
for this test was obtained from a more highly educated population
than the sample included in this study.
Trade-off between The committee noted that there was no evidence from the van den
benefits and harms Dungen trial to suggest that participants or their carers received any
benefit or sustained harm from case finding as judged by their
reported quality of life outcomes. However, the committee noted that
a diagnosis of dementia can be upsetting and a proportion of people
may be wrongly diagnosed (false positive) due to diagnostic test
inaccuracy, leading to inappropriate treatment and unnecessary
stress. In addition, case finding may consume resources that could
be better used elsewhere. The committee noted that a case finding
programme would be particularly wasteful if a large proportion of
people at high risk of dementia refused to be subsequently tested, as
was the case in the van den Dungen trial. Thus the committee was
unwilling to recommend case finding and proposed that it should only
be used as part of an appropriately designed research study that
includes an intervention for people diagnosed with dementia following
identification.
The committee noted that in the context of the acute hospital there
are already recommendations to screen for delirium. This is different
from screening for dementia and should still be carried out in
accordance with the NICE delirium guideline
The committee agreed that the lack of evidence regarding the
effectiveness of case finding in dementia necessitated a research
recommendation for studies that include an intervention component
for the people identified as having dementia. The committee agreed
that this recommendation should be for structured case finding with
defined criteria for the selection of high risk patients.
Consideration of health The committee agreed that case finding was not resource neutral,
benefits and resource and that in the absence of subsequent treatment strategies with
use proven benefits for people who have been identified through case
finding, it is not a good use of resources.
Quality of evidence The committee noted the shortage of evidence to address the
effectiveness of case finding for dementia.
The committee discussed whether the van den Dungen trial met the
study selection criteria sufficiently. In particular the committee
commented that the RCT used clinician recollection and medical
records to identify people with possible cognitive impairment to form
their high risk group. The committee noted that while case finding
required the targeted screening of people at high risk of dementia, a
more scientific basis would be needed to define these groups for
testing in practice.
The committee commented that the control arm of this trial also
involved participants identified as having possible dementia by their
126
primary care physicians and that as a result of this identification
process there may have been an increase in the diagnosis of
dementia and MCI in this group as well. This could account for the
lack of difference in the new diagnoses of dementia between the
control and intervention groups.
The committee noted that diagnosis of MCI, but not dementia, was
increased in the intervention group compared with the control group.
The committee commented that that the dementia management
intervention was likely to have had less impact on the people
diagnosed with MCI, and this may have contributed to the similarity in
participant and carer outcomes between the control and intervention
groups.
The committee noted that the RCT had several additional limitations:
the low incidence of dementia and MCI in the intervention arm
despite selecting people with possible cognitive impairment, the
problems of nurses and physicians failing to comply with the study
protocol in the intervention arm and the low level of consent for stage
2 of the trial by people with possible dementia. This lack of
compliance by the medical staff and people with possible dementia
would also be expected in real life, and therefore this trial may be
accurately replicating the difficulties that could be encountered in
introducing case finding interventions to general practice.
Other considerations The committee agreed that it was particularly important to ensure that
people with learning disabilities were included in the research
recommendation as a group of people at high risk of dementia.
The committee noted that case finding for dementia was introduced
in acute hospitals as part of the Commissioning for Quality and
Innovation (CQUIN) 2013/14 guidance, and that the committee’s
recommendations should be interpreted as applying to situations
outside this existing guidance.
32. Only conduct case finding for suspected dementia as part of a clinical trial that
also provides an intervention to people diagnosed with dementia.
3. What is the effectiveness of structured case finding (including a subsequent

intervention for people identified as having dementia) in people at high risk of
dementia, following up both people identified as having or not having dementia?
appendix L.
127
6 Involving people living with dementia in
decisions about care
When faced with a progressive medical condition most people will be interested in how that
condition will be managed in the future. To be successful, advanced care planning must have
the person and their wishes at its centre. Professionals may face challenges as they offer
support and guidance throughout this process.
Each person’s situation is unique, and therefore some people living with dementia will want
as much practical support and information as is available in their location; others may deny
they have problems and reject help; and again others may prefer to defer all decisions to the
authority of the health or social care professional. Professionals will also need to adapt their
approach when supporting and communicating as the dementia progresses, and this can
become increasingly problematic if a person’s mental capacity fluctuates. Many people living
with dementia and their carers or families would prefer the health or social care professional
to initiate the conversation about advance planning and discuss how to get access to advice.
Providing good quality, timely information for the person living with dementia can help
increase their involvement in key decisions and help them to have a share in the decision
making process. We should not assume that a person living with dementia will lack mental
capacity, and even if a person living with dementia does lack the mental capacity to make a
certain decision, this does not mean that they will lack it with regards to other decisions they
face. The Mental Capacity Act 2005 Code of Practice provides clear instruction as to how
people without the capacity can be supported to make a decision or when others have to
make decisions for the person, such as during a best interest meeting.
Each person’s family dynamic is unique. During discussions about advanced care planning
various family members, and especially the primary carer, can feel conflicted; attempting to
balance the needs of the person they are caring for, their own needs, and those of the rest of
the family. At times families can become divided as to the best way forward, which can lead
to disagreements, or there may be specific cultural needs which for each member of the
family carry their own set of values and considerations. With all this in mind, professionals
will need to ensure they are working to maintain the person living with dementia’s own social
network, as well as provide support that helps to maintain family cohesion.
Each organisation will have its own set of policies and protocols and there are a number of
legal requirements that professionals must adhere to. We must ensure that these regulations
are used to work for the person living with dementia, not against them. This will often require
working closely with other agencies in partnership to ensure the person living with dementia,
their families and carers receive high quality support.
128
6.1 Barriers and facilitators to involvement in decision making
for people living with dementia
Review questions
 What barriers and facilitators have an impact on involving people living with dementia in
decisions about their present and future care?
 What barriers and facilitators have an impact on how people living with dementia can
make use of advance planning?
6.1.1 Introduction
Table 26: Review summary: Barriers and facilitators to accessing care

Population  People (aged 40 years and over) living with dementia
 Carers of people (aged 40 years and over) living with dementia
Phenomena of interest  Equity of access (financial, physical or geographic restrictions)
 Behaviours and attitudes of professionals
 Communication
 Loss of autonomy
 Information needs
Outcomes  Experiences and satisfaction of people living with dementia
 Experiences and satisfaction of carers of people living with dementia
Qualitative studies and qualitative evidence syntheses were included if they explored the
barriers and facilitators to involving people living with dementia in decisions about their
present and future care, and making use of advance planning. Studies needed to contain
participants from the UK, report the views of either people living with dementia or their
carers, and match the criteria given in Table 26. For full details of the review protocol, see
Appendix C.
. Papers were excluded if they:

 did not report the views people living with dementia or their carers in the UK
 included only quantitative analysis of the collected information
 were not in English language
 were abstracts, conference proceedings and other unpublished studies.
A single search was conducted for all the qualitative questions included in this guideline,
which returned a total of 10,085 references. References were screened based on their titles
and abstracts, and the full texts of 61 references that were potentially relevant to these
review questions were requested. Ten qualitative studies exploring the barriers and
facilitators to involving people living with dementia in decisions about their care were
included in the review. The included studies are summarised in Table 27. The 51 excluded
papers, with reasons for exclusion, are presented in Appendix F.
A summary of the included primary studies for this review question is given in Table 27. In
addition, one systematic review of qualitative studies exploring barriers and facilitators for
carers making decisions for people with dementia (Lord 2011) was also included. For the full
129
evidence tables and full CERQual profiles please see Appendix E and Appendix G.
References for the included studies are given in appendix I.
Table 27: Summary of included primary studies

Study
details Study population Methods Outcome
Bisson People with symptoms of In-depth interviews Development of a care
(2009) Huntington’s disease, carers, pathway for advanced
asymptomatic people who decisions for people with
had the altered Huntington’s Huntington’s disease.
disease gene, clinicians, a
medical ethicist, two advisors
and a lawyer
Dening 6 people living with dementia Intervention: Asking The opinions of people
(2017) and 7 carers dyads questions living with dementia and
about their past, their carers
present and future
healthcare decision
making
Method of data
collection: semi-
structured interviews
Goodman People with dementia or Guided conversations An exploration of the
(2013) people considered to have preferences and priorities
dementia by the care home of care for people with
manager dementia.
Livingston Carers of people living with Focus groups and An identification of the
(2010) dementia individual interviews common difficult decisions
made by family carers on
behalf of people with
dementia, and facilitators
of and barriers to such
decisions.
Mackenzie Carers of people with Semi-structured An investigation of East
(2006) dementia interviews European and South Asian
family carers with regards
to how they negotiate the
stigma of dementia.
Murphy People with dementia and A cross-over trial A study of how useful
(2013) their carers involving narrative Talking Mats were for
interviews and a people with dementia and
questionnaire their carers with regards to
making decisions together.
Parveen 20 family carers of South Intervention: The opinions of familial
(2017) Asian people living with Information carers of South Asian
dementia. There were an Programme for South people living with dementia
additional 22 people who Asian families
were not carers but were (IPSAF)
family members. Method of data
collection: focus
group interviews
Poppe People with mild dementia, In-depth interviews Evaluation of the
(2013) carers and members of staff Advanced Care Planning
in Early Dementia tool.
Samsi People with dementia and Longitudinal interview A study of how people with
(2013) their carers study dementia and their carers
make decisions.
130
As this review question was qualitative in nature, it was not appropriate to conduct a search
for economic literature.
6.1.4.1 Barriers to decision making
The following barriers were identified to involving people living with dementia in decision
about their current and future care:
 Individuals denying they have problems – a barrier to advance planning on the part of the
people with dementia and carers was difficulty for some people with dementia or carers to
accept the diagnosis (high confidence)
 Individuals rejecting help - people will often reject help, either because they feel they do
not need it or because accepting help would involve psychologically acknowledging the
severity of their problems (high confidence)
 Individuals having a deference towards the authority of healthcare professionals - knowing
that they had dementia affected confidence in expressing opinions, self-esteem and
whether they thought their views were worth listening to (very low confidence)
 Individuals having a poor relationship with their formal or informal carers (very low
confidence)
 Healthcare professional not recognising the problems people have and their need for
support - healthcare professionals may not recognise people need additional assistance
to be involved in decision-making particularly when people are not open about difficulties
they are having (high confidence)
 Late diagnoses of dementia - if the diagnosis of dementia is delayed, this can make it
difficult for all the necessary advance discussions to be had before capacity issues start to
occur (high confidence)
 Lack of quality information given in a timely fashion, and available whenever best suits the
individual (high confidence)
 Confidentiality issues preventing carers having the information they feel they need to
support decision-making (high confidence)
 Staff sticking to protocols and policies, rather than having individualised discussions (high
confidence)
 Carers feeling conflict between the different roles they have to fulfil (high confidence)
 Friend carers felt less able to support decision-making than family carers (low confidence)
 Carer guilt about decisions made - feelings of anguish and guilt over decisions made.
Journey towards a decision was directed by a mixture of fatigue and a lack of obvious or
available alternatives. Feelings of guilt and failure were particularly strong for people
obliged to cope alone (high confidence)
 Conflict within families - when the person living with dementia was involved in decision-
making, they usually expressed reluctance to move to a care home. This often led the
carer either to delay the decision or exclude the person living with dementia from decision-
making (high confidence)
 Rigidity of healthcare system, and difficulty in changing decisions made - people felt that
once a decision was reached, it was then difficult to change this decision if circumstances
changed, and this led to a reluctance to make initial decisions (high confidence)
 An inability to plan due to unpredictability of condition and waiting lists for interventions –
people struggle with knowing when to seek care home placement due to dementia being
131
unpredictable and wait lists of institutions. Some patients find discussing the future difficult
without knowing what the future will bring (high confidence)
 Often there was one partner more dominant in decision-making. (low confidence)
 Fear of stigma prevented carers and people living with dementia from seeking help. (low
confidence)
 Becoming the main decision-maker for some carers was wearisome and felt like a burden
(medium confidence)
 Limited knowledge of the legal system to support decision-making when capacity was lost,
including advance care planning and Lasting Powers of Attorney (low confidence)
6.1.4.2 Facilitators for decision making
The following facilitators were identified to involving people living with dementia in decision
about their current and future care:
 Reconceptualisation and adjustment to altered circumstances, and presentation of
decision-making as trying to maximise independence - allowing services to develop slowly
(high confidence)
 Practical support and information provided by healthcare professionals - suggesting
interventions to facilitate agreement, or structured approaches to decision making.
Collaboration with staff helped carers with decision-making, and this was facilitated by a
trusted healthcare professional who consulted them and advocated effectively (high
confidence)
 Healthcare professionals initiating conversation about advance planning - people felt that
clinician’s raising these discussions helped them with decision-making (high confidence)
 Access to legal and financial advice (high confidence)
 Structured decision support and discussion tools - open-ended, structured tools may be
useful to guide discussions around advance planning. Staff who had not yet conducted
any advance care planning discussions themselves were unsure how to initiate the
discussion with those people with dementia who had not raised the issue themselves, but
saw the tool as a potential way of facilitating this (low confidence)
 Carers accompanying patients on visits to healthcare professionals (high confidence)
 Shared decision-making approaches - carers found it helpful to hear the perspectives of
other members of the family or professionals when making decision on behalf of the
person living with dementia – they felt it “gave permission” to make decisions (high
confidence)
 Family cohesion and support (high confidence)
 Social support networks - extended family, voluntary and community networks (high
confidence)
 Alternative communication strategies - discussing care was facilitated by using Talking
Mats. Talking Mats helped the participants living with dementia to be aware of what their
family members were doing for them, and were seen an enjoyable activity which improved
communication between the person living with dementia and his/her family (low
confidence)
6.1.4.3 Issues identified in Huntington’s disease
The following facilitators were identified in a study specifically in people with Huntington’s
disease
 Importance of information provision - easy-to-follow, consistent verbal and written
information was desired, which should be Huntington’s disease specific (low confidence)
132
 Importance of therapeutic relationship between individual and an expert in Huntington’s
disease - an established therapeutic relationship with an expert in Huntington’s disease.
Personal qualities such as being approachable, caring and sensitive with good
communication skills were felt to be important (low confidence)
 Early introduction to advance decisions - opinions of patients with Huntington’s disease
were different to professionals. Professionals were reluctant to approach service users too
early, particularly asymptomatic individuals with the altered Huntington’s disease gene, for
fear of causing distress (low confidence)
 Importance of advance decision forms (low confidence)
 Importance of discussions on power of attorney (low confidence)

Relative value of different The committee agreed that it was useful to have evidence of both
outcomes barriers and facilitators to decision making for people living with
dementia. However, evidence involving facilitators was of greater
value because it translated more easily to recommendations, whilst
where a barrier was identified without an accompanying facilitator, it
was not always easy to think of a practical solution to overcome that
barrier.
Trade-off between The committee noted that the evidence demonstrated that most
benefits and harms people living with dementia expressed a clear preference for being
offered information early, usually soon after diagnosis. However
where people living with dementia or their carers did not want to
receive the information soon after diagnosis the committee agreed
that the person’s wishes should be respected but it is important that
the information is offered on an ongoing basis or when requested by
the person. The committee also noted that people diagnosed with
dementia may need more than 1 appointment to be able to process
and understand the information. Therefore a recommendation was
made that people living with dementia be offered information and
opportunities for discussion on an ongoing basis.
The committee agreed with the finding in the evidence that some
people, following a diagnosis, may not initially want follow-up
appointments or referral to other services, but that these people may
change their minds later. Therefore, the committee agreed that
professionals should consider providing these people the opportunity
to be contacted at a specified future time, when they decide they do
want to access more information and support. If this is not available,
there is a risk that people may live without appropriate support for a
considerable period of time before things become sufficiently bad that
services once again become involved.
The committee agreed with the finding in the evidence that that how
some professionals currently interpret patient confidentiality guidance
may have the unintended consequence of reducing standards of
care. If appropriate information is not shared with informal carers this
can make it difficult for carers to provide the necessary support for
people living with dementia. Therefore, the committee recommended
that when dementia is diagnosed, the person’s consent should be
sought for information sharing with their carers and/or family
members. The reason that this conversation should happen at
diagnosis is because the person living with dementia is more likely at
that time to have capacity to decide who they would like their
133
information to be shared with. The committee noted that if consent for
information sharing has been obtained, this information should be
recorded in the person’s records. This should help to enable carers
and/or family members to provide a better standard of appropriate
care.
The committee agreed that the data protection training offered to staff
who work with people who have dementia should specifically cover
the benefits of sharing information where permission has been given.
Sharing information enables other health and social care
professionals to step in and work effectively with the person.
The committee agreed with the finding in the evidence that advocacy
and voluntary support services are important for people living with
dementia and the committee agreed that the advocacy and voluntary
support service recommendations from the previous guideline should
be retained (informing people about the services that are available).
However the evidence presented identified that people living with
dementia and their carers benefit from access to financial and legal
services and these services were added to the list of services to
inform people about.
The committee agreed it was important to discuss both advance
statements and advance directives with people living with dementia
for as long as they have the ability to be involved in decision making.
It was noted that some people living with dementia feel discouraged
from making advance decisions because they are concerned they
would not be able to change these decisions in the future. Therefore,
it was agreed that a recommendation be made that people living with
dementia are offered regular opportunities to make changes to their
advance statements and advance directives. To further support this,
the committee noted that people living with dementia should be
advised upfront that they will be able to change their advance
statements and advance directives in the future, to ensure that fears
that advanced statements and advanced directives cannot be
changed do not act as a barrier to advance planning.
The committee agreed that advance planning forms should be
standardised as far as possible, to maximise their level of
transferability. This is to ensure that the wishes of the person living
with dementia can be understood by everyone who needs to read
them. Whilst it was noted that it was not possible to ensure national
standardisation, it was agreed that at a local level this should be a
more achievable goal. The recommendation made by the committee
on this subject is reported in section 7.1.6 on coordinating care for
people living with dementia.
The committee agreed that training staff in managing difficult and
emotional conversations should enable them to have the confidence
to initiate and support discussions on advance planning. The
evidence presented to the committee suggests that many people
living with dementia and their carers would prefer staff to initiate such
conversations. However, the evidence also demonstrates that it is
common for staff to lack confidence in their ability to discuss advance
planning because staff feel that advance planning involves difficult
and emotional conversations. The committee noted this training
would also be important for people involved in the diagnosis of
dementia, as this was another time where emotionally challenging
conversations would take place.
The evidence suggests that people living with dementia can have a
lack of confidence in the value of their own opinions. This is a
significant barrier for people with dementia to make decisions about
their care, as healthcare professionals may not be able to recognise
the problems people have and their need for support. Therefore, the
committee agreed that it was important that carers and staff were
134
aware of possible low self-esteem or confidence in people living with
dementia so they can try to overcome this barrier. The committee
agreed that people living with dementia should be encouraged and
enabled to give their own views and opinions. The word ‘enable’ was
included in the recommendations because it is proactive.
The evidence presented further suggests that alternative
communications strategies (such as pictorial communication tools)
are a facilitator to involving people living with dementia in decision
making. Therefore, the committee recommended that their use
should be considered by carers and healthcare professionals.
Consideration of health The committee agreed that for the majority of the recommendations
benefits and resource made, any changes in resource use were likely to come from
use potentially needing more time to complete appointments. However,
because the recommendations provide advice on potential
approaches rather than prescribe specific actions, the committee
agreed they were not likely to lead to a significant increase in
resource use.
The only recommendations the committee agreed would have a
specific cost attached to them were the 2 recommendations made
around staff training. However, the committee were confident that the
costs are justified by the long-term benefits of better-trained staff.
Quality of evidence The committee agreed that overall the evidence was of good quality
and the issues identified agreed with their practical experience. It
was, however, noted that the reliance on qualitative evidence meant
it was not possible to offer prescriptive recommendations in many
areas, but rather more general guidance.
For the purposes of developing dementia guidelines, the study in
Huntington’s disease was agreed to be of lower value. This is
because some people in the study had the Huntington’s disease
gene and had not yet developed symptoms at the time the research
was conducted. Nevertheless, it was agreed this study was still
valuable because it provided a different perspective; that of a
younger population.
Other considerations The committee noted that there are many subgroups of people living
with dementia who may have very different information needs (e.g.
younger people, those with comorbidities, people with rarer dementia
subtypes). They agreed it was important that the information provided
be tailored to these different circumstances, rather than only general
information about dementia being provided.
The committee also agreed it was appropriate to cross-refer in this
section to both the NHS Accessible Information Standard (which it is
a requirement for people to follow) and the NICE guidelines on
patient experience in adult NHS services and people’s experience in
adult social care services, which provides more advice on providing
appropriate information, and making that information accessible.
Sections of recommendations referring to people in or looking for
work were also informed by the evidence review on the specific
needs of younger people living with dementia (section 17).
The committee noted that there was considerable interest from many
people living with dementia in being involved in research, but people
were often unaware of opportunities to do so. The committee
therefore agreed it was appropriate to make a recommendation that
people living with dementia be informed about opportunities to
participate in research.
Information provision
135
33. Provide people living with dementia and their family members or carers (as
appropriate) with information that is relevant to their circumstances and the stage
of their condition.
34. Be aware of the obligation to provide accessible information as detailed in the

NHS Accessible Information Standard. For more guidance on providing
information and discussing people’s preferences with them, see the NICE
guideline on patient experience in adult NHS services and people’s experience in
adult social care services.
35. At diagnosis, offer the person and their family members or carers (as appropriate)
oral and written information that explains:
 what their dementia subtype is and the changes to expect as the
condition progresses
 which healthcare professionals and social care teams will be involved in
their care and how to contact them
 if appropriate, how dementia affects driving, and that they need to tell
the Driver and Vehicle Licensing Agency (DVLA) and their car insurer
about their dementia diagnosis
 their legal rights and responsibilities
 their right to reasonable adjustments (in line with the Equality Act 2010)
if they are working or looking for work
 how the following groups can help and how to contact them:
 local support groups, online forums and national charities
 financial and legal advice services
 advocacy services.
36. If it has not been documented earlier, ask the person at diagnosis:
 for their consent for services to share information
 which people they would like services to share information with (for
example family members or carers)
 what information they would like services to share.
Document these decisions in the person’s records.
37. After diagnosis, direct people and their family members or carers (as appropriate)
to relevant services for information and support (see recommendations 47 and 48
on care coordination).
38. For people who do not want follow-up appointments and who are not using other
services, ask if they would like to be contacted again at a specified future date.
39. Ensure that people living with dementia and their carers know how to get more
information and who from if their needs change.
40. Tell people living with dementia (at all stages of the condition) about research
studies they could participate in.
Advance care planning
136
41. Offer early and ongoing opportunities for people living with dementia and people
involved in their care (see recommendation 36) to discuss:
 the benefits of planning ahead
 lasting power of attorney (for health and welfare decisions and property
and financial affairs decisions)
 an advance statement about their wishes, preferences, beliefs and
values regarding their future care
 advance decisions to refuse treatment
 their preferences for place of care and place of death.
Explain that they will be given chances to review and change any advance
42. At each care review, offer people the chance to review and change any advance
Involving people in decision-making
43. Encourage and enable people living with dementia to give their own views and
opinions about their care.
44. If needed, use additional or modified ways of communicating (for example visual
aids or simplified text).
Staff training
45. Ensure that all health and social care staff are aware of:
 The extent of their responsibility to protect confidentiality under data
protection legislation and
 any rights that family members, carers and others have to information
about the person’s care (see recommendation 48 on information sharing
between different care settings).
46. Health and social care professionals advising people living with dementia
(including professionals involved in diagnosis) should be trained in starting and
holding difficult and emotionally challenging conversations.
137
7 Care planning, review and co-ordination
In order to deliver outcome focused person centred care it is essential that a comprehensive
care and support plan is in place which is co-ordinated and reviewed regularly. For this
reason, it should be easily identified and accessible.
Person centred care is a broad term and involves a wide range of elements. It can range
from life story work with people living with dementia to an organisational framework that
values all people within an organisation, both those providing care and those being cared for.
Brooker and Latham (2016) in Person Centred Dementia Care discuss these elements within
their influential book. They state that person centred care should be a constant thread
throughout the whole cycle of dementia services from memory clinics and initial diagnosis to
end of life care for people living with dementia.
It is important that there is a person who is responsible for co-ordinating all aspects of health
and social care. Whilst this person isn’t necessarily responsible for the delivery of all aspects
of the care and support plan, someone needs an overview to ensure that services and care
are delivered in a co-ordinated and timely manner, without duplication. This ensures that the
person living with dementia or the person’s family have the security of knowing who to
contact, and stops people having to constantly repeat their history and care needs.
The time after diagnosis can be a very difficult and emotional time for people newly
diagnosed with dementia and responses vary across the spectrum from denial to a desire to
know and understand everything about dementia. Therefore, people working in services
need to understand that people respond to diagnosis in varied ways and post diagnostic
support must be person centred to each individual as opposed to a standard pathway. This
means accessing post-diagnosis support in a timely way, which fits the individual needs of
the person with dementia and their carer.
Post-diagnosis reviews should also be undertaken in ways and at intervals which reflect the
individual needs of the patient. To ensure the maximisation of benefits for the person with
dementia and their carer, the review should be meaningful, holistic in approach with actions
and outcomes being clearly recorded in the care and support plan.
138
7.1 Health and social care co-ordination
Review questions
 What are the most effective methods of care planning, focussing upon improving
outcomes for people with dementia and their carers?
 How should health and social care be co-ordinated for people living with dementia?
7.1.1 Introduction
These questions considered randomised controlled trials and qualitative evidence on

effective models of care planning and co-ordination for improving the care and experiences
of people living with dementia and their carers. Full details of the review protocol are
available in appendix C.
Table 28: Review summary: qualitative evidence

Phenomena of interest  Methods and models of care planning for people living with
dementia
 Models of health and social care co-ordination, which may include
features such as:
 Configuration and integration of services
 Timing and delivery of services (e.g. transfers, referral pathways)
 Staff communication
 Location of services
Table 29: Review summary: quantitative evidence

Interventions  Methods and models of care planning for people living with
dementia
Comparator  Each other
 Standard care
Outcomes  Clinical outcomes including cognitive, functional and behavioural
ability
 Access to health and social care support
 Patient and carer wellbeing, experience and satisfaction
 Patient and carer health-related quality of life
Randomised controlled trials, qualitative studies and systematic reviews of randomised

controlled trials or qualitative studies were included if they explored how health and social
care should be co-ordinated or explored the most effective methods of care planning and
focussed upon improving outcomes for people with dementia and their carers. Studies
needed to report the views of either people living with dementia or their carers, and match
the criteria given in either Table 28 or Table 29. Papers were excluded if they:
139
7.1.2.1 Qualitative evidence
A single search was conducted for all the qualitative questions included in this guideline, which returned a total of 10,085 references. References
were screened based on their titles and abstracts, and the full texts of 90 references that were potentially relevant to these review questions were
requested. Eighteen qualitative studies exploring care coordination were included in the review. The included studies are summarised in Table 30.
For the full evidence tables and full CERQual profiles please see Appendix E and Appendix G. The 73 excluded papers, with reasons for
exclusion, are presented in Appendix F. References for the included studies are given in appendix I.
7.1.2.1.1 Description of included studies

Study details Study population Methods Outcomes
Brooker (2017) 9 people living with dementia Intervention: post-diagnostic psychosocial The opinions of people living with dementia
and 6 carers interventions and their carers
Method of data collection: focus groups.
Bunn (2017) 28 people living with dementia Intervention: management of comorbidities The opinions of people living with dementia
alongside dementia and their carers
Method of data collection: semi-structured
interviews and focus groups.
Dayson (2016) 9 carers of people living with Intervention: resilience service The opinions of carers of people living with
dementia Method of data collection: interviews dementia
Faith 2015 6 people living with mild Intervention: self-management course for The opinions of people living with dementia
dementia people living with dementia.
Method of data collection: Interviews
Gethin-Jones 2014 20 familial carers of people living Intervention: outcome-focused care for The opinions familial carers of people living
with dementia who were living people with dementia who are living alone. with dementia who are living alone
alone Method of data collection: Semi-structured
interviews
Gibson 2007 10 people living with dementia Interventions: community-based vs clinic- The opinions of people living with dementia
and their carers based memory service and their carers
140
Method of data collection: in-depth
qualitative interviews
Gladman 2007 15 carers of people living with Intervention: Daisy Chain: a commercial The opinions of carers of people living with
dementia person-centred dementia service that seems dementia
to have some elements of case management
Methods of data collection: semi-structured
interviews, observation and focus groups
Górska 2013 12 people living with dementia Intervention: a dementia service in Scotland. The opinions of people living with dementia
and 19 carers Method of data collection: semi-structured and their carers
interviews
Hean 2011 An unspecified number of Intervention: an integrated memory The opinions of people living with dementia
people living with dementia and assessment and support service. and their carers
their carers. Method of data collection: interviews
Iliffe 2014 6 people living with dementia Intervention: case management The opinions of people living with dementia
and 10 carers Method of data collection: interviews and their carers
Innes 2014 6 people living with dementia Intervention: post-diagnostic support in a Participant experiences of post-diagnostic
and 12 family carers remote and rural region in Scotland. support
Method of data collection: Semi-structured
interviews
Kelly 2016 8 people living with dementia Intervention: post-diagnostic support. Participant experiences of post-diagnostic
and 8 family carers Method of data collection: Semi-structured support
interviews
Moore 2011 An unspecified number of carers Intervention: self-directed support. The opinions of carers of people living with
of people living with dementia Method of data collection: Interviews dementia
Popham 2012 25 people living with dementia in Intervention: residential care home, which The opinions of people living with dementia
residential care homes and 11 includes aspects of case and residential care home managers with
carers coordination/management regards to what extent the care home
Method of data collection: Interviews using environment met the requirements of
open questions. residents. Sheffield Care Environment
Assessment Matrix (SCEAM)
Rothera 2008 27 people living with dementia Intervention: a specialist multiagency home The opinions of carers of people living with
and 18 family carers support service. dementia
141
Method of data collection: semi-structured
interviews
Sonola 2013 9 people living with dementia Intervention: case coordination service The opinions of carers of people living with
and 9 carers Method of data collection: interviews dementia
Toms 2015 13 people living with early stage Intervention: self-management The opinions of people living with dementia
dementia and 11 carers Method of data collection: semi-structured and their carers
interviews
Willis 2011 16 people living with dementia Intervention: memory service that was also a The opinions of people living with dementia
and 15 carers one-stop shop for case and carers
coordination/management.
Method of data collection: interviews
7.1.2.2 Quantitative evidence
A single search was conducted that returned a total of 5,735 references. These references were screened for RCTs that evaluated case
management/planning/coordination interventions. A key feature of the references identified was that they involved a single specified individual
being responsible for managing/planning/coordinating care. References were screened based on their titles and abstracts, and the full texts of 79
references that were potentially relevant to the review question were requested. Of these, 53 were excluded on full text review, with reasons for
exclusion presented in Appendix F. Twenty six studies were analysed individually and in the following groups: by country, the profession of the
individual case managing/planning/coordinating, frequency of follow-up, and the contact method when following-up. For the full evidence tables
and full GRADE profiles please see Appendix E and Appendix G. References for the included studies are given in appendix I.
Table 31: Summary of included studies – RCT evidence

Bass (2015) 114 people living with dementia Intervention: care-coordination program Cognitive impairment, behavioural symptoms,
and their carers. Control: usual care. hospital admissions
Bass (2014) 194 people living with dementia Intervention: care-coordination program Unmet need, embarrassment about memory
and their carers. Control: usual care. problems, isolation, relationship strain,
depression.
142
Bass (2013) 718 people living with dementia Intervention: care-coordination program Unmet needs, carer strains, depression, support
and their carers. Control: usual care. resources.
Bass (2003) 157 family carers of people Intervention: integrating Alzheimer’s Health service usage, carer satisfaction, carer
living with dementia. Association care consultation service with depression and strain.
health care services offered by a large
managed care system.
Control: usual care.
Callahan (2006) 114 people living with dementia Intervention: Interdisciplinary team led by an Patient Neuropsychiatric inventory, Cornell Scale
and their carers. advanced practice nurse working with the for Depression, Telephone interview for
patient’s family carer and integrated within cognition, ADLs, carer Neuropsychiatric
primary care. Inventory, carer patient health questionnaire.
Chien (2008) 88 people living with dementia Intervention: Case manager who together with Family Caregiving Burden Inventory, World
and their carers. another nurse, prioritised problem areas and Health Organization Quality of
formulated a multidisciplinary education Life Scale, 6-item Social Support Questionnaire,
program for each family on effective dementia MMSE, 12-item Neuro-psychiatric Inventory,
care. Institutionalization over the past 6 months, Family
Control: usual care. Support Services Index.
Chien (2011) 92 people living with dementia Intervention: Case manager who together with Family Caregiving Burden Inventory, World
and their carers. another nurse, prioritized the problems and Health Organisation Quality of Life Scale, Six
formulated an individualized education and item
support programme for each family. Social Support Questionnaire, MMSE,
Control: usual care. Neuropsychiatric Inventory, Family Support
Services Index.
Chodosh (2015) 43 people living with dementia Intervention: Care management delivered in ZARIT Caregiver Burden: 22-question scale,
and their carers. person and by telephone. Revised Memory and Behaviour Problem
Control: Care management delivered by Checklist, carer depression, Carer Quality of Life,
telephone only. Advance directive discussed or completed and
documented, Carer involved in care plan
development, received services or information,
Safe return program (for wandering) discussed or
recommended, aware of identification items,
enrolled in safety programs, receipt of caregiving
assistance or respite/support services
143
Chu (2000) 75 people living with dementia Intervention: Comprehensive Geriatric Depression Scale-Short Form,
and their carers. home care program functional performance, The Burden Interview,
Control: usual care. Memory and Behaviour Checklist, level of
depressive symptoms of the carers.
Dias (2008) 59 people living with dementia Intervention: a trained Community Team that Carer mental health, Zarit Burden Score, distress
and their carers. focused on supporting the carer through due to behavioural disturbances, behavioural
information on dementia, guidance on problems in the subject and activities of daily
behaviour management, a single psychiatric living.
assessment and psychotropic medication if
needed.
Eloniemi-Sulkava 100 people living with dementia Intervention: systematic, comprehensive Time to institutionalization (period in community
(2001) and their carers. support by a dementia family care coordinator care) from enrolment of patients in the study to
Control: usual care. their placement in long-term institutional care.
Eloniemi-Sulkava 125 people living with dementia Intervention: a multicomponent intervention Time from enrolment to institutionalization of
(2009) and their carers. program with a family care coordinator, a spouses with dementia and use of services and
geriatrician, support groups for carers, and service expenditure of couples.
individualized services.
Fortinsky (2009) 84 people living with dementia Intervention: standardised assessment tool Nursing home admission rate, self-efficacy for
and their carers. and a care consultant who had monthly symptom management, self-efficacy for
contact with each family carer. A monthly care accessing support services, depressive
plan was organised. symptoms, carer burden, and physical
Control: educational and community resource symptoms.
information but no care consultation.
Jansen (2011) 99 people living with dementia Intervention: case management by district Carer’s sense of competence, carer’s quality of
and their carers. nurses. life, physical component summary, carer’s
Control: usual care. depressive symptoms, carer’s burden, care
recipient’s Quality of life, care received.
Kwak (2011) 94 people living with dementia Intervention: Tailored Caregiver Assessment Service recommendation, compliance, and use.
and their carers. and Referral (TCARE®) protocol, a care Carer identity discrepancy, carer burden,
management process designed to help family depressive symptoms.
carers, on care planning and carer outcomes.
144
Lam (2010) 92 people living with mild Intervention: case management by MMSE, Cornell Scale for Depression in Dementia
dementia and their carers. occupational therapist. score, Neuropsychiatric Inventory score,
Control: usual care. Personal Wellbeing Index for Intellectually
Disabled, Zarit Burden Interview score, Personal
Wellbeing Index for Adult, General Health
Questionnaire score.
Meeuwsen (2012) 153 people living with dementia Intervention: Coordination of care by memory Quality of life, caregiving burden, geriatric
and their carers. clinic. depression scale, neuropsychiatric inventory,
Control: Coordination of care by GP. carer’s personality, carer depression and anxiety,
carer’s mastery, carer’s neuropsychiatric
inventory, social support.
Miller (1999) 8095 people living with dementia Intervention: case management and for an Nursing home entry rates.
and their carers. 80% discount on community care benefits, up
to about $600 per month.
Newcomer (1999) 1906 people living with dementia Intervention: case management and for up to Carer burden, carer depression.
and their carers. $699 per month in community care benefits.
Samus (2014) 188 people living with dementia Intervention: 18-month care coordination Time to transfer from home and total percent of
and their carers. intervention to systematically identify and unmet care needs.
address dementia-related care needs through
individualized care planning; referral and
linkage to services; provision of dementia
education and skill-building strategies; and
care monitoring by an interdisciplinary team.
Schoenmakers 46 people living with dementia Intervention: a care counsellor, coordinating Depression in the family carer, coping behaviour
(2010) and their carers. care in quasi-unstructured way. of the carer, anxiety of the carer, carer burden,
Control: usual care. activities of daily living, neuropsychiatric
symptoms.
Shelton (2001) 412 people living with dementia Intervention: registered nurses as case Likelihood of hospitalization, ADL limitations.
and their carers. managers and operating within a multispecialty
physician group practice and a vertically
integrated healthcare system
145
Tanner (2015) 171 people living with dementia Intervention: MIND at Home, a community- Carer unmet needs, carer burden, depression,
and their carers. based, multicomponent, care coordination QOL.
intervention.
Van Mierlo (2015) 49 informal carers of people with Intervention: The DEMentia Digital Interactive Camberwell Assessment of Needs for the
dementia, supported by 19 Social Chart (DEM-DISC) is an ICT tool to Elderly.
randomised case support customised disease management in
Managers participated in the dementia.
study Control: usual care.
Vickrey (2006) 290 people living with dementia Intervention: Disease management program Connection of patient-carer dyad with community
and their carers. led by care managers agencies, receipt of caregiving assistance or
Control: usual care. respite/support services, patient health-related
quality of life, quality of patient’s health care,
carer confidence in caregiving, carer health-
related quality of life, social support, received as
much help as needed with behaviour problem.
Xiao (2016) 61 family carers of people living Intervention: personalised carer support Sense of competence, severity of BPSD, carer
with dementia. All were from 10 Control: usual care. distress, usage of respite care, usage of carer
minority groups. support group, usage of dementia helpline,
satisfaction with service providers, usage of
community aged care, community aged care at
home.
146
total of 5,127 citations was returned. Details of the literature search are provided in Appendix
D. Following review of titles and abstracts, no full-text studies were retrieved for detailed
consideration. Therefore, no relevant cost–utility analyses were identified. However, 1 cost–
utility analysis that was relevant to this review question was identified in a literature search
for another review question. Therefore 1 study was included in this review.
7.1.3.1 Case management
Vroomen et al. (2016) conducted a trial-based cost-utility analysis alongside a Dutch 2-year
prospective, observational cohort study with 521 informal carers and community-dwelling
persons with dementia. The study compared case management provided within 1 care
organisation (intensive case management model [n=234], ICMM), case management where
care was provided by different care organizations within 1 region (linkage model [n=214],
LM), and a group with no access to case management (control) (n=73). For further details,
please see the economic evidence profile in Appendix M.
Cost diaries were used to estimate costs from a societal perspective; where possible, costs
inconsistent with the NICE reference case were excluded from the results presented here.
Table 32 shows the costs that we have included in our analysis. Costs were adjusted to price
year 2010 and expressed in Euros (€).
Health-related quality of life was measured using the EQ-5D-3L (carer-rated for the person
living with dementia).
Table 32: Costs from Vroomen et al. (2016) with societal costs removed
ICMM LM Control
Person living with dementia: QALYs 1.25 1.18 1.27
Carer+person living with dementia QALYs 2.9 2.9 3.0
General practice € 1,279 € 1,362 € 1,088
Hospital and outpatient clinics € 2,642 € 3,336 € 4,835
Overnight care € 313 € 227 € 318
Day centre € 6,135 € 7,190 € 10,506
Long term institutionalization € 6,017 € 5,688 € 11,227
Welfare services € 3,043 € 4,050 € 20,784
Medications € 2,220 € 1,867 € 1,766
Case management costs € 3,120 € 2,469 -
Home care
Home-making services Excluded
Informal care costs
Total costs € 24,769 € 26,189 € 50,524
The mean incremental costs and effects are presented in Table 33.
147
Table 33: Costs and effects calculated for Vroomen et al. (2016)
Strategy Cost Effect Cost Effect a ICER
ICMM €24,769 2.9 QALYs
LM €26,189 2.9 QALYs €1,420 -0.02 QALYs Dominated
Control €50,524 3.0 QALYs €25,755 -0.0004 QALYs Dominated
a difference adjusted for baseline utility
Assuming QALYs are valued at £20–30,000, the QALY losses observed with both ICMM and
LM are small enough to justify the large savings associated with these forms of case
management.
The authors report probabilistic sensitivity analyses suggesting a 0.992 probability that ICMM
has an ICER of €30,000 or better compared with control, and a 0.977 probability that ICMM
has an ICER of €30,000 or better compared with LM. However, these analyses include costs
outside the NICE reference case that cannot be disaggregated.
The authors concluded that their study provides preliminary evidence that the ICMM is cost
effective compared with the control group and LM.
7.1.4.1.1 Self-management intervention
The following themes were identified for the self-management intervention for people living
with dementia and their carers:
 Although people living with dementia said that they could not recall all of the activities,
they had enjoyed the training program (low confidence)
 The participants felt empowered: training programs encouraged people living with
dementia to continue with their hobbies and goals. Access to a budget provided a sense
of empowerment (moderate confidence)
 Peer support, such as provided by support groups, was considered valuable by
participants (low confidence)
 Additional support, such as a support group, was available, but these were often time-
limited, which led both carers and people with dementia to the question of what happened
when such support ended (low confidence)
 Respondents thought that professional support was important for effective self-
management, and valued this resource. They thought that this help was necessary
because not everything could be self-managed within the family (low confidence)
 Many respondents were unsure how to access the services that were available, and
reported finding them limited and poorly integrated. This made it harder to self-manage
the condition (low confidence)
 The approach of normalising difficulties was evident in many interviews (low confidence)
 A sense of stoicism, often expressed when respondents gave their ideas about self-
management, was evident in many interviews, and this seemed to be a form of
psychological management (low confidence)
148
7.1.4.1.2 Outcome-focused/needs-led care vs standard care
The following themes identified for outcome-focused/needs-led care vs standard care for
people living with dementia and their carers:
 Standard care: The most common concern of familial carers is the feeling of not being
able to cope (non-comparative questioning) (moderate confidence)
 Standard care: The sense of isolation expressed by the participants came over very
strongly. This isolation appeared to come from their sense that they were on the outside
with little control because the care was planned by the other professionals. Family carers
felt that they were isolated as they had all the responsibility and in their eyes and
potentially all the blame when things went wrong (moderate confidence)
 Outcome-focussed care: There was an improvement in the carers’ self-reported subjective
well-being, after the outcome-focused homecare intervention had been implemented (high
confidence)
 Outcome-focussed care: All the carers felt the subjective well-being of their relative had
improved after the six month outcome-focused care intervention (moderate confidence)
7.1.4.1.3 Community-based case management
The following themes were identified for community-based case management for people
living with dementia and their carers:
 Meeting health and social care professionals at home was more relaxing and less
stressful compared with using the memory service (moderate confidence)
 Being at home facilitated communication with health and social care professionals
(moderate confidence)
 The case manager was good at identifying needs and providing the right support
 Carers expected case managers to provide information about dementia and services
 Case managers should be proactive in asking carers and people living with dementia if
they feel they need assistance. This is because participants frequently expressed a
reluctance to initiate contact with the case manager, which undermines the concept that
they could ask for help when needed (moderate confidence)
 A common reason why people living with dementia and their carers do not initiate contact
with case managers is because they associate case managers with assisting with ‘major’
problems such as arranging residential care homes. They do not associate case
managers with assisting with day-to-day issues (moderate confidence)
 People living with dementia and their carers preferred to have their case manager based
at their GP’s surgery. This is because there was the perception that their GP’s surgery
would then be a ‘one-stop shop’. In addition, having the case manager at the GP’s surgery
provided an additional opportunity to talk to the case manager while visiting the GP’s
surgery (moderate confidence)
 For some, exposure to others at more severe stages of the illness within the clinic was a
potent contributor towards anxiety, illustrating what could be expected as the disease
progresses. Appointments at home removed this exposure (moderate confidence)
 Case management made access to services easier including GPs, benefit checks and
links to other services (moderate confidence)
 A key aspect of case management valued by people living with dementia and their carers
was the idea of background support that could easily be called on at a time of need
149
 For people living with dementia and their carers to feel comfortable about contacting the
case manager in the event of difficulties, there needed to be time and opportunities to
develop a deeper relationship (moderate confidence)
 Face-to-face and telephone contact were both considered acceptable, although face-to-
face contact was often preferred as it facilitated relationship building better than telephone
contact (moderate confidence)
7.1.4.1.4 Memory-clinic case management
The following themes were identified for memory clinic case management for people living
with dementia and their carers:
 For memory services that do not have post-diagnostic support, many participants
expressed feelings of abandonment or ‘being sent away’ by professionals on receipt of
diagnosis (moderate confidence)
 For memory services that do have post-diagnostic support, people with dementia and their
carers explained the value of having support as soon after diagnosis as possible and the
importance of skilled, knowledgeable, sensitive project workers to deliver support
 Carers frequently reported positively on the help received from the project workers with
claiming benefits (moderate confidence)
 Carers spoke of receiving support with arranging Power of Attorney and valued the input
from project workers in negotiating the process (moderate confidence)
 The service and nature of the staff made carers and people living with dementia feel
supported and reassured. (Having a named person to contact in times of crisis, and the
security that they would not left to manage alone.) (high confidence)
 People living with dementia felt pressure of time because the psychiatrist was busy (very
low confidence)
 There were accounts of receiving no information, or insufficient or inappropriate
information following diagnosis (moderate confidence)
 Some carers expressed discomfort with some of the information they received. Some felt
that it was too much to face too soon. Many participants stated that a ‘one size fits all’
approach was not what they wanted (moderate confidence)
 Participants valued that information was delivered by the project workers on a one-to-one
basis and specifically targeted to individual needs and wishes (moderate confidence)
 People living with dementia and their carers liked seeing the same person throughout
treatment (high confidence)
 People living with dementia and their carers recognised the one stop shop aspect of the
memory service. Ten participants described the memory service as a central point of
access to all necessary services (low confidence)
 People living with dementia and their carers valued transport that was arranged by case
managers/project workers (high confidence)
 Memory service post-diagnostic support when individualised and one-to-one, causes
people with dementia to re-engage socially or with old hobbies (moderate confidence)
7.1.4.1.5 Daisy Chain: a commercial person-centred dementia service that seems to have some
elements of case management
The following themes were identified for Daisy Chain:

 The person-centred community-based dementia service was well received, provides a
personalised service and helped carers to cope (low confidence)
150
 There are sometimes differences of opinion between people living with dementia, paid
carers and familial carers (low confidence)
7.1.4.1.6 Non-specified case management style(s) in predominantly remote and rural areas in
Scotland
The following themes were identified for non-specified case management style(s) in
predominantly remote and rural areas in Scotland:
 The lack of alternative options sometimes led to provision of no support at all (very low
confidence)
 Poor coordination of services. The participants particularly emphasized poor
communication between existing services, which results in unsatisfactory case
management and delays in service provision. The need for a single point of access to
information and service coordination was expressed as a means to manage these
challenges and to facilitate more efficient and effective service delivery. Participant reports
also highlighted inconsistencies in care provision and suggested the need for well-defined
care pathways (high confidence)
 Some experienced lack of continuity of care. This can lead to poor communication and is
confusing (high confidence)
 There were high satisfaction levels with the support received from the Community Mental
Health Team (moderate confidence)
 Participants discussed the importance of staff building a rapport with the person living with
dementia. This facilitates communication (very low confidence)
 When it was available, a carers’ group (carer support) was appreciated (very low
confidence)
 Practical support was important to most carers who received help from private or
voluntary services regularly. Carers perceived this type of support as an opportunity to
take a respite from caregiving responsibilities. Many used the respite time to rest, run
errands which required getting out, or to attend carers meetings (very low confidence)
 Information was not always in a format appropriate for the person living with dementia or
their carer (moderate confidence)
 The way information was delivered was important. Participants preferred a direct
approach with the opportunity to ask questions (moderate confidence)
 Care managers should be proactive in anticipating the needs of people living with
dementia and their carers and provide relevant information (very low confidence)
7.1.4.1.7 Case management in residential care homes
The following theme was identified for case management in residential care homes:
 Participants spoke about having the freedom to be able to carry out normal everyday
activities and domestic chores (moderate confidence)
7.1.4.2.1 Care coordination/management using a protocol/action plan
Moderate-quality evidence from up to 8 RCTs containing 2,474 people living with dementia
found improvements in quality of life and rates of entry in to long stay care for the person
living with dementia and carer burden for people offered case management versus usual
care, but could not differentiate cognition, depressive symptoms or behavioural and
psychological symptoms of dementia for the person living with dementia, or carer depressive
symptoms or quality of life. Across the studies, larger gains were seen in studies with more
frequent follow-up, studies where the case manager was a nurse, studies where contact was
151
made face-to-face in the person’s home, and studies conducted in Hong Kong. However,
these differences were difficult to interpret as there was considerable correlation between
these features across the studies.
7.1.4.2.2 Care coordination/management using a specific structured protocol vs care

coordination/management that is unstructured
Moderate- to low-quality evidence from 1 RCT containing 73 people living with dementia
could not differentiate outcomes for carers between people offered care
coordination/management using a specific structured protocol vs care
coordination/management that is unstructured.
7.1.4.2.3 Care coordination by telephone vs care coordination in-person
Care coordination by telephone ('experimental') vs care coordination in-person

('control'). Follow-up frequency was monthly for the first 3 months and quarterly
thereafter
Moderate-quality evidence from 1 RCT containing 43 people living with dementia found that
care coordination in-person improved carer quality of life (benefits of caregiving) compared
with follow-up by telephone. However, this study could not differentiate quality of life of the
person living with dementia, the total number of problems, carer depressive symptoms, carer
quality of life (spirituality and faith) or carer burden.
7.1.4.2.4 Follow-up organised by memory clinic vs GP
Follow-up organised by memory clinic vs GP
High-quality evidence from 1 RCT containing 153 people living with dementia found that
follow-up by GP was associated with less carer depressive symptoms and anxiety compared
with follow-up by memory clinic. However, this study could not differentiate outcomes relating
to the person living with dementia, carer quality of life, carer social involvement, carer
neuropsychiatric index, carer personality or carer mastery.
7.1.4.2.5 Medicare Alzheimer's disease demonstration
The Medicare Alzheimer's disease demonstration (care coordination/management with

unspecified follow-up frequency) vs usual care
Moderate-quality evidence from 1 RCT containing 412 people living with dementia found that
care coordination/management with unspecified follow-up frequency reduces the likelihood of
any carer hospitalisation compared with usual care. However, this study could not
differentiate emergency department visits.
Low-quality evidence from 1 RCT containing 7,803 people living with dementia could not
differentiate rates of entry into residential care between people offered care
coordination/management with an unspecified follow-up frequency
Low-quality evidence from 1 RCT containing 1,906 people living with dementia could not
differentiate carer outcomes between people offered care coordination/management with an
unspecified follow-up frequency.
152
7.1.4.2.6 DEM-DISC
Care coordination/management using DEM-DISC vs care coordination/management

without DEM-DISC
Very low- to low-quality evidence from 1 RCT containing 49 people living with dementia could
not differentiate outcomes for people living with dementia between people offered care
coordination/management using DEM-DISC compared with care coordination/management
without DEM-DISC.
7.1.4.2.7 Personalised carer support for minority groups
Personalised carer support for minority groups vs usual care for minority groups
Moderate- to low-quality evidence from 1 RCT containing 61 people living with dementia
found that personalised carer support for minority groups compared with usual care for
minority groups improves carer sense of competence, carer quality of life (mental), care
recipient’s behavioural problems, carer distress, and carer’s satisfaction with service
providers. However, personalised carer support increased usage of respite care. This study
could not differentiate carer quality of life (physical) nor usage of community aged care.
7.1.4.3 Health economics - Two forms of case management
One partially applicable cost–utility analysis with very serious limitations conducted alongside
a cohort study explored the cost-effectiveness of 2 forms of case management. Both forms of
case management resulted in a very small loss of QALYs compared with control, but resulted
in significant cost savings. An intensive case management model was found to dominate a
linkage model of case management.

Relative value of different The aim of this review question was to identify the most effective
outcomes methods of care planning, focussing upon improving outcomes for
people with dementia and their carers. Therefore, the committee
agreed that studies with interventions and outcomes benefitting
people living with dementia and their carers would be more relevant
compared with studies that just looked at carers alone.
Trade-off between The importance of linking care coordination recommendations
benefits and harms to the diagnosis recommendations
The committee noted that occasionally people living with dementia
are diagnosed as having dementia but are then ‘forgotten’ by the
system. The harm of this is that by the time a care and support plan
is made, that individual’s control over planning their future care might
be lost forever. Therefore, the committee agreed that the care
coordination recommendations will be linked into the ‘at diagnosis’
recommendations because care management and formulating a care
and support plan should start from diagnosis. However it was noted
that no matter what stage of dementia a person has, they should
always have a care and support plan.
The importance of having recommendations that will lead to
better coordination of care
The committee agreed that studies that had good results (particularly
Chien 2008, Chien 2011) had components of care that already exist
throughout the NHS. However, these studies had relatively better
coordination compared with most NHS current practice. The
committee agreed that current NHS practice is more fragmented,
potentially because funding comes from different places, and that
153
better coordination would be likely to lead to better outcomes for
people living with dementia and their carers.
A single named individual responsible for coordinating care
The committee agreed that the positive overall findings from the
studies on case management provided robust evidence that there
should be a single person responsible for coordinating care.
Otherwise, it is common for health and social care professionals to
assume that other members within the team are coordinating care
when they are not. The majority of the studies that involved care
coordination had one person as the single named individual
responsible for coordinating care, and these studies showed an
improvement in the quality of life of the person living with dementia,
and reductions in both burden and depression for the carer.
The initial assessment of needs should be face to face whenever
possible
The committee agreed that the initial assessment of needs should be
face to face whenever possible. The studies that have particularly
good results (Chien 2008, Chien 2011) involve face to face
assessments of needs, and the committee agreed this was in line
with their experience. They also agreed that it was important for the
recommendations to be pro-active in identifying potential problems
rather than reactive. In addition, it was noted that it was not always
possible to identify in advance which individuals would benefit from a
face to face assessment.
The committee agreed that the person conducting a face-to-face
assessment does not have to be the person coordinating care. For
example, in some situations the district nurse coordinates the care
but the general practitioner provides much of the care.
Therefore, the committee agreed that health and social care
professionals should ensure that people living with dementia who
have healthcare or support needs have a single named individual
responsible for co-ordinating their care, who should ensure there is
an initial assessment of their needs.
Offering information about available services
The committee noted that studies with good results usually involve
the care coordinator offering information to the informal carers on
available services and how to access them, and noted this was in line
with their own experience. The committee agreed that this should be
incorporated into the recommendations.
Involving carers, agreeing the care and support plan and
reviewing it
The committee noted that studies which have good results (for
example, Chien 2008, Chien 2011) involve a documented and
regularly reviewed care and support plan. These studies often involve
the person’s carers and family members in support and decision
making. Therefore, the committee agreed that the person’s carers
and family members (if appropriate) should be involved in support
and decision making. In addition, the care and support plan should be
agreed with the individual and their carers and/or family members (as
appropriate). A review date should also be agreed and there should
be a discussion about how that plan will be reviewed, and a copy of
the plan should be provided to the person living with dementia and
their carers (if appropriate).
People living with dementia who refuse assistance
The committee noted that if the person does not have capacity to
make decisions about their care, special consideration should be
given to the individual’s views, in line with the principles of the Mental
Capacity Act. They noted that the term ‘special consideration’ has a
specific meaning with the Act, and was therefore the correct term to
154
include in the guideline. Related to this, they also felt it was important
to inform people about the availability of local advocate services, and
in particular their rights to an independent mental capacity advocate,
should they meet the criteria for needing one.
Recording progress
The committee agreed that when reviewing care and support plans
with the person living with dementia, progress should be evaluated
and recorded against the objectives in the original care and support
plan, in coordination with all professionals involved in the person’s
care. They noted this was in line with best practice in current UK
services.
Services to ensure that the maximum number of people living
with dementia can access them
The committee noted that people who live alone who do not have
regular contact with a carer are at risk of poor outcomes. In addition,
carers can have comorbidities that make it more difficult for both
them and the person they are caring for to access services. The
committee agreed that situations like these can lead to equity of
access problems. Therefore, the committee agreed that service
providers should design services to ensure that the maximum
number of people living with dementia can access them. This
includes people who do not have an informal carer, people whose
informal carer is not able to support them to access services, and
people who do not have access to affordable transport or find
transport difficult to use. It would also include groups with
comorbidities that might limit access to services, or who are from
groups who may be less likely to access health and social care
services, and the committee felt it appropriate to specify learning
disabilities, sensory impairment, and people from black, Asian and
minority ethnic groups as examples.
Ensuring access to information
The committee agreed that people living with dementia and their
carers should know where, who from and how to obtain information
or if their needs change. Evidence for this came from Iliffe 2014, Kelly
2016, Gorska 2013, Innes 2014 and Moore 2017, and was in line with
the committee’s own experience.
Transferring information
The committee noted that the interventions in Samas 2014 and
Vickrey 2006 involve using special software to assist coordination.
The committee agreed that the beneficial results of these studies
could be partly because of the more reliable transfer of information.
The committee noted that when people living with dementia are
transferred from their home to residential care, their information is
often not sent with them. As a result, when the person living with
dementia is transferred to a residential care home, their care and
support plan often has to be created again. In addition, when
information is not sent with the person to the residential care home,
established personal routines are sometimes not respected.
Therefore, the committee agreed that service providers should
ensure that information about people living with dementia (including
care and support plans and advanced care and support plans) can be
easily transferred between different care settings (including between
home, community and residential care), including requesting consent
for these to be transferred when they are produced. The committee
agreed that the evidence did not allow them to be more specific about
how this should be ensured, and that different local areas may adopt
different solutions to this problem.
Maximising the consistency and efficiency of care
155
The committee agreed that staff delivering care and support should
maximise continuity and consistency of care. This statement refers to
ensuring continuity of the people delivering care and consistency of
approach. If this is not possible, the committee noted that it is
important to ensure that relevant information is shared across
services/people providing care. This information should be recorded
in the person’s care and support plan. This approach will reduce
duplication of questions asked by professionals and increase
efficiency. Evidence for this was found in Gorska 2013 and Innes
2014.
Consideration of health The economic evidence for 2 forms for case management supported
benefits and resource the idea that case management did not make patient’s dementia any
use worse in terms of QALYs, but could result in significant cost savings.
The committee agreed that the recommendations resulting from other
evidence considered by the committee should not incur extra costs
as the components of care recommended already exist. To the
contrary, ensuring that there is an initial assessment of needs
(identifying potential problems) and ensuring that information is
transferred, with consent, should result in cost savings. Transferring
information should avoid duplication of effort and expense, which can
be considerable – for example, saving hours of time for professionals
for each person living with dementia. Furthermore, having a single
named individual responsible for coordinating care for patients with
dementia should improve efficiency in the use of resources.
Quality of evidence The quality of the evidence was sufficient to recommend care
coordination, the formulation of a care and support plan and transfer
of information. However, the committee noted that none of the RCTs
included people living with dementia who do not have an informal
carer. Therefore, the committee took care to include
recommendations that would help people living with dementia who do
not have informal carers.
The committee noted that the evidence base primarily came from
people living in the community, and there was a lack of evidence on
people in residential care settings. They agreed on the basis of this
evidence that recommendations should be made to cover the whole
population of people living with dementia, as the key principles of
well-coordinated care are likely to be similar across cares settings,
and it was agreed that it was inappropriate to exclude people from
recommendations as a result of where they live/are receiving care.
Other considerations The committee noted that sometimes people who live with dementia
do not realise they have dementia and refuse care. This can place a
great deal of burden and stress on their informal carers. They
committee noted it was important in these situations for the carer to
continue to receive appropriate support.
Research recommendations
The benefits of high intensity case management for the UK are
difficult to estimate on the basis of the data available because of the
differences between countries with regards to primary care provision
and the thresholds for entry to long-stay care. The committee agreed
that UK-based RCTs should be conducted to determine the impact of
higher intensity case management compared with usual care on
institutionalisation and quality of life. These should measure the
quality of life of both the person living with dementia and their
carer(s). From a commissioning perspective, reducing entry to long-
stay care is a key aim and would be a priority. Ideally, studies should
also include people at various stages of dementia.
The committee noted that none of the RCTs involved care planning
for people in residential care settings. Therefore, the committee
agreed that research should be conducted to find the most effective
methods of care planning for people in residential care settings.
156
The committee felt that it was important to include a research
question to find the most effective methods of care planning for
people who do not have regular contact with an informal carer. This is
because it is common for people living with dementia to not have
regular contact with a carer and these people are often the ones with
the greatest needs and are frequently left out. The committee agreed
that whilst it may be more difficult to recruit people without carers
than those with in to research studies, it was important these people
are not forgotten when research is undertaken.
Sections of recommendations referring to younger people were also
informed by the evidence review on the specific needs of younger
people living with dementia (section 17).
Care coordination
47. Provide people living with dementia with a single named health or social care
professional who is responsible for coordinating their care.
48. Named professionals should:

 arrange an initial assessment of the person’s needs, which should be
face to face if possible.
 provide information about available services and how to access them.
 involve the person’s family members or carers (as appropriate) in
support and decision-making.
 give special consideration to the views of people who do not have
capacity to make decisions about their care, in line with the principles of
the Mental Capacity Act 2005
 ensure that people are aware of their rights to and the availability of local
advocacy services, and if appropriate to the immediate situation an
independent mental capacity advocate
 develop a care and support plan, and:
 agree and review it with the involvement of the person, their family
members or carers (as appropriate) and relevant professionals
 specify in the plan when and how often it will be reviewed
 evaluate and record progress towards the objectives at each review
 ensure it covers the management of any comorbidities
 provide a copy of the plan to the person and their family members or
carers (as appropriate).
49. When developing care and support plans and advance care and support plans,
request consent to transfer these to different care settings as needed.
50. Service providers should ensure that information (such as care and support plans
and advance care and support plans) can be easily transferred between different
care settings (for example home, inpatient, community and residential care).
51. Staff delivering care and support should maximise continuity and consistency of
care. Ensure that relevant information is shared and recorded in the person’s care
and support plan.
157
52. Service providers should design services to be accessible to as many people
living with dementia as possible, including:
 people who do not have a carer or whose carer cannot support them on
their own
 people who do not have access to affordable transport, or find transport
difficult to use
 people who have responsibilities (such as work, children or being a carer
themselves)
 people with learning disabilities, sensory impairment (such as sight or
hearing loss) or physical disabilities
 people who may be less likely to access health and social care services,
such as people from black, Asian and minority ethnic groups.
4. What is the effectiveness and cost effectiveness of high-intensity case

management compared with usual care on quality of life (for the person living with
dementia and for their carer) and the timing of entry to long-term care?
5. What are the most effective methods of care planning for people in residential
care settings?
6. What are the most effective methods of care planning for people who do not have
regular contact with an informal carer?
For more details on the research recommendations made, and the rationale behind them,
see appendix L.
158
7.2 Post diagnosis review for people living with dementia
Review question
 How should people living with dementia be reviewed post diagnosis?
7.2.1 Introduction
The aim of this review was to identify how and where people living with dementia should be
reviewed post diagnosis, and also to identify any harms caused by failures in, or
inappropriate models of, post diagnosis review for people living with dementia. The review
focused on identifying studies that fulfilled the conditions specified in Table 34. For full details
of the review protocol, see Appendix C. This question specifically focused on studies where
the intervention was a method of, location for, or individual responsible for reviewing a
person living with dementia, rather than a treatment or service being provided to the
individual. The specified trial intervention needed to be an element of how the person is
reviewed.
Table 34: Review summary: review post diagnosis for people living with dementia
Population  People (aged 40 years and over) living with dementia and admitted
to hospital
Intervention Models of post diagnosis review for people living with dementia, which
may include features such as:
 Review of mental health (memory, mood, challenging behaviours)
 Review of physical health (including co-morbidities)
 Review of functional ability
 Nutrition and hydration (swallowing)
 Lifestyle advice
 Medication review (including co-prescribing)
 Information needs
 Driving safety review
 Financial advice
 Future care planning needs
 Carer support and assessment
 Usual care
ability
 Process outcomes (e.g. adherence of staff to review protocols)
 Patient and carer experience and satisfaction
 Equity of access to services
 Adverse events (medication)
159
7.2.2 1 Evidence review
2 A systematic literature search was carried out to identify any comparative quantitative study designs (RCT’s, non-randomised controlled trials,
3 before and after studies and cohort studies). A total of 8,678 references were screened at the title and abstract level, with 33 potentially relevant
4 references being ordered for full text review. Of these references, 5 were selected for inclusion based on their relevance to the review protocol.
5 The excluded studies are listed, with reasons for their exclusion, in Appendix F. For the full evidence tables and full GRADE profiles please see
6 Appendix E and Appendix G.
7.2.2.1 7 Description of included studies
8 The characteristics of the included studies are summarised in Table 35. References for the included studies are given in appendix I.
9 Table 35: Included studies
Study reference Study design Study population Intervention & comparator Relevant outcomes Comments
Bass (2003) Randomised 157 people living with Intervention: Care consultations  No of Emergency department Location: USA
controlled trial dementia comprising use of managed visits Follow up: 12
health services in partnership with  Hospital admissions months
use of Alzheimer’s associations
 Physician visits
services
Comparison: Usual managed care  Case management visit
services only  Use of direct care community
services
 Use of non-Association
support services
Crotty (2004) Randomised 154 residents with Intervention: Multidisciplinary  Medication Appropriateness Location: Australia
controlled trial medication problems case conferences Index (MAI) Follow-up: 3 months
and/or challenging Comparison: Usual care  Behaviour (Nursing Home
behaviours Behaviour Problem Scale)
160
Kohler (2014) Randomised 235 community living Intervention: Multidisciplinary  Cognition (MMSE) Location: Germany
controlled trial dyads of people living regional dementia network  Functional (NAA; IADL) Follow-up 6-12
with dementia/ carers Comparison: Usual care months
 Quality of life (EQ5D; QOL-
AD)
Meeuwsen Randomised 175 community living Intervention: Post-diagnosis  Quality of life (QoL-AD) Location:
(2012) controlled trial dyads of people living dementia care in memory clinic  Depression (GDS) Netherlands
with dementia/ carers Comparison: Post-diagnosis Follow-up:12
 Functional (Interview for
dementia care by GP deterioration in daily living in months
dementia)
Nourhashemi Randomised 574 people living with Intervention: Comprehensive  Functional decline (ADCS- Location: France
(2010) controlled trial Alzheimer’s disease standardised evaluation every six ADL) Follow-up 24
months  Mean time to admission months
Comparison: Usual care  Risk of admission to
residential care
 Risk of mortality
 Reason for admission
(worsening medical
conditions)
 Reason for admission (carer
related reasons)
161

evaluating how people living with dementia should be reviewed post-diagnosis. In total, 3,291
articles were returned, of which 3 were selected as potentially relevant and retrieved for full
text review. Of these studies, 1 study considering memory clinics compared with GP follow-
up was deemed relevant and included.
7.2.3.1 Memory clinic versus GP follow up
Meeuwsen et al. (2013) conducted a cost–utility analysis alongside the Dutch AD-Euro RCT
summarised above, comparing the cost effectiveness of memory clinics with GP care for the
review and coordination of care of people with mild-to-moderate dementia. The primary
outcome measures were QALYs and costs over 12 months after diagnosis. For further
details, please see the economic evidence profile in Appendix M.
productivity loss and informal care costs; however, disaggregated results are reported,
enabling the recalculation of results with a perspective that is consistent with the NICE
reference case. Information about resources used was derived from a case report form
provided by the carer. The study also used the hospital information system, GP electronic
medical records and information from different healthcare workers involved to estimate
resources used. Cost prices were based on standard Dutch sources. All prices were
converted to the year 2009 and expressed in Euros.
Utilities were generated for both patient and carer using the EQ-5D, with Dutch utility
weights.
Base-case results (Table 36) suggested memory clinics are associated with cost savings of
€512 per person compared with the GP group. Aside from GP and memory clinic contacts,
only 1 variable was significantly different between the 2 arms: there were 8 hospital
admissions in the memory clinic group compared with 16 in the GP group – there was no
discussion or further information provided as to why this was the case. A QALY benefit of
0.025 in favour of the GP group was found. Taken together, these results suggest that
memory clinics save money that may be considered sufficient to justify the QALY losses with
which they are associated; however, both cost and QALY differences are relative small and
extremely uncertain.
Table 36: Analysis from Meeuwsen et al. (2013) adjusted to remove productivity loss
and informal care costs
Strategy Cost Effect Cost Effect (95%CI) ICER
Memory clinic €17,912 NR
0.025 QALYs (−0.114 to €20,480
GP €18,424 NR €512 0.064) /QALY
The authors’ probabilistic sensitivity analysis (from which it was not possible to disaggregate
costs not considered by the NICE reference case) showed that memory clinics were cost
saving, compared with GP follow-up, in 59% of cases and the probability that they are
associated with an ICER of €50,000 or better was around 50%.
162
The authors concluded that no evidence was found that memory clinics were cost effective
compared with GPs for post-diagnosis review and coordination of care of people with
dementia in the first year after diagnosis. The authors used UK utility weights in a scenario
analysis; the results were deemed similar to the Dutch analyses and were therefore not
presented by the authors.
7.2.4.1 Care reviews and consultations in partnership with Alzheimer’s associations services
versus usual managed care services only
Moderate- to high-quality evidence from 1 RCT of 157 people living with Alzheimer’s disease
found the use of case management was significantly greater at 12 months follow up for
people receiving the intervention but could not differentiate number of hospital admissions,
number of emergency department visits, number of physician visits, use of direct care
community services or use of non- Association information and support for people who were
reviewed using managed health services in partnership with Alzheimer’s Association
services compared with usual managed care services only.
7.2.4.2 Multidisciplinary case conferences versus usual care
Low- to moderate-quality evidence from 1 RCT of 154 aged care residents with pain and
dementia related problem behaviours and medication problems found significant
improvements at 3 months in the appropriate use of medicines (Medicines Appropriation
Index), but could not differentiate number of medicines (Medicines Appropriation Index),
number of drugs, changes in the number of drugs, or behaviour (Nursing Home Behaviour
Problem Checklist) for people receiving a medication advisory case conference compared
with usual care.
7.2.4.3 Network multidisciplinary monitoring and care versus usual care
Low-quality evidence from 1 RCT of 235 people living with dementia could not differentiate
between functional outcomes (NAA; IADL), patient reported health outcomes (EQ5D-VAS),
cognition (MMSE), quality of life (QOL-AD), or carers’ health related quality of life (EQ5D-
VAS; SF36 physical health, SF36 mental health) for people being reviewed using a network
of multidisciplinary care compared with usual care.
7.2.4.4 Memory clinic review and monitoring versus GP review and monitoring
Moderate- to high-quality evidence from 1 RCT of 175 people newly diagnosed with mild to
moderate dementia could not differentiate between quality of life (QOL-AD, carer report;
QOL-AD, patient report), neuropsychiatric symptoms (NPI), functional outcomes (Interview
for deterioration in dementia) or depressive symptoms (GDS) for people being reviewed in a
memory clinic compared with GP care.
Outcomes for carers found higher levels of depressive symptoms (CES-D), and anxiety
(STAI trait) for people receiving care in a memory clinic compared with GP care but could not
differentiate quality of life (QOL-AD); social support (ISB), sense of competence, emotional
problems (NPI), mastery (PMS) or personality (EPQ).
163
7.2.4.5 Standardised evaluation and assessment in memory clinic versus usual care in
memory clinic
Very low- to moderate-quality evidence from 1 RCT of 574 people living with Alzheimer’s
disease found that admissions due to worsening conditions were significantly reduced and
admissions due to carer reasons were significantly greater at 2 years follow up but could not
differentiate between functional decline at 2 years (ADCS-ADL), mean time of admission to
care at 2 years, risk of admission to care or risk of mortality at 2 years for people being
reviewed by specialist care in a memory clinic compared with usual care in a memory clinic.
7.2.4.6.1 Memory clinic versus GP follow up
One partially applicable trial-based cost–utility analysis with potentially serious limitations
explored the cost effectiveness of memory clinics compared with general practitioner care for
the review and coordination of care for people with mild-to-moderate dementia in the first
year after diagnosis. When costs not relevant to the NICE reference case were excluded,
memory clinics were cheaper than GP care, and associated with a decrease in QALYs,
resulting in a point-estimate ICER of €20,480 saved per QALY forgone with memory clinics
instead of GP care. Results were subject to substantial uncertainty: at a 95% confidence
level, the data were consistent with cost savings and/or QALY gains with either approach.

Relative value of different The committee agreed that the purpose of adopting a specific
outcomes strategy for reviewing people living with dementia would be to
improve their clinical outcomes, and so the same patient outcome
measures would be relevant here as for questions on treatment
(cognition, quality of life etc.) It also agreed that these improvements
would be achieved through more effective reviewing strategies better
identifying individual needs, and therefore leading to a more efficient
and effective use of other services. Therefore, measures like ‘people
given appropriate and timely access to services’, or ‘reductions in
inappropriate service or medicine use’ would also be valuable
outcomes.
Trade-off between The committee noted that for typical patients, only having routinely
benefits and harms scheduled appointments and standard structured assessments may
not deal with the reality of situations experienced by people living with
dementia. It agreed that reviewing people living with dementia
requires a flexible approach and is dependent upon individual needs,
rather than reliant upon a prescriptive approach. The committee
recognised that a flexible approach would have different implications
at each stage of the dementia trajectory. People living with more
severe dementia may also be living with multiple comorbidities, which
may require more rapid reviews and more frequent follow up. The
committee agreed that continuity of care is of paramount importance
in considering the evolving needs of people living with dementia.
The committee recognised there was wide variation in the
arrangement of memory service facilities in the UK and agreed it
would not be appropriate to recommend a specific service model for
reviewing people living with dementia. It highlighted the importance of
not losing people in the system and recognised that every service
contact could be used as an opportunity to provide dementia care (for
example, during GP annual reviews). For this reason, the committee
agreed it would be more important to allow people living with
dementia to have access to a multidisciplinary service, involving both
health, social care and volunteer services, in order to address issues
on a more emergent basis.
164
Consideration of health The committee acknowledged that Meeuwsen (2012) did not find any
benefits and resource significant or clinically meaningful differences in health outcomes
use between the intervention (memory clinics) and control group (GP
care). The economic evaluation alongside the trial (Meeuwsen et al
2013) found that the memory clinics intervention compared with GP
care produced a small saving in terms of costs, but also produced
fewer QALYs. However, as the author’s economic evaluation
considered costs which were not relevant to the NICE reference
case, an analysis was conducted where these costs were removed.
Under these circumstances, it was found that the memory clinics
intervention was showing a smaller saving in costs and an equal loss
of QALYs compared with the author’s economic evaluation.
The committee agreed that it would not be appropriate to make a
recommendation based on this evidence showing insignificant
savings and loss of QALYs.
Quality of evidence The committee agreed that the evidence presented did not directly
address the full complexity of the issues that relate to reviewing a
person living with dementia, although there were certain aspects that
could be taken from the included studies.
The committee noted that although the evidence they had seen was
located oversees, similar practices could be observed in UK practice.
For example it is not unusual to observe collaborations between care
management services and volunteer services or charities in the UK,
similar to that cited in Bass (2003).
In addition, the use of scheduled (bi-yearly) appointments had been
considered in Nourhashemi (2010). That study had reported no
significant benefit for clinical and functional outcomes, for people who
had received scheduled appointments and who had been reviewed in
a structured or prescriptive manner. The committee concluded that a
more beneficial service would be likely to reflect current practice and
focus more upon an emergent and flexible approach to reviewing
people living with dementia and addressing person-centred need
based upon a multidisciplinary co-ordination of care.
Other considerations The committee acknowledged that a post-diagnosis review of people
living with dementia impacts upon all members of the family. It
therefore agreed that all recommendations would be relevant to both
people living with dementia and their cares and family members.
53. After a person is diagnosed with dementia, ensure they and their family members
or carers (as appropriate) have access to a memory service or equivalent hospital-
or primary-care-based multidisciplinary dementia service.
54. Memory services and equivalent hospital- and primary-care-based

multidisciplinary dementia services should offer a choice of flexible access or
prescheduled monitoring appointments.
55. When people living with dementia or their carers have a primary care appointment,
assess for any emerging dementia-related needs and ask them if they need any
more support.
165
Inpatient care
8 Inpatient care
At any one time up to 25% of acute hospital beds are occupied by people living with
dementia. People with dementia often experience longer durations of hospital admission,
delays in leaving hospital and reduced levels of independent functioning (CQC 2017, DAA
2016). Over recent years there have been a range of initiatives to focus action and attention
on improving the experience and outcomes of hospital care for people with dementia (DAA,
Dementia Friendly Hospitals Charter & Dementia CQUIN). Acute hospital admission can be a
time of distress, confusion and delirium for someone with dementia. These factors may
contribute to a decline in global functioning and reduced ability to return home to
independent living. The achievement of improved or at least maintained levels of
independent functioning is a minimum expectation following a period of acute care. This is a
key opportunity in the persons journey with dementia for holistic comprehensive care
planning to be undertaken with the person and their significant others.
Acute hospital admission has been identified as a key opportunity for people with previously
undiagnosed dementia to access appropriate assessment & diagnosis of dementia to;
improve their care and treatment while in hospital, facilitate appropriate early discharge and
enable access to a full range of post-diagnostic support and interventions.
The extent to which the needs of people with dementia experiencing an acute hospital
admission are understood and effectively met remains variable across the country (CQC
2014 & 2017, RcPsych 2016). Many examples of improvements in this area of care over
recent years have been identified, strengthened by local commissioning arrangements and
good clinical leadership, but there remains significant local variation in how effectively people
experiencing dementia in a hospital setting are diagnosed and then provided with appropriate
tailored individual support and discharge planning.
166
Inpatient care
8.1 Caring for people living with dementia who are admitted to
hospital
Review question
 How should people living with dementia be cared for when admitted to hospital?
8.1.1 Introduction
The aim of this review was to identify the most appropriate ways to care for people living with
dementia when they are admitted to hospital and to identify any harms that may be caused
by failures in or inappropriate models of hospital care for people living with dementia. The
review focused on identifying studies that fulfilled the conditions specified in Table 37. For full
details of the review protocol, see Appendix C.
Table 37: Review summary: inpatient care for people living with dementia
Population  People (aged 40 years and over) living with dementia and admitted
to hospital
Intervention Models of hospital care for people living with dementia, which may
include elements such as:
 Additional support from hospital staff/others
 Information needs (both information for the person living with
dementia and the information needs of the hospital staff)
 Person-centred assessment
 Assessment for hospital discharge (timing of discharge)
 Family/carer information needs, access and involvement in care
 Types of ward
 Environmental design
 Comprehensive geriatric assessment
 Medicines reconciliation and review
ability
 Process outcomes (e.g. adherence of staff to care protocols)
 Staff wellbeing and job satisfaction, skill levels
 Co-patient experience
 Adverse events
 Equity of access to services
A systematic literature search was carried out to identify any comparative quantitative study
designs (for example RCTs, non-randomised controlled trials, before and after studies and
cohort studies). A total of 8,857 references were screened at the title and abstract level, with
46 potentially relevant references being ordered for full text review. Of these references, 5
were selected for inclusion based on their relevance to the review protocol. The excluded
studies are listed, with reasons for their exclusion, in Appendix F. Evidence tables for the
included studies are presented in Appendix E, with GRADE profiles in appendix G.
167
Inpatient care
2 The characteristics of the included studies are summarised in Table 38. References for the included studies are given in appendix I.

Intervention &
Study reference Study design Study population comparator Relevant outcomes Comments
Baldwin (2004) RCT 153 medically ill people  Intervention: Multi-  Health of Nation outcome scale Study location UK
with cognitive impairment faceted nurse led  Geriatric Depression Scale General hospital
and/or depression in a intervention
 MMSE
district general hospital  Comparator: Usual care
 Length of stay in hospital (days)
 Readmissions at 3 months
 Death at 3 months
Boltz (2015) Non-randomised 86 patient/carer dyads of  Intervention: Family  Patient outcomes: Study location:
controlled trial hospitalised people living centred function o Hospital readmission USA
with dementia focused care o Occurrence of delirium General hospital
 Comparator: Usual care o Activities of daily living
o Utilisation of post-acute
rehabilitation
 Carer outcomes
o Preparedness for caregiving
o Anxiety
o Depression
o Strain
Campbell (2004) Prospective cohort 52 people with end-stage  Intervention: Proactive  Hospital and ICU length of stay Study location USA
dementia (defined by FAS case finding and  Use of non-beneficial resource ICU
&NHO guidelines) treated collaboration between
 Establishment of do not resuscitate
in an ICU palliative care service
goals
and ICU staff
 Comparator: Usual care
168
Inpatient care
Intervention &
Study reference Study design Study population comparator Relevant outcomes Comments
Goldberg (2013) RCT 600 people aged over 65  Intervention: Specialist  Quality of life (EQ-5D (short Study location
NIHR TEAM trial years with confusion medical and mental London handicap); DEMQOL; UK
(delirium/ cognitive health unit EuroQoL) Acute General
impairment) admitted for  Comparator: Standard  Physical disability hospital
acute medical care care  Cognitive Impairment
 Carer strain
 Carer psychological wellbeing
Villars (2013) Before and after 390 people with  Intervention: Geriatric  Rate of re-hospitalisations 1 month Study location
study Alzheimer’s disease team unit post discharge France
hospitalised in a special  Comparator: Usual care  Vocally disruptive behaviours Special acute care
acute care unit unit
169
Inpatient care

(CUAs) evaluating how people living with dementia should be cared for when admitted to
hospital. In total, 3,367 articles were returned, of which 1 was selected as potentially relevant
and retrieved for full-text review, after which it was deemed relevant and included.
Medical and mental health unit
Tanajewski et al. (2015) compared the cost effectiveness of a dedicated medical and mental
health unit (MMHU) with usual care (acute geriatric medical wards and general (internal)
medical wards). The authors conducted a cost-–utility analysis alongside the trial of an
elderly acute care medical and mental health unit (NIHR TEAM trial; Goldberg et al., 2013)
(n=600), an RCT conducted between 2010 and 2012 in the UK, and collected health care
utility data using the EQ-5D-3L with societal weights. Primary outcome measures were
QALYs and costs over 90 days. For further details, please see the economic evidence profile
in Appendix M.
Service-use data included health and social care costs. Rates of resource use were taken
from electronic administrative records systems used to record patient care and were costed
using standard reference costs. Data were collected for 3 months post-hospital admission
and 1 year pre-admission. The mean results are presented in Table 39.
Table 39: Base-case cost–utility results – Tanajewski et al., 2015

Costs Effects Costs Effects
Treatment (95% CI) (95% CI) (95% CI) (95% CI) ICER
£7,862 0.108 QALYs
Standard care
(7,758, 7,965) (0.101, 0.114)
Medical and mental £7,714 0.109 QALYs £-149 0.001 QALYs
Dominant
health unit (MMHU) (7,606, 7,822) (0.102, 0.116) (-298, 4) (-0.006, 0.008)
Over a period of 90 days, MMHUs resulted in monetary savings and an increased number of
QALYs, making MMHUs a dominant strategy. Probabilistic analysis showed that there is a
58% probability of the MMHU being dominant, and a 94% probability of cost effectiveness, if
QALYs are valued at £20,000 each. The probability of the MMHU being cost-saving with
QALY loss (SW quadrant of the cost–utility plane) was 39%.
The authors concluded that the specialist unit for people with delirium and dementia did not
demonstrate convincing benefits in health status over usual hospital care, as no significant
effect on QALY gain was observed. However, the results did show a trend towards cost
savings and a high probability of cost effectiveness from a combined health and social care
perspective, when usual criteria were applied.
8.1.4.1 Nurse-led mental health liaison service versus usual care
Very low- to low-quality evidence from an RCT containing 153 participants could not
differentiate levels of depressive symptoms (Geriatric Depression Scale), cognition (MMSE),
general health (Health of Nations Outcome Scale), length of stay, the number of psychotropic
medications prescribed at discharge, number readmitted to hospital and number of deaths in
170
Inpatient care
hospital between hospitalised people living with dementia, cognitive impairment or

depressive symptoms being cared for by a nurse-led mental health liaison service compared
with usual care.
8.1.4.2 Family-centred function focused care versus usual care
Very-low to low-quality evidence from 1 non-randomised controlled trial containing 86

participants found improved levels of activities of daily living (Barthel Index) and walking
performance and reduced delirium incidence and delirium severity in hospitalised people
living with dementia receiving family-centred function focused care compared with usual
care, but could not differentiate length of stay, number of hospital readmissions, utilisation of
post-acute rehabilitation at discharge, gait and balance (Tinetti Scale) or carer related
outcomes.
8.1.4.3 Proactive case finding with palliative care service versus usual care
Very-low to low-quality evidence from 1 cohort study conducted in the USA containing 52
participants found reduced length of stay in hospital and ICU and reduced levels of ICU
workload after DNR rules were established in people in hospitalised people living with
dementia who had been proactively identified in liaison between ICU staff and the palliative
care service versus usual care, but could not differentiate levels of mortality, length of time
from admission until do not resuscitate goals were established, length of stay from
establishment of do not resuscitate goals until discharge or ICU workload before DNR rules
were established.
8.1.4.4 Specialist medical and mental health unit versus usual care
Very low- to moderate-quality evidence from 1 randomised controlled trial containing 600
participants could not differentiate cognition (MMSE), activities of daily living (Barthel Index),
quality of life (DEMQOL self-report, DEMQOL proxy, EQ5D self-report, EQ5D proxy), general
health measures (London Handicap Scale), number returning home from hospital within 90
days, overall mortality, readmissions or carer strain (Carer Strain Index) between
hospitalised people living with dementia and/or delirium receiving care at a specialist medical
and mental health unit versus usual care.
8.1.4.5 Follow-up individualised care plan versus usual care
Very low-quality evidence from 1 observational study containing 390 participants could not
differentiate early ER rehospitalisation rates at discharge and after 3 months or early
rehospitalisation rates in any ward at discharge and after 3 months in hospitalised people
living with dementia receiving an individualised follow-up care plan versus usual care.
One directly applicable trial-based cost–utility analysis with minor limitations found that,
compared with usual care, a dedicated medical and mental health unit resulted in cost
savings of £149 per person and were associated with a small gain in QALYs of 0.001,
rendering the strategy dominant. Probabilistic analysis showed a 58% probability of the
dedicated unit being dominant and a 94% probability of cost effectiveness, when QALYs
were valued at £20,000 each.

Relative value of different The committee recognised the relevance of all studies included in the
outcomes evidence base, although they agreed, given the inclusion of an
171
Inpatient care
economic evaluation alongside the study, that the NIHR TEAM trial
was the most relevant. They also agreed that the outcome measures
used in this study were reasonable, and would be expected to
capture any major differences found between the two groups.
Trade-off between The committee agreed that none of the interventions tested showed
benefits and harms consistent evidence of benefits for either patients or carers, and
therefore it was not appropriate to make any specific
recommendations based on these trials.
Although the committee agreed the NIHR funded TEAM trial
(Goldberg, 2013) had demonstrated an appropriate model of care
within a UK based hospital, they acknowledged there would be
practical considerations in service delivery if this model was applied
at a nationwide level. In particular, the committee recognised this
model would be difficult to roll out nationally, as it would require a
major reorganisation of staff in many UK hospitals. For this reason,
the committee agreed there were not compelling clinical reasons to
write a recommendation in support of this service delivery model.
The committee agreed that, despite the lack of evidence found for
specific interventions to improve hospital care for people living with
dementia, there were nonetheless specific issues people with
dementia faced in hospital. In particular, they agreed it was often not
appropriate for people living with dementia to be treated on general
hospital wards, and felt that a geriatric ward was usually a more
appropriate location. Whilst these wards are not dementia specific, a
high enough proportion of people passing through them are likely to
have dementia (simply based on the underlying prevalence in the
population) and therefore the staff are likely to be better trained and
more experienced with people living with dementia than those on a
general hospital ward.
The committee also agreed that because the hospital population
fluctuates, there are times when there will be a higher proportion of
people living with dementia than at other times. Therefore, it would
not be viable for the NHS to arrange units for older aged care into
separate units specifically for people who are living with dementia
and those who do not have dementia. The presence of some such
units, but without the capacity to accept all people with dementia,
risked creating a culture of inequity, whereby patients unable to
access a specialist unit would potentially be treated as “being in the
wrong place”, and therefore be likely to get sub-optimal care. The
committee agreed the correct approach was rather to take elements
of best care found in specialist units and apply these to all geriatric
units, thereby raising the overall standard of care.
Consideration of health The committee acknowledged that TEAM trial did not find any
benefits and resource significant or clinically meaningful differences in health outcomes
use between the intervention (MMHU - medical and mental health unit)
and control group (acute geriatric and general medical wards) and
therefore the economic evaluation (Tanajewski 2015) shifted the
focus onto how savings were made with the MMHU intervention
compared with control group (usual care). The committee expressed
concern about the level of breakdown of costs provided in the paper
as it was not clear where the cost savings for MMHU had come from,
and the committee agreed it did not seem likely that such savings
would be achieved in practice from what is a more staff-intensive
model of care. The committee agreed the only way such savings
would be possible in such a model would be either if the extra
permanent staffing led to substantial reductions in spending on
agency/temporary staff, or if increased staff knowledge led to a
reduction in inappropriate interventions (e.g. unnecessary
investigations). In the absence of sufficient detail in the paper to
address these issues, the committee agreed that it would not be
172
Inpatient care
appropriate to make a recommendation solely based on evidence of

cost saving that is unlikely to be achievable in practice.
Quality of evidence The committee did not have any real concerns about the overall
quality of the evidence they had seen. Although the results of the
TEAM trial (Goldberg, 2013) demonstrated non-significant outcomes
and were unable to differentiate outcomes when looking at care
provided by a specialised service unit compared with usual care, the
committee acknowledged it was a well-controlled trial, which had
included a substantial sample of 600 individuals.
The committee recognised there were differences in care between
the populations reported in the US papers and populations that are
cared for in the UK. This was particularly relevant to interpretation of
the service delivery model reported in Campbell (2004). The patient
population described within the Intensive Care Unit would have been
very different to patient populations in an ICU within a UK setting. For
this reason the committee were cautious about highlighting that
model of service organisation.
The committee acknowledged there was a mixed population of
people included in two papers (Goldberg 2013; and Baldwin 2004)
where people with delirium (Goldberg) and depression (Baldwin)
were included alongside people living with dementia. The committee
agreed these would not have had a negative influence upon
interpretation of results because in practice the structural
organisation of in-patient care for people living with delirium and
dementia were likely to be similar. The committee did agree,
however, that people living with dementia were at substantially
increased risk of delirium when in hospital, and therefore it was
appropriate to cross-refer to the sections of the delirium guideline on
interventions to both prevent and treat delirium.
Other considerations The committee noted there were other forms of evidence, and
although these did not directly fulfil the inclusion criteria as defined in
the review protocol for this question, they may provide further insight
into the relevance of certain models of care. In particular, the
committee highlighted the qualitative evidence that was conducted
alongside the TEAM trial to provide insights into experiences of those
involved in that model of care. The committee also recognised that
Rapid Assessment, Interface and Discharge protocols are applied in
current practice for more general older aged populations. The
committee also highlighted there are other established national
standards set out for healthcare practitioners to follow, in particular
advice from the Royal College of Psychiatrists for managing older
people in acute hospital settings.
56. Be aware of the increased risk of delirium in people living with dementia who are
admitted to hospital. See the NICE guideline on delirium for interventions to
prevent and treat delirium.
173
Care setting transitions
9 Care setting transitions

Improper/poorly managed discharges from a service/environment (home, care home,
hospital or respite care) can lead to increased stress and anxiety, both for people living with
dementia and those caring for them. This uncertainty of transition can amplify negative
feelings and cause unnecessary distress. Poor transition/planning between services can lead
to increased likelihood of re-hospitalisation, delayed discharges, failed discharges,
inappropriate placements and carer breakdown (Naylor 2008).
There is much documentation surrounding poor communication and planning when

transitioning from one setting to another. Completing multi-disciplinary discharge meetings
and ensuring all relevant parties are included in such decisions is vital in maintaining good
communication and positive outcomes. Working in a collaborative manner increases positive
outcomes by ensuring that everyone is aware of the support required and where this can
best be achieved.
The Care Act 2014 discusses the responsibility of those working in adult care to ensure a
person’s wellbeing when managing and supporting their care, respecting the individual’s
wishes and the things that matter to them. The principal purpose is to ensure that everyone
has support to meet their individual needs, rather than a one size fits all style of care.
When transitioning from one environment/setting to another the fundamental principles that
apply are: planning, communication, collaboration and person centred support.
174
9.1 Managing the transition between different settings for

people living with dementia
Review question
 What are the most effective ways of managing the transition between different settings
(home, care home, hospital, and respite) for people living with dementia?
9.1.1 Introduction
The aims of this review were to establish the most effective ways of managing the transition
between different settings for people living with dementia, and their carers. The review
focused on identifying studies that fulfilled the conditions specified in Table 40. For full details
of the review protocol, see Appendix C.
Table 40: Review summary: Managing the transition between different settings
Population People (aged 40 years and over) living with dementia
Factors/Interventions Policies or systems for managing transfers between settings for
people living with dementia, which may include elements such as:
 Assessment of person’s needs and living environment (destination
environment)
 Systems for monitoring and adjusting plans as needs change
 Ways of confirming required services are in place pre-transfer
 Involvement of family members and carers
 Transfer of information (continuity of care)
 Maintaining relationships
 Timing of transfer
ability
 Rates of delayed discharge
 Adverse events
Randomised controlled trials and systematic review of randomised controlled trials were
included if they compared different methods of managing transitions between care settings.
Papers were excluded if they:
 were not in the English language
 considered transitions into or out of inpatient hospital settings: these transitions are
covered by another NICE guideline. (NG26: Transition between inpatient hospital settings
and community or care home settings for adults with social care needs)
 considered transitions into or out of inpatient mental health settings: these transitions are
covered by another NICE guideline. (NG53: Transition between inpatient mental health
settings and community or care home settings)
 considered aspects of medicines-related communication systems when patients move
between care settings: this is covered by another NICE guideline. (NG5: Medicines
optimisation: the safe and effective use of medicines to enable the best possible
outcomes)
175
A total of 3,451 unique citations were identified through a systematic search, of which 46
were retrieved for full-text appraisal. Four of these studies ultimately met the criteria for
inclusion, with the remaining 42 studies excluded, with reasons for exclusion given in
Appendix F. The included studies are summarised in Table 41. For the full evidence tables
and full GRADE profiles please see Appendix E and Appendix G. References for the

Study details Study population Interventions Outcomes
Davies (2011) 56 carers of people living Psychosocial  Carer burden
with dementia who have intervention (TIFF-  Carer depression
recently moved to a care NH) versus usual
 Carer satisfaction with
home care
facility
Gaugler (2011) 406 carers of people living Psychosocial  Carer burden
with dementia who are intervention (NYU  Carer depression
transitioning to a care carer intervention)

home versus usual care
Gaugler (2015) 36 carers of people living Psychosocial  Carer burden
with dementia who have intervention  Carer stress
recently moved to a care (RCTM) versus
 Carer depression
home usual care
facility
role
McGilton (2003) 32 people living with Way finding  Agitation
Alzheimer’s disease who (reorientation)  Spatial orientation
have recently changed intervention versus
care home usual care
Standard health economic filters with social care outcome terms were applied to the clinical
search for this question, and a total of 2,974 citations was returned. Following review of titles
and abstracts, no full text studies were retrieved for detailed consideration. Therefore, no
relevant cost–utility analyses were identified for this question.
9.1.4.1 Intervention for people living with dementia
Low-quality evidence from 1 RCT containing 32 people could not differentiate levels of
agitation or spatial disorientation between people with Alzheimer’s disease relocated to a
new nursing home facility intervention who were offered or not offered a reorientation
intervention.
9.1.4.2 Interventions for carers
Very low to low-quality evidence from 1 RCT containing 406 people found lower levels of
depressive symptoms in carers of people living with dementia transitioning to nursing homes
176
who were offered comprehensive psychosocial support (New York University Carer
Intervention) compared with those not offered support, but could not differentiate levels of
carer burden.
Very low-quality evidence from 2 RCTs containing 82 people could not differentiate levels of
carer burden, stress, depressive symptoms, satisfaction with the care facility or role
satisfactions between carers of people living with dementia transitioning to nursing homes
who were offered psychosocial interventions compared with those not offered interventions.

Relative value of different The committee agreed that, in order to recommend specific transfer
outcomes interventions, data from quantitative studies (particularly randomised
controlled trials) would be necessary. They agreed that data from
qualitative studies alone would be unlikely to be sufficient to
recommend potentially expensive interventions
The committee also discussed the existing NICE guidelines on
transfers between different settings, to assess their applicability for
Trade-off between The committee agreed that the following NICE guidelines on transfer
benefits and harms are relevant to this guideline: Transition between inpatient mental
health settings and community or care home settings (NG53),
Transition between inpatient hospital settings and community or care
home settings for adults with social care needs (NG27) and the
recommendations specifically on transfer in section 1.2 of Medicines
optimisation: the safe and effective use of medicines to enable the
best possible outcomes (NG5). The committee therefore agreed that
it would be appropriate to cross-refer to these pieces of guidance.
The reason why other sections of NG5 were not referred to in this
section is because they do not relate to the transfer of people.
The committee agreed that all further recommendations should then
focus on areas of transfer particularly relevant to people living with
dementia, over and above standard transfers.
The main distinct feature of people who live with dementia were
identified as being the difficulty in ensuring that needs and wishes
(including any Do Not Attempt Resuscitation (DNAR) documentation)
are reviewed. In the experience of the committee, this is very
important information that is commonly missing or not understood.
For example, when a person living with dementia is transferred from
one environment to another, staff are often unsure as to whether a
pre-existing DNAR form is still relevant because potentially the
person’s needs or wishes have changed.
The committee wanted the phrase “needs and wishes” to be
included, rather than refer to further specific documents. The aim of
this is to get staff to think about what the needs and wishes of people
are and to enable life-story documentation to be reviewed. In
addition, there are already separate recommendations on what
information should be documented in the palliative care and barriers
and facilitators to involvement in decision-making sections of this
guideline.
The committee agreed it was important to include the phrase “after
any transitions” because this is when information is reviewed by the
new staff taking care of the person.
177
The committee therefore made the recommendation that after any

transition, to ensure that the person’s needs and wishes (including
any DNAR documentation) are reviewed.
The committee also agreed that the same principles underlying good
practice for transfers to and from inpatient settings would also be
applicable for transfers within the community, and therefore agreed it
would be appropriate to make recommendations that these principles
also be applied in the community setting. Whilst no specific evidence
was identified for community transitions, the committee agreed that
the evidence identified for inpatient transfers that led to the
development of recommendations for those guidelines could
reasonably be extrapolated to the community setting as well.
Consideration of health The committee agreed that the following additional recommendation
benefits and resource should incur no additional cost. This is because this recommendation
use should be current standard practice, and the problems caused by
information not being appropriately shared are likely to have a higher
resource burden than the cost of ensuring appropriate transfer of
information.
Quality of evidence The committee agreed that none of the RCTs on specific transfer
interventions provided strong enough evidence to make specific
additional recommendations, over and above those in the other NICE
guidelines on transitions between care settings.
Other considerations The committee agreed that future research should be done on the
questions of appropriately managing transitions for people living with
dementia, as it is an important issue for service providers that it not
currently addressed in the research literature. The committee advised
that future RCTs should involve people living with dementia and
compare a structured transfer plan to standard care. Examples of
useful outcomes are: quality of life measures, narrative opinions,
costs and adverse events.
57. For guidance on managing transition between care settings for people living with
dementia, see:
 the NICE guideline on transition between inpatient hospital settings and
community or care home settings for adults with social care needs
 the NICE guideline on transition between inpatient mental health settings
and community or care home settings
 section 1.2 of the NICE guideline on medicines optimisation.
Follow the principles in these guidelines for transitions between other settings
(for example from home to a care home or respite care).
58. Review the person's needs and wishes (including any care and support plans and
advance care and support plans) after every transition.
7. What is the effectiveness of structured transfer plans to ease the transition

between different environments for people living with dementia and their carers?
appendix L.
178
Modifying risk factors for dementia progression
10 Modifying risk factors for dementia

progression
Several risk factors have been described for dementia, and most also apply to the main
subtype of Alzheimer’s disease. These include factors such as: advancing age, female sex,
low education, possession of certain genetic risk factors, such as apolipoprotein E4, family
history of dementia, history of depression, history of head injury, pre-existing history of
learning difficulties, especially Down’s syndrome. However, in addition, there has been much
recent interest in several vascular risk factors which have also emerged as risk factors for
dementia and also for Alzheimer’s disease. These include smoking, hypertension,
hypocholesterolaemia, diabetes, ischemic heart disease, obesity, lack of exercise and atrial
fibrillation. The main clinical implication of these vascular risk factors, as opposed to the
other risk factors described, is that all are potentially modifiable. Epidemiological evidence
has largely informed our knowledge about risk factors, but such evidence cannot indicate
whether modification of any or all of these risk factors will either prevent the development of
dementia or, more relevantly for the current guideline, delay the progression of cognitive or
functional impairment in those with established dementia. This question concerns whether
modifying risk factors may have an effect on slowing the progression of dementia in those
with recognised dementia.
The mechanism by which such modification may delay the progression of dementia, if they
are effective, is not entirely clear. The two broad hypotheses would be, firstly, that such
modification will reduce the accumulation of additional vascular burden in such subjects, an
effect which is known to accelerate the expression of dementia even in those with
Alzheimer’s disease, for example by reducing the occurrence of subcortical vascular disease
(white matter lesions and lacunar infarcts). The second is potentially through direct
modification of degenerative (Alzheimer’s-type, i.e. plaque and tangle) pathology, since there
is some evidence both from the animal literature and limited human studies that vascular
factors such as hypertension and diabetes, as well as ischemic damage which might be
secondary to vascular changes, can hasten the spread of Alzheimer-type changes.
179
10.1 Risk factors for dementia progression

Review question
 What effect does modifying risk factors have on slowing the progression of dementia?
10.1.1 Introduction
The aim of this review question is to assess whether interventions targeting underlying risk
factors for progression can be used to slow progression of dementia, or its sequelae, after
diagnosis. The interventions in question are based on known modifiable risk factors that have
not been addressed in other sections of this guideline. The review identified studies that
fulfilled the conditions specified in Table 42. For full details of the review protocol, see
appendix C.
Table 42: Review summary: modifying risk factors for dementia progression
Interventions  Interventions to modify risk factors for dementia progression.
Potentially modifiable risk factors may include:
 Alcohol consumption
 Smoking
 Obesity
 Diabetes
 Hypertension
 Hypercholesterolaemia
 Diet
 Non-steroidal anti-inflammatory drugs
 Antipsychotics
Comparator  No intervention
Outcomes  Rates of dementia progression
 Clinical outcomes including cognitive, functional and behavioural
ability
 Adverse events
A systematic literature search for systematic reviews and RCTs identified 3,217 references.
These were screened at title and abstract level, with 43 papers (9 systematic reviews and 34
RCTs) ordered as potentially relevant. Of these studies, 6 RCTs assessing the effect of risk
factor modification on dementia progression were included. Fifteen additional references
were identified through assessment of the bibliographies of excluded systematic reviews and
RCTs, all of which were included. In total, 20 studies (reported in 21 publications) were
included, with 36 excluded at full-text review. The original protocol for this question specified
that it should include studies of at least 12 months duration. However, due to the relatively
low number of RCTs identified, this was expanded to include trials of at least 6 months
duration. The studies excluded at full-text review, and the reasons for exclusion, are given in
appendix F.
180
2 Randomised controlled trials assessing 4 different types of interventions were found:

3  Two studies evaluated the efficacy of antidiabetic drugs for reducing cognitive decline (namely, rosiglitazone). In these studies rosiglitazone was
4 given to people living with Alzheimer’s disease who did not have diabetes.
5  Ten trials were identified that assessed the safety and efficacy of non-steroidal anti-inflammatory drugs (NSAIDs) for reducing cognitive decline
6 in people living with Alzheimer’s disease. NSAIDs included naproxen, aspirin, indomethacin, tarenflurbil (an R-enantiomer of flurbiprofen),
7 ibuprofen, diclofenac, celecoxib and rofecoxib. It was noted that rofecoxib was withdrawn from the market in 2004; however, the identified
8 evidence was included in analyses to explore the class effect of NSAIDs.
9  Four trials were identified that assessed the efficacy of statins (atorvastatin and simvastatin) for reducing cognitive decline in people living with
10 Alzheimer’s disease.
11  Four trials were identified that assessed the efficacy of antihypertensive drugs for reducing cognitive decline. Three of these studies included
12 people living with Alzheimer’s disease whereas 1 study included people with subcortical vascular dementia. It was noted that some of these
13 studies included people who did not have hypertension. Antihypertensive drugs included telmisartan, nimodipine amlodipine, perindopril,
14 captopril, enalapril imidapril, nifedipine and nilvadipine.
15 For the purpose of this review, studies which assessed the efficacy of dietary supplements, exercise and antipsychotics for treating aggression or
16 psychosis, were not included. This is because the evidence on these interventions was considered elsewhere in the guideline. For the full evidence
17 tables and full GRADE profiles please see Appendix E and Appendix G. References for the included studies are given in appendix I.
18 Table 43: Summary of included studies: antidiabetic medicines

Gold (2010) 581 people with probable Rosiglitazone 2 mg, rosiglitazone 8  Cognitive outcomes (MMSE; ADAS-cog)
Alzheimer’s disease. Living status mg, donepezil 10 mg or placebo.  Functional ability (Disability Assessment of Dementia test)
not specified.
 Clinical global assessment (CIBIC+)
 Behavioural/Neuropsychological outcomes (NPI)
 Adverse events
Risner (2006) 511 people with mild to moderate Rosiglitazone 2 mg, rosiglitazone 4  Cognitive outcomes (ADAS-cog)
Alzheimer’s disease. Living status mg, rosiglitazone 8 mg or placebo.  Clinical global assessment (CIBIC+, collected but not reported)
not specified.
 Adverse events
181
1 Table 44: Summary of included studies: NSAIDs

Aisen (2003) 351 people with probable Naproxen 220 mg bid., rofecoxib 25  Cognitive outcomes (ADAS-cog)
Alzheimer’s disease. Living status mg, or placebo.  Functional ability (ADCS-ADL)
not specified.
 Dementia severity (CDR-SB)
 Quality of life (QoL-AD)
 Adverse events
Bentham 310 people with probable Aspirin 75 mg or aspirin avoidance.  Cognitive outcomes (MMSE)
(2008) Alzheimer’s disease. Living at  Functional ability (BADLS)
home.
 Behavioural/Neuropsychological outcomes (NPI, collected but
not reported)
 carer outcomes (GHQ)
 Adverse events
De Jong (2008) 51 people with probable Indomethacin 50 mg bid. or  Cognitive outcomes (ADAS-cog; MMSE)
Alzheimer’s disease. Living either placebo.  Clinical global assessment (CIBIC+)
at home or at a nursing home.
 Functional ability (IDDD)
 Carer outcomes (NPI-D)
 Adverse events
Green (2009) 1,649 with mild to moderate Tarenflurbil 400 mg bid., tarenflurbil  Cognitive outcomes (ADAS-cog; MMSE)
Alzheimer’s disease. Living in the 800 mg bid. or placebo.  Functional ability (ADCS-ADL)
community.
 Dementia severity (CDR-SB); Quality of life (QOL-AD)
 Carer outcomes (CBI)
 Adverse events
Pasqualetti 132 people with probable Ibuprofen 400 mg bid. or placebo.  Cognitive outcomes (ADAS-Cog; MMSE)
(2009) Alzheimer’s disease. Living status  Clinical global assessment (CIBIC+)
not specified.
 Functional ability (BADLS)
182

 Depression (BDI; GDS)
 Carer outcomes (STA1-Y1; STA1-Y2)
 Adverse events
Reines (2004) 692 people with probable or Rofecoxib 25 mg or placebo  Cognitive outcomes (ADAS-Cog; MMSE)
possible Alzheimer’s disease.  Clinical global assessment (CIBIC+)
Living status not specified
 Functional ability (ADCS-ADL)
 Adverse events
Rogers (1993) 44 people with probable Indomethacin (Dosage adjusted to  Cognitive outcomes (ADAS-Cog; MMSE; BNT; Token test)
Alzheimer’s disease. Living status weight) or placebo.  Adverse events
not specified.
Scharf (1999) 41 people with probable Diclofenac plus misoprostol  Cognitive outcomes (ADAS-Cog; ADAS-Noncog; MMSE)
Alzheimer’s disease. Living status (Dosage not specified) or placebo.  Clinical global assessment (GDS; CGIC)
not specified.
 functional ability (IADL; PSMS)
 Carer outcomes (cGIC)
 Adverse events, collected but insufficiently reported
Soininen 461 people with probable or Celecoxib 200 mg bid. or placebo.  Cognitive outcomes (ADAS-Cog; MMSE)
(2007) possible Alzheimer’s disease.  Clinical global assessment (CIBIC+)
Living status not specified.
 Nurses' Observation Scale For Geriatric Patients [NOSGER])
 Behavioural/Neuropsychological outcomes (Behave-AD2)
 Depression (MADRS)
 Adverse events
Wilcock (2008) 189 people with probable Tarenflurbil 400 mg bid., tarenflurbil  Cognitive outcomes (ADAS-Cog)
Alzheimer’s disease. Living in the 800 mg bid. or placebo.  Functional ability (ADCS-ADL)
community.
 Adverse events
183
1 Table 45: Summary of included studies: statins

Feldman 640 people with probable Atorvastatin 40 mg bid. or placebo.  Cognitive outcomes (ADAS-Cog; MMSE)
(2010) Alzheimer’s disease, most of  Functional ability (ADFACS)
whom were living in the
 Clinical global assessment (ADCS-CGIC)
community.
 Carer outcomes
 Healthcare resource
Sano (2011) 406 people with probable Simvastatin (20 mg per day for 6  Cognitive outcomes (ADAS-Cog; MMSE)
Alzheimer’s disease. Living status weeks, and simvastatin 40 mg  Functional ability (ADCS-ADL)
not specified. thereafter) or placebo.
 Caregiving hours
 Adverse events
Simons (2002) 44 people with probable Simvastatin or placebo.  Cognitive outcomes (ADAS-Cog; MMSE)
Alzheimer’s disease. Living status  Lipid concentrations
not specified.
Sparks (2005) 63 people with probable or Atorvastatin 40 mg bid. or placebo.  Cognitive outcomes (ADAS-Cog; MMSE)
& Sparks possible Alzheimer’s disease.  Functional ability (ADCS-ADL, collected but not reported)
(2006) Living status not specified.
 Clinical global assessment (CGIC)
 Depression (GDS)
2 Table 46: Summary of included studies: antihypertensive medicines

Kume (2012) 20 people with probable Telmisartan 40 to 80 mg or  Cognitive outcomes (MMSE; ADAS-Cog; WMS-R logical
Alzheimer’s disease. Living status amlodipine 5 to 10 mg. memory test)
not specified.  Blood pressure changes
 Cerebral blood flow
184

Ohrui (2004) 162 people with probable Brain-penetrating ACE inhibitors  Cognitive outcomes (MMSE)
Alzheimer’s disease. Living status (perindopril 2 mg or captopril 37.5
not specified. mg), non-brain-penetrating ACE
inhibitors (enalapril 5 mg or
imidapril 5 mg) or a calcium-
channel blocker (nifedipine 20 mg
or nilvadipine 4 mg).
Morich (2012) 1,648 people with Alzheimer’s Nimodipine 90 mg, nimodipine 180  Cognitive outcomes (MMSE; ADAS-Cog; ADAS-total score;
disease. Living at home. mg, or placebo. BSR; GERRI)
 Clinical global assessment (GDS; CGI-S; CGI-I)
 Adverse events
Pantoni (2005) 242 people subcortical dementia. Nimpdodipine 30 mg tid. or  Cognitive outcomes (MMSE; SCAG test; set test; digit span
Living status not specified. placebo. test for working memory)
 Clinical global assessment (NOSGER)
 Verbal fluency (Zahlen-Verbingdungs test; lexical production)
 Depression (HAM-D)
 Motor performance
185
Standard filters plus social care extras were applied to the clinical search for this question,
and a total of 1,455 citations was returned. Following review of titles and abstracts, no full
text studies were retrieved for detailed consideration. Therefore, no relevant cost-utility
analyses were identified for this question.
10.1.4.1 Antidiabetic drugs versus placebo
Very low to low-quality evidence from up to 2 RCTs, including 882 people living with
Alzheimer’s disease, found no meaningful differences in cognition, global assessment,
behavioural and psychological symptoms, adverse events, serious adverse events or
adverse events leading to discontinuation between people offered rosiglitazone or placebo.
10.1.4.2 NSAIDs versus placebo
Very low- to moderate-quality evidence from up to 8 RCTs, including 3,284 people living with
Alzheimer’s disease, found no meaningful differences in cognition, global assessment,
behavioural and psychological symptoms, dementia severity, quality of life, serious adverse
events or mortality between people offered NSAIDs or placebo.
Low-quality evidence from up to 7 RCTs, including 2,989 people living with Alzheimer’s
disease, found better functional ability in people offered NSAIDs compared with those who
received placebo, but these differences were either not clinically significant or did not persist
in a sensitivity analysis removing rofecoxib (a treatment that has been withdrawn from the
market).
Low-quality evidence from up to 6 RCTs, including 3,533 people living with Alzheimer’s
disease, found higher levels of adverse events leading to discontinuation in people offered
NSAIDs compared with those receiving a placebo.
10.1.4.3 Statins versus placebo
Very low- to moderate-quality evidence from up to 4 RCTs, including 1,084 people living with
Alzheimer’s disease, found no meaningful differences in cognition, behavioural and
psychological symptoms, adverse events, serious adverse events or mortality between
people offered statins or placebo.
Moderate-quality evidence from 1 RCT, including 639 people living with Alzheimer’s disease,
found higher levels of adverse events leading to discontinuation in participants who were
offered statins compared with those who received placebo.
10.1.4.4 Antihypertensive drugs
10.1.4.4.1 Calcium-channel blockers versus placebo
Low-quality evidence from 1 RCT, including 1,442 people living with Alzheimer’s disease,
found no meaningful differences in cognition, measured by ADAS-cog scores, between
people who received calcium-channel blockers and those who received placebo. However,
moderate-quality evidence from the same trial found a smaller decline in cognition, measured
186
by MMSE scores, in people treated by calcium-channel blockers compared with those who
received placebo.
Low-quality evidence from 1 RCT, including 1,636 people living with Alzheimer’s disease,
found no meaningful differences in global assessment, adverse events, or discontinuation
due to adverse events between people who received calcium-channel blockers and those
who received placebo. Moderate-quality evidence from the same trial found higher levels of
serious adverse events in people who received calcium-channel blockers compared with
those who received placebo.
10.1.4.5 Angiotensin II receptor antagonist versus calcium-channel blocker
found no meaningful differences in cognition between people treated by and angiotensin II
receptor antagonist or a calcium-channel blocker.
10.1.4.6 Brain-penetrating angiotensin converting enzyme (ACE) inhibitor versus calcium-

channel blocker
found smaller declines in cognition in people treated by brain-penetrating ACE inhibitors
compared with those treated by calcium-channel blockers.
10.1.4.6.1 Non-brain-penetrating angiotensin converting enzyme (ACE) inhibitor versus calcium-

channel blocker
found smaller declines in cognition in people treated by non-brain-penetrating ACE inhibitors
compared with those treated by calcium-channel blockers.

Relative value of different The committee agreed that, since the aim of these interventions was to
outcomes modify the underlying progression of dementia, the outcome measures
that would provide the most information were measures of cognition.
They noted that comparatively small differences in cognition may prove
to be meaningful, provided they are sustained over a long period of
time.
The original protocol specified that only trials of at least 12 months
duration would be included. However, due to the relatively low number
of RCTs identified, the committee agreed it was appropriate to expand
this to include trials of at least 6 months duration.
Trade-off between Antidiabetic drugs
benefits and harms The committee agreed the included studies did not provide any
evidence that antidiabetic drugs (specifically rosiglitazone) were
effective in slowing the progression of Alzheimer’s disease. The
committee discussed whether the evidence on rosiglitazone was
sufficient to make generic recommendations about all antidiabetic
drugs. It was agreed that clinicians would not offer a drug to patients
without robust assessments of its disease modifying effects. As a result,
the committee felt that a “do not offer recommendation” would apply to
all antidiabetic drugs, as there is no evidence of clinical effectiveness.
187
NSAIDs
The committee noted there was weak evidence of a potentially small
effect on some outcomes with NSAIDs for people with Alzheimer’s
disease. Specifically, there was a small improvement in functional ability
at 12 months. However, the magnitude of this benefit was below the
level defined as being clinically significant, and there was no evidence
of an effect on cognition or other clinical outcomes. The committee
agreed this small difference may have been a direct result of the anti-
inflammatory effects of NSAIDs, and there was no evidence it was
mediated through changes in disease progression.
The committee noted that although absolute rates of adverse events
were similar between people who were treated with NSAIDs and those
who received placebo, more adverse events leading to discontinuation
were observed in people who were given NSAIDs. The committee
agreed that the trend is not uncommon in trials which assess the safety
of NSAIDs. This is because there are certain red-flag adverse events
associated with NSAID treatment which will automatically lead to people
being taken out of trials.
The committee agreed that the evidence was consistent with a class
effect for NSAIDs and therefore it was appropriate that a negative
recommendation be made for the entire class. The committee agreed it
was appropriate to specifically mention aspirin within the
recommendation, as it was included within the evidence base but is not
always recognised as being an NSAID.
Statins
The committee agreed the included studies did not provide any
evidence that statins were effective in slowing the progression of
Alzheimer’s disease. The committee agreed that the evidence was
consistent with a class effect for statins and therefore it was appropriate
that a negative recommendation be made for the entire class. As with
NSAIDs, there was evidence of higher levels of adverse events leading
to discontinuation with treatment, which was agreed to be consistent
with what would be expected in trials of statins in people without
dementia.
Antihypertensive drugs
The committee agreed there was no robust evidence of improvements
in cognition for people living with dementia treated with antihypertensive
drugs. Whilst there was a small positive benefit on the MMSE at 6
months, this effect was not replicated at 12 months, nor was the same
benefit found on the other measure of cognition in the study (the ADAS-
cog). Significant differences in cognition were found between ACE
inhibitors and calcium-channel blockers, but this was based on a small
study in which it was not clear if participants or assessors were blinded.
The committee therefore agreed the balance of the evidence did not
suggest a clear positive benefit with antihypertensive drugs, and
therefore it was appropriate to include them within the ‘do not offer’
recommendation made.
Future research
The committee noted that there is currently ongoing research in a
number of the drug classes included in this review, looking at long-term
effects on cognition in people living with dementia. They therefore
agreed it was appropriate to add a caveat to the recommendation,
allowing for their use as part of randomised controlled trials. Since much
of this research is already ongoing, the committee did not feel it was
appropriate to make a specific research recommendation, as no specific
intervention was found to be promising enough to justify a positive
recommendation for research.
188
Trade-off between net No positive recommendations were made for this review questions, and
health benefits and therefore the committee was not concerned by the lack of economic
resource use evidence identified. Furthermore, they agreed that the
recommendations made would not result in any increase in resource
use.
Quality of evidence The committee noted that all but one of the identified studies only
included people living with Alzheimer’s disease. The committee
considered that it was unlikely that trial participants had mixed
dementias because the diagnostic criteria for Alzheimer’s disease
excludes other types of dementia. Furthermore, some of the studies
explicitly stated that people were excluded if they had modified
Hachinski score greater than 4, ruling out people with a high likelihood
of vascular dementia. As a result, it was considered there was
insufficient evidence to make recommendations on other types of
dementia.
In other areas of this guideline, in particular in areas around non-
pharmacological interventions, evidence has been extrapolated from
Alzheimer’s disease to other forms of dementia, as the committee
agreed that there are some situations where people with similar
symptoms need the same kinds of support, regardless of the underlying
cause of the dementia. However, since the intention of these
interventions is specifically to modify disease progression, and this
effect is likely to differ based on the underlying disease, it was not
appropriate to extrapolate the evidence to other types of dementia in
this context.
The committee noted the absence of studies evaluating non-
pharmacological or behavioural interventions for modifying risk factors
like poor diet, obesity, alcohol consumption and smoking.
Other considerations The committee noted that studies that assessed the effect of
antidiabetic drugs on dementia progression included people without
diabetes at baseline. Abnormal insulin signalling has been identified as
a feature of Alzheimer’s disease. As a result, it was considered that
these studies focused on poor insulin signalling as a risk factor, rather
than diabetes. The committee also noted that studies which assessed
whether statins affected cognitive decline in people with Alzheimer’s
disease included people without hypercholesterolaemia. Furthermore,
some of the studies which assessed the efficacy of antihypertensive
medicines included normotensive people (without primary
hypertension).
The committee agreed it was important to specify in the
recommendation that the negative recommendations made only
considered the use of these treatments for the purposes of slowing
dementia, and people who needed to be treated for a co-morbidity
should continue to receive treatment as normal. To clarify this, the
committee agreed it was appropriate to cross-refer to the relevant NICE
guidelines for the diagnosis and management of diabetes,
hypercholesterolaemia and hypertension.
59. Do not offer the following specifically to slow the progress of Alzheimer's disease,
except as part of a randomised controlled trial:
 diabetes medicines
 hypertension medicines
 statins
 non-steroidal anti-inflammatory drugs (NSAIDs), including aspirin.
189
Cholinesterase inhibitors and memantine for dementia
11 Cholinesterase inhibitors and memantine

for dementia
The three cholinesterase inhibitors, donepezil, rivastigmine, galantamine, and the NMDA
receptor antagonist, memantine, are currently the only effective licensed treatments for
dementia (O’Brien 2017). They are thought to be largely symptomatic agents and although
effects on the underlying disease process have been proposed, there is no convincing
evidence that they modify the disease process in Alzheimer’s or any other type of dementia.
There have been several studies demonstrating their efficacy in Alzheimer’s disease, leading
to their licensing and NICE HTA approval for use in mild to moderate Alzheimer’s disease
(for donepezil, rivastigmine, galantamine) and moderate to severe Alzheimer’s disease (for
memantine). There have also been some studies in other disorders, including dementia with
Lewy bodies and Parkinson’s disease dementia, where cholinesterase inhibitors have been
shown to be effective (Stinton 2015), and frontotemporal and vascular dementia, where they
have generally not (O’Brien 2015, O’Brien 2017). Memantine has been trialled in dementia
with Lewy bodies and Parkinson’s disease dementia, without clear evidence of efficacy as
yet (Stinton 2015).
Historically, initiation of these drugs has been by specialists in dementia, usually based in
secondary care. There has often been ongoing specialist review of these drugs, and often
recommendations that treatment be withdrawn when a certain stage of illness is reached.
However, these drugs have been licenced for over a decade and there is now substantial
experience in their use. They are generally safe and side effects are well recognised.
Although it is important that the drugs are only started following an appropriate, accurate,
specialist diagnosis, the actual initiation and monitoring of these drugs by specialists in
secondary care might no longer be needed.
This chapter includes four important and closely related review questions: 1) who should
start and review the three cholinesterase inhibitors and memantine in people with
Alzheimer’s disease; 2) when and if treatment with these drugs should be withdrawn for
those with Alzheimer’s disease; 3) whether there is evidence for the efficacy of co-
prescription of cholinesterase inhibitors and memantine (which do have different actions);
and 4) whether there is evidence of effectiveness for cholinesterase inhibitors and
memantine in Parkinson’s disease and in other dementias.
190
11.1 Cholinesterase inhibitors and memantine for people living

with Alzheimer’s disease
Review question
 Who should start and review the following pharmacological interventions: (donepezil,
galantamine, rivastigmine, memantine) for people with Alzheimer's disease and how
should a review be carried out?
11.1.1 Introduction
The aim of this review question was to determine which clinicians should prescribe and
review donepezil, galantamine, rivastigmine or memantine for the cognitive symptoms of
dementia in people diagnosed with Alzheimer’s disease. It includes a partial update of the
existing NICE technology appraisal guidance TA217 (Donepezil, galantamine, rivastigmine
and memantine for the treatment of Alzheimer's disease).
A systematic evidence search was conducted (see appendix D) which identified 6344
articles. The titles and abstracts were screened and 66 full-text papers were identified for
inclusion. Sixty three papers were subsequently excluded because they did not fit the
inclusion criteria. Two studies described in 3 papers were presented to the committee
(Aupperle et al., 2000; Aupperle et al., 2003; Watanabe et al., 2012).
A review flowchart is provided in appendix K, and the excluded studies (with reasons for
exclusion) are shown in appendix F.
Table 47: Review summary

Population  People aged 40 years and over with a diagnosis of Alzheimer’s disease
Intervention  The initiation and review of donepezil, galantamine, rivastigmine, memantine
by non-specialists in any setting (for example secondary care; mental health
services; community mental health services, including memory clinics; GP
outreach clinics; primary care)
 Shared-care prescribing protocols
Comparator  The initiation and review of donepezil, galantamine, rivastigmine, memantine
by psychiatrists (including old age psychiatrists and those specialising in
learning disability), neurologists, and physicians specialising in the care of
older people
Outcome  Clinical outcome including cognitive functional and behavioural ability
 Over-prescribing/under-prescribing and potentially avoidable adverse effects
(including hospital admission)
 Medication errors
 Adherence
For full details of the review protocol please see appendix C.
There was no restriction on study design for inclusion in the evidence review. However, it
was anticipated that the most useful study types would be observational designs including
prospective/ retrospective cohort studies. It was expected that the most appropriate design
191
would be a study that compares non-specialist prescribing of these interventions with

specialist prescribing.
The committee was interested in identifying evidence relating to both the prescribing and
reviewing of AChEs and memantine. This is because it was expected that the prescribing of
these medications for people living with Alzheimer’s disease may be carried out by a different
health professional to the person undertaking the review. Evidence associated with these
practices was identified independently.
Two observational studies were included in the evidence review. One study presented in 2
papers provided evidence on the prescribing of donepezil for people living with Alzheimer’s
disease and 1 paper was identified as evidence for reviewing treatment with donepezil.
The quality of evidence for each outcome was considered using the approach recommended
by the Grading Recommendations, Assessment, Development and Evaluation (GRADE)
working group. Due to variations in the way the outcome data were reported by each study,
the evidence statements were presented by intervention/study rather than by outcome.
For a summary of included studies please see Table 48 (for the full evidence tables and full
GRADE profiles please see Appendix E and Appendix G). References for the included
studies are given in appendix I.
A literature review was undertaken by applying standard health economic filters to the clinical
literature searches. In total, 1,049 records were returned; 0 were retained as cost–utility
analyses that addressed the review question.
192
2 Table 48: Summary of included studies

Outcomes of Study
Author (year) Study type Participant details Comparisons interest Length of follow up location
Prescribing donepezil
Aupperle et al. Retrospective Patient characteristics:  Participants being  Clinical outcome  1 year (2000) USA
(2000); Aupperle cohort All patients had received an seen by a primary (including cognitive,  2 years (2003)
et al. (2003) initial evaluation and diagnosis care physician functional,
of Alzheimer’s disease from a  Compared with: behavioural ability)
university diagnostic clinic  Participants being  Access to health
Evaluable total: seen by a member care and social
Original population receiving of a geriatric care support
diagnosis (N=80) psychiatry facility  Concordance and
Participants with 1-year compliance
follow up data:  Patient and carer
(N= 58) experience and
mean age 78.8 years satisfaction
MED (n=31); 
mean age = 82.9 years
GERO (n=27);
mean age = 80.4 years
Participants with 2-year
follow up data:
(N= 39)
mean age 78.4 years
MED (n=22);
mean age = not reported
GERO (n=17);
mean age = not reported
Reviewing donepezil
193
Outcomes of Study
Author (year) Study type Participant details Comparisons interest Length of follow up location
Watanabe et al. Observational Patient characteristics:  Participants  Concordance and  4 weeks Japan
(2012) before-and- The records of patients enrolled into a compliance
after study diagnosed with AD or mixed donepezil  Patient and carer
AD/VaD were followed up with outpatient advisory experience and
the GP service after it was satisfaction
Evaluable total: established
(DOCS)
Total sample (N=111)
Non DOCS (n=59);  Compared with:
mean age = 79.0 years  Participants
enrolled before a
DOCS (n=52);
donepezil
mean age = 77.2 years outpatient advisory
service was
established (non
DOCS)
194
11.1.4.1 Prescribing donepezil (speciality versus non-speciality prescribing)
Very low to low-quality evidence from 1 observational study conducted in the USA in the
1990s with 57 participants found at 1 year follow up the number of people receiving a
prescription of donepezil was significantly lower for people being seen by a primary care
physician compared with those seen by a geriatric psychiatrist.
At 1 year follow up, the study reported a mean Clinical Dementia Rating significantly higher
(indicating more severe dementia) for people being seen by a primary care physician
compared with those being seen by a geriatric psychiatrist. The use of health and social care
support (including number of hospitalisations, use of home health aides and dementia day
care programs), and the mean carer distress rating were not significantly different for people
being seen by a primary care physician compared with those being seen by a geriatric
psychiatrist.
In the same study, at 2 year follow up, (39 participants), the number of people receiving a
prescription of donepezil and the use of health and social care support (including number of
hospitalisations, use of assisted living and residence in nursing homes) were not significantly
different for people being seen by a primary care physician compared with those being seen
by a geriatric psychiatrist.
11.1.4.2 Reviewing donepezil (advisory service versus no advisory service)
Very low-quality evidence from 1 before-and-after study conducted in Japan with 111
participants reported the number of people living with Alzheimer’s disease who were
continuing to use donepezil after 1 year was significantly greater for people using an advisory
consultation service compared with those who had not used this service. The mean duration
of donepezil treatment and mean level of understanding for patients and carers was also
significantly higher for people using the advisory consultation service compared with those
who had not used the service.

Relative value of different The Guideline committee agreed it was important that included
outcomes outcomes considered the impact of medication changes on access to
health and social care support and also reflected outcomes for both
people living with dementia and their carers. The committee recognised
that the outcomes presented in the evidence review were limited and
felt this was consistent with the very low quality of the evidence (see
‘Quality of evidence’ below). The committee noted that the processes
for issuing and dispensing prescriptions differ across primary and
secondary care settings. For example, it was perceived that the issuing
of repeat prescriptions in primary care is likely to be more reactive to
requests from the person living with dementia, whereas the issuing of
prescriptions in specialist services is perceived to be more proactive
when treatment is initiated.
The committee noted that the number of prescriptions dispensed may
not necessarily equate to adherence with prescribed medication, as it
does not indicate whether people take the medicines dispensed.
195
The committee observed for the outcome concordance and compliance,

that the relative risk associated with the number of prescriptions at 2
year follow up did not identify a significant effect; however the primary
data indicated a large difference. The committee considered this
magnitude of effect as potentially important regardless of statistical
significance.
Although, in Aupperle et al. 2003, the authors did not report standard
deviation at 2 year follow up the committee noted that, participants who
were seen by a geriatric psychiatrist experienced an overall slight
improvement in Clinical Dementia Rating (CDR) over 2 years. The
committee thought this would be very unusual, as Alzheimer’s disease
is a degenerative condition.
The committee agreed that the outcomes reported were in line with their
own clinical experience. It was noted that people with dementia in non-
specialist settings may be more likely to stop medications.
Quality of evidence The committee agreed that the evidence presented was very low quality
and noted the methodological limitations of the identified studies.
The committee agreed that the identified research evidence would not
necessarily reflect current practice in the UK for the use of AChEs. It
was noted that the studies were conducted during the 1990s, when
clinicians were much less familiar with AChEs. The included studies
were also conducted overseas where healthcare systems and services
differ to practice in the UK.
For Aupperle et al. (2000) and Aupperle et al. (2003) the committee
noted that the observational design of the studies meant that there was
a high risk of selection bias. The committee further noted that the
observational design of Aupperle et al. (2003) meant there was a lack of
interpretable findings on reasons for attrition, making it difficult to infer
whether attrition was a consequence of adverse effects or lack of
efficacy.
The committee acknowledged the limitations of the Watanabe et al.
(2012) study. They agreed the observational design, small sample size,
short follow up and selective reporting reflected that the study was very
low quality.
Trade-off between The committee discussed the evidence base and agreed that they
benefits and harms would be unable to make recommendations based solely on the
reported outcomes.
The committee raised concerns about the lack of evidence identified in
relation to the initiation of AChEs and memantine but agreed that
initiation is implicitly linked to diagnosis. It noted that recommendation
1.1 and 1.2 of TA217 imply that a diagnosis is needed before treatment
can be initiated. The committee agreed that the purpose of memory
clinics is not solely to prescribe AChEs and memantine but to provide
specialist assessment in diagnosing, treating and supporting people
living with dementia. The committee noted that the current guideline
suggests that diagnosis should be made by a specialist (CG42 1.4.3.1),
and that this will be subject to a separate evidence review as part of the
ongoing update.
The committee acknowledged the practical issues around the
mechanisms for prescribing, dispensing and monitoring medication
adherence. Committee members raised concerns that people may have
to wait for a diagnosis before they can start treatment but the committee
agreed that there should not be an artificial barrier preventing the
transfer of care between specialist and non-specialist healthcare
settings. The committee was keen to ensure that AChEs and
memantine were only initiated following a diagnosis and those treatment
recommendations are made by a clinician with appropriate specialist
expertise. However, it acknowledged the difficulties that sometimes
arise where the diagnosing clinician is required to issue the first
prescription for an AChE or memantine. The committee acknowledged
196
that licensing for AChEs and memantine (as set out in each product’s
Summary of Product Characteristics; SPC) is clear about initiation and
supervision of these drugs and therefore agreed that it was appropriate
to reflect this in the recommendations. The committee noted that the
SPCs for each of the AChEs and memantine make reference to
initiation and supervision of treatment by specialist physicians.
However, it noted that the wording of these SPCs pre-dates legislative
changes, in the early 2000s, which authorised the use of non-medical
prescribers. The committee agreed that any interpretation of the
recommendations would need to take account of this different
prescribing environment. For this reason, the committee thought it was
not necessary to stipulate that treatment should be initiated by
physicians (i.e. doctors) alone, and preferred to emphasise that the
prescriber starts treatment on advice from a healthcare professional
with specialist experience, regardless of professional label. The
committee also agreed it was important to note that once a decision has
been made to initiate therapy, then prescriptions can be made in
primary care.
The committee discussed their concerns over communication of
information between specialist and non-specialist settings and agreed
that reference to NICE’s Medicines Optimisation guideline (NG5) would
be helpful. The committee discussed recommendation 1.2 in the NICE
Medicines Optimisation guideline which considers medicines-related
communication systems where patients move between care settings,
which is of particular relevance. However, following further discussion, it
was agreed that reference to all of NG5 would be more appropriate.
When considering the monitoring and review of these drugs, the
committee noted and agreed that an annual dementia review is
mandated. They agreed that these drugs should be part of the annual
dementia review as opposed to a standard medicines review. The
committee noted again that it would be appropriate to refer to the
Medicines Optimisation NICE guidance with regard to medication
review, and the arrangements that should be in place between different
care settings (in this instance, secondary and primary care).
Trade-off between net No published health economic evidence was identified for this review
health benefits and question. The committee noted that, in the past (including when TA217
resource use was published), the medicines under consideration all had proprietary
status, but they are all now available in generic formulations. This
change has been accompanied by a significant fall in the acquisition
costs of the drugs. The committee felt that, if cost containment had been
a motivating factor in restricting prescribing to people with specialist
experience of Alzheimer’s disease, this was no longer such a
substantial concern. However, the committee emphasised that other
reasons for involving specialists remain relevant.
Other considerations It was noted that the current recommendations make reference to
carers’ views. The committee agreed that this is an important
consideration. However following discussion it was agreed that it could
be adequately addressed by cross-reference to NICE’s Medicines
Optimisation guideline (NG5), which gives detailed guidance on the
need to involve carers in the diagnosis, management and treatment of
individuals. It was also agreed to be appropriate to add addition cross-
references to the NICE guidelines on managing medicines for adults
receiving social care in the community, and managing medicines in care
homes, in particular because these guidelines includes specific
recommendations around covert administration, which will be relevant
for some people living with dementia. This guideline will be relevant for
people with all dementia subtypes, not only those with Alzheimer’s
disease included in this review.
This section also includes the recommendations directly incorporated in
to this guideline from NICE technology appraisal 217.
197
Recommendations shaded in grey are taken directly from NICE technology appraisal
guidance 217 and were not updated as part of this guideline, and therefore no changes to
these recommendations can be made.
60. The three acetylcholinesterase (AChE) inhibitors donepezil, galantamine and

rivastigmine as monotherapies are recommended as options for managing mild to
moderate Alzheimer's disease under all of the conditions specified in 62 and 63.
61. Memantine monotherapy is recommended as an option for managing Alzheimer's

disease for people with:
 moderate Alzheimer's disease who are intolerant of or have a
contraindication to AChE inhibitors or
 severe Alzheimer's disease.
Treatment should be under the conditions specified in recommendation 6.
62. Treatment should be under the following conditions:

 For people who are not taking an AChE inhibitor or memantine,
prescribers should only start treatment with these on the advice of a
clinician who has the necessary knowledge and skills. This could
include:
 secondary care medical specialists such as psychiatrists, geriatricians
and neurologists
 other healthcare professionals (such as GPs, nurse consultants and
advanced nurse practitioners), if they have specialist expertise in
diagnosing and treating Alzheimer’s disease.
 Once a decision has been made to start cholinesterase inhibitors or
memantine, the first prescription may be made in primary care.
 Ensure that local arrangements for prescribing, supply and treatment
review follow the NICE guideline on medicines optimisation.
63. If prescribing an AChE inhibitor (donepezil, galantamine or rivastigmine),

treatment should normally be started with the drug with the lowest acquisition
cost (taking into account required daily dose and the price per dose once shared
care has started). However, an alternative AChE inhibitor could be prescribed if it
is considered appropriate when taking into account adverse event profile,
expectations about adherence, medical comorbidity, possibility of drug
interactions and dosing profiles.
64. When using assessment scales to determine the severity of Alzheimer's disease,
healthcare professionals should take into account any physical, sensory or
learning disabilities, or communication difficulties that could affect the results and
make any adjustments they consider appropriate. Healthcare professionals should
also be mindful of the need to secure equality of access to treatment for patients
from different ethnic groups, in particular those from different cultural
backgrounds.
65. When assessing the severity of Alzheimer's disease and the need for treatment,
healthcare professionals should not rely solely on cognition scores in
circumstances in which it would be inappropriate to do so. These include:
198
 if the cognition score is not, or is not by itself, a clinically appropriate tool

for assessing the severity of that patient's dementia because of the
patient's learning difficulties or other disabilities (for example, sensory
impairments), linguistic or other communication difficulties or level of
education or
 if it is not possible to apply the tool in a language in which the patient is
sufficiently fluent for it to be appropriate for assessing the severity of
dementia or
 if there are other similar reasons why using a cognition score, or the
score alone, would be inappropriate for assessing the severity of
dementia.
In such cases healthcare professionals should determine the need for initiation or
continuation of treatment by using another appropriate method of assessment.
66. For guidance on managing medicines (including covert administration), see the
NICE guidelines on managing medicines for adults receiving social care in the
community and managing medicines in care homes.
199
11.2 Co-prescription and withdrawal of cholinesterase inhibitors

and memantine in Alzheimer’s disease
Review questions
 How effective is the co-prescription of cholinesterase inhibitors and memantine for the
treatment of Alzheimer’s disease?
 When should treatment with donepezil, galantamine, rivastigmine, memantine be
withdrawn for people with Alzheimer’s disease?
11.2.1 Introduction
The aim of these review questions was to determine the effectiveness and cost-effectiveness
of memantine plus a cholinesterase inhibitor for cognitive enhancement in Alzheimer’s
disease and to determine the clinically appropriate points to withdraw treatment with
donepezil, galantamine, rivastigmine and memantine for people with Alzheimer’s disease.
The review identified studies that fulfilled the conditions specified in Table 49 and Table 50.
For full details of the review protocols, see Appendix C.
Table 49: Review summary: co-prescription of cholinesterase inhibitors and

memantine for the treatment of Alzheimer’s disease
Population People with a diagnosis of Alzheimer’s disease
Interventions Memantine plus a cholinesterase inhibitor
Comparator  Memantine
 Cholinesterase inhibitors
 Placebo
 Dose escalation is a possible alternative to co-prescription
ability
 Adverse events
Table 50: Review summary: withdrawal of cholinesterase inhibitors and memantine for
people living with Alzheimer’s disease
Population People aged (40 years and over) with a diagnosis of Alzheimer’s
disease and currently being treated with donepezil, galantamine,
rivastigmine and/or memantine
Interventions Withdrawal of pharmacological treatment
Explicit stopping rule for pharmacological treatment
Comparator Continuation of previous treatment
Change of treatment drug (to another of the specified 4 drugs)
Change of treatment dose
Alternative stopping rules
Outcomes Clinical outcomes including cognition, function, behaviour and
neuropsychiatric symptoms
Adverse events
Patient and carer experience and satisfaction
Patient and carer health-related quality of life
200
Resource use and costs
Two separate systematic searches were undertaken to address the 2 questions in this
section.
For the co-prescription of cholinesterase inhibitors and memantine a systematic search

identified 1,914 references. Fifty three references were ordered for full text review after
screening upon title and abstract and 8 papers were included in the final review.
Five papers (Howard, 2012; Tariot 2004; Dysken 2014; Porsteinsson 2008 and Grossberg
2013) recruited people who were currently receiving a cholinesterase inhibitor as treatment
for their Alzheimer’s disease and involved randomisation to co-prescription with memantine
or placebo memantine. To assist with analysis, additional data was sought from the authors
of the DOMINO-AD trial (Howard, 2012; please see Appendix E for this information). Two
papers (Araki 2014; Choi 2011) compared co-prescription of a cholinesterase inhibitor to
monotherapy with the same cholinesterase inhibitor. One paper (Shao 2015) randomised
people with mild to moderate Alzheimer’s disease to different treatment arms for each
cholinesterase inhibitor and compared these to placebo. For a detailed list of excluded
studies and reasons for exclusion see Appendix F.
For discontinuation of cholinesterase inhibitors and memantine, the search identified 1,242
references. The references were screened on their titles and abstracts and 35 references
were ordered for full text review. Thirty two papers were subsequently excluded because
they did not fit the inclusion criteria (see Appendix F for a detailed list of excluded studies
and reasons for their exclusion). Three papers were included in the evidence review. A
further 3 papers included outcomes for people discontinued from cholinesterase inhibitors,
but these did not meet the criteria for the review as they did not include a population of
people taking a cholinesterase inhibitor at start of the study, but rather people were included
in the study, started on treatment and then had the treatment withdrawn as part of the same
study.
All 3 included papers involved randomisation to withdrawal of a cholinesterase inhibitor and

replacement with placebo. One study reported in 2 papers (Howard et al., 2012; Howard et
al., 2015) additionally randomised people within each arm to receive active or placebo
memantine, creating a 2x2 factorial trial design. The other study (Herrmann et al., 2016)
recruited people on a range of cholinesterase inhibitors (rivastigmine, galantamine,
donepezil) and reported outcomes without stratifying by the specific cholinesterase inhibitor
which had been continued or withdrawn.
A summary of the characteristics of the included studies is provided in and Table 51 and
Table 52. Data from the included studies were extracted into evidence tables. See Appendix
E for the full evidence tables, and for the full GRADE profiles see Appendix G. References
for the included studies are given in appendix I.
Results for co-prescription versus cholinesterase inhibitor monotherapy are presented in 2

separate ways. The first analysis uses the full dataset, and stratifies the analysis into mild-to-
moderate, and moderate-to-severe populations, as each of the included trials recruits 1 of
those 2 subpopulations. The second analysis subdivides the population into separate mild,
moderate and severe subgroups, using additional information obtained from a Cochrane
review of memantine in dementia, and from the authors of the DOMINO-AD trial. It was not
possible to obtain data for all outcomes of all trials in this format, and therefore this analysis
contains only a subset of the participants included in the full analysis.
201
2 Table 51: Included studies for co-prescription of cholinesterase inhibitors and memantine
Study Intervention and
reference Study type Study population comparator Relevant outcomes Comments
Donepezil plus memantine versus donepezil plus placebo
Howard 2012 Randomised controlled Moderate to severe Combination of donepezil  SMMSE Follow up 52 weeks
trial Alzheimer’s disease 10mg / day plus memantine  Bristol ADL scale Location: UK
5mg/ day increasing to
 NPI
20mg/day versus placebo
 DEMQOL proxy
 GHQ-12
Tariot 2004 Randomised controlled Moderate to severe Participants currently  ADCS-ADL Follow up 24 weeks
trial Alzheimer’s disease receiving donepezil  CIBIC plus Location: USA
additionally received
 NPI
memantine 5mg/day
increasing to 20mg/day  Severe impairment Battery
versus placebo  Behaviour rating scale for
Geriatric Patients
Any Cholinesterase inhibitor plus memantine versus placebo
Dysken 2014 Randomised controlled Mild to moderate One arm of TEAM AD (trial of  ADCS-ADL Follow up ranged
trial possible or probable vitamin A and memantine in  MMSE from 6 months to 2
Alzheimer’s disease Alzheimer’ disease) years
 ADAS-cog
Participants currently Location USA
receiving cholinesterase  NPI
14 VA centres
inhibitors additionally received  Caregiver Activity survey
memantine (titrated to 20 mg/  Any adverse events
day) versus placebo
Porsteinsson Randomised controlled Mild to moderate Participants currently  ADAS-Cog Follow up 24 weeks
2008 trial probable Alzheimer’s receiving cholinesterase  ADCS-ADL Location: USA
disease inhibitors additionally received
 NPI
memantine 20mg/ day
versus placebo  MMSE
 Adverse events
202
Study Intervention and

reference Study type Study population comparator Relevant outcomes Comments
Grossberg Randomised controlled Moderate to severe MEM-MD-50 Participants  ADCS-ADL Follow up 24 weeks
2013 trial probable Alzheimer’s currently receiving  Severe Impairment Battery Location :
disease cholinesterase inhibitors multinational
 CIBIC plus
additionally received once Study dose
daily (28mg) extended  NPI
memantine was a
release memantine versus  Verbal fluency test higher dose
placebo  Any adverse events extended release
formulation
Donepezil plus memantine versus donepezil only
Araki 2014 Randomised controlled Moderate to severe Participants currently  MMSE Follow up 24 weeks
trial Alzheimer’s disease receiving donepezil  NPI Location: Japan
additionally received
 Japanese Zarit Burden
memantine 5mg/ day
Interview
increasing to 20mg/day
versus donepezil only  Clinical Global Impression-
Improvement
Rivastigmine plus memantine versus rivastigmine only
Choi 2011 Randomised open label Moderate to severe Combination of rivastigmine  Korean MMSE Follow up 16 weeks
trial probable Alzheimer’s transdermal 10cm patch plus  ADAS-Cog Location: South
disease (NINCDS- memantine 5mg/ day Korea
 NPI (carers assessment)
ADRDA) increasing to 20mg/day Study conducted in
versus rivastigmine 10cm  Frontal Assessment Battery
26 centres
transdermal patch  ADCS- ADL
monotherapy  CDR-SB
 Koran CMAI
 Safety and tolerability
Donepezil or galantamine or rivastigmine plus memantine versus memantine only
Shao 2015 Randomised controlled Mild to moderate Memantine plus donepezil  MMSE Follow up 24 weeks
trial Alzheimer’s disease Memantine plus rivastigmine  ADCS ADL Location: China
Memantine plus rivastigmine  Incidence of adverse events
Versus memantine plus
placebo
203
1 Table 52: Included studies for withdrawal of cholinesterase inhibitors or memantine

Study Study Intervention and Relevant
reference Study type population comparator outcomes Comments
Donepezil withdrawal
Howard 2012 2x2 factorial RCT Community-dwelling Group 1: donepezil  Standardised MMSE 52-week study
people with continuation (10mg/day) plus  Bristol Activities of Daily Living
moderate to severe initiation of placebo Scale
Alzheimer’s disease memantine (from week 1)
 Neuropsychiatric Inventory
 DEM-QOL-proxy
Group 2: donepezil
discontinuation (donepezil at  Carer health status (GHQ-12)
Howard 2015 5mg for weeks 1-4, placebo  Nursing home placements Up to 4 years’ follow-
(Further donepezil thereafter ) plus up
analysis of initiation of placebo
Howard 2012) memantine (from week 1)
Group 3: donepezil
discontinuation (donepezil at
5mg for weeks 1-4, placebo
donepezil thereafter) plus
initiation of memantine (5mg
from week 1, increasing in 5
mg increments to reach 20
mg from week 4 onwards)
Group 4: donepezil
continuation plus initiation of
memantine (5mg from week
1, increasing in 5 mg
increments to reach 20 mg
from week 4 onwards)
Cholinesterase inhibitor withdrawal (assorted treatments)
Herrmann Randomised controlled Institutionalised Continuation of existing  Clinician's Global Impression 8-week study
2016 trial people with cholinesterase inhibitor of Change (CGI/CGI-C) recruiting from 2
(rivastigmine, galantamine,
204
Study Study Intervention and Relevant

reference Study type population comparator outcomes Comments
moderate to severe donepezil) vs withdrawal and  Standardised Mini Mental long-term care
Alzheimer’s disease switch to placebo State Examination (sMMSE) facilities
 Severe Impairment Battery
(SIB)
 Neuropsychiatric inventory -
Nursing Home version (NPI-
NH)
 Cohen-Mansfield Agitation
Inventory (CMAI)
 Apathy Evaluation Scale
(AES)
 Alzheimer's Disease Co-
operative Study - Activities of
Daily Living Inventory,
modified for severe AD
(ADCS-ADL-sev)
 Quality of Life in Late-Stage
Dementia (QUALID)
205
As for the cost-effectiveness searches, 2 separate systematic searches were undertaken to

address the 2 questions in this section. For full details of the search strategies, please see
Appendix D.
11.2.3.1 Co-prescription of cholinesterase inhibitors and memantine
The search identified 820 references. After screening on title and abstract, 12 papers were
ordered for full-text review; following detailed perusal, 5 cost–utility analyses (CUAs) were
included in the final review.
Three CUAs were closely related: Lachaine et al. (2011) developed a Markov model
comparing co-prescription of cholinesterase inhibitors (AChEs) and memantine with AChE
monotherapy from a Canadian perspective, whilst Pfeil et al. (2012) and Touchon et al.
(2014) replicated the analysis from Swiss and French perspectives, respectively. The models
adopted a 3-state structure: the modelled cohorts began in community-based care, with a
proportion progressing to full-time residential care and the possibility of transition to death
from either state. The probability of requiring full-time residential care was estimated from an
observational (retrospective cohort) study, based on a US population, which found that
people who had received combination therapy were significantly less likely to enter care than
people who had received AChE monotherapy (Lopez et al., 2014). Economic evidence
profiles for all three CUAs are available in Appendix M.
All 3 analyses were judged to be partially applicable with very serious limitations. Each
concluded that combination therapy dominates AChE monotherapy (that is, it is both less
expensive and more effective). All 3 CUAs report setting-specific costs from a ‘healthcare’
perspective and a ‘societal’ perspective. Neither of these is consistent with the NICE
reference case, as the ‘healthcare’ perspective omits relevant social care costs, and the
‘societal’ perspective includes these costs but also encompasses estimates of informal carer
costs. However, results are not qualitatively different between the 2 perspectives, so it can
be inferred that a perspective that sits between the 2 would reach the same conclusions.
Weycker et al. (2007) developed a patient-level cohort ‘micro-simulation’ model, simulating

the costs and effects of combination therapy with memantine and donepezil compared with
donepezil monotherapy in moderate-to-severe Alzheimer’s disease from a US perspective.
Treatment effects for combination therapy were based on severe impairment battery (SIB)
results from an RCT comparing memantine and donepezil with donepezil monotherapy that
is included in our clinical review (Tariot et al. 2004; see above). However, the effects of
donepezil were not drawn from the same RCT; rather, they were based on the donepezil arm
of a different, placebo-controlled RCT (Feldman et al., 2001). The authors justify this decision
by arguing that the benefits of donepezil observed in the Tariot et al. RCT ‘are probably an
artefact of trial participation and not donepezil therapy per se’. SIB results in these 2 trial
arms are used to estimate costs, utility and probability of entry to full-time care, via a
mapping to MMSE which is, in turn, used to approximate CDR (and, in the case of HRQoL,
mapped once again to HUI3).
The CUA was assessed as partially applicable with very serious limitations. It concluded that,
over their lifetime, an average patient in the cohort would spend around 2 years in a
community setting, followed by a little under 4 years in full-time care, and such a person
would accrue a negative QALY aggregate over this period (that is, their remaining quality-
adjusted life expectancy should be considered a substantially worse prospect than
206
immediate death). Incremental results suggest that combination therapy dominates donepezil
monotherapy (that is, it is both less expensive and more effective).
The final CUA, reported by Knapp et al. (2016), was a UK trial-based analysis conducted
alongside the DOMINO-AD RCT, which is described above (Howard et al., 2012). The
analysis used directly reported carer-rated EQ-5D results to estimate QALYs over the 1-year
duration of the trial. Resource-use data were collected using a comprehensive inventory of
health and social care support. Although the trial randomised people to 4 mutually exclusive
strategies (donepezil, memantine, donepezil and memantine combination, or placebo), the
paper reports incremental results for 3 comparisons that the investigators had pre-specified
as of interest: donepezil or combination versus memantine or placebo, combination versus
donepezil and memantine versus placebo. The second comparison is of interest for the co-
prescription review (see below for analyses of interest to the discontinuation question).
As for the effectiveness review, the investigators shared results stratified by baseline
severity, which enabled the committee to consider separate (moderate and severe
Alzheimer’s disease [AD]) populations. It also made it possible to assess evidence for
comparisons not reported in the paper – for example, combination therapy compared with
memantine monotherapy (which is a particularly relevant comparison in people with severe
AD, given that memantine is licensed in this population, but donepezil is not). Having access
to the stratified data had the substantial advantage that the committee could consider
analyses of direct relevance to the decision problems with which it was faced. However, it
came with the disadvantage that, as data had been subdivided into smaller subsets, it was
not feasible to adjust the analyses for the baseline characteristics of the participants, as had
been done in the published analysis.
The CUA was assessed as directly applicable with minor limitations. It had the following
results:
 In moderate disease, combination therapy was associated with increased costs
(£1,310 [95%CI: −£3,021 to £5,641]) and reduced quality of life (−0.07 [95%CI: −0.22 to
0.08]) compared with donepezil monotherapy. However, these estimates were subject to
substantial uncertainty, as can be seen in the broad confidence intervals with which the
disaggregated values are associated.
 In the severe AD stratum,
o combination therapy reduced costs (−£1,658 [95%CI: −£6,399 to £3,082]) and
increased QALYs (0.11 [95%CI: −0.06 to 0.28]) compared with memantine
monotherapy
o combination therapy reduced costs (−£240 [95%CI: −£4,759 to 4,279]) and increased
QALYs (0.11 [95%CI: −0.06 to 0.28]) compared with donepezil alone
Once again, these estimates were subject to substantial uncertainty, as seen in the broad
confidence intervals.
 Probabilistic analysis is only available in the published analysis, which combines
moderate and severe AD. This suggested that there is an exactly 50:50 chance that
combination therapy or donepezil monotherapy is optimal, if QALYs are valued at £20,000
each.
11.2.3.2 Discontinuation of cholinesterase inhibitors and memantine
The search identified 346 references. After screening on title and abstract, 8 papers were
ordered for full-text review; following detailed perusal, 1 was included in the final review.
The one included CUA was Knapp et al.’s, within-trial analysis for the DOMINO-AD RCT
(2016), as described above. For the discontinuation question, the comparisons of interest are
207
donepezil vs placebo and memantine vs placebo in both the moderate and severe
subgroups.
The CUA was assessed as directly applicable with minor limitations. It had the following
results:
 In moderate disease,
o donepezil monotherapy reduced costs (£974 [95%CI: −£2,383 to £4,332]) and
increased QALYs (0.26 [95%CI: 0.08 to 0.45]) compared with placebo
o memantine monotherapy reduced costs (−£1,472 [95%CI: −£4,273 to £1,329]) and
increased QALYs (0.18 [95%CI: −0.02 to 0.37]) compared with placebo
 In severe disease,
o donepezil monotherapy reduced costs (−£5,711 [95%CI: −£19,015 to £7,592]) and
o memantine monotherapy reduced costs (−£4,293 [95%CI: −£17,677 to £9,091]) and
These results were typically associated with substantial uncertainty. The one finding that
appeared reliable, at a 95% confidence level, is that discontinuing donepezil, in people with
moderate AD, results in a loss of QALYs.
11.2.4.1 Cholinesterase inhibitor plus memantine versus cholinesterase inhibitor plus placebo
11.2.4.1.1 Full population
Low- to high-quality evidence from up to 5 RCTs containing 1,875 people living with
Alzheimer’s disease found global functioning (CIBIC plus), behavioural and psychological
symptoms (NPI), care dependency (Behaviour rating scale for geriatric patients- care
dependency subscale), verbal fluency (VFT) and activities of daily living (ADCS-ADL/BADLS)
were significantly improved in people treated with combination therapy of cholinesterase
inhibitors plus memantine compared with treatment with a cholinesterase inhibitor plus a
placebo, but could not differentiate cognition (ADAS-cog; MMSE), global assessment (SIB),
health-related quality of life (DEMQOL), global health (Global health questionnaire), carer
activity (Caregiver activity survey), adverse events, serious adverse events, discontinuations
due to adverse events, entry to long term care or mortality.
11.2.4.1.2 Mild to moderate
Low- to moderate-quality evidence from up to 3 RCTs containing 740 people living with mild
to moderate Alzheimer’s disease could not differentiate cognition (ADAS-cog; MMSE),
activities of daily living (ADCS-ADL/BADLS), global functioning (CIBIC plus), behavioural and
psychological symptoms (NPI), health-related quality of life (DEMQOL), adverse events,
serious adverse events, discontinuations due to adverse events or mortality between people
treated with combination therapy of cholinesterase inhibitors plus memantine compared with
treatment with a cholinesterase inhibitor plus a placebo.
11.2.4.1.3 Moderate to severe
Very low- to high-quality evidence from up to 4 RCTs containing 1,166 people living with
moderate to severe Alzheimer’s disease found activities of daily living (ADCS-ADL/BADLS),
global functioning (CIBIC plus), behavioural and psychological symptoms (NPI), care
dependency (Behaviour rating scale for geriatric patients- care dependency subscale) and
verbal fluency (VFT) were significantly improved in people treated with combination therapy
208
of cholinesterase inhibitors plus memantine compared with treatment with a cholinesterase

inhibitor plus a placebo, but could not differentiate cognition (MMSE), global assessment
(SIB), health related quality of life (DEMQOL), global health (Global health questionnaire),
adverse events, serious adverse events, discontinuations due to adverse events or carer
outcomes (Caregiver activity survey).
Economic evidence
One directly applicable cost–utility analysis with minor limitations found that combination
therapy is associated with small, uncertain increases in both costs and QALYs, with a base-
case ICER of £19,967. The probability that either approach is optimal is 50%, if QALYs are
valued at £20,000 each.
11.2.4.1.4 Mild only
Low- to moderate-quality evidence from up to 2 RCTs containing 315 people living with mild
Alzheimer’s disease could not differentiate global assessment, cognitive function or activities
of daily living between people treated with combination therapy of cholinesterase inhibitors
plus memantine compared with treatment with a cholinesterase inhibitor plus a placebo.
11.2.4.1.5 Moderate only
Low- to moderate-quality evidence from up to 4 RCTs containing 663 people living with
moderate Alzheimer’s disease found cognitive function was significantly improved in people
treated with combination therapy of cholinesterase inhibitors plus memantine compared with
treatment with a cholinesterase inhibitor plus a placebo, but could not differentiate global
assessment, activities of daily living, behavioural and psychological symptoms (NPI), health-
related quality of life (DEMQOL) or global health (Global health questionnaire).
Economic evidence
One directly applicable cost–utility analysis with minor limitations found that donepezil
monotherapy may be associated with lower costs and more QALYs than combination
therapy with donepezil and memantine. However, this result is subject to substantial
uncertainty, such that the data are consistent with either approach providing better value
than the other.
11.2.4.1.6 Severe only
Moderate- to high-quality evidence from up to 3 RCTs containing 218 people living with
severe Alzheimer’s disease found cognitive function, activities of daily living and behavioural
and psychological symptoms (NPI) were significantly improved in people treated with
combination therapy of cholinesterase inhibitors plus memantine compared with treatment
with a cholinesterase inhibitor plus a placebo, but could not differentiate global assessment,
health-related quality of life (DEMQOL) or global health (Global health questionnaire).
Economic evidence
One directly applicable cost–utility analysis with minor limitations found that combination
therapy with donepezil and memantine may be associated with lower costs and more QALYs
than either agent as monotherapy. However, this result is subject to substantial uncertainty,
such that the data are consistent with any approach providing best value.
209
11.2.4.1.7 Cholinesterase inhibitor plus memantine versus cholinesterase inhibitor monotherapy
Very low- to moderate-quality evidence from up to 2 RCTs containing 183 people living with
Alzheimer’s disease found global assessment (Clinical Global Impression-Improvement),
cognition (Clock drawing Test), behavioural and psychological symptoms (NPI) and carer
burden (Zarit Burden Interview) were significantly improved in people treated with
combination therapy of cholinesterase inhibitors plus memantine compared with
cholinesterase inhibitor monotherapy, but could not differentiate cognition (MMSE; ADAS-
cog), behavioural and psychological symptoms (carer administered), dementia severity
(Clinical Dementia rating–sum of boxes; Frontal Assessment Battery), agitation (Cohen
Mansfield Agitation Inventory), any adverse events or any serious adverse events.
When data was stratified by baseline dementia severity, cognition (MMSE) was significantly
improved for people living with moderate–severe Alzheimer’s disease, but could not be
differentiated for people living with mild-to-moderate Alzheimer’s disease.
Economic evidence
Four partially applicable cost–utility analyses with very serious limitations found that, in
people with moderate–severe AD, combination therapy with a cholinesterase inhibitor and
memantine dominates cholinesterase inhibitor monotherapy (that is, combination therapy is
associated with lower costs and more QALYs). All 4 analyses were funded by the
manufacturer of memantine, and each derived its estimates of effect from non-randomised
comparisons.
11.2.4.1.8 Cholinesterase inhibitor plus memantine versus memantine plus placebo
Very low to low-quality evidence from 1 RCT with 88 people living with mild to moderate
Alzheimer’s disease could not differentiate cognition (MMSE); activities of daily living (ADCS-
ADL) or number of adverse events for people treated with combination therapy of
cholinesterase inhibitors plus memantine compared with people treated with memantine plus
a placebo.
Economic evidence
One directly applicable cost–utility analysis with minor limitations found that, in people with
severe AD, combination therapy with donepezil and memantine may be associated with
lower costs and more QALYs than memantine monotherapy. However, this result is subject
to substantial uncertainty, such that the data are consistent with either approach providing
better value than the other.
11.2.4.2 Cholinesterase inhibitor withdrawal
Low- to moderate-quality evidence from up to 2 RCTs containing 148 people living with
moderate to severe Alzheimer’s disease could not differentiate cognition, activities of daily
living, behavioural and psychological symptoms, quality of life or rates of entry to long term
care between people continuing with or discontinuing from cholinesterase inhibitors.
Significant benefits from continuation were found for cognition in people with moderate
Economic evidence
One directly applicable cost–utility analysis with minor limitations found that
210
 in people with moderate AD, switching from donepezil to placebo reduces QALYs and
may also increase costs, meaning it is likely to be dominated by continued therapy
 in people with severe AD, switching from donepezil to placebo may reduce QALYs and
may also increase costs, meaning it may be dominated by continued therapy
11.2.4.3 Cholinesterase inhibitor switch to memantine
Low- to moderate-quality evidence from 1 RCT containing 105 people living with moderate to
severe Alzheimer’s disease could not differentiate cognition, activities of daily living,
behavioural and psychological symptoms, quality of life or rates of entry to long term care
between people continuing with cholinesterase inhibitors or switching to memantine.
Significant benefits from switching were found for behavioural and psychological symptoms
in people with severe Alzheimer’s disease.
Economic evidence
One directly applicable cost–utility analysis with minor limitations could not differentiate costs
and QALYs between switching from donepezil to memantine or continuing donepezil, in
people with severe AD.
11.2.4.4 Memantine withdrawal
No evidence was identified looking at the outcomes of memantine withdrawal in people living
with dementia.

Relative value of different The committee noted that the primary outcome measures in most of the
outcomes trials was cognitive function (using one of the MMSE, ADAS-cog and
SIB). They recognised that in designing trials, these instruments are
used as a measure of identifying disease severity. However, in practice,
clinicians are more likely to consider global functioning, quality of life,
and the impacts on people’s daily activities. In deciding whether to
continue treatment, the committee agreed it was important to consider
the overall benefit of treatment, rather than simply focusing upon
cognitive benefit.
The committee also agreed that it was important to consider evidence
on behavioural symptoms, and in particular agitation. There is evidence
from the literature on memantine monotherapy that it may have positive
effects on agitation, and some evidence that cholinesterase inhibitors
may worsen agitation in some individuals. These potential effects are
also important to consider.
Trade-off between Discontinuation of cholinesterase inhibitors
benefits and harms The committee noted there was clear evidence of harm from
discontinuing cholinesterase inhibitors in people with moderate
Alzheimer’s disease, with a substantial worsening in cognitive function.
The committee noted that the trials effectively used an MMSE score as
the trigger for discontinuation, and that decisions in clinical practice
were likely to be more complex than this, and cover a range of
outcomes for the person living with Alzheimer’s disease. They therefore
agreed the evidence demonstrated that it was not possible to use a set
cut-off for disease severity to decide when treatment should be
discontinued (as no effective cut-off to use was found). This was
supported by the primary analysis of the DOMINO-AD study, which
found significant worsening on cholinesterase inhibitor withdrawal. The
committee therefore agreed it was appropriate to recommend that
211
disease severity should not be used as a reason for treatment

discontinuation.
The committee agreed that in practice there may be other reasons why
treatments are discontinued (e.g. the person is unable or unwilling or
distressed by the process of taking the medicine, they are suffering side
effects, or they are entering a terminal phase of the illness and harms
are felt to outweigh the benefits of treatment). However, it was agreed
these discussions were part of normal clinical decision-making, it was
not necessary to make any specific recommendations about them.
Co-prescription of cholinesterase inhibitors and memantine
The committee agreed that is was important to separate the population
into those with mild Alzheimer’s disease and those with
moderate/severe disease when considering this question, as the
evidence on memantine monotherapy suggests that memantine is not
effective in mild Alzheimer’s disease.
The committee agreed this same pattern was present in the evidence
on co-prescription, with no evidence of effect in the mild subgroup, but
improvements in cognitive function and global functioning in both the
moderate and severe populations (and additionally, improvements in
activities of daily living in the severe population). The magnitudes of the
effects were approximately half of those seen for cholinesterase
inhibitor monotherapy versus placebo in treatment naive individuals, but
were still at a level that would be likely to be clinically meaningful,
particularly given that there was no evidence of an increase in adverse
events from co-prescription.
The committee therefore agreed it was appropriate to make positive
recommendations for co-prescription in both moderate and severe
Alzheimer’s disease. A stronger recommendation was made for severe
Alzheimer’s disease compared with moderate disease for two reasons:
firstly, the magnitudes of effects seen were larger in people with more
severe Alzheimer’s disease; and secondly, the trials in moderate to
severe Alzheimer’s disease only tended to recruit people at the more
severe end of the moderate category, and therefore the evidence of
benefit is less clear in people at the milder end of the moderate
category.
Switching from cholinesterase inhibitors to memantine
The committee considered the risks and benefits of transferring from
monotherapy with cholinesterase inhibitors to monotherapy with
memantine. They concluded that, in practice, there may be a risk of
doing harm by switching (by moving people off a drug they are known to
tolerate), when you have no particular indication to stop. There may be
a danger that if a transfer to memantine is not tolerated, people may be
left without any prescription, when continuing therapy with
cholinesterase inhibitors may have had some benefit. Therefore, the
committee agreed the balance of evidence was in favour of co-
prescription as opposed to treatment switching in people with severe
Trade-off between net The committee noted that, of the economic evaluations identified, only
health benefits and the DOMINO-AD study was directly applicable and subject to only minor
resource use limitations: it was the only evidence from the UK, based on an RCT, and
that reflected the current prices of the drugs, which have fallen
substantially since generic formulations became available. Therefore,
the committee agreed it provided more robust evidence than the other
studies. The committee noted that the other studies were all judged to
be partially applicable with very serious limitations.
It agreed that, as the cost of all of these medicines was now extremely
low, any intervention that was clinically effective (in improving quality of
life) was almost certain to be cost effective. This was supported by
evidence from DOMINO-AD, where, with the exception of 1 analysis,
interventions associated with more QALYs were also associated with
212
lower costs. The committee considered this a predictable finding

because, if the intervention is effective in maintaining people’s cognition
and/or functional independence, the low costs of the drugs themselves
will be outweighed by money saved in support required. The committee
also noted that the other, lower-quality CUAs were consistent with this
pattern, showing that combination therapy dominated monotherapy with
a cholinesterase inhibitor alone, resulting in decreased costs and
increased QALYs. The committee did not consider these studies to
provide convincing evidence, in themselves, that combination therapy
provides better value than monotherapy; however, it agreed with the
emerging principle that any strategy, in this area, that improves quality
of life is also likely to reduce overall costs.
Discontinuation of cholinesterase inhibitors
The committee considered analyses from the DOMINO-AD study
comparing donepezil monotherapy versus placebo and memantine
monotherapy versus placebo in both moderate and severe AD
subgroups. The committee agreed the findings of the analyses agree
with the findings of the clinical review that monotherapy in both
moderate and severe subgroups with either donepezil or memantine
was effective compared with placebo, and resulted in decreased costs
and increased QALYs. The committee were confident in the finding that
discontinuing donepezil in people with moderate/severe AD resulted in a
loss of QALYs, and therefore concluded that donepezil should be not
discontinued from people with Alzheimer’s disease solely on the basis
of disease progression. There was somewhat more uncertainty in the
costs associated with these analyses; however, the committee did not
think this was a serious issue, as the clinical data were robust, and the
cost data clearly pointed in a direction that reflected the previously
stated argument that any effective treatment is likely to be cost effective.
Combination therapy
The analyses found that, in moderate AD, combination therapy was
associated with increased costs and reduced quality of life, albeit with
substantial uncertainty.
In the case of severe AD, however, combination therapy appeared to be
associated with reduced costs and increased QALYs compared with
monotherapy with either memantine or donepezil alone. The committee
agreed the level of uncertainty in the evidence meant it was hard to
draw definitive conclusions from the economics alone, but again felt that
the clear clinical benefits justified making recommendations for co-
prescription. The stronger evidence for cost effectiveness in the severe
population meant an ‘offer’ recommendation was made for that group,
whilst only a ‘consider’ recommendation was made for moderate
Quality of evidence The committee agreed that for the co-prescription review question, the
evidence that was most relevant for UK practice came from the trials
that compared cholinesterase inhibitors plus memantine against
cholinesterase inhibitors plus placebo, in people already taking a
cholinesterase inhibitor at baseline. The evidence from these trials was
general of moderate to high quality, and therefore the committee was
comfortable using this evidence to make strong recommendations. The
committee agreed that the additional trials of cholinesterase inhibitors
plus memantine versus cholinesterase inhibitor monotherapy (without a
placebo) were at much higher risk of bias and therefore represented a
much lower standard of evidence. They also noted that these trials were
excluded from the recent Cochrane review on this topic, and therefore
felt it appropriate that these trials not be included as part of the primary
analysis looking at the effect of co-prescription.
The committee noted that there was also evidence on the effectiveness
of adding cholinesterase inhibitors to people already taking memantine
at baseline, but agreed this comparison was of much less relevance to
213
the UK. Specifically, the technology appraisals for memantine (TA217)

state that it is an option in moderate Alzheimer’s disease only if
cholinesterase inhibitors are contraindicated, and therefore people
should only be started on memantine monotherapy ahead of
cholinesterase inhibitor monotherapy if they are contraindicated for
cholinesterase inhibitors.
The committee agreed the DOMINO-AD trial (Howard 2012) was a well-
conducted and robust trial that provided evidence directly applicable to
the UK on the question of withdrawal of cholinesterase inhibitors in
people with moderate to severe Alzheimer’s disease. They also noted
that this trial was specifically designed to answer questions raised by
the 2006 NICE dementia guideline, and therefore was likely to remain
the best evidence available to address this question.
No evidence was found that looked at the effect of withdrawing
memantine in people with severe Alzheimer’s disease, and therefore the
committee agreed it was not appropriate to make any recommendations
on this topic.
Other considerations The committee considered the practical arrangements for co-
prescription. They agreed it was important to apply the
recommendations made for these review questions within the context of
the published NICE technology appraisal guidance (TA217) related to
the use of cholinesterase inhibitors and/or memantine for the treatment
of Alzheimer’s disease. They also considered the wording of these
recommendations alongside the other TA217 recommendations
updated as part of this guideline, on the appropriate people to initiate
treatment with cholinesterase inhibitors and/or memantine.
The committee agreed that those recommendations (which specify that
treatment should only be started on the advice of someone with
expertise in diagnosing and treating dementia) were to ensure that
prescription is made following a correct diagnosis of dementia.
However, this is no longer an important issue when considering the
addition of a second medication in people who have an established
dementia diagnosis and are living with more advanced stages of
disease. For this reason, the committee was happy for co-prescription of
these medications to be initiated in primary care, in people who already
have an established diagnosis of Alzheimer’s disease and are already
taking cholinesterase inhibitors.
The committee agreed that, whilst the majority of the included evidence
was on donepezil, it was reasonable to assume a class effect for
cholinesterase inhibitors, and therefore recommendations were made
generally for cholinesterase inhibitors.
67. Do not stop AChE inhibitors in people with Alzheimer’s disease because of
disease severity alone.
68. For people with an established diagnosis of Alzheimer’s disease who are already
taking an AChE inhibitor:
 consider memantine in addition to an AChE inhibitor if they have
moderate disease
 offer memantine in addition to an AChE inhibitor if they have severe
disease.
69. For people with an established diagnosis of Alzheimer’s disease who are already
taking an AChE inhibitor, primary care prescribers may start treatment with
214
memantine (see recommendation 68) without taking advice from a specialist

clinician.
215
11.3 Pharmacological management of dementia with Lewy

bodies
Review question
 What is the comparative effectiveness of donepezil, galantamine, memantine and
rivastigmine for cognitive enhancement in dementia associated with Parkinson’s disease?
11.3.1 Introduction
Dementia (the progressive loss of global cognitive function) is common in Parkinson’s

disease; 48% to 80% of people develop dementia at some point in their condition.
Traditionally, dementia developing more than 1 year after the onset of the motor symptoms
of Parkinson’s disease is referred to as Parkinson’s disease dementia (PDD). Dementia
developing within 1 year of the onset of motor symptoms is referred to as dementia with
Lewy bodies (DLB).
The relationship between DLB and PDD is unclear, but they have many common clinical
features and there is some opinion that they may be the same condition. Therefore, the
committee agreed that the population included in this review question should cover people
with DLB and PDD. Studies that included people with mild cognitive impairment were
excluded.
The aim of this review question was to assess the comparative efficacy of pharmacological
interventions for cognitive enhancement in dementia associated with Parkinson’s disease,
compared with placebo or other active comparator(s). This updates the evidence reviews on:
 Cholinesterase inhibitors for cognitive enhancement in Parkinson’s disease from the 2006
guideline on Parkinson’s disease (CG35).
 Cholinesterase inhibitors or memantine for the treatment of cognitive symptoms of
Dementia with Lewy bodies from the 2006 guideline on Dementia (CG42).
 Cholinesterase inhibitors or memantine for the treatment of non-cognitive symptoms of
dementia with Lewy bodies from the 2006 guideline on Dementia (CG42).
This updated review incorporates some studies that were included in the previous guidelines
together with newly published evidence.
The review focused on identifying studies that fulfilled the conditions specified in Table 53.
Table 53: Review summary: effectiveness of pharmacological for cognitive

enhancement in dementia associated with Parkinson’s disease
Population People with a diagnosis of PDD or DLB
Interventions  Donepezil
 Galantamine
 Memantine
 Rivastigmine1
 Memantine plus cholinesterase inhibitor
Comparators  Placebo
 Each other
 Combination of memantine plus cholinesterase inhibitor
Outcomes  Cognitive outcomes, including:
o Mini Mental State Examination (MMSE)
216
o Alzheimer's Disease Assessment Scale – cognitive subscale (ADAS-cog)

o Montreal Cognitive Assessment (MoCA)
 Global outcomes, including:
o Unified Parkinson’s Disease Rating Scale (UPDRS)
o Global impression of change
 Activities of daily living (ADL), including:
o Unified Parkinson’s Disease Rating Scale – activities of daily living scale
(UPDRS-ADL)
o Measures used in DLB research (including those derived from Alzheimer’s
disease measures)
 Other non-cognitive outcomes, including:
o Neuropsychiatric outcomes, such as the Neuropsychiatric Inventory (NPI)
o Motor symptoms, such as tremor and rigidity
 Adverse events, such as hallucinations
 Study withdrawal
 Health-related quality of life
 Carer-reported outcomes
 Time to institutionalised care
1 Rivastigmine capsules are currently the only intervention that is licensed for mild to moderate dementia in
Parkinson’s disease. No treatments are licensed for dementia with Lewy bodies.
For full details of the review protocol, please see Appendix C. Randomised controlled trials
(RCTs) were considered to be the most appropriate study design to derive treatment effect
metrics, and were therefore considered to be the highest quality within a GRADE framework.
All other study designs were excluded from this review, including case–control studies,
cohort studies, and case reports.
A systematic search of the literature was conducted (see appendix D) which identified 1,152
references. This search was restricted to studies published from 2005 onwards to avoid
duplicates of studies considered in the previous Parkinson’s disease guideline (CG35). After
removing duplicates the references were screened on their titles and abstracts and full
papers of 130 references were obtained and reviewed against the inclusion and exclusion
criteria in the review protocol (see appendix C).
Overall, 121 studies were excluded as they did not meet the eligibility criteria, such as not
utilising a randomised-control design. The 9 remaining published papers met the eligibility
criteria and were included in the review. A list of excluded studies and reasons for their
exclusion is provided in appendix F.
Five RCTs included in previous guidelines on Parkinson’s disease (CG35) and Dementia
(CG42) were reviewed. Of these, 2 RCTs were already included from the search (McKeith et
al., 2000, Ravina et al., 2005) and 2 RCTs (Aarsland et al., 2002; Emre et al., 2004) met the
present inclusion and exclusion criteria and were included. The remaining RCT (Leroi et al.,
2004) was excluded as people in the study had mild cognitive impairment associated with
Parkinson’s disease.
Systematic reviews identified in the literature search were also analysed to identify any
published papers meeting the eligibility criteria that had not been identified in the search. No
further studies were identified. 5 published papers were identified through a rerun of the
literature search in June 2016, none of which were included. Therefore, a total of 11 RCTs
were included in the evidence review.
217
Data were extracted into detailed evidence tables (see appendix E), and further summarised
in GRADE profiles (appendix G).
See Table 54 for a summary of included studies. References for the included studies are
given in appendix I.
11.3.2.1.1 Pharmacological interventions in DLB
3 double-blind placebo-controlled RCTs assessed the effectiveness of a cholinesterase

inhibitor in people with DLB:
 donepezil (Ikeda et al., 2015, Mori et al., 2012)
 rivastigmine (McKeith et al., 2000).
1 double-blind placebo-controlled RCT (Emre et al., 2010) assessed the effectiveness of

memantine in people with DLB.
No studies assessed the effectiveness of a combination of cholinesterase inhibitor plus

memantine in people with DLB.
11.3.2.1.2 Pharmacological interventions in PDD
4 double-blind placebo-controlled RCTs (reported in 5 publications) assessed the

effectiveness of a cholinesterase inhibitor in people with PDD:
 donepezil (Aarsland et al., 2002, Dubois et al., 2012, Ravina et al., 2005)
 rivastigmine (Emre et al., 2004, Dujardin et al., 2006 [secondary publication]).
1 open-label RCT (Emre et al., 2014) assessed the effectiveness of rivastigmine capsules
compared with rivastigmine patches in people with PDD.
2 double-blind placebo-controlled RCTs, reported in 3 publications (Emre et al., 2010; Leroi

et al., 2009, Leroi et al., 2014 [secondary publication]) assessed the effectiveness of
memantine in people with PDD.
No studies assessed the effectiveness of a combination of cholinesterase inhibitor plus

memantine in people with PDD.
11.3.2.1.3 Mixed population (PDD or DLB)
1 double-blind placebo-controlled RCT assessed the effectiveness of memantine in a mixed

population of people with PDD or DLB (Aarsland et al., 2009).
11.3.2.1.4 Prioritisation of outcomes
A large number of outcomes were reported in the studies, particularly those measuring
cognitive function. Some outcomes were reported frequently (for example, MMSE) while
others were reported only in a single small RCT. Therefore, the committee prioritised some
key critical outcomes for the analyses.
Key critical outcomes prioritised by the committee were:

 Adverse events
 Cognitive function, measured by:
218
o Mini Mental State Examination (MMSE)

o Alzheimer's Disease Assessment Scale – cognitive subscale (ADAS-cog)
o Mattis Dementia Rating Scale (MDRS)
o Delis-Kaplan Executive Functions System verbal fluency test (D-KEFS)
o 10-point clock drawing test
o Cognitive Drug Research computerised assessment system (CDR)
o Brief test of attention (BTA)
 Global assessment
 Activities of daily living
 Other non-cognitive outcomes, including
o Neuropsychiatric Inventory (NPI)
o Unified Parkinson’s Disease Rating Scale – motor subscale (UPDRS III)
Analyses
The following analyses were conducted:

 pharmacological interventions in people with DLB:
o cholinesterase inhibitors versus placebo
o memantine versus placebo
 pharmacological interventions in people with PDD:
o rivastigmine patches versus capsules
 combined analyses – pharmacological interventions in a mixed population (PDD or DLB)
o network meta-analyses of pharmacological interventions versus placebo
The combined analyses were only carried out for outcomes when data were available for
both PDD and DLB populations.
For studies which had more than one active treatment arm, for example different doses, the
active treatment arms were combined together to give an overall effect.
The evidence across outcomes was appraised using the GRADE framework and forest plots
are presented where appropriate (see appendix G and appendix H).
219

Study Population Intervention Comparison Prioritised outcomes
Parkinson’s disease dementia (PDD)
Aarsland et al. People aged 45-95 years with Donepezil 5mg daily, Placebo  Adverse events
(2002) cognitive impairment increased to 10mg daily after  Cognitive outcome: MMSE
associated with PD (MMSE 6 weeks if well tolerated
 Global outcome: CIBIC+
score 16 to 26 inclusive [mean
20.8])  Non-cognitive outcomes: NPI, UPDRS III
Dubois et al. People aged 40 years and Donepezil 5mg or 10mg daily Placebo  Adverse events
(2012) older with PDD (MMSE score  Cognitive outcomes: ADAS-cog, MMSE,
10 to 26 inclusive [mean 21.4]) D-KEFS verbal fluency test, BTA
 Global outcomes: CIBIC+
 ADL: DAD
 Non-cognitive outcomes: NPI, UPDRS III
Emre et al. People aged at least 50 years Rivastigmine 1.5mg twice Placebo  Adverse events
(2004) old with PDD (MMSE 10 to 24 daily, increasing to a  Cognitive outcomes: ADAS-cog, MMSE,
[mean 19.3]) maximum well tolerated dose D-KEFS verbal fluency test, CDR, 10-point
(up to 6mg twice daily) clock drawing test
 Global outcome: ADCS-CGIC
 ADL: ADCS-ADL
Emre et al. People aged 50 years and Memantine 5mg daily, Placebo  Adverse events
(2010)1 older with PDD (MMSE score increasing to a maintenance  Global outcome: ADCS-CGIC
10 to 24 inclusive [mean 21.1]) dose of 20mg daily
 ADL: ADCS-ADL
 Carer-reported outcome: ZBI
Emre et al. People aged 50 to 85 years Rivastigmine 4.6mg/24h Rivastigmine 1.5mg  Adverse events
(2014) with PDD (MMSE score 10 to patch, increasing to twice daily, increasing to  Cognitive outcome: MDRS
26 inclusive [mean 20.9]) 9.5mg/24h patch a maximum well
 ADL: ADCS-ADL
tolerated dose (up to
6mg twice daily)  Non-cognitive outcome: NPI
220

Leroi et al. People with PDD (MMSE score Memantine 20mg daily Placebo  Adverse events
(2009) 10 to 27 [mean 19.1])  Cognitive outcomes: MMSE, DRS
Ravina et al. People aged 40 years and Donepezil 5mg daily or 5mg Placebo  Adverse events
(2005) older with PDD (MMSE score twice daily  Cognitive outcomes: ADAS-cog, MMSE,
17 to 26 inclusive [mean 22.2]) MDRS
 Global outcomes: CGIC, UPDRS (total
score)
 Non-cognitive outcomes: UPDRS III
Dementia with Lewy bodies (DLB)
Emre et al. People aged 50 years and Memantine 5mg daily,  Adverse events
(2010)1 older with DLB (MMSE score increasing to a maintenance  Global outcome: ADCS-CGIC
10 to 24 inclusive [mean 20.4]) dose of 20mg daily
 ADL: ADCS-ADL
 Carer reported outcome: ZBI
McKeith et al. People with DLB (MMSE score Rivastigmine 1.5mg twice Placebo  Adverse events
(2000) over 9 [mean 17.9]) daily, increasing to a  Cognitive outcome: MMSE
maximum well tolerated dose
 Global outcome: CGC+
(up to 6mg twice daily)
Ikeda et al. People aged 50 years and Donepezil 5mg or 10mg daily Placebo  Adverse events
(2015) older with DLB (MMSE score  Cognitive outcome: MMSE
10 to 26 inclusive [mean 20.4])
Mori et al. (2012) People aged 50 years and Donepezil 3mg, 5mg or Placebo  Adverse events
older with DLB (MMSE score 10mg daily  Cognitive outcome: MMSE
10 to 26 inclusive [mean 19.6])
221

Mixed population (PDD or DLB)
Aarsland et al. People with PDD or DLB Memantine 5mg daily, Placebo  Adverse events
(2009) (MMSE score 12 or above increasing to a maintenance  Cognitive outcomes: MMSE
[mean 20.0]) dose of 10mg twice daily
 Global outcome: CGIC
 ADL: DAD
1 Study included people with PDD and DLB; data for PDD, DLB and the mixed population was presented separately in the paper
1 ADAS-cog; Alzheimer's Disease Assessment Scale – cognitive subscale

2 ADCS-ADL; Alzheimer's Disease Assessment Scale – Activities of Daily Living subscale
3 ADCS-CGIC; Alzheimer’s disease Cooperation Study – Clinical Global Impression of Change
4 ADL; Activities of daily living
5 BTA; Brief test of attention
6 CDR; Cognitive Drug research computerised assessment system
7 CGC-plus; Clinical Global Change-plus
8 CGIC; Clinical Global Impression of change
9 CIBIC+; Clinician’s interview based impression of change
10 DAD; Disability assessment for dementia
11 D-KEFS; Delis-Kaplan Executive Functions System
12 MDRS; Mattis Dementia Rating Scale
13 MMSE; Mini Mental State Examination
14 NPI; Neuropsychiatric Inventory
15 UPDRS; Unified Parkinson’s Disease Rating Scale
16 ZBI; Zarit caregiver Burden Interview
222
Literature searches were undertaken to find any existing cost–utility analyses (CUAs)
assessing pharmacological interventions for cognitive enhancement in dementia associated
with Parkinson’s disease. In total, 344 articles were returned, of which 2 were selected as
potentially relevant and retrieved for full text review. Both were included. Studies were
assessed using the quality appraisal criteria as outlined in the NICE guidelines manual
(NICE, 2012).
Willan et al. (2006) compared rivastigmine with placebo in people with mild PDD (MMSE
20–24), based on evidence from the EXPRESS RCT (Emre et al. 2004). The analysis
concentrated solely on short-term cognitive effect, as measured by MMSE at 24 weeks,
which was translated to health-related quality of life (EQ-5D) using a mapping function based
on a Scandinavian population with Alzheimer’s disease (Jönsson, 2003). For further details,
please see the economic evidence profile in Appendix M. The authors’ base case adopted a
broad societal perspective, including an attempt to value carer time; however, disaggregated
results are reported, enabling the recalculation of results with a perspective that is consistent
with the NICE reference case (that is, NHS and PSS costs only). This suggests that
rivastigmine is associated with an ICER of around £58,600 per QALY gained. However, this
analysis comes from a time when rivastigmine was only available as a proprietary product;
since then, it has become available generically and costs have decreased substantially.
Therefore, to approximate the results of this CUA from a present-day perspective, the
developer recalculated results by:
 removing costs borne by patients and carers;
 re-estimating rivastigmine drug cost, assuming the overall change is proportional to the
change in price of a 28 x 3 mg pack (£2004=£34.02 [BNF 47]; £2016=£2.57 [NHS Drug
Tariff Feb 2016]; reduction of 92.4%);
 inflating all other costs from £2004/05 to £2015/16 using PSSRU hospital & community
health services inflators.
This analysis estimated an ICER of approximately £16,000 per QALY gained.
Gustavsson et al. (2009) simulated a population with DLB (from which people with PDD
were explicitly excluded) receiving unspecified AChE inhibitors. For further details, please
see the economic evidence profile in Appendix M. The authors drew treatment effects from a
UK observational audit for the first 4 months, and extrapolated these to 5 years using a
Scandinavian longitudinal study in Alzheimer’s disease (Wallin et al., 2007). Additional non-
cognitive symptoms (extra-pyramidal symptoms and psychosis) were assumed for DLB. The
authors used 3 separate models, and compared results. The first was a reconstruction of the
Southampton Alzheimer’s disease model (Loveman et al., 2006); the second was a micro-
simulation model; and the third was a Markov model with 4 discrete MMSE states. When
applied to people with all severities of dementia, ICERs of between £2,700 and £46,800 per
QALY were estimated; when the population was limited to people with moderate dementia
(MMSE 10–20), AChE inhibitors were dominant in all 3 models (that is, they were predicted
to save money and improve health). Again, it was possible to estimate present-day results for
these analyses, by:
 re-estimating AChE inhibitor drug costs, assuming the original model used the cost of
donepezil 10 mg daily and assumed 2 monitoring visits per year, and that the overall
change in drug costs is proportional to the change in price of a 28 x 10 mg pack of
donepezil (£2005=£89.06 [BNF 49]; £2016=£1.45 [NHS Drug Tariff Feb 2016]; reduction
of 98.4%);
223
 inflating all other costs from £2005/06 to £2015/16 using PSSRU hospital & community
health services inflators
This recalculation estimated that treatment with AChE inhibitors is less costly and more
effective than placebo in all analyses, regardless of population modelled or model preferred.
11.3.4.1 Evidence statements – Dementia with Lewy bodies
11.3.4.1.1 Adverse events
Cholinesterase inhibitors
Moderate-quality evidence from 3 RCTs could not differentiate the risk of any adverse
events, serious adverse events or adverse events requiring treatment withdrawal between
cholinesterase inhibitors and placebo.
Memantine
Moderate-quality evidence from 1 RCT could not differentiate the risk of any adverse events,
serious adverse events or adverse events requiring treatment withdrawal between
memantine and placebo.
11.3.4.1.2 Cognitive function
High-quality evidence from 3 RCTs suggests that, compared with placebo, cholinesterase
inhibitors significantly improve cognitive function as assessed by the MMSE.
Memantine
Moderate-quality evidence from 1 RCT could not differentiate the effect on cognitive function
between memantine and placebo, as assessed by the 10-point clock drawing.
11.3.4.1.3 Global assessment
High-quality evidence from 1 RCT suggests that, compared with placebo, donepezil
significantly improves global response as assessed by CIBIC+.
High-quality evidence from 1 RCT suggests that, compared with placebo, donepezil
significantly improves global response as assessed by at least minimal improvement in
CIBIC+.
Memantine
Moderate-quality evidence from 1 RCT could not differentiate the effect on global response
between memantine and placebo, as assessed by ADCS-CGIC.
224
11.3.4.1.4 Activities of daily living
Memantine
Moderate-quality evidence from 1 RCT could not differentiate the effect on activities of daily
living between memantine and placebo, as assessed by ADCS-ADL.
11.3.4.1.5 Carer-reported outcomes
High-quality evidence from 2 RCTs suggests that, compared with placebo, donepezil
significantly improves carer burden as assessed by the Zarit caregiver burden interview.
Memantine
Moderate-quality evidence from 1 RCT could not differentiate the effect on carer burden
between memantine and placebo, as assessed by the Zarit caregiver burden interview.
11.3.4.1.6 Other non-cognitive outcomes
Low-quality evidence from 3 RCTs could not differentiate the effect on neuropsychiatric
symptoms between cholinesterase inhibitors and placebo, as assessed by the NPI-10 item
score.
inhibitors significantly improve neuropsychiatric symptoms (hallucinations, delusions,
dysphoria and apathy) as assessed by the NPI-4 item score.
Low-quality evidence from 2 RCTs could not differentiate the effect on neuropsychiatric
symptoms (hallucinations, cognitive fluctuation) between donepezil and placebo, as
assessed by the NPI-2 item score.
Low-quality evidence from 2 RCTs could not differentiate the effect on motor symptoms
between cholinesterase inhibitors and placebo, as assessed by UPDRS III.
Memantine
Moderate-quality evidence from 1 RCT could not differentiate the effect on neuropsychiatric
symptoms between memantine and placebo, as assessed by the NPI-12 item score.
Moderate-quality evidence from 1 RCT could not differentiate the effect on motor symptoms
between memantine and placebo, as assessed by UPDRS III.
11.3.4.1.7 Economic evidence statements
One partially applicable cost–utility analysis with very serious limitations concluded that, in all
people with DLB, AChEs improve QALYs at increased cost, with ICERs ranging from £2,700
to £46,800, depending on modelling assumptions. In a subgroup of people with moderate
DLB, AChEs were found to be cost-saving. An approximation to 2016 costs suggests that,
now generic AChEs are available at lower cost, treatment would be dominant in all models
and all populations. The study undertook no exploration of uncertainty.
225
11.3.4.2 Evidence statements – Parkinson’s disease dementia
Moderate-quality evidence from 4 RCTs suggests that, compared with placebo,

cholinesterase inhibitors significantly increase the risk of any adverse events.
Moderate-quality evidence from 2 RCTs could not differentiate the risk of serious adverse
events between cholinesterase inhibitors and placebo.

cholinesterase inhibitors significantly increase the risk of study withdrawal due to adverse
events.

cholinesterase inhibitors significantly reduce the risk of hallucinations.
Low-quality evidence from 1 RCT could not differentiate the risk of any adverse events,
serious adverse events, study withdrawal due to adverse events or hallucinations between
rivastigmine patches and rivastigmine capsules.
Memantine
Moderate-quality evidence from 2 RCTs could not differentiate the risk of any adverse
events, serious adverse events or study withdrawal due to adverse events between
inhibitors significantly improve cognitive function as assessed by ADAS-cog.
Low- to moderate-quality evidence from 1 RCT could not differentiate the effect on cognitive
function between rivastigmine patches and rivastigmine capsules at 24 weeks, as assessed
by the MDRS total score, but there was a significant benefit for rivastigmine capsules at 76
weeks.
Memantine
Low-quality evidence from 1 RCT could not differentiate the effect on cognitive function
between memantine and placebo, as assessed by the MMSE and by the 10-point clock
drawing test.
226

cholinesterase inhibitors significantly improve global function as assessed by different
measures.
inhibitors significantly improve global response as assessed by different measures of at least
minimal improvement.
Memantine
Moderate-quality evidence from 1 RCT could not differentiate the effect on global function
between memantine and placebo, as assessed by ADCS-CGIC.
Low-quality evidence from 1 RCT could not differentiate the effect on global response
between memantine and placebo, as assessed by at least minimal improvement in CIBIC+.
inhibitors significantly improve activities of daily living as assessed by different ADL
measures.
Low- to moderate-quality evidence from 1 RCT could not differentiate the effect on activities
of daily living between rivastigmine patches and rivastigmine capsules at 24 weeks, as
assessed by ADCS-ADL, but there was a significant benefit for rivastigmine capsules at 76
weeks.
Memantine
Moderate-quality evidence from 1 RCT could not differentiate the effect on activities of daily
living between memantine and placebo, as assessed by ADCS-ADL.
Memantine
Moderate-quality evidence from 2 RCTs could not differentiate the effect on carer burden
inhibitors significantly improve neuropsychiatric symptoms as assessed by the NPI-10 item
score.
227
Low- to moderate-quality evidence from 1 RCT could not differentiate the effect on
neuropsychiatric symptoms between rivastigmine patches and rivastigmine capsules at 24
weeks, as assessed by the NPI-10 item score, but there was a significant benefit for
rivastigmine patches at 76 weeks.
between donepezil and placebo, as assessed by UPDRS III.
Low-quality evidence from 1 RCT could not differentiate the effect on motor symptoms
between rivastigmine patches and rivastigmine capsules, as assessed by UPDRS III.
Memantine
Moderate-quality evidence from 2 RCTs could not differentiate the effect on neuropsychiatric
symptoms (NPI-10 item or NPI-12 item scores) or motor symptoms (UPDRS III) between
11.3.4.2.7 Economic evidence statements
One partially applicable cost–utility analysis with very serious limitations explored
proprietarily-priced rivastigmine for the treatment of PDD. It concluded that rivastigmine is
likely to improve quality-adjusted life expectation and may reduce overall costs. However,
when an NHS and PSS perspective is adopted, rivastigmine is no longer cost-saving, with an
ICER of £58,600/QALY. An approximation to 2016 costs suggests that, now generic
rivastigmine is available at lower cost, it would be associated with an ICER of around
£16,000/QALY.
11.3.4.3 Evidence statements – mixed population (PDD or DLB)
inhibitors significantly increase the risk of any adverse events.
Moderate-quality evidence from 5 RCTs could not differentiate the risk of serious adverse
events between cholinesterase inhibitors and placebo.
inhibitors significantly increase the risk of adverse events requiring treatment withdrawal.
Memantine
Low- to moderate-quality evidence from 2 RCTs could not differentiate the risk of any
adverse events, serious adverse events or study withdrawal due to adverse events.
228
Memantine
Low-quality evidence from 2 RCTs could not differentiate the effect on cognitive function
between memantine and placebo, as assessed by the MMSE.

cholinesterase inhibitors significantly improve global function as assessed by different
measures.
inhibitors significantly improve global response as assessed by different measures of at least
minimal improvement.
Memantine
Moderate-quality evidence from 2 RCTs suggests that, compared with placebo, memantine
significantly improves global function as assessed by different measures.
Memantine
Moderate-quality evidence from 2 RCTs could not differentiate the effect on activities of daily
living between memantine and placebo, as assessed by different ADL measures.
Memantine
Moderate-quality evidence from 2 RCTs could not differentiate the effect on carer burden
inhibitors significantly improve neuropsychiatric symptoms as assessed by the NPI-10 item
score.
between donepezil and placebo, as assessed by UPDRS III.
Memantine
Moderate-quality evidence from 3 RCTs could not differentiate the effect on neuropsychiatric
symptoms between memantine and placebo, as assessed by the NPI-10 item or NPI-12 item
scores.
229
Moderate-quality evidence from 3 RCTs could not differentiate the effect on motor symptoms
between memantine and placebo, as assessed by UPDRS III.
11.3.4.3.7 Network meta-analyses
Moderate- to high-quality evidence from a network meta-analysis of 9 RCTs showed that

cholinesterase inhibitors are associated with a significant increase in any adverse events,
compared with placebo, but the data could not differentiate between memantine compared
with placebo or cholinesterase inhibitors.
Moderate-quality evidence from a network meta-analysis of 7 RCTs could not differentiate

the rates of serious adverse events between any treatment alternative compared with
placebo, or between cholinesterase inhibitors and memantine.
Low- to high-quality evidence from a network meta-analysis of 8 RCTs showed that

cholinesterase inhibitors are associated with a significant increase in treatment withdrawal
due to adverse events, compared with placebo, but the data could not differentiate between
memantine compared with placebo or cholinesterase inhibitors.
Low- to high-quality evidence from a network meta-analysis of 10 RCTs showed that

cholinesterase inhibitors are associated with a significant improvement in cognitive function
assessed by the MMSE, compared with placebo, but the data could not differentiate between
memantine compared with placebo or cholinesterase inhibitors.

cholinesterase inhibitors are associated with a significant improvement in global function,
compared with placebo, but the data could not differentiate between memantine compared
with placebo or cholinesterase inhibitors.

cholinesterase inhibitors are associated with a significant improvement in neuropsychiatric
symptoms, compared with placebo, but the data could not differentiate between memantine
compared with placebo or cholinesterase inhibitors.
Low- to moderate-quality evidence from a network meta-analysis of 7 RCTs could not

differentiate the effect on motor symptoms between any treatment alternative compared with
placebo, or between cholinesterase inhibitors and memantine.

Relative value of Cognitive outcomes were critical to decision-making for this review question.
different Many different cognitive outcomes were reported in the studies; therefore the
outcomes committee prioritised those outcomes where more data were available to
inform their decision-making. MMSE and ADAS-cog were the most frequently
reported cognitive outcomes. However, they recognised the limitations of, for
example, MMSE, as a measure of the effectiveness of medication. Frequently,
clinicians may be looking for stability, rather than an actual improvement in
cognitive function. The committee also recognised that treatments for dementia
may have important benefits in non-cognitive outcomes, such as global
function, activities of daily living, carer burden and behavioural symptoms.
Trade-off The committee agreed that the evidence overall suggests that the
between benefits effectiveness of pharmacological interventions is similar in people with PDD
and harms and DLB. This supports their original assertion about the similarity between
these 2 conditions. The effectiveness of these interventions also appears to be
broadly consistent with the effects observed in Alzheimer’s disease (AD). Most
RCTs ranged from 12 to 24 weeks, which the committee recognised was a
230
short duration for a long-term degenerative disease. The committee was also
aware that no pharmacological interventions are licensed for managing DLB.
In the previous guideline, the committee was aware that only 1 RCT was
identified (McKeith et al., 2000) which showed no significant improvement in
cognitive function with rivastigmine, compared with placebo. The new evidence
identified for this guideline update shows a significant improvement in cognitive
function and other important outcomes with cholinesterase inhibitors,
compared with placebo. The committee recognised that caution was needed
interpreting the outcomes of RCTs in isolation and patient and clinician factors
also needed to be considered.
Overall, evidence was identified for donepezil and rivastigmine; no significant
differences were observed between the 2 treatments for any of the outcome
measures. The committee discussed whether these results were generalisable
for all cholinesterase inhibitors. They were concerned that no efficacy or safety
data were available for galantamine in people with DLB, but were aware of
data to support its use in AD.
The committee did not expect significant differences to be observed on
pharmacological grounds with galantamine compared with either donepezil or
rivastigmine. However, they did have concerns about making a
recommendation that included galantamine in the absence of evidence, and
therefore a first line recommendation was made for donepezil and rivastigmine,
with galantamine only given as an option if these agents are not tolerated.
The committee discussed their experience of differences between the
cholinesterase inhibitors in their clinical practice. The committee’s experience
suggests that donepezil is generally better tolerated than rivastigmine.
Rivastigmine is generally better in treating neuropsychiatric symptoms. This is
also supported by trends observed in the evidence review, although possible
differences observed did not reach conventional levels of statistical
significance. In the committee’s view, galantamine is not widely used in
practice, compared with donepezil and rivastigmine. Furthermore, clinicians
may take acquisition cost into account when making decisions with people
about the choice of treatment. Galantamine is significantly more costly that
donepezil or rivastigmine.
The committee discussed the importance of appropriate dose titration when
taking cholinesterase inhibitors. Donepezil has a simpler dose titration regime,
which may be an important consideration for individual patients. The committee
had concerns that many people did not have the initial dose of cholinesterase
inhibitor titrated up to the maximum tolerated dose. As cholinesterase inhibitors
are being used ‘off label’ in people with DLB, there is no recommended dose.
However, the committee agreed that they would expect it to be consistent with
the doses licensed for Alzheimer’s disease. To reflect the evidence base, the
committee agreed that the dose of cholinesterase inhibitor should be titrated up
to the maximum tolerated dose.
The committee recognised that the evidence identified was in people with mild
to moderate DLB. In their clinical experience, the committee was aware of
cholinesterase inhibitors being started in people with mild or moderate DLB
and subsequently stopped in some patients because they had reached the
severe stage of the disease. They agreed that treatment should not be stopped
on this basis alone. The committee were concerned about the detrimental
effects observed in many people in clinical practice when cholinesterase
inhibitors were stopped. Although, they were also mindful that some people
stay on cholinesterase inhibitors indefinitely without appropriate review.
Some people present with DLB in the advanced stages of the disease. The
committee recognised this required careful discussion and consideration on a
case-by-case basis, weighing up the possible risks and benefits of treatment.
The committee emphasised the importance of medicines being considered
appropriately at the right time and right stage of disease.
The RCT (Emre et al., 2014) which compared rivastigmine patches with
rivastigmine capsules found that the long-term (76-week) effect on cognitive
231
function was significantly better with capsules. However, the committee felt that
patient factors such as medicines adherence need to be considered on an
individual patient basis. There were no other clinically meaningful differences
between patches and capsules, including the risk of adverse effects.
Therefore, the committee concluded that patches may be an option to consider
for some people, but could not make a recommendation specifically on their
use.
The committee was confident that there is clear evidence of benefit with
donepezil and rivastigmine in improving cognition, global function, activities of
daily living, carer burden and neuropsychiatric symptoms at a cost that is
dominant over placebo. The committee concluded that an ‘offer’
recommendation should be made to reflect the evidence-base. The
recommendation to offer treatment applies to people with mild to moderate
DLB as there was no evidence of starting treatment in people with severe DLB.
The committee also agreed that the recommendation should inform clinicians
that donepezil and rivastigmine are not licensed for DLB.
While the committee could not be certain about the effect of galantamine in
people with mild to moderate DLB, they agreed that galantamine may be
considered for people with mild to moderate DLB if donepezil or rivastigmine
are not tolerated. The committee also agreed that the recommendation should
inform clinicians that galantamine is not licensed for DLB.
Furthermore, although no RCT evidence was identified, the committee
discussed and agreed by consensus that a consider recommendation should
be made for donepezil and rivastigmine in people with severe DLB. They noted
that although no evidence was found, there was no biological or
pharmacological reason to expect that the effect would be less in people with
severe dementia, and it was therefore appropriate to extrapolate the evidence
to that population.
Memantine
The committee recognised that there were far less data for memantine versus
placebo, compared with cholinesterase inhibitor versus placebo. The
committee was concerned that memantine was only significantly better than
placebo on the global assessment scales. However, the committee recognised
the limitations of the available data. The committee did agree, however, that it
was appropriate to make a ‘consider’ level recommendation for memantine, but
only if cholinesterase inhibitors are not tolerated or are contraindicated.
Combination treatment
Although no studies were identified where participants were randomised to
combination treatment with a cholinesterase inhibitor and memantine, the
committee recognised that this option was being used in practice. From their
clinical experience, some people do respond to combination treatment. As
there was no evidence, the committee felt this was an important priority for
research and therefore made a research recommendation.
Trade-off The committee agreed that the economic evidence presented had very serious
between net limitations, and lacked direct applicability to the question, particularly because
health benefits they took place at a time before the generic versions of the drugs were
and resource available. However, they also noted that, once appropriate adjustments had
use been made to the price of the drugs, the fact that cholinesterase inhibitors
came out as consistently either cost-effective or cost-saving compared with
placebo added additional evidence to support the recommendations made.
Quality of Based on the clear and consistent findings for donepezil and rivastigmine, the
evidence committee were confident in making an ‘offer’ recommendation for people with
mild to moderate DLB. The evidence-base for memantine was of lower quality
and despite a trend towards improvement the committee could not be confident
of the effectiveness of memantine.
Other The committee noted that the recommendations made here were broadly
considerations consistent with those in the NICE Parkinson’s disease guideline on managing
Parkinson’s disease dementia. This was agreed to be important as the
evidence did not suggest there was any reason to expect the treatments to
232
have different levels of effectiveness in the two groups, and agreed it was
appropriate to add a cross-reference to those recommendations in the
guideline.
70. Offer donepezil or rivastigmine to people with mild to moderate dementia with
Lewy bodies.a
71. Only consider galantamineb for people with mild to moderate dementia with Lewy
bodies if donepezil and rivastigmine are not tolerated.
72. Consider donepezil or rivastigmine for people with severe dementia with Lewy
bodies.a
73. Consider memantinec for people with dementia with Lewy bodies if AChE
inhibitorsd are not tolerated or are contraindicated.
74. For guidance on pharmacological management of Parkinson’s disease dementia,

see Parkinson’s disease dementia in the NICE guideline on Parkinson’s disease.
8. What is the effectiveness of combination treatment with a cholinesterase inhibitor

and memantine for people with dementia with Lewy bodies if treatment with a
cholinesterase inhibitor alone is not effective or no longer effective?
appendix L.
a At the time of publication (June 2018), donepezil and rivastigmine did not have a UK marketing authorisation for
this indication. The prescriber should follow relevant professional guidance, taking full responsibility for the
decision. Informed consent should be obtained and documented. See the General Medical Council's
Prescribing guidance: prescribing unlicensed medicines for further information
b At the time of publication (June 2018), galantamine did not have a UK marketing authorisation for this indication.
The prescriber should follow relevant professional guidance, taking full responsibility for the decision. Informed
consent should be obtained and documented. See the General Medical Council's Prescribing guidance:
prescribing unlicensed medicines for further information
c At the time of publication (June 2018), memantine did not have a UK marketing authorisation for this indication.
prescribing unlicensed medicines for further information.
d At the time of publication (June2018), the AChE inhibitors donepezil, rivastigmine and galantamine did not have
a UK marketing authorisation for this indication. The prescriber should follow relevant professional guidance,
taking full responsibility for the decision. Informed consent should be obtained and documented. See the
General Medical Council's Prescribing guidance: prescribing unlicensed medicines for further information
233
11.4 Cholinesterase inhibitors and memantine for types of

dementia other than typical Alzheimer’s disease
Review question
 How effective are cholinesterase inhibitors and memantine for types of dementia other
than typical Alzheimer’s disease?
11.4.1 Introduction
The aim of this review question was to determine the comparative effectiveness of donepezil,
galantamine, rivastigmine and memantine for cognitive enhancement in dementia types other
than typical Alzheimer’s disease. The use of cholinesterase inhibitors and memantine for
Parkinson’s disease dementia is covered in the Parkinson’s disease guideline and dementia
with Lewy bodies has previously been considered in section 11.3. Therefore these 2 types of
dementia were excluded from the evidence review for this question.
The review identified studies that fulfilled the conditions specified in Table 55. For full details
of the review protocols, see Appendix C.
Table 55: Review summary: cholinesterase inhibitors and memantine for non-
Alzheimer’s dementia
Population People with a diagnosis of dementia other than typical Alzheimer’s
disease or Lewy-body dementia
Interventions  Donepezil
 Galantamine
 Memantine
 Rivastigmine
 Memantine plus cholinesterase inhibitor
 Combination of memantine plus cholinesterase inhibitor
 Placebo
Outcomes Cognitive outcomes, including:
 Mini Mental State Examination (MMSE)
 Alzheimer's Disease Assessment Scale –cognitive subscale (ADAS-
cog)
 Montreal Cognitive Assessment (MoCA)
Global outcomes, including:
 Global impression of change
Activities of daily living
Non-cognitive outcomes, e.g.
 NPI
 Adverse events
 Study withdrawal
 Health-related quality of life
 Entry to long stay care
234
A systematic search identified 1,772 references. The references were screened on their titles
and abstracts and 99 references were ordered for full text review. Eighty-three papers were
subsequently excluded because they did not fit the inclusion criteria (see Appendix F for a
detailed list of excluded studies and reasons for their exclusion). Sixteen randomised
controlled trials were included in the evidence review.
All included papers considered treatment versus placebo. Nine studies were included for
vascular dementia (3 studies compared donepezil versus placebo, 2 studies compared
galantamine versus placebo, 2 studies compared rivastigmine versus placebo and 2 studies
compared memantine versus placebo).
Three studies were included for frontotemporal dementia (1 paper compared galantamine
versus placebo and 2 papers compared memantine versus placebo). The data was stratified
into participants with behavioural variant frontotemporal dementia and those with semantic
variant frontotemporal dementia.
Three studies were included for cognitive impairment in people with multiple sclerosis (1
study compared donepezil versus placebo, 1 paper considered rivastigmine versus placebo
and 1 paper considered memantine versus placebo).
One study was included for Huntington’s disease and compared rivastigmine versus placebo.
A summary of the characteristics of the included studies is provided in Table 56. Data from
the included studies were extracted into evidence tables. Please see Appendix E for the full
evidence tables, and for the full GRADE profiles please see Appendix G. References for the
A total of 381 citations was returned from the search for this question. Following review of
titles and abstracts, the full texts of 10 studies were retrieved for detailed consideration, but
none met the inclusion criteria. One study, Wong et al. (2009) was considered to be
potentially relevant, presenting dominant results from a cost-effectiveness analysis,
permissible in some circumstances by Developing NICE guidelines (2014). However, its
short time horizon (24-28 weeks) and limited costing perspective (only intervention costs and
physician costs), in addition to the absence of QALYs, led to the conclusion that the study
would not be appropriate to support decision-making. Therefore, no cost–utility analyses
were identified for this question.
235

Study Study Intervention Relevant Authors’
reference Study type population and comparator outcomes conclusions Comments
Vascular Dementia
Galantamine versus placebo
Auchus (2007) RCT 786 participants with Intervention: Cognition (ADAS-cog- Greater improvement in Location: USA
GAL-INT-26 probable vascular Galantamine 4mg 11, EXIT-25) ADAS-cog/11 at 26 weeks Follow up 26
dementia according to twice daily Functional ability for galantamine treated weeks
NINDSA-AIREN increasing to 8mg (ADCS-ADL) group
or 12mg twice daily NPI No difference in ADCS/ADL
after 4 weeks Adverse events, at 26 weeks
Control: serious adverse Mean change in baseline of
Placebo events, mortality and EXIT-25 was significantly
discontinuation due to greater for galantamine
adverse events group
Small (2003) RCT 1,954 participants with Intervention: Cognition (ADAS-cog- Significant improvement in Location:
vascular dementia Galantamine 24mg 11) ADAS-cog/11 at 6 months Multicentre
according to NINDSA- per day for galantamine group international
AIREN Control: Follow up 6
Placebo months plus
additional 6
months open label
extension
Post hoc sub
analysis1 of
Erkinjuntti (2002,
2003)
Rivastigmine versus placebo
Ballard (2008) RCT 5,723 participants with Intervention: Cognition (ADAS-cog, Greater improvement in Location:
vantagE study vascular dementia Rivastigmine MMSE) cognitive performance Multicentre
1.5mg twice daily (VaDAS; ADAS-cog; MMSE) international
236

according to NINDSA- dose escalation by Global assessment at 24 weeks for rivastigmine Follow up 24
AIREN 1.5mg every 4 (VaDAS, ADCS- group weeks
weeks over 16 CGIC, GDS) All other outcomes non-
week period Functional ability significant
Control: (ADCS-ADL)
Placebo NPI-12
Serious adverse
events, mortality and
discontinuation due to
adverse events
Mok (2007) RCT 40 participants with Intervention: Cognition (MMSE) No statistical mean Location: China
Vascular dementia Rivastigmine Global assessment difference in in any efficacy Follow up 26
according to a 1.5mg twice daily (CDR-SB) measures weeks
modification of dose escalating to Functional ability All outcome
NINDSA-AIREN 3mg twice daily (IADL, FAB) measured
after 4 weeks NPI assessed by
Control: Adverse events, Chinese versions
Placebo mortality and of tools
adverse events
Donepezil versus placebo
Black (2003) RCT 4,783 participants with Intervention: Cognition (ADAS-cog, Greater improvement in Location-
Donepezil 307 Vascular dementia Donepezil 5mg per MMSE) cognitive performance international
according to NINDSA- day or donepezil Global assessment (ADAS-cog) at 24 weeks for Follow up 24
AIREN 10mg per day (CDR-SB) donepezil 5mg and 10mg weeks
Control: Functional ability group
Placebo (ADFACS) Greater improvement in
Adverse events, global functional 24 weeks
serious adverse (CIBICplus) for donepezil
events, mortality and 5mg (CDR-SB) donepezil
discontinuation due to 10mg group
adverse events Significant improvement in
ADL (ADFACS) at 24 weeks
237

for donepezil 5mg and 10mg
groups)
Roman 2010 RCT 8,183 participants with Intervention: Cognition (ADAS-cog, Significant improvement in Location: Multi
Vascular dementia Donepezil 5mg per VaDAS-cog, MMSE, cognition (VaDAS-cog) for centre
according to NINDSA- day EXIT-25) donepezil group international
AIREN Control: Functional ability No significant change in
Placebo (DAD) CIBICplus
Global assessment
(CDR-SB)
Adverse events,
serious adverse
events and mortality
Wilkinson RCT 616 participants with Intervention: Cognition (ADAS-cog) Significant improvements in Location: Multi
(2003) Vascular dementia Donepezil 5mg per Global assessment cognition (ADAS-cog) centre
Donepezil 308 according to NINDSA- day or donepezil (CDR-SB) Significant benefits of global international
AIREN 10mg per day Functional ability function (CIBICplus and Follow up 24
Control: (ADFACS, IADL) CDRSB) weeks
Placebo Adverse events, Significant functional
serious adverse benefits (ADFACS)
events and mortality
Memantine versus placebo
Orgogozo RCT 2,883 participants with Intervention: Cognition (ADAS-cog, Significant improvement in Location:
(2002) Vascular dementia Memantine 20mg MMSE) cognitive performance Multicentre
MMM300 according to NINDSA- per day or Global assessment (ADAS-cog and MMSE) for (France,
AIREN Control: (CIBICplus) memantine group with Switzerland,
Serious adverse significant decline in placebo Belgium)
Placebo
events group Follow up 28
No significant difference in weeks
mean change of global
assessment (CIBICplus)
Wilcock (2002) RCT 548 participants with Intervention: Cognition (ADAS-cog) Significant improvement in Location:
MMM500 MMM500 Vascular Dementia Memantine 20mg Adverse events cognitive performance Multicentre UK
per day (ADAS-cog)
238

according to NINDS- Control: No significant difference in Follow up 28
AIREN Placebo mean change of global weeks
assessment (CGI-C)
Frontotemporal dementia
Boxer (2013) RCT 64 participants with Intervention: Cognition (MMSE, No significant difference in Location: USA
Frontotemporal Memantine 10mg EXIT-25) outcomes at 26 weeks Follow up 26
dementia twice per day Global assessment weeks
Control: (CIGIC, CDR-SB)
Placebo NPI
UPDRS
Serious adverse
events
Vercelletto RCT 42 participants with Intervention: Cognition (MMSE, No significant difference in Location:
(2011) Frontotemporal Memantine 10mg MDRS) outcomes at 52 weeks Multicentre France
dementia and mean twice per day Global assessment Follow up 52
MMSE at baseline of Control: (CIBIC-plus) weeks
≥19 Carer burden (ZBI)
Placebo
NPI
Adverse events,
serious adverse
events, mortality and
adverse events
Galantamine versus placebo
Kertesz (2008): RCT 36 participants with Intervention: Cognition (MMSE, No significant difference in Location: USA
Frontotemporal Galantamine 16- DRS) outcomes at 26 weeks Follow up 26
dementia and Primary 24mg per day Functional ability weeks
Progressive Aphasia Control: (FAB, ADCS-ADL) Participants
Placebo NPI completed an
initial 18 week
open label period `
239

Adverse events and
adverse events
Multiple Sclerosis
Donepezil versus placebo
Krupp (2011) RCT 120 participants with Intervention: Cognition (total recall No significant difference in Location USA
MS and memory 10mg per day on SRT; PASAT2&3) neuropsychological Follow up 24
impairment Control: SDMT outcomes at 24 weeks weeks
Placebo
Maurer (2012) RCT 86 participants with Intervention: Domain-specific No significant difference in Location:
MS and cognitive Rivastigmine cognition (SRT, SRT total recall score at 16 Germany
impairment (defined by patches 4.6mg PASAT) weeks Follow up 16
FST ≥ 3 and/or MUSIC (5cm2) per day Global outcomes weeks
score≤ 19 escalating to 9.4mg (CGIC)
(10cm2) per day MS relapse
after 4 weeks Adverse events,
Control: serious adverse
Placebo events and
adverse events
Saint-Paul RCT 62 participants with Intervention: Domain-specific No significant difference in Location: France
(2016) MS and cognitive Memantine 10mg cognition (PASAT) PASAT scores at 52 weeks Multicentre
EMERITE impairment (defined by per day escalating MS progression No significant difference in Follow up 16
DRS ≥130 to 20 mg per day (EDSS) EDSS scores weeks
over 3 weeks Adverse events and
Control: discontinuation due to
Placebo adverse events
Huntington’s Disease
240

Sesok (2014) RCT 18 participants with Intervention: Domain-specific No significant difference in Location: Slovenia
Huntington’s disease Rivastigmine cognition (SDMT, outcomes Follow up 6
1.5mg twice per RFCT, RFFT, TOL) months
day increasing to
3mg twice a day
after 3 months
Control:
Placebo
Abbreviations: NINDSA-AIREN= National Institute of Neurological Disorders and Stroke-AIREN criteria; ADAS-cog= Alzheimer’s disease Assessment cognitive
subscale; CIBICplus=Clinician’s Interview based Impression of Change plus caregivers assessment; ADCS-ADL= Alzheimer’s diseases Cooperative study –
Activities of Daily Living; ADCS-CGIC= Alzheimer’s disease Cooperative study- Clinician’s Global Impression of Change; VaDAS= Vascular Dementia
Assessment Scale; MMSE= Mini mental state evaluation; GDS= Global Deterioration scale; NPI= Neuropsychiatric inventory; WMS= Wechsler Memory Scale;
DRS = Dementia Rating Scale. PASAT =Paced Auditory Serial Addition Test; EDSSS = Expanded Disability Status Scale; FST= Faces Symbol Test; MUSIC=
Multiple Sclerosis Inventarium Cognition Score; SRT= Spatial Recall Test; MDRS = Mattis Dementia Rating Scale; FBI = Frontal Behavioural Inventory; ZBI =
Zarit Burden Interview; DAD; Disability assessment Daily; FAQ=Functional Activities Questionnaire; SDMT= Symbol digit modalities test; CVLT= California
verbal learning test; RFFT= Ruff figure & fluency test; TOL= Tower of London test of learning and memory; TFLS= Texas Functional Living Scale; 1Post hoc
analysis of population sub group with VaD - original trial based on VaD/mixed treatment
241
11.4.4.1 Vascular dementia
11.4.4.1.1 Cholinesterase inhibitors versus placebo
Moderate- to high-quality evidence found global cognition (measured by the MMSE, ADAS-
Cog, ADAS-Cog-11, VaDAS-cognitive subscale) was significantly better in people receiving
cholinesterase inhibitors compared with placebo, but low-quality evidence found it could not
be differentiated when measured by the EXIT-25.
High-quality evidence found neuropsychiatric symptoms (measured by the NPI) were

significantly worse in people receiving cholinesterase inhibitors compared with placebo, but
moderate-quality evidence found they could not be differentiated when measured by the NPI-
12.
Moderate- to high-quality evidence found global assessment (measured by the Clinician’s

Global Impression of Change, Clinical Dementia Rating-sum of boxes) was significantly
better in people receiving cholinesterase inhibitors compared with placebo, but could not be
differentiated when measured by the Vascular Dementia Assessment Scale or Global
Deterioration Scale.
High-quality evidence found functional ability (measured by the Alzheimer’s Disease

Functional Assessment and Change Scale) was significantly better in people receiving
cholinesterase inhibitors compared with placebo, but very low- to moderate-quality evidence
found it could not be differentiated when measured by the ADCS-ADL, Instrumental Activities
of Daily Living, Functional Assessment Battery or the Disability assessment for Dementia.
High-quality evidence found the numbers of adverse events and discontinuations due to
adverse events were significantly higher in people receiving cholinesterase inhibitors
compared with placebo, but low- to moderate-quality evidence could not differentiate the
numbers of deaths or serious adverse events.
11.4.4.1.2 Memantine versus placebo
High-quality evidence found global cognition (measured by the MMSE, ADAS-Cog) was
significantly better in people receiving memantine compared with placebo.
Moderate-quality evidence could not differentiate behavioural symptoms (measured by the

Nurses’ Observational Scale for Geriatric Patients) between people receiving memantine
compared with placebo.
Moderate-quality evidence could not differentiate global assessment (measured by the

Gottfries-Bråne-Steen scale or CIGIC) between people receiving memantine compared with
placebo.
Low- to high-quality evidence could not differentiate the numbers of adverse events or
serious adverse events between people receiving memantine compared with placebo.
Moderate- to high-quality evidence found cognition (measured by the MMSE, ADAS-Cog)

was significantly better in people receiving either cholinesterase inhibitors or memantine
242
compared with placebo, but could not differentiate scores between cholinesterase inhibitors
and memantine.
Moderate-quality evidence found the numbers of adverse events was higher in people
receiving cholinesterase inhibitors than placebo, but could not differentiate numbers between
memantine and placebo or cholinesterase inhibitors and memantine.
Moderate-quality evidence could not differentiate the numbers of serious adverse events
between cholinesterase inhibitors, memantine and placebo.
11.4.4.2 Behavioural variant frontotemporal dementia
Low-quality evidence could not differentiate global cognition (measured by the MMSE or
DRS), functional ability (measured by the Functional Assessment Battery or ACDS-ADL) or
neuropsychiatric symptoms (measured by the NPI) between people receiving cholinesterase
inhibitors compared with placebo.
Low-quality evidence could not differentiate the numbers of adverse events or

discontinuations due to adverse events between people receiving cholinesterase inhibitors
compared with placebo.
11.4.4.3 Memantine versus placebo
Low- to moderate-quality evidence could not differentiate global cognition (measured by the
MMSE, Mattis Dementia Rating Scale, EXIT-25), neuropsychiatric symptoms (measured by
the NPI), global assessment (measured by CIBIC, CGIC, CDR-SB), motor function
(measured by the Unified Parkinson’s disease rating scale) or carer burden (measured by
the ZBI) between people receiving memantine compared with placebo.
Very low to low-quality evidence could not differentiate the numbers of adverse events,
serious adverse events, deaths or discontinuations due to adverse events between people
receiving memantine compared with placebo.
11.4.4.4 Network meta-analyses
Moderate-quality evidence could not differentiate global cognition (measured by the MMSE),
neuropsychiatric symptoms (measured by the NPI) or the numbers of adverse events or
discontinuations due to adverse events between cholinesterase inhibitors, memantine and
placebo.
11.4.4.5 Semantic variant frontotemporal dementia
Low-quality evidence could not differentiate global cognition (measured by the MMSE, EXIT-
25), neuropsychiatric symptoms (measured by the NPI), global assessment (measured by
the CIBIC, CGIC, CDR-SB) or motor function (measured by the Unified Parkinson’s disease
rating scale) between people receiving memantine compared with placebo.
Low-quality evidence could not differentiate the number of serious adverse events between
people receiving memantine compared with placebo.
243
11.4.4.6 Cognitive impairment in people with multiple sclerosis
Moderate-quality evidence could not differentiate global cognition (measured by the Selective
Reminding Test, Multiple Sclerosis Inventarium Cognition Score), domain-specific cognition
(measured by the Symbol Digit Modalities Test, Paced Auditory Serial Addition Test, Faces
Symbol Test) or depressive symptoms (measured by the Montgomery-Åsberg Depression
Rating Scale) between people receiving cholinesterase inhibitors compared with placebo.
Low to moderate-quality evidence could not differentiate the numbers of adverse events,
serious adverse events, discontinuations due to adverse events or multiple sclerosis
relapses between people receiving cholinesterase inhibitors compared with placebo.
Moderate-quality evidence could not differentiate domain-specific cognition (measured by the

Paced Auditory Serial Addition Test) or multiple sclerosis progression (measured by the
Expanded Disability Status Scale) between people receiving memantine compared with
placebo.
High-quality evidence found the number of adverse events was significantly higher in people
receiving memantine compared with placebo, but low-quality evidence could not differentiate
the number of serious adverse events.
High-quality evidence found significantly higher numbers of adverse events in people

receiving memantine compared with either cholinesterase inhibitors or placebo, but
moderate-quality evidence could not differentiate cognition (measured by the Paced Auditory
Serial Addition Test) or the numbers of discontinuations due to adverse events.
11.4.4.7 Huntington’s disease
Moderate-quality evidence found domain-specific cognition (measured by the Tower of

London total times score) was worse in people receiving cholinesterase inhibitors compared
with placebo, but low-quality evidence found it could not be differentiated when measured by
the Symbol Digit Modalities Test, Tower of London total moves score, Rey Complex Figure
Test - delayed recall, Rey Complex Figure Test - immediate recall or Ruff Figural Fluency
Test - unique designs.

Relative value of different The committee agreed that the following measures (where available)
outcomes would be prioritised across the various outcome domains:
 Global cognition (MMSE, ASAS-cog)
 Global assessment (Clinical Dementia Rating scale)
 Functional ability (Disability Assessment for Dementia)
 Neuropsychiatric symptoms (NPI)
244
 Dementia-specific quality of life (DEMQOL)

 Generic health-related quality of life (EQ-5D)
No outcome was prioritised for domain-specific cognition, as it was
agreed that the different outcome measures were often measuring
different aspects of cognition, and therefore it would not be appropriate
to prioritise 1 measure.
Trade-off between Vascular dementia
benefits and harms The committee recognised there was some evidence of cognitive
benefits with treatment, although these were small and often below
levels considered clinically significant, such as 1.4 points on the MMSE.
Additionally, information on how these differences affect functional
ability or quality of life was limited, and the group noted that NPI scores
were worse with cholinesterase inhibitors. It was noted that many of the
studies (those using a definition of possible or probable vascular
dementia) may have included people with underlying Alzheimer’s
disease, and the committee agreed any effect is likely to be greater in
this population than in people with pure vascular dementia, a finding
supported by 1 study reporting data on Alzheimer’s disease with
cerebrovascular legions (which showed an improvement of 3.2 points in
the ADAS-cog with treatment, compared with a 1.6 point improvement
in people with pure vascular dementia). The committee agreed therefore
that it was only clinically justified to consider these drugs in groups of
people with vascular dementia where there is clinical suspicion of there
also being the presence of another form of dementia where these drugs
have been shown to be effective and are recommended by NICE
(Alzheimer’s disease, Parkinson’s disease dementia, dementia with
Lewy Bodies). This recommendation was agreed to be important as
people may initially be diagnosed with one form of dementia (such as
vascular dementia), and then not receive a formal diagnosis of another
dementia subtype at a later date. The committee discussed the
evidence base and agreed that it was not sufficiently robust to enable
making any recommendations about the comparative effectiveness of
cholinesterase inhibitors and memantine.
Frontotemporal dementia
The committee noted the evidence base for this population was far
smaller than in the population with vascular dementia, and there was no
biological hypothesis as to why these treatments would be expected to
provide a benefit (there is not usually a cholinergic deficit in people with
frontotemporal dementia). No evidence was found of benefit for the use
of cholinesterase inhibitors or memantine in behavioural variant or
semantic variant frontotemporal dementia, and therefore the committee
agreed that, owing to the potential adverse effects associated with these
drugs, their use could not be justified in this group. Hence a ‘do not use’
recommendation was made; however, the committee recognised that it
was unlikely that cholinesterase inhibitors or memantine were
commonly used in this group, and that the drugs are not licensed for
frontotemporal dementia. The committee agreed it was appropriate to
extend the recommendation to the third subtype of frontotemporal
dementia (progressive non-fluent aphasia) as there is also no
underlying cholinergic deficit in this group, and therefore there is no
reason to expect that the drugs would be more effective in this group
than the other subtypes of frontotemporal dementia.
Multiple sclerosis
The committee discussed whether cognitive impairment associated with
multiple sclerosis should be included in this section of the guideline. The
committee recognised that the definition of cognitive impairment may
not be as clearly defined as in other conditions. They noted there were
additional challenges in interpreting the evidence due to the physical
impairments associated with the condition when symptoms of cognitive
impairment appear. The committee noted that the population in the
245
included studies was younger than other subtypes but agreed it was
important to include people with multiple sclerosis because the evidence
associated with adverse events meant it was important to prevent
clinicians from prescribing treatment with cholinesterase inhibitors or
memantine to people in this group. They therefore agreed to make a ‘do
not use’ recommendation.
Huntington’s disease
The committee noted that the evidence was drawn from a small pilot
study and therefore agreed that it would not be appropriate to make any
recommendations for this population.
Trade-off between net No economic evidence was identified for this review question and health
health benefits and economic modelling was not prioritised. The committee noted that all of
resource use the named drugs are now generic and off patent and therefore it is
unlikely that any significant resource implications would arise from the
recommendations made for their use. They also noted that
cholinesterase inhibitors and memantine have been found to be cost
effective in Alzheimer’s disease, Parkinson’s Disease dementia and
dementia with Lewy Bodies, where they have shown clinical benefit.
Therefore, it is reasonable to infer that, in any cases where using
cholinesterase inhibitors is clinically effective, the cost of the drugs will
be justified.
Quality of evidence The committee considered making a research recommendation around
vascular dementia (now we are better able to diagnose the subtype of
people with pure vascular dementia). However, large RCTs have
already been conducted in this group (even if these did not always
measure the outcomes we would want nowadays – functional ability and
quality of life) so it was felt unlikely that such studies would be
conducted.
The committee agreed the quality of evidence for vascular dementia
was based upon a number of large clinical trials but recognised the
evidence was poor for other subtypes, mainly due to the size of the
studies.
Other considerations The committee agreed that there is generally a need for more research
in vascular dementia, but felt the focus is now more on disease
modifying agents rather than symptom focused and therefore agreed
this was not the forum to pursue such research, as there is not yet
sufficient evidence to know which medicines are worth testing for
disease modifying purposes.
75. Only consider AChE inhibitorse or memantinef to people with vascular dementia if
they have suspected comorbid Alzheimer’s disease, Parkinson’s disease
dementia or dementia with Lewy bodies.
76. Do not offer AChE inhibitors or memantine for people with frontotemporal
dementia.g
e At the time of publication (June 2018), the AChE inhibitors donepezil, rivastigmine and galantamine did not have
a UK marketing authorisation for this indication. The prescriber should follow relevant professional guidance,
taking full responsibility for the decision. Informed consent should be obtained and documented. See the
General Medical Council's Prescribing guidance: prescribing unlicensed medicines for further information.
f At the time of publication (June 2018), memantine did not have a UK marketing authorisation for this indication.
prescribing unlicensed medicines for further information.
g Note that logopenic aphasia, which has previously been included in some diagnostic guidelines for
frontotemporal dementia, has now been shown to most commonly be caused by Alzheimer’s disease.
246
77. Do not offer AChE inhibitors or memantine to people with cognitive impairment
caused by multiple sclerosis.
247
Drugs that may worsen cognitive decline
12 Drugs that may worsen cognitive decline

Dementia is defined by the WHO as a condition where there is a deterioration in cognitive
function that adversely affects many aspects of life including memory, understanding,
learning capacity, language, behaviour and judgement. Certain classes of drugs, including
anticholinergics, have been shown to be associated with an increased risk of dementia in
clinical trials and they appear to have more extreme effects in older adults.
Anticholinergics are used to treat a range of conditions including psychosis, depression,

urinary incontinence, allergies and COPD. Anticholinergics work by blocking the action of
acetylcholine in the nervous system, preventing signal transmission between nerve cells in
the brain or nerve cells and muscles in the body. Other side effects include dry mouth,
drowsiness, blurred vision, urinary retention and constipation. The mechanism by which
anticholinergics could act to increase the risk of dementia remains unclear.
Dementia is usually progressive and chronic with few potentially modifiable risk factors,
however, it is possible to alter the level of anticholinergics prescribed to people with dementia
and this may improve their cognitive functioning to some extent. The cumulative effect of
taking several medicines containing anticholinergics, known as the anticholinergic burden,
also increases the risk of cognitive decline and thus it is essential to carefully monitor the use
of anticholinergics in older people and those with dementia.
This chapter focuses on identifying which anticholinergics are commonly used for people with
dementia with the aim of reducing the level of prescription of these drugs where possible. In
addition, it seeks to address whether there are tools which would allow the identification of
specific drugs that may be causing cognitive decline.
248
12.1 Drugs that may cause cognitive decline

Review questions
 What drugs that may worsen cognitive decline are commonly prescribed in people
diagnosed with dementia?
 What are the most effective tools to identify drugs that may be causing cognitive decline?
12.1.1 Introduction
The aim of the first specified review question was to identify drugs that may cause cognitive
decline that are commonly prescribed in people living with dementia. It was agreed that the
main class of drugs that is a cause of concern (anticholinergics) was clear, and the harms of
these drugs were well established. Therefore, the key question to answer was not what the
harms of these drugs are, but rather which of these drugs are commonly in use, as this may
help to focus efforts to reduce levels of inappropriate prescribing.
The aim of the second review question was to identify whether there are appropriate tools
available to identify medicines that may be the cause of cognitive decline in a person
suspected of having dementia, and therefore prevent false positive diagnoses. The review
identified studies that fulfilled the conditions specified in Table 57. The full review protocol is
available in Appendix C.
Table 57: Review summary: drugs that may cause cognitive decline
dementia
Diagnostic variables  Standardised tools assessments, instruments and protocols used to
identify drugs that cause cognitive decline
 Anticholinergic burden scale
 Clinical history
 Change in prevalence of appropriate polypharmacy
 Potentially avoidable hospital admissions
No literature review was undertaken to identify commonly prescribed drugs that may cause
cognitive decline in people with dementia, as it was decided that such a review would only be
able to provide data on the well-established harms of certain medicines, rather than identify
which of these medicines are currently commonly used in the UK. Therefore, this question
was instead supported by evidence provided from an expert witness, the UK Prescribing
Observatory for Mental Health. Data were provided on current prescribing patterns in mental
health trusts, with the evidence presented to the committee summarised in Appendix N.
A systematic literature search was carried out to identify diagnostic accuracy studies, or
systematic reviews of diagnostic accuracy studies for the question considering the tools used
to identify drugs that may cause cognitive decline. A total of 6,337 references were screened
at the title and abstract level, with 37 potentially relevant references being ordered for full text
review. No evidence was identified in a population of people living with dementia, but 1
systematic review of anticholinergic scales in older people (Salahudeen 2015) was included.
The excluded studies, with reasons for their exclusion, are listed in Appendix F. Evidence
249
tables for Salahudeen (2015) and the 7 studies included in that review are presented in
Appendix E.
The characteristics of the studies included in Salahudeen (2015) are summarised in Table
58. References for the included studies are given in appendix I.
Table 58 Summary of studies included in Salahudeen (2015)

Methods No of
Study Study Assessment used to medicines
details population scales devise scale Outcome(s) identified
Ancelin 372 people aged Anticholinergic Based on Cognitive 27
(2006) >60 years living Burden Serum function
in the Classification Anticholinergic
community (ABC) Activity (SAA)
and expert
opinion
Boustani 87 nursing home Anticholinergic Based on Quality of life 88
(2008) residents with Cognitive Burden published
dementia aged Score (ACB) data and
≥65 years expert opinion
Carnahan 279 long term Anticholinergic Based on SAA 117
(2006) care residents Drug Score (ADS)a ranking of
identified
anticholinergic
drugs and
expert opinion
Ehrt 78 people with Anticholinergic Based on Cognitive 99
(2010) Parkinson’ Activity Scale existing function
(AAS) evidence and
expert opinion
Han 544 men aged Clinician’s rated Based on Cognitive 60
(2008) ≥65 years living Anticholinergic published function
in the Scale (CrAS)a anticholinergic Functional
community scales and assessment
expert opinion
Rudolph 132 people aged Anticholinergic Based on Central 49
(2008) ≥65 years in Risk Score (ARS) literature adverse
hospital and 117 review and effects
people aged ≥65 expert opinion (confusion,
years living in dizziness,
the community falls)
Sittironarit 211 people with Anticholinergic Based on Executive 49
(2011) Alzheimer’s Loading Scale published function
disease; (ACL) anticholinergic Psychomotor
133 people with scales and speed
MCI; 768 expert opinion
healthy controls
aADS is a modified version of CrAS
Standard health economic filters were applied to the clinical search for these questions, and
a total of 1,062 citations was returned. Following review of titles and abstracts, no full text
250
studies were retrieved for detailed consideration. Therefore, no relevant cost–utility analyses
were identified for these questions.
Moderate-quality evidence from 1 systematic review of 7 observational studies with 2,325

people (including 297 people diagnosed with dementia) found 7 validated anticholinergic
scales developed by expert opinion detected an association between higher scores on
anticholinergic scales and harms caused by anticholinergic medicines in older aged
populations.

Relative value of different The committee agreed there was a need to raise awareness that there
outcomes are certain groups of drugs that may influence cognitive function and
that any recommendations should acknowledge both the minimisation of
drugs causing anticholinergic activity and raise awareness of the scales
that may be used to detect anticholinergic activity or burden. The
committee recognised that there was a clinical issue that older aged
populations include a substantial proportion of people with
multimorbidities, where more than one condition may be treated with a
medicine that has an anticholinergic effect, and the use of multiple
medications with anticholinergic burden has a cumulative effect.
Two separate settings were identified where anticholinergic burden may
be an important factor to consider. The first is when considering a
diagnosis of dementia, where the presence of a substantial
anticholinergic burden may mimic the symptoms of dementia and
therefore lead to false diagnoses. The second, in people with a known
diagnosis of dementia, is that the use of anticholinergics may
exacerbate the symptoms of cognitive decline, and therefore their use
should be carefully monitored.
Trade-off between At the time of diagnosis
benefits and harms The committee agreed that although the evidence they had seen was in
relation to a generally older aged population, there should be little
difference for people living with dementia. Although the effects of
anticholinergic burden may differ by population, the specific drugs that
cause the largest anticholinergic burden are likely to remain the same. It
noted that there may be implications for community prescribers to
assess cognition before prescribing medications, in order to recognise
that some drugs may affect anticholinergic burden.
The committee agreed that, at the time of considering a diagnosis of
dementia, it would be appropriate to reduce the level of anticholinergic
drugs being used, if possible alternatives were available, in order to rule
out potential false diagnoses, and a ‘consider’ recommendation was
made on this point.
The committee agreed that, whilst many of the classes of drugs with
high anticholinergic activity were understood widely by clinicians, there
were other individual drugs where the high level of anticholinergic
activity was not well known, particularly when these come from a class
of drugs not usually associated with it. Therefore, it agreed that
validated, structured tools to assess anticholinergic burden would be
useful, as they would make clinicians undertaking reviews aware of
drugs they might not otherwise have considered.
251
The committee considered the types and classes of drugs listed in each
scale. It noted that different drugs may be rated differently in each scale,
with some drugs scoring higher for anticholinergic activity than others
(this is likely to result from the different methodologies by which the
scales were constructed). It agreed that there was currently no evidence
to recommend the use of one scale in preference to another, so agreed
that it was appropriate only to make clinicians aware of the existence of
these scales, rather than make a specific recommendation that one
should be used. However, the committee agreed it would be helpful to
provide a link to an example of one of these scales, to ensure that non-
specialist clinicians could understand the sorts of tools that exist. The
Anticholinergic Cognitive Burden Scale was chosen as the most
appropriate to reference because it uses standard UK names for drugs,
and because it has been updated more recently (2012) than some of
the scales identified. It did however, want the recommendation to re-
emphasise that this was simply one tool available for determining the
anticholinergic activity of specified drugs and not the only tool available
for consideration.
The committee recognised that awareness of anticholinergic burden
should form part of the full patient pathway and acknowledged there
were other areas of the guideline (for example when undertaking
diagnostic assessments) where this issue would be revisited.
After diagnosis
The committee agreed it remained important to continuously assess the
level of anticholinergic burden in people living with dementia, and that
an assessment should be made of anticholinergic burden as part of
medication reviews. The committee agreed that, again, possible
alternatives should be sought for drugs with a high anticholinergic
burden, if these are available. It noted that the audit data presented
showed that high levels of anticholinergic medicines were still
prescribed in people living with dementia, and that while much of this
prescribing is likely to be appropriate and necessary, it is likely that
there is still inappropriate prescribing of medications with an
anticholinergic effect when alternatives without this effect are available.
No evidence was identified about how reviews should be conducted and
what tools should be used, and therefore the committee agreed the
most appropriate action was to cross-reference the NICE guideline on
medicines optimisation, which provides guidance on how to undertake
medication reviews.
Trade-off between net The committee agreed that there was potentially a high cost associated
health benefits and with the inappropriate prescription of drugs causing a high level of
resource use anticholinergic burden, both due to the side effects they can cause and
the costs associated with inaccurate diagnoses. Therefore, appropriate
reviews of anticholinergic medicines, both at diagnosis and reviews,
would be likely to be cost-saving if it reduced levels of inappropriate
prescribing.
Quality of evidence The committee agreed that the most appropriate method of validating
the scales would have been to consider measures of diagnostic
accuracy, such as sensitivity and specificity, but they noted the
evidence they had seen did not report validity in this way. This meant it
was difficult to assess the overall utility of each scale. After considering
the methodologies in each scale, the committee agreed there was an
absence of evidence to identify one single well-validated tool over
another.
The committee noted that the audit data presented on commonly
prescribed anticholinergic medicines came from mental health trusts,
and therefore there was still a gap in the evidence for which are the
most commonly prescribed anticholinergic medicines in primary care.
However, the committee agreed the recommendations made for
252
medication reviews would still be appropriate in a primary care setting,

as the adverse effects to the person treated will be the same.
Other considerations The committee acknowledged there is a gap in the current evidence
base for considering whether reducing anticholinergic burden can
improve the cognitive outcomes for people who have cognitive
impairment, as the currently available studies are either cross-sectional
or look at populations with a stable or increasing anticholinergic burden
over time. It was therefore agreed that randomised control trials
comparing a strategy of actively lowering anticholinergic burden, versus
usual care, would be useful to fill this gap in the evidence.
78. Be aware that some commonly prescribed medicines are associated with
increased anticholinergic burden, and therefore cognitive impairment.
79. Consider minimising the use of medicines associated with increased

anticholinergic burden, and if possible look for alternatives:
 when assessing whether to refer a person with suspected dementia for
diagnosis
 during medication reviews with people living with dementia.
80. Be aware that there are validated tools for assessing anticholinergic burden (for
example, the Anticholinergic Cognitive Burden Scale), but there is insufficient
evidence to recommend one over the others.
81. For guidance on carrying out medication reviews, see medication review in the
NICE guideline on medicines optimisation.
9. Does actively reducing anticholinergic burden in people living with dementia

improve cognitive outcomes compared with usual care?
appendix L.
253
Non-pharmacological interventions for people living with dementia
13 Non-pharmacological interventions for

Dementia is a progressive, long-term condition that people live with, often for many years.
Enabling and supporting people to live as well as possible with the condition is an important
priority.
Living with dementia brings many challenges. The symptoms of dementia make it harder to
participate in activities and engage socially, to maintain independence, to communicate
effectively, to feel in control and, ultimately, to care for oneself. Experiencing symptoms of
dementia has consequences, such as loss of confidence or tensions in family relationships,
which compound the original disability. All of this can profoundly threaten a person’s sense of
identity and security, especially where the person’s environment is not well-adapted or the
surrounding community is not inclusive, and can impact on the ability of families to provide
care. There is potential for people with dementia to live meaningful and satisfying lives and to
experience a good quality of life, but this requires support both to promote inclusion and to
manage disability.
In recent years a growing social movement has focused on changing public attitudes,
inspired the creation of dementia-friendly communities, and promoted inclusion of people
with dementia and awareness of the rights of people with dementia, for example through the
development of peer support and advocacy groups. Involvement can provide enormous
benefits for those people with dementia who wish to engage in this way.
Alongside this, there has been an increasing focus on enabling individuals with dementia and
their families through providing interventions, suitable for the stage of dementia and tailored
to personal needs and preferences, that promote functioning and well-being and help to
sustain positive family relationships. A wide range of interventions has been proposed with
these general aims in mind. At the time of diagnosis, supportive interventions offer an
opportunity to process emotions, adjust to the condition and plan for the future. In the mild to
moderate stages of dementia, personalised support with maintaining independence and
managing everyday activities aims to ensure continued engagement and participation, while
bringing people together in groups aims to enhance functioning as well as providing
important opportunities for social contact. At all stages of the condition, engaging in
enjoyable, creative and health-enhancing activities offers a means of promoting well-being.
Nevertheless, there remains a significant gap in provision of non-pharmacological
interventions which is compounded by inequity of access to services, for example for people
living in rural areas or areas with poor transport links.
This chapter focuses on the range of non-pharmacological interventions for which rigorous
evidence from randomised controlled trials is available. The evidence covered here excludes
studies of interventions targeted at people experiencing specific illness-emergent non-
cognitive symptoms such as depression, anxiety, or agitation; these are covered in section
14. This chapter reviews evidence from studies involving broader groups of people with
dementia, at different stages of the condition, and aiming to produce benefits in cognition,
functioning or well-being. In translating this evidence to practice it is vital to acknowledge that
each person is different, and hence a one-size-fits-all approach is inappropriate. It is
essential to seek the views of people living with dementia and their carers about the kinds of
interventions they consider to be feasible and acceptable. Decisions about which non-
pharmacological interventions are likely to be suitable and helpful for a given individual need
to be based on an understanding of that individual’s unique set of life experiences,
circumstances, preferences, strengths and needs, with interventions personalised and
tailored as far as possible to each individual and, where relevant, each family.
254
13.1 Pre-, peri- and post-diagnostic counselling and support for

people living with dementia and their families
Review question
 How effective are pre-, peri- & post-diagnostic counselling and support on outcomes for
people living with dementia and their families?
13.1.1 Introduction
The aim of this review question was to determine the effectiveness and cost-effectiveness of
pre-, peri- and post-diagnostic counselling and support for improving outcomes for people
living with dementia and their families and to identify the most effective and cost-effective
ways of providing it. The review identified studies that fulfilled the conditions specified in
Table 59. For full details of the review protocol, see appendix C.
Table 59: Review summary: post-diagnostic counselling and support

Population People (aged 40 years and over) living with dementia and having
been diagnosed within the previous 12 months
Interventions Counselling and support for people living with dementia and their
families, which may include elements such as:
 Diagnostic counselling
 Psychosocial support
 Peer support groups
 Information and advice
 Signposting
 Standard care
ability
A systematic literature search for systematic reviews and RCTs identified 2,662 references.
These were screened at title and abstract level, with 4 systematic reviews ordered as
potentially relevant and 17 ordered as potentially relevant RCTs of counselling and support
interventions for people living with dementia and their families. Finally, because of the limited
RCT data available, a separate search was conducted to look for comparative observational
studies. This identified 1,942 references, of which 9 were judged to be potentially relevant
and ordered for full-text review. Six additional references were also identified from excluded
systematic reviews, the search for health economics evidence, and a related included
reference. In total, 4,604 references were identified, of which 36 were judged to be potentially
relevant and ordered for full-text review.
Three RCTs were included (reported in 4 publications), with 32 excluded at full-text review.
Excluded studies are listed, with reasons for exclusion, in appendix F, and evidence tables
are available in appendix E.
255
Two different types of post-diagnostic support interventions were found aimed at people who
were recently diagnosed with dementia (within the previous 12 months):
 Psychosocial interventions – These interventions included elements of counselling,
psychosocial support, information and advice to people living with dementia and their
carers.
 Self-management interventions – These interventions included elements of psychosocial
support, information and advice to people living with dementia and their carers.
No evidence was found on pre- and peri-diagnostic counselling and support for people living
with dementia and their families.
13.1.2.1 Analyses
Data from different studies were meta-analysed where possible, with GRADE tables and
meta-analysis results given in appendices G and H, respectively. References for the included
studies are given in appendix I.
A minimal clinically important difference of ≥0.03 points was considered by Koivisto (2016)
for quality of life as measured with the 15D questionnaire (Sintonen 2001).
A difference of 3 points was considered by Laakkonen (2016) as clinically important for the
physical component summary of the RAND-36 survey of health-related quality of life (Hays
and Morales 2001).
256

Study population
Study Intervention Location
reference Control Follow-up Included participants Outcomes of interest
Koivisto (2016) Intervention: Psychosocial Study population: People with n=236 dyads (patient- People with dementia:
Control: Basic counselling Alzheimer’s disease diagnosis for carer)  Quality of life (QoL-AD and VAS)
at diagnosis an average of 5 months and n=84 allocated to  Cognitive impairment (MMSE)
clinical dementia rating of 0.5 (very intervention
mild) or 1.0 (mild) and their carers.  Memory disorder severity(CDR-SOB)
n=152 allocated to
Setting: Brain Research and control  Activities of daily living (ADCS-ADL)
Rehabilitation Centre  Behavioural disturbances (NPI)
Location: Finland  Nursing home placement
Follow-up: 36 months.  Mortality
Carers:
 Quality of life (15D and VAS)
 Psychological distress during caregiving
(GHQ)
 Orientation to life (SOC)
 Depression (BDI)
Laakkonen Intervention: Self- Study population: People with a n=136 couples People with dementia:
(2016) management recent diagnosis of dementia and n=67 allocated to  Health-related quality of life (15D)
Control: Usual care their spouses. intervention  Cognitive function (CDR, VF, CDT)
(including basic post- Setting: Primary care and memory n=69 allocated to control Spouses:
diagnostic counselling if clinics
needed)  Health-related quality of life (RAND-36 PCS)
Location: Finland
Follow-up: 9 months.
Waldorff (2012) Intervention: Psychosocial Study population: People n=330 patient-carer Patients:
Phung (2013) Control: Follow-up support diagnosed with Alzheimer’s dyads  Quality of life (QoL-VAS and QoL-AD)
(including basic post- disease, mixed Alzheimer’s n=163 allocated to  Cognitive function (MMSE)
diagnostic counselling if disease or Lewy body dementia intervention
within the previous 12 months of  Activities of daily living (ADCS-ADL)
needed)
257
Study population
Study Intervention Location
reference Control Follow-up Included participants Outcomes of interest
recruitment in the trial and their n=167 allocated to  Behavioural disturbances (NPI-Q)
primary carers. control  Depression (CDS)
Setting: Study centres (hospitals;
 Nursing home placement
community health centre)
 Mortality
Location: Denmark
Carers:
Follow-up: 12 months
 Quality of life (QoL-VAS)
Follow-up: 36 months.
 Depression (GDS)
15D: 15-dimensional health-related quality of life instrument; ADCS-ADL: Alzheimer’s Disease Cooperative Study-Activities of Daily Living; BDI: Beck
Depression Inventory; CDR-SOB: clinical dementia rating global and the sum of boxes score; CDS: Cornell scale for depression in dementia; CDT: clock
drawing test; GDS: geriatric depression scale; GHQ: general health questionnaire; MMSE: mini-mental state examination; NPI: neuropsychiatric inventory;
NPI-Q: brief clinical form of the neuropsychiatric inventory; QoL-AD: quality of life in Alzheimer’s disease; RAND-36 PCS: Research and Development
Corporation 36 item health survey physical component survey; SOC: sense of coherence; VAS: visual analogue scale; VF: verbal fluency test.
258

(CUAs) evaluating the effectiveness of pre-, peri- and post-diagnostic counselling and
support on outcomes for people living with dementia and their families. In total, 1,392 articles
were returned, of which 2 were selected as potentially relevant and retrieved for full text
review. Of these studies, 1 study considering psychosocial intervention was deemed relevant
and included.
13.1.3.1 Psychosocial interventions
Søgaard et al. (2014) conducted a cost–utility analysis alongside the Danish Alzheimer's
Intervention Study (DAISY) (n=300), an RCT conducted in 2004 in Denmark, which collected
health care utility data using the EQ-5D. They compared the cost effectiveness of a
psychosocial intervention with control support (usual care). Details of the psychosocial
intervention and control support (usual care) can be found in the reported Waldorff et al.
(2012) study. Primary outcome measures were QALYs and costs over 36 months. For further
informal care and associated production loss costs; however, disaggregated results are
reported, enabling the recalculation of results with a perspective that is consistent with the
NICE reference case (that is, NHS and PSS costs only). All costs were valued according to
2008 prices and converted into euros (€). Healthcare costs were based on national registers
and Danish governmental tariffs.
The EQ-5D was administered to carers at baseline and at 6, 12 and 36 months of follow-up
for reporting of carers’ health-related quality of life and for proxy reporting of participants’
health-related quality of life. Lifetime was measured in days based on register data from the
national registry of death causes.
Base-case results (Table 61) suggest that, over a period of 36 months, psychosocial
intervention resulted in decreased costs and a loss of QALYs.
Table 61: Base-case cost–utility results from Søgaard et al. (2014), based on multiple
imputation-based analysis over 36 months with informal care costs and
production loss costs excluded
Treatment Cost Effect Cost Effect a ICER
3.26
Psychosocial intervention €35,040 QALYs
3.46 0.09 €49,256 /
Control €39,473 QALYs €4,433 QALYs QALY
a difference adjusted for baseline utility
The results of the complete case analysis showed a relatively smaller savings of cost
compared with the multiple imputation-based analysis of -€958 and the loss of 0.38 QALYs
(when adjusted for baseline utility) for the psychosocial intervention compared with usual
care. This results in an ICER of €2,521 saved per QALY forgone.
Amongst the authors’ probabilistic sensitivity analyses, 1 set of results represents a ‘formal
care’ perspective that is most closely comparable to the NICE reference case. This suggests
259
that there is a 78% probability that the intervention reduced health and social care costs and
a 60% chance that the intervention is associated with an ICER of €30,000/QALY or better
(arising because the amount of money it is predicted to save would, in a majority of cases, be
judged to outweigh the harm it is predicted to cause).
The authors concluded that a psychosocial intervention is unlikely to be cost effective in a

Danish setting because it did not generate additional QALYs and it led to higher average use
of informal care. When the costs of such care (which are not consistent with the NICE
reference case) are removed, it appears that, although the use of psychosocial interventions
results in fewer QALYs than usual care, it may save sufficient money to offset this. These
findings are subject to considerable uncertainty, and should be understood in the context that
any cost savings that may be associated with the intervention arose by shifting cost burden
from health and social care budgets to informal carers.
No evidence was found on pre- and peri-diagnostic counselling and support for people living
with dementia and their families.
13.1.4.1 Psychosocial interventions
13.1.4.1.1 People living with dementia
Moderate- to high-quality evidence from 2 RCTs containing 566 people diagnosed with
dementia within the last year found that people offered psychosocial interventions had better
quality of life (measured with a dementia specific scale) at 12 months compared with people
offered follow-up support. However they could not differentiate quality of life (measured with
a visual analogue scale) at 12 months, or using a quality of life instrument at 36 months.
Moderate- to high-quality evidence from 2 RCTs containing 601 people diagnosed with
dementia within the last year found that people offered psychosocial interventions had lower
levels of depressive symptoms at 12 months (measured with the Cornell depression scale)
compared with people offered follow-up support, but could not differentiate levels of
depressive symptoms at 36 months.
High-quality evidence from 1 RCT containing 236 people diagnosed with dementia within the
last year found that people offered psychosocial interventions had more severe memory
disorder at 36 weeks compared with people offered basic counselling.
High-quality evidence from 2 RCTs containing 566 people diagnosed with dementia within
the last year found people offered psychosocial interventions had worse functioning in their
daily activities at 36 months compared with people offered follow-up support or basic
counselling at diagnosis, but could not differentiate activities of daily living at 12 months.
Low- to moderate-quality evidence from 2 RCTs containing 566 people diagnosed with
dementia within the last year could not differentiate numbers of nursing home placements,
mortality, cognition or behavioural and psychological symptoms of dementia at 12 or 36
months between people offered psychosocial interventions and people offered follow-up
support or basic counselling at diagnosis.
13.1.4.1.2 Carers
Low- to moderate-quality evidence from 2 RCTs containing 566 carers for people diagnosed
with dementia within the last year could not differentiate quality of life, psychological distress
260
during caregiving, orientation to life, and depressive symptoms at either 12 or 36 months

between people offered psychosocial interventions compared with follow-up support or basic
counselling at diagnosis.
13.1.4.2 Self-management interventions
13.1.4.2.1 People with dementia
Moderate-quality evidence from 1 RCT containing 134 people diagnosed with dementia
within the last year found that people offered self-management interventions had better
cognitive function (measured with the verbal fluency and clock drawing tests) compared with
people offered usual care.
Low-quality evidence from 1 RCT containing 134 people diagnosed with dementia within the
last year could not differentiate quality of life and cognitive function (measured with the
clinical dementia rating scale) between people offered self-management interventions or
usual care.
13.1.4.2.2 Carers
Low-quality evidence from 1 RCT containing 134 carers for people diagnosed with dementia
within the last year could not differentiate quality of life between people offered self-
management interventions or usual care.
13.1.4.3.1 Psychosocial interventions
One partially applicable trial-based cost–utility analysis with potentially serious limitations
explored the cost effectiveness of psychosocial intervention over 36 months. When
compared with usual care, the intervention resulted in a loss of 0.09 QALYs and a decrease
in health and social care costs of €4,433 per person, resulting in an ICER of €49,255 saved
per QALY forgone. In probabilistic sensitivity analysis, the intervention has a 60% probability
of being considered cost effective, if QALYs are valued at €30,000 each.

Relative value of different The committee agreed that long-term outcomes capturing the
outcomes experiences of people living with dementia and their carers, including
quality of life and cognitive function, were the most important to address
this question. It noted that, since these studies were recruiting people
with comparatively minor problems at baseline, studies may need to
have long follow-up periods in order to detect meaningful differences,
whereas in a more advanced population symptomatic benefit may be
detectable over a shorter timeframe.
The committee agreed that differences in mortality and entry into full
time care would also be important, but that it would be unlikely that
studies recruiting people at or around the time of diagnosis would be
able to detect differences in these outcomes.
Trade-off between The committee agreed that the evidence presented did not provide clear
benefits and harms evidence of benefits for either of the post-diagnostic interventions
identified (psychosocial support or self-management). Whilst there were
short-term improvements in quality of life, cognition and depression as
measured by some of the instruments included in the trials, these
effects did not persist, nor were they replicated on other instruments in
trials measuring similar outcomes. The committee therefore agreed
261
there was not sufficient confidence in the evidence to make a positive

recommendation for either of the interventions considered.
Conversely, the committee also agreed it would not be appropriate to
make a negative recommendation, as only a small number of RCTs
were identified, covering only a narrow range of the different types of
interventions that have been developed and tested for people living with
dementia over the whole lifetime of the condition. The committee agreed
that even though the particular interventions considered in these trials
had not been shown to be effective, this could not be used as evidence
that other interventions would not prove to be so. Therefore, the
committee agreed the current evidence did not enable them to make
either positive or negative recommendations.
The committee agreed the lack of effectiveness in the trials meant there
was no evidence that the period around diagnosis should be treated as
a separate, discrete phase in the dementia pathway, requiring
specifically different interventions. It agreed that the severity of
dementia was likely to be a more important factor in which interventions
were effective, rather than the time since diagnosis. In particular, if an
intervention was shown to be effective for people with mild dementia, it
was likely to also be effective for people with mild dementia around the
time of diagnosis. The committee therefore agreed that the population
considered in this section would be covered by recommendations made
for mild dementia in the section of this guideline on non-
pharmacological interventions for cognition, functional ability and
wellbeing.
Trade-off between net The committee noted that the DAISY trial did not find any significant or
health benefits and clinically meaningful differences in health outcomes between the
resource use intervention (psychosocial intervention) and control groups. The
economic evaluation conducted alongside the trial (Søgaard et al.,
2014) found that the psychosocial intervention was a dominated
strategy (as it cost more money than control and produced fewer
QALYs). However, as Søgaard et al (2014) considered costs that were
not relevant to the NICE reference case, we conducted an analysis
where these costs were removed. Under these circumstances, it was
found that the psychosocial intervention was cost saving compared with
the control group, but still produced fewer QALYs. The committee
agreed that it would not be appropriate to make a recommendation
based on evidence which showed that the psychosocial intervention
produced fewer treatment benefits than the control, when any cost
savings expected from the intervention were (a) extremely uncertain
and (b) appear to be achieved by shifting costs from health and social
care budgets to people living with dementia and their carers.
Quality of evidence No evidence was identified about pre- and peri-diagnostic counselling
and support for people living with dementia and their families. The
committee agreed that this was not surprising because although this is
recognised as being good practice, there has been little research in this
area.
The committee highlighted an issue with the control arms in the studies.
The control arm participants would very likely receive an enhanced
version of usual care (often a limited version of the intervention), and
this would reduce the estimated effect sizes.
The committee also highlighted that a possible confounding factor in the
included studies was whether or not people had received pre-diagnostic
counselling and support before entering the included interventions for
post-diagnostic support, as it is not known what the earlier part of the
process was like for the participants. Whilst appropriate randomisation
should mean that the level of pre-diagnostic support is approximately
balanced between the trial arms, the committee agreed that high levels
of pre-diagnostic support may reduce the incremental benefit of the
262
intervention, as people having already received support would have less

potential to gain.
Other considerations The committee highlighted that the use of 1 year post-diagnosis as a
cut-off for this question was an arbitrary figure, but it was not possible to
robustly define the transition from the post-diagnostic phase to more
general support services, as this trajectory would differ considerably
between individuals. It also highlighted that as people with dementia are
now being diagnosed earlier than has been the case historically, this
means that over time the populations in the studies may no longer be
representative of the real world post-diagnosis population.
The committee agreed the key issue for many individuals was not about
the specifics of post-diagnostic support, but rather around continuity of
care as people move from diagnosis in to more general support
services. These issues are considered in the section of this guideline on
the co-ordination of health and social are services.
No recommendations were made
263
13.2 Interventions to promote cognition, independence and

wellbeing
Review questions
 What are the most effective non-pharmacological interventions for supporting cognitive
functioning in people living with dementia?
 What are the most effective non-pharmacological interventions for supporting functional
ability in people living with dementia?
 What are the most effective non-pharmacological interventions to support wellbeing in
people living with dementia?
 What are the most effective methods of supporting people living with dementia to reduce
harm and stay independent?
13.2.1 Introduction
The aim of these review questions was to determine the most effective and cost-effective
non-pharmacological interventions for supporting people living with dementia. This review
covered separate questions, looking at interventions to support cognition, functional ability,
independence and wellbeing, with all the interventions identified considered across all these
categories.
of the review protocol, see Appendix C.
Table 62: Review summary: non-pharmacological interventions to support

independence, cognition, functional ability and wellbeing in people living
with dementia
Interventions Non-pharmacological interventions which may have a positive impact
on cognitive functioning
 Standard care
ability
 Admissions to hospitals/care homes
 Adverse events
The identification of studies was conducted in two stages for these review questions. First, a
systematic review of systematic reviews was conducted by the York Health Economics
consortium, looking to identify the most recent high-quality systematic reviews published on
non-pharmacological interventions for people living with dementia. Where a newer
systematic review was available that completely covered the subject of an earlier review, the
earlier review was not included. In total, 33 relevant reviews were identified, and these
reviews were then used as a source of primary RCTs. In addition, a separate RCT search
was conducted to identify trials published after the search dates of the included reviews,
264
which found a total of 204 records, which were then screened on title and abstract, as
described in the separate sections for each intervention below. Full details of this review are
available in appendix O.
Finally, an additional systematic literature search was carried out to identify any randomised
controlled trials or systematic reviews of randomised controlled trials, where the stated
intention of the intervention was to promote independence in people living with dementia. A
total of 5,152 references were screened at the title and abstract level, with 51 potentially
relevant references being ordered for full text review. Of these references, 15 were selected
for inclusion based on their relevance to the review protocol. The excluded studies are listed,
with reasons for their exclusion, in Appendix B. All data included in the review was extracted
from the primary studies, regardless of from which source the primary paper was identified.
Data was extracted for the outcomes of cognition, activities of daily living, behavioural and
psychological symptoms, global assessment, health-related quality of life, dementia severity,
depression, agitation, carer burden and mortality. In each case, a standardised mean
difference was used to combine all outcomes measures on that scale, and additionally data
were reported on the mean difference scale for the pre-specified primary outcome measure
in each domain. These primary outcomes were:
 Cognition – MMSE
 Behavioural and psychological symptoms – NPI
 Activities of daily living – ADCS-ADL
 Global assessment – CIBIC+
 Health-related quality of life – QoL-AD (dementia-specific), EQ-5D (generic)
 Dementia severity (CDR)
 Depression (CSDD)
 Agitation (CMAI)
 Carer burden (ZBI)
Where papers did not contain any of the outcome domains of interest in an extractable
format, the paper was excluded. When studies reported outcome measures at multiple time
points, data were extracted post-intervention (the first measurement time after completion of
the intervention) and at long-term follow-up (the final time point of the study).
13.2.2.1 Cognitive stimulation, cognitive training and cognitive rehabilitation
There is considerable inconsistency in the terminology used in the literature around cognitive
stimulation, cognitive training and cognitive rehabilitation for people living with dementia. In
particular, the terms cognitive training and rehabilitation have often been used
interchangeably despite arising from different disciplines and having very different intentions
in what they are trying to achieve. For the purposes of the guideline, the following definitions
of these interventions have been used:
 Cognitive stimulation: Engaging in a range of activities and discussions (usually in a

group) that are aimed at general improvement of cognitive and social functioning.
 Cognitive training: Guided practice on a set of standard tasks that are designed to
reflect particular cognitive functions. There may be a range of difficulty levels, to fit
the tasks to each person’s level of ability.
 Cognitive rehabilitation: Identifying functional goals that are relevant to the person
living with dementia, and working with them and their family members or carers to
achieve these. The emphasis is on improving or maintaining functioning in everyday
265
life, building on the person’s strengths and finding ways to compensate for
impairments, and supporting independence. Cognitive rehabilitation does not aim to
improve cognition, but addresses the disability resulting from the impact of cognitive
impairment on everyday functioning and activity. Rehabilitation is sometimes referred
to as ‘reablement’.
In total, 86 potentially relevant papers were identified for full-text review. Of these, 44 were
excluded (with reasons for exclusion given in appendix B), and 40 were included for the three
relevant interventions. Of these, 25 reported data on cognitive stimulation, 13 on cognitive
training and 5 on cognitive rehabilitation (1 paper reported data on both cognitive training and
cognitive rehabilitation). Evidence tables for these studies are given in appendix E, and
GRADE profiles in appendix G.
13.2.2.2 Self-management groups
excluded (with reasons for exclusion given in appendix B), and 3 were included. These
papers provided data on self-management groups for people living with dementia. Evidence
tables for these studies are given in appendix E, and GRADE profiles in appendix G.
13.2.2.3 Reminiscence therapy
papers provided data on either individual or group reminiscence therapy for people living with
dementia. An additional paper was identified at the re-run stage which included both group
and individual therapy and was included. Therefore, 12 papers were included in total. Some
interventions only focused on the person living with dementia, whilst others were dyadic and
also included a component for carers. Evidence tables for these studies are given in
appendix E, and GRADE profiles in appendix G.
13.2.2.4 Occupational therapy
papers provided data on occupational therapy for people living with dementia. Evidence
tables for these studies are given in appendix E and GRADE profiles in appendix G.
13.2.2.5 Psychotherapy
papers provided data on psychotherapy for people living with dementia, where the trials did
not require people to have a diagnosis of a non-cognitive symptom (such as depression or
anxiety) at baseline. Trials looking at the treatment of anxiety or depression in people living
with dementia are covered in the section of this guideline on managing non-cognitive
symptoms. Evidence tables for these studies are given in appendix E, and GRADE profiles in
appendix G.
13.2.2.6 Exercise
excluded (with reasons for exclusion given in appendix B), and 22 were included. Of these, 1
primarily focused on tai chi, 2 on dance therapy, 2 on non-aerobic exercise, 5 on aerobic
266
exercise, 9 on combined aerobic/non-aerobic exercise, and 3 on multimodal interventions

with a primary exercise component, but also containing other interventions such as
occupational therapy or cognitive behavioural therapy. In all analyses including multimodal
interventions, a sensitivity analysis was also conducted excluding those trials, to estimate the
effect of studies with exercise only as the intervention. Evidence tables for these studies are
given in appendix E, and GRADE profiles in appendix G.
13.2.2.7 Nutrition
excluded (with reasons for exclusion given in appendix B), and 39 were included. Studies
comparing nutritional supplements with either placebo or usual care were included, but
studies comparing to pharmacological interventions (such as cholinesterase inhibitors or
memantine) were excluded.
Evidence tables for these studies are given in appendix E, and GRADE profiles in appendix
G.
13.2.2.8 Music therapy
excluded (with reasons for exclusion given in appendix B), and 16 were included. Nine
publications specified non-cognitive symptoms such as anxiety as an inclusion criteria.
Twelve publications compared music therapy to standard care with one also including an
active comparator arm. Four publications compared music therapy to active comparator
sessions, such as cooking. Music therapy varied widely across the publications; differences
were observed for example in the type (active or receptive therapy), the delivery setting
(individualised or small group sessions), number of sessions (1 to around 60), as well as the
length of sessions (10 min to 90 min) and experimental period (10 min to 4 months). Eight
publications included a follow-up period which varied from 1 hour after treatment session to 6
months. Evidence tables for these studies are given in appendix E, and GRADE profiles in
appendix G.
13.2.2.9 Aromatherapy
papers provided data on aromatherapy, in particular Melissa and Lavendula, for people living
with dementia and non-cognitive symptoms. Evidence tables for the included studies are
13.2.2.10 Light therapy
papers provided data on light therapy for people living with dementia. Evidence tables for the
included studies are given in appendix E, and GRADE profiles in appendix G.
13.2.2.11 Motor-cognitive dual-task training
In total, 3 potentially relevant papers were identified looking at motor-cognitive dual-task

training (which involves the addition of cognitive tasks to mobility interventions aimed at
domains such as gait to balance) for full-text review. All 3 of these papers were excluded
(with reasons for exclusion given in appendix F), and 0 were included.
267
13.2.2.12 Non-invasive brain stimulation
excluded (with reasons for exclusion given in appendix B), and 6 were included. Five of the
included publications reported on people with Alzheimer’s disease of varying severity and 1
on people with mild vascular dementia. In 4 publications the investigators used rTMS
(repetitive transcranial magnetic stimulation) of 1 Hz, 10 Hz or 20 Hz intensity; in 2 studies
tDCS (transcranial direct current stimulation), of 2 mA, was used. All publications included
multiple sessions with the length of active phase varying from 4 days to 6 weeks. All
publications included a follow-up period which varied from 18 days to 4.5 months with 1
study including maintenance treatment during the follow-up. Evidence tables for these
studies are given in appendix E, and GRADE profiles in appendix G.
13.2.2.13 Acupuncture
excluded (with reasons for exclusion given in appendix B) and 2 were included. These
papers provided data on acupuncture for people living with dementia. Evidence tables for the
13.2.2.14 Assistive technology
A 2017 Cochrane review identified no randomised controlled trials studying the effectiveness
of assistive technology in people living with dementia. This review looked for studies
evaluating the efficacy of assistive technology for supporting memory, daily performance of
personal and instrumental activities of daily living (ADL) independence, behavioural and
psychological symptoms, need for informal and formal care, quality of life and carer burden.
13.2.2.15 Assisted animal therapy
One paper was identified during the re-run stage and was included in the review. This paper
evaluated the effectiveness of animal assisted therapy by interaction with a dog compared
with usual treatment. Reported outcomes were: depression, anxiety and quality of life.
Evidence tables for the included studies are given in appendix E, and GRADE profiles in
appendix G.
13.2.2.16 Robotic pet therapy
evaluated the effectiveness of robotic pet therapy using PARO (personal robot), a FDA-
approved device designed to look like a baby harp seal. Individual interaction by the
participants was encouraged at group sessions. Evidence tables for the included studies are
13.2.2.17 Adapted mindfulness program
evaluated the effectiveness of an adapted mindfulness program plus treatment as usual
versus treatment as usual alone and reported outcomes on cognition, quality of life and
depression. Evidence tables for the included studies are given in appendix E, and GRADE
profiles in appendix G.
268
13.2.2.18 Home safety toolkit
One paper was identified, evaluating the effectiveness of a home safety toolkit for promoting
independence and reducing harm in people living with dementia. Evidence tables for the
13.2.3.1 Systematic review of published economic evaluations

(CUAs) evaluating non-pharmacological interventions for people living with dementia. In total,
3,229 articles were returned, of which 57 were selected as potentially relevant and retrieved
for full-text review. Additionally, a committee member made available a pre-publication
manuscript detailing an NIHR-funded RCT. A final literature search performed to identify any
newly indexed papers found 1 new paper. In total, 4 publications were judged to be at least
partially applicable to the review questions and were therefore included. Of these studies, 1
evaluated cognitive rehabilitation, 1 evaluated maintenance cognitive stimulation therapy, 1
evaluated joint reminiscence group therapy and 1 evaluated exercise. Details of the literature
search are provided in Appendix D.
13.2.3.1.1 Cognitive rehabilitation
Clare et al. (in press) conducted a cost–utility analysis alongside the GREAT RCT. The
inclusion criteria were that participants had an ICD-10 diagnosis of Alzheimer’s disease,
vascular or mixed dementia with mild to moderate cognitive impairment (MMSE score ≥ 18).
Cognitive rehabilitation was administered in additional to usual treatment. It was delivered by
therapists with experience of rehabilitative interventions in 10 individual sessions over
3 months, followed by 4 maintenance sessions over the next 6 months. For further details,
To cost the intervention, the Client Services Receipt Inventory was completed by the person
with dementia and carer together. Costs of health and social care were calculated by
applying relevant, nationally generalisable unit costs, including data from NHS reference
costs and PSSRU.
Cost–utility analysis was undertaken separately for participants with dementia and their
carers using QALYs generated from the EQ-5D-3L for carers and DEMQOL-U for the person
with dementia. Cases included imputed costs and QALYs where data were missing.
Base-case results (Table 63) suggested that cognitive rehabilitation was associated with
increased health and social care costs (though the data were also consistent with no
difference at a 95% confidence level), but no additional benefit could be detected. For carers,
there was no difference in utility scores between the control group and the cognitive
rehabilitation group.
269
Table 63: Base-case cost–utility results (person with dementia) from Clare et al. (in
press)
Effect
Treatment Cost [95%CI] Effect [95%CI] Cost [95%CI] [95%CI] ICER
£4,286 0.45 QALYs
Usual care
[3,353 to 5,672] [0.44 to 0.46]
Cognitive £5,397 0.45 QALYs £1,110 0.001 QALYs £1,110,000
rehabilitation [4,563 to 6,388] [0.44 to 0.46] [-382 to 2,187] [-0.01 to 0.01] /QALY
In probabilistic analysis, the probability that cognitive rehabilitation is cost effective was very
low, regardless of the value that was assumed for QALYs, from the health and social care
perspective
The authors noted that the attainment of personally set goals did not bring about changes in
those domains that are measured in the dementia-specific quality of life measure
(DEMQOL), nor did it bring about changes in carer health-related quality of life (measured by
EQ5D). They authors concluded that, for commissioning purposes, they did not find that
cognitive rehabilitation is cost-effective when gauged against QALY gains for either
participants with dementia or carers.
13.2.3.1.2 Maintenance cognitive stimulation therapy
D’Amico et al. (2015) conducted a cost–utility analysis alongside the Orrell et al. (2014) RCT
investigating 24 weeks of weekly maintenance cognitive stimulation therapy (CST) versus no
additional therapy following a 7-week, 14-session initial CST programme (see 13.2.2, above).
The authors’ base case adopted a health and social care perspective. The Client Service
Receipt Inventory was used to capture resource use. The cost of maintenance CST itself
included a 1-day training course for facilitators, materials and equipment, costs of 2
co-facilitators and transport costs for participants.
Unit costs were taken from the PSSRU and BNF and from market sources for equipment and
adaptations. Costs were expressed in 2011 British pounds.
QALYs were calculated from both generic and dementia-specific quality of life measures,
using both participant- and proxy-reported measures, by ‘area under the curve’ analysis, with
linear interpolation between assessment points.
Base-case results (Table 64) show that maintenance CST was associated with a non-
significant increase in both costs and QALYs. The ICER approached levels that might be
considered a reasonable use of health and social care funds only when QALYs were
estimated using the proxy-rated EQ-5D. Probabilistic sensitivity analysis showed that, using
this measure, there was a 40% chance that maintenance CST was associated with an ICER
of £20,000 or better. There was no evidence that maintenance CST would be considered
cost effective when any of the other QALY measures was used.
270
Table 64: Base-case cost–utility results from D’Amico et al. (2015)

Treatment Cost Effect Cost [95%CI] Effect [95%CI] ICER
EQ-5D
Usual care NR NR
£475 0.0013 QALYs £365,276
Maintenance CST NR NR
[−314 to 1,264] [−0.020 to 0.022] /QALY
Proxy-rated EQ-5D
Usual care NR NR
£474 0.0176 £26,835
[−315 to 1,263] [−0.005 to 0.040] /QALY
DEMQOL
Usual care NR NR
£518 0.0039 QALYs £132,539
[−347 to 1383] [−0.009 to 0.017] /QALY
Proxy-rated DEMQOL
Usual care NR NR
£402 0.0062 QALYs £64,785
[−442 to 1,245] [−0.005 to 0.017] /QALY
13.2.3.1.3 Joint reminiscence group therapy
Woods et al. (2016) conducted a cost–utility analysis alongside the REMCARE RCT. To be
included in the study, patients had to have mild/moderate dementia diagnosed using DSM-IV
criteria.
Joint reminiscence group therapy followed the ‘Remembering Yesterday, Caring Today’
manual. Group sessions were held weekly over 12 consecutive weeks, followed by 7 monthly
maintenance group sessions. Sessions were led by 2 trained facilitators in each centre,
supported by trained volunteers.
The authors undertook a micro-costing of the intervention by recording the types and
quantities of resource input including: staff time, materials, room rental, training and
supervision of staff. Appropriate national salary scales were used for staff who were NHS or
university employees. Costs were adjusted to the price year 2010 and expressed in British
pounds.
Cost–utility analysis was undertaken separately for participants with dementia and their
carers using QALYs generated from the self-completed EQ-5D-3L. Carers completed the
measure from their own perspective and for the person with dementia, who would also
complete it whenever possible.
Base-case results (Table 65) showed that reminiscence therapy was associated with
substantial extra costs of over £1,000 per participant, but did not produce any meaningful
QALY gains for people living with dementia or carers.
271
Table 65: Base-case cost–utility results from Woods et al. (2016)

Cost Effect Cost Effect ICER
Treatment (£) (QALYs) (£) (QALYs) (£/QALY)
Person living with dementia
Usual care 4,309 0.643
Reminiscence 5,853 0.644 1,544 0.001 1,544,000
Carer
Usual care 1,359 0.633
Maintenance CST 2,495 0.632 1,136 -0.001 dominated
The authors stated that, whilst a full probabilistic analysis had been planned, the results
showed that generating cost-effectiveness acceptability curves would not be meaningful. The
authors concluded that the REMCARE trial does not support the clinical effectiveness or
cost-effectiveness of joint reminiscence group therapy.
13.2.3.1.4 Exercise
Sopina et al. (2017) conducted a cost–utility analysis alongside a Danish RCT (Hoffmann et
al., 2013). To be included in the study, participants had to have a confirmed diagnosis of
Alzheimer’s disease and an MMSE of 20 or more. The control group received treatment as
usual while the intervention group performed 1 hour of supervised moderate-to-high intensity
aerobic exercise 3 times weekly for 16 weeks.
Health-related quality of life was measured using the 5-level version of the EQ-5D (EQ-5D-
5L), with ratings elicited from participants and the primary caregivers as proxy respondents.
The study had relatively short follow-up period and did not include cost of health and social
care used by the participants.
Base-case cost–utility results (Table 66) suggested that the intervention was associated with
significant cost increases but very small QALY benefits, leading to high ICERs in excess of
€100,000/QALY.
Table 66: Base-case cost–utility results from Sopina et al. (2017)

Treatment Cost Effect Cost [95%CI] Effect ICER
EQ-5D-5L
Usual care NR NR
0.00313 €158,520
NR NR
Exercise €496 [495 to 497] QALYs /QALY
Proxy-rated EQ-5D-5L
Usual care NR NR
0.00411 €120,790
NR NR
Exercise €496 [495 to 497] QALYs /QALY
In probabilistic analysis, the authors found that, using participant-rated EQ-5D-5L, the
chance of the intervention being cost effective only reached 50% when QALYs were valued
at €175,000 or greater each.
272
13.2.3.2 Original economic analysis
13.2.3.2.1 Methods
Owing to the paucity of published health economic studies for non-pharmacological

interventions in dementia, the GDG prioritised these questions for original economic
modelling.
A series of simple cost–utility models was developed that sought to simulate the average
patient receiving each intervention of interest, compared with usual care. The models used a
simple area-under-the-curve method to estimate differences, over time, between a person
receiving the intervention and one receiving usual care in multiple clinical outcomes that
could then be used to estimate health-related quality of life (and, consequently, QALYs).
The systematic reviews undertaken for this chapter (see 13.2.2, above) were used to identify
non-pharmacological interventions for which sufficient data were available for modelling.
Single summary estimates for each continuous variable of interest were drawn from the
meta-analyses, which were then substituted into published models (2 univariable and
1 multivariable), drawn from a review of available literature, to estimate the health-related
quality of life that could be expected for the typical person living with dementia receiving the
intervention in question. The constraints of available utility models dictated that clinical
outcomes of interest were effects in cognitive, functional and behavioural domains. Data
synthesised as standardised mean differences (SMDs) were re-expressed in units needed
for the utility models (e.g. cognitionMMSE, functionalDAD, behaviouralNPI), using
pooled standard deviations from the assembled evidence-base.
Two datapoints were used to estimate treatment effect: change at the end of the intervention
and (where available) change at post-intervention follow-up. For each outcome-within-
intervention, these timepoints were defined by a weighted average of intervention length and
follow-up length, respectively, in each trial contributing to the effect estimate, with weights
defined by the relevant meta-analysis.
To extrapolate any observed benefits of treatment beyond the empirical data, advice from the
Guide to the methods of technology appraisal (2013) was followed. This stipulates that a
range of assumptions should be explored, ranging from no additional benefit to an indefinite
preservation of gains achieved over the course of the intervention. For the base case, a
scenario between these extremes was adopted: patients revert to natural history 6 months
after the longest follow-up period available in a linear fashion.
Figure 1 provides a schematic depiction of the method, using the example of change in
MMSE (where the multivariable utility model was used, similar analyses were performed for
other relevant outcomes and the joint effect of intervention on all domains estimated).
273
30 INTERVENTION FOLLOW-UP CONVERGENCE
meta-analysed mean meta-analysed mean Intervention

25 difference between difference between
Control
intervention and control intervention and control
at end of intervention at last follow-up
20
MMSE
15
10
weighted mean
length of
intervention in RCTs
5
weighted mean length of follow-up in RCTs 6 months

0
100
200
500
600
700
300
400
0
Days
Model shown reflects base-case assumptions; alternative scenarios (including immediate onset of treatment
benefit, and longer and shorter durations of post-follow-up benefit) were explored in sensitivity analysis
NB some quantities have been exagerrated, compared with empirical data, for schematic clarity
Figure 1: Schematic depiction of model
In the base case, the effect difference between any 2 time points occurred in a linear fashion.
In instances where the average follow-up period exceeded 18 months, patients in the control
and interventions arms were assumed to converge at 24 months (730 days), the maximal
time horizon for the model.
Costs included in the analyses only related to the costs of delivering the interventions
themselves. While it may be expected that effective interventions may reduce other health
and social care costs (for example, improvements in functional ability might reduce
requirement for domiciliary support), there was no evidence of significant differences in costs
between treatment and control in any of the within-trial analyses summarised above
(13.2.3.1). Therefore, it was assumed that, aside from the expense incurred in delivering the
intervention in question, there would be no difference in total costs between people who do
and do not receive the intervention. Resource use for each of the interventions was
estimated, where possible, using evidence from the assembled RCTs. This included the
number of sessions, length of sessions, grade of staff required to deliver the intervention.
Where data were not available from clinical papers, the guideline committee was consulted
to provide estimates of resource use in the English NHS setting. Where unit cost data were
not available from study papers and PSSRU unit costs, the guideline committee were
consulted to provide estimates of unit costs in the English NHS setting. Resource use and
unit cost data were combined to produce a cost for each intervention modelled.
Probabilistic sensitivity analyses and one-way sensitivity analyses were conducted to

examine the effects of model input parameters that were subject to uncertainty. Scenario
analyses were also conducted to examine the effects of varying several parameters at the
same time. A full description can be found in Appendix J.
274
The analyses used a patient perspective for outcomes and an NHS+PSS perspective for
costs, in line with Developing NICE guidelines (2014). Costs and outcomes were discounted
at the rate of 3.5% per annum.
Some of the key assumptions of the model include:

 Non-pharmacological interventions are unable to alter the disease process or mortality
rates in patients with dementia.
 The maximal effects of the intervention are likely to be limited to the duration for which the
patient receives it.
 Clinical measures – and, by extension, utility scores – change in a linear fashion between
points at which measurements are estimated.
 The model does not consider any difference of resource use (i.e. hospital inpatient stays,
GP appointments etc.) or disutility as a result of interventions.
More information about the model can be found in appendix J.
13.2.3.2.2 Results
Group cognitive stimulation therapy
The base-case model suggested that group CST was associated with a benefit of a little over
0.033 QALYs relative to control, at an additional cost of £653, leading to an ICER of
£19,966/QALY (Table 67). One-way sensitivity analysis found that the model was extremely
sensitive to almost all parameters in the model: varying any parameter within plausible range
generates results lying on either side of a £20,000/QALY threshold.
Table 67: Incremental costs and effects for group cognitive stimulation therapy versus
control
Absolute Incremental Ceiling £ for
Costs QALYs Costs QALYs ICER this benefita
Base case (multivariable model)
Control £0 1.164
Intervention £653 1.197 £653 0.033 £19,966 £654
Univariable model (MMSE)
Control £0 1.069
Intervention £653 1.080 £653 0.010 £62,973 £207
Univariable model (ADCS-ADL)
Control £0 1.007
Intervention £653 1.029 £653 0.022 £29,986 £435
a The maximum intervention cost at which benefits of the magnitude estimated here would lead to an ICER of
£20,000/QALY or better
Probabilistic sensitivity analysis (Figure 2) suggested that the probability that intervention is
cost-effective is around 50%, if QALYs are valued at £20,000 each, or 70%, if a higher
threshold of £30,000/QALY is used.
In a scenario analysis where costs relating to staff training, staff travel, venue, and
administration were set to zero (that is, only staff time and participant travel expenses were
accounted for), group CST was found to cost an additional £554 compared with control, and
produced an ICER of £16,942/QALY.
275
Figure 2: Cost-effectiveness acceptability curve – group cognitive stimulation versus

usual care
Cognitive rehabilitation (individual)
The base-case model suggested that cognitive rehabilitation was associated with a benefit of
a little over 0.027 QALYs relative to control, at an additional cost of £1,827, leading to an
ICER of £66,863/QALY (Table 68). One-way sensitivity analysis found that the model was
most sensitive to the SMD value for BPSD at long term follow-up; however, no parameter
variations resulted in an ICER lower than £20,000/QALY.
Table 68: Incremental costs and effects for individual cognitive rehabilitation versus
control
Control £0 1.164
Intervention £1,827 1.191 £1,827 0.027 £66,863 £546
Control £0 1.069
Intervention £1,827 1.113 £1,827 0.044 £41,900 £872
Control £0 1.031
Intervention £1,827 1.101 £1,827 0.070 £26,006 £1,405
276
1
0.9
Probability cost effective

0.8
0.7
0.6
CR (Individual)
0.5
Usual care
0.4
0.3
+0.050 +0.100 +0.150 0.2
0.1
0
£0K £10K £20K £30K £40K £50K
Incremental QALYs Value of 1 QALY
Figure 3: CEAC for individual cognitive rehabilitation versus control
Cognitive training for groups
The base-case model suggested that cognitive training for groups was associated with a
benefit of a little over 0.003 QALYs relative to control, at an additional cost of £653, leading
to an ICER of £254,615/QALY. One-way sensitivity analysis found that the model was most
sensitive to the use of the univariable MMSE model and the SMD for cognition at the end of
the intervention. If the MMSE values in the univariable MMSE model and the SMD for
cognition at the end of the intervention were increased to their highest plausible value,
cognitive training may be a cost-effective treatment as the incremental net monetary benefit
would be greater than zero.
Table 69: Incremental costs and effects for group cognitive training versus control
Control £0 1.164
Intervention £653 1.166 £653 0.003 £251,615 £52
Control £0 1.069
Intervention £653 1.163 £653 0.094 £6,978 £1,871
Control £0 1.031
Intervention £653 1.039 £653 0.008 £78,324 £167
277
1
0.9

0.8
0.7
0.6
CT (Group)
0.5
Usual care
0.4
0.3
+0.050 +0.100 +0.150 0.2
0.1
0
£0K £10K £20K £30K £40K £50K
Figure 4: CEAC for group cognitive training versus control
Reminiscence therapy in a group setting
Reminiscence therapy in a group setting relative to control was dominated in the base case,
and had high ICERs in the univariable MMSE and ADCS-ADL models (Table 70). One-way
sensitivity analysis found that the model was most sensitive to a lower cost per participant
per course; however, variations to any single parameter did not result in ICERs below a
£20,000/QALY threshold.
Table 70: Incremental costs and effects for group reminiscence therapy versus control
Control £0 1.164
Intervention £964 1.160 £964 -0.004 dominated -£74
Control £0 1.069
Intervention £964 1.087 £964 0.018 £52,853 £365
Control £0 1.031
Intervention £964 1.032 £964 0.001 £809,456 £24
278
1
0.9

0.8
0.7
0.6
Reminiscence therapy
0.5 (Group)
0.4 Usual care
0.3
+0.050 +0.100 +0.150 0.2
0.1
0
£0K £10K £20K £30K £40K £50K
Figure 5: CEAC for group reminiscence therapy versus control
One-to-one exercise therapy
The base-case model suggested that one-to-one exercise therapy was associated with a
benefit of a little over 0.023 QALYs relative to control, at an additional cost of £1,776, leading
to an ICER of £76,678/QALY(Table 71). One-way sensitivity analysis found that the model
was most sensitive to a lower cost per course of individualised exercise therapy, but would
still not make individualised exercise therapy a cost-effective treatment at the £20,000/QALY
threshold.
Table 71: Incremental costs and effects for one-to-one exercise therapy versus control
Control £0 1.164
Intervention £1,776 1.187 £1,776 0.023 £76,678 £463
Control £0 1.069
Intervention £1,776 1.101 £1,776 0.032 £55,573 £639
Control £0 1.031
Intervention £1,776 1.058 £1,776 0.027 £65,402 £543
administration were set to zero – and only staff time and participant travel expenses were
accounted for, one-to-one exercise therapy was found to cost an additional £1,445 compared
to control, and produced an ICER of £62,382/QALY.
279
1
0.9

0.8
0.7
0.6
Exercise (Individual)
0.5
Usual care
0.4
0.3
0.2
0.1
+0.050 +0.100 +0.150 0
£0K £10K £20K £30K £40K £50K
Figure 6: CEAC for one-to-one exercise therapy versus control
Group exercise therapy
The base-case model suggested that group exercise therapy was associated with a benefit
of a little over 0.042 QALYs relative to control, at an additional cost of £1,727, leading to an
ICER of £41,359/QALY (Table 72). One-way sensitivity analysis found that the model was
most sensitive to the cost per group session but only the indefinite long-term extrapolation
scenario resulted in an ICER lower than £20,000/QALY.
Table 72: Incremental costs and effects for group exercise versus control
Control £0 1.164
Intervention £1,727 1.206 £1,727 0.042 £41,359 £835
Control £0 1.069
Intervention £1,727 1.124 £1,727 0.054 £31,791 £1,087
Control £0 1.031
Intervention £1,727 1.049 £1,727 0.019 £92,373 £374
accounted for, group exercise therapy was found to cost an additional £1,529 compared to
control, and produced an ICER of £36,600/QALY.
280
1
0.9

0.8
0.7
0.6
0.5 Exercise (Group) Usual care
0.4
0.3
0.2
0.1
+0.050 +0.100 +0.150 0
£0K £10K £20K £30K £40K £50K
Figure 7: CEAC for group exercise versus control
Group exercise therapy for people with severe dementia
The base-case model suggested that group exercise therapy for people with severe
dementia was associated with a benefit of a little over 0.05 QALYs relative to control, at an
additional cost of £1,510, leading to an ICER of £329,685/QALY (Table 73). One-way
sensitivity analysis found that the model was most sensitive to the cost per patient per
course, but no parameter variations suggested that the intervention would cost effective at a
£20,000/QALY threshold.
Table 73: Incremental costs and effects for group exercise therapy for people with
severe dementia versus control
Control £0 1.164
Intervention £1,510 1.168 £1,510 0.005 £329,685 £92
Control – –
Intervention – – – – – –
Control £0 1.031
Intervention £1,510 1.045 £1,510 0.014 £105,987 £285
accounted for, group exercise therapy for people with severe dementia was found to cost an
281
additional £1,367 compared to control, and produced an ICER of £298,488/QALY.
1
0.9

0.8
0.7
0.6
Exercise (Group -- Severe)
0.5
Usual care
0.4
0.3
0.2
0.1
+0.050 +0.100 +0.150 0
£0K £10K £20K £30K £40K £50K
Figure 8: CEAC for group exercise therapy for people with severe dementia versus
control
Group music therapy (participatory)
The base-case model suggested that participatory group music therapy was associated with
a benefit of a little over 0.016 QALYs relative to control, at an additional cost of £434, leading
to an ICER of £26,944/QALY (Table 74). One-way sensitivity analysis found that plausible
variations to 8 parameters resulted in ICERs lower than £20,000/QALY, including those
relating to long-term extrapolation of treatment effects, lower costs of treatment, and
treatment effects at the upper 95% confidence interval of synthesised estimates.
Table 74: Incremental costs and effects for group music therapy versus control
Control £0 1.164
Intervention £434 1.180 £434 0.016 £26,944 £322
Control £0 1.069
Intervention £434 1.083 £434 0.014 £31,369 £276
Control £0 1.031
Intervention £434 1.059 £434 0.028 £15,599 £556
282
accounted for, participatory group music therapy was found to cost an additional £364s
compared to control, and produced an ICER of £22,623/QALY.
1
0.9

0.8
0.7
0.6
Music therapy
0.5 (Group -- Active)
0.4 Usual care
0.3
+0.050 +0.100 +0.150 0.2
0.1
0
£0K £10K £20K £30K £40K £50K
Figure 9: CEAC for group music therapy versus control
One-to-one music therapy
The base-case model suggested that one-to-one music therapy was associated with a
benefit of a little over 0.019 QALYs relative to control, at an additional cost of £1,010, leading
to an ICER of £52,970 (Table 75). One-way sensitivity analysis found that the model was
most sensitive to the long-term extrapolation scenario, with indefinitely projected benefit
producing ICERs below £20,000/QALY.
Table 75: Incremental costs and effects for one-to-one music therapy versus control
Control £0 1.164
Intervention £1,010 1.183 £1,010 0.019 £52,970 £381
Control £0 1.069
Intervention £1,010 1.145 £1,010 0.076 £13,243 £1,525
Control – –
283
accounted for, one-to-one music therapy was found to cost an additional £883 compared to
1
0.9

0.8
0.7
0.6
Music therapy
0.5 (Individual)
0.4 Usual care
0.3
+0.050 +0.100 +0.150 0.2
0.1
0
£0K £10K £20K £30K £40K £50K
Figure 10: CEAC for one-to-one music therapy versus control
Occupational therapy
The base-case model suggested that occupational therapy was associated with a benefit of a
little over 0.010 QALYs relative to control, at an additional cost of £1,241, leading to an ICER
of £130,349/QALY (Table 76). One-way sensitivity analysis found that the model was most
sensitive to the MMSE and ADL variables, but no sensitivity analysis resulted in an ICER
lower than £20,000/QALY.
Table 76: Incremental costs and effects for occupational therapy versus control
Control £0 1.164
Intervention £1,241 1.173 £1,241 0.010 £130,249 £191
Control – –
Control £0 1.031
Intervention £1,241 1.055 £1,241 0.025 £50,509 £491
284
accounted for, occupational therapy was found to cost an additional £1,176 compared to
1
0.9

0.8
0.7
0.6
Occupational Therapy
0.5 (Individual)
0.4 Usual care
0.3
+0.050 +0.100 +0.150 0.2
0.1
0
£0K £10K £20K £30K £40K £50K
Figure 11: CEAC for occupational therapy versus control
13.2.4.1 Cognitive stimulation therapy
Moderate-quality evidence from up to 23 RCTs containing 1,398 participants found a

clinically meaningfully improvement in cognition in people living with mild/moderate dementia
offered cognitive stimulation therapy versus usual care.
Low- to high-quality evidence from up to 11 RCTs containing 895 participants could not
detect clinically meaningful differences in activities of daily living, behavioural and
psychological symptoms, depressive symptoms, quality of life or carer burden between
people living with mild/moderate dementia offered cognitive stimulation therapy versus usual
care.
13.2.4.1.1 Economic evidence
One directly applicable original cost–utility model with potentially serious limitations
comparing high-intensity group cognitive stimulation therapy with usual care suggested that
group cognitive stimulation therapy is associated with an ICER of approximately
£20,000/QALY. However, one-way sensitivity analysis found that this finding was extremely
sensitive; varying almost any parameter within a plausible range generates results lying on
either side of a £20,000/QALY threshold. A scenario analysis where costs relating to staff
training, staff travel, venue and administration were set to zero resulted in an ICER of
£17,000/QALY.
One directly applicable cost–utility analysis with minor limitations conducted alongside a 6-
month RCT explored the cost effectiveness of maintenance cognitive stimulation therapy in
patients in England. Only where QALYs were calculated from proxy EQ-5D was maintenance
CST associated with an ICER of less than £30,000 per QALY. When QALYs were calculated
using the person living with dementia’s own rating, the intervention was not cost-effective at
6 months.
285
13.2.4.2 Cognitive training
Very low- to moderate-quality evidence from up to 12 RCTs containing 608 participants could
not detect clinically meaningful differences in cognition, activities of daily living, behavioural
and psychological symptoms, depressive symptoms, quality of life or carer burden between
people living with mild/moderate dementia offered cognitive training versus usual care.
comparing high-intensity group cognitive training with usual care suggested that the
intervention is associated with a base-case ICER of around £250,000/QALY. Alternative
model assumptions resulted in a much more favourable of ICER of around £7,000/QALY;
however, this was based on an estimated effect in the cognitive domain that was heavily
influenced by 1 small RCT’s extremely large but uncertain estimate of benefit. Other
sensitivity analyses suggested a low probability of the intervention being associated with an
ICER of £20,000/QALY or better.
13.2.4.3 Cognitive rehabilitation
Moderate-quality evidence from up to 3 RCTs containing 527 participants found a clinically

meaningfully improvement in activities of daily living in people living with mild/moderate
dementia offered cognitive rehabilitation versus usual care.
not detect clinically meaningful differences in cognition, behavioural and psychological
symptoms, depressive symptoms, quality of life or carer burden between people living with
mild/moderate dementia offered cognitive rehabilitation versus usual care.
comparing high-intensity cognitive rehabilitation with usual care showed that cognitive
rehabilitation was associated with an ICER of around £67,000/QALY. One-way and
probabilistic sensitivity analyses showed there is a very low probability of the intervention
being associated with an ICER of £20,000/QALY or better.
One directly applicable cost–utility analysis with minor limitations conducted alongside an
RCT explored the cost effectiveness of cognitive rehabilitation compared with usual care. For
persons with dementia, the intervention was associated with substantial additional costs and
negligible QALY gains, leading to an ICER in excess of £1m/QALY. There was no difference
in quality of life between the cognitive rehabilitation group and the control group for carers of
the person with dementia.
13.2.4.4 Self-management groups
Low- to moderate-quality evidence from up to 3 RCTs containing 291 participants could not
detect clinically meaningful differences in cognition, depressive symptoms or quality of life
between people living with mild dementia offered access to self-management groups versus
usual care.
286
13.2.4.5 Reminiscence therapy
Very-low to moderate-quality evidence from up to 8 RCTs containing 1,432 participants

found clinically meaningful post-intervention improvements in cognition and depressive
symptoms in people living with dementia offered reminiscence therapy versus usual care, but
the effects on depressive symptoms did not persist at long-term follow-up.
Low- to moderate-quality evidence from up to 5 RCTs containing 1,071 participants could not
detect clinically meaningful differences in activities of daily living, behavioural and
psychological symptoms, quality of life, agitation or carer burden between people living with
dementia offered reminiscence therapy versus usual care.
suggested that high-intensity group reminiscence therapy was a dominated strategy as it
cost more and produced fewer QALYs than usual care. One-way and probabilistic sensitivity
analyses showed no probability of the intervention being associated with an ICER of
£20,000/QALY or better.
One directly applicable cost–utility analysis with minor limitations conducted alongside an
RCT found that, when compared with usual care, joint reminiscence group therapy was
associated with substantial extra costs of over £1,000 per participant, but did not produce
any meaningful QALY gains for people living with dementia or carers, leading to an ICER in
excess of £1m/QALY.
13.2.4.6 Occupational therapy
Low- to high-quality evidence from up to 4 RCTs containing 491 participants found clinically
meaningful post-intervention improvements in depressive symptoms and quality of life in
people living with dementia offered occupational therapy versus usual care, but the effect on
quality of life did not persist at long-term follow-up.
not detect clinically meaningful differences in activities of daily living, agitation or carer
burden between people living with dementia offered occupational therapy versus usual care.
comparing occupational therapy with usual care suggested that the intervention was
associated with a base-case ICER of around £130,000/QALY. One-way, probabilistic and
scenario analyses showed no probability of the intervention being associated with an ICER of
£20,000/QALY or better.
Moderate-quality evidence from up to 3 RCTs containing 125 participants found a clinically

meaningful post-intervention improvement in depressive symptoms in people living with
dementia offered psychotherapy versus usual care, but these effects did not persist at long-
term follow-up.
287

detect clinically meaningful differences in cognition, activities of daily living, or quality of life
between people living with dementia offered psychotherapy versus usual care.
13.2.4.8 Exercise
Low- to high-quality evidence from up to 16 RCTs containing 1,474 participants found

clinically meaningful post-intervention improvements in cognition, activities of daily living,
global assessment and behavioural and psychological symptoms in people living with
dementia offered exercise interventions versus usual care, but these effects did not persist at
long-term follow-up.
not detect clinically meaningful differences in depressive symptoms, quality of life or carer
burden between people living with dementia offered exercise interventions versus usual care.
comparing high-intensity group exercise therapy with usual care suggested that the
intervention was associated with a base-case ICER of around £41,000/QALY. Probabilistic
sensitivity analysis showed less than 20% probability of the intervention being associated
with an ICER of £20,000/QALY or better. A scenario analysis where costs relating to staff
training, staff travel, venue, and administration were set to zero resulted in an ICER of
£36,600/QALY. The only deterministic sensitivity analysis reducing the ICER to below
£20,000/QALY was when the benefits of exercise were assumed to last indefinitely.
comparing high-intensity one-to-one exercise therapy with usual care suggested that the
intervention was associated with a base-case ICER of around £77,000/QALY. One-way,
probabilistic, and scenario analyses showed little probability of the intervention being
associated with an ICER of £20,000/QALY or better.
comparing high-intensity group exercise therapy for people with severe dementia with usual
care suggested that the intervention was associated with a base-case ICER of over
£300,000/QALY. One-way, probabilistic and scenario analyses showed no probability of the
intervention being associated with an ICER of £20,000/QALY or better.
One partially applicable cost-utility analysis with potentially serious limitations examined the
cost-effectiveness of a moderate-to-high intensity aerobic exercise for patients with mild
Alzheimer’s disease in Denmark. The study found that exercise was associated with
significant cost increases but very small QALY benefits, leading to high ICERs in excess of
€100,000/QALY.
13.2.4.9 Nutrition
13.2.4.9.1 Ginkgo biloba
detect clinically meaningful differences in cognition, activities of daily living or global
assessment between people living with mild to moderate Alzheimer’s disease offered ginkgo
biloba versus placebo.
288
Low- to moderate-quality evidence from up to 6 RCTs containing 1,922 participants found

clinically meaningful differences in cognition, activities of daily living, behavioural and
psychological symptoms, global assessment and quality of life in people living with mild to
moderate Alzheimer’s disease or vascular dementia offered ginkgo biloba compared with
placebo. The evidence was primarily from a population who had defined non-cognitive
symptoms at baseline.
13.2.4.9.2 Omega-3 fatty acids
Moderate-quality evidence from up to 3 RCTs containing 604 participants could not detect
clinically meaningful differences in cognition, activities of daily living, behavioural and
psychological symptoms or dementia severity between people living with dementia offered
omega-3 fatty acid supplementation versus placebo.
13.2.4.9.3 Souvenaid
detect clinically meaningful differences in cognition, activities of daily living, quality of life or
dementia severity between people living with dementia offered souvenaid versus placebo.
13.2.4.9.4 Huperzine A
Very low- to low-quality evidence from up to 7 RCTs containing 648 participants found
clinically meaningful improvements in cognition and activities of daily living in people living
with mild to moderate Alzheimer’s disease offered Huperzine A versus placebo or no
intervention.
Moderate-quality evidence from 1 RCT containing 210 participants could not detect clinically
meaningful differences in behavioural and psychological symptoms between people living
with mild to moderate Alzheimer’s disease offered Huperzine A versus placebo or usual care.
13.2.4.9.5 Tailored nutritional guidance
Moderate-quality evidence from 1 RCT containing 78 participants found a clinically

meaningful post-intervention improvement in quality of life in people living with dementia
offered tailored nutritional guidance versus usual care.
13.2.4.9.6 Other nutritional interventions
Moderate-quality evidence from up to 3 RCTs containing 226 participants could not detect
clinically meaningful differences between people living with dementia offered ginseng,
vitamin E supplements or other nutritional or herbal formulations and those offered placebo
or no intervention.
13.2.4.10 Music therapy
13.2.4.10.1 Music therapy versus usual care
Very low- to moderate-quality evidence from up to 5 RCTs containing 322 participants found
clinically meaningful post-intervention improvements in cognition and activities of daily living
in people living with mild/moderate dementia offered music therapy versus usual care. These
effects were consistent when trials only recruiting people with non-cognitive symptoms at
baseline were excluded, but did not persist at long-term follow-up.
Very low- to moderate-quality evidence from up to 2 RCTs containing 236 participants found
clinically meaningful long term follow-up improvements in agitation, quality of life and carer
289
burden in people living with mild/moderate dementia offered music therapy versus usual
care. These effects were consistent when trials only recruiting people with non-cognitive
symptoms at baseline were excluded, but were not found in post-intervention measurements.
Very low- to low-quality evidence from up to 2 RCTs containing 124 participants could not
detect clinically meaningful differences in behavioural, psychological or depressive
symptoms, or, between people living with mild/moderate dementia offered music therapy
versus usual care.
Economic evidence
comparing group participatory music therapy with usual care showed that participatory group
music therapy was associated with a base-case ICER of around £27,000/QALY. One-way
sensitivity analysis found that varying several parameters within plausible ranges generated
ICERs lying on either side of a £20,000/QALY threshold. Probabilistic sensitivity analysis
showed a 26% probability that the intervention is associated with an ICER of £20,000/QALY
or better. A scenario analysis where costs relating to staff training, staff travel, venue and
administration were set to zero resulted in an ICER of £23,000/QALY.
comparing one-to-one music therapy with usual care suggested that the intervention was
associated with a base-case ICER of around £53,000/QALY. Some alternative model
assumptions resulted in much more favourable ICERs; however, probabilistic sensitivity
analyses showed very little probability of the intervention being associated with an ICER of
£20,000/QALY or better. A scenario analysis where costs relating to staff training, staff travel,
venue and administration were set to zero resulted in an ICER of £46,000/QALY.
13.2.4.10.2 Music therapy versus active control
Very low- to low-quality evidence from up to 3 RCTs containing 104 participants could not
detect clinically meaningful differences in cognition, behavioural and psychological
symptoms, depressive symptoms, agitation, quality of life or carer burden between people
living with mild/moderate dementia offered music therapy versus an active control
intervention.
13.2.4.11 Aromatherapy
Low-quality evidence from 1 RCT containing 56 participants found a clinically meaningfully

improvement in depressive symptoms in people living with dementia offered aromatherapy
versus usual care.
Very low to low-quality evidence from up to 3 RCTs containing 190 participants could not
detect clinically meaningful differences in behavioural and psychological symptoms,
agitation, activities of daily or quality of life between people living with dementia offered
aromatherapy versus usual care.
13.2.4.12 Light therapy
Very-low to low-quality evidence from up to 2 RCTs containing 103 participants could not
detect clinically meaningful differences in cognition, behavioural and psychological
symptoms, depressive symptoms, agitation or activities of daily living between people living
with dementia offered bright light therapy versus usual care.
290
13.2.4.13 Non-invasive brain stimulation
13.2.4.13.1 Alzheimer’s disease
Very-low to low quality evidence from up to 5 RCTs containing 105 participants could not
detect clinically meaningful differences in cognition, activities of daily living, or depressive
symptoms between people living with mild, moderate or severe dementia offered non-
invasive brain stimulation versus usual care.
13.2.4.13.2 Vascular dementia
Very-low quality evidence from 1 RCT containing 21 participants could not detect clinically
meaningful differences in cognition between people living with mild vascular dementia
offered non-invasive brain stimulation versus usual care.
13.2.4.14 Acupuncture
Very-low to low-quality evidence from up to 2 RCTs containing 223 participants could not
detect clinically meaningful differences in cognition or activities of daily living between people
living with dementia offered acupuncture versus no treatment.
13.2.4.15 Animal assisted therapy
Very-low to moderate-quality evidence from 1 RCT containing 50 participants found a

clinically meaningful improvement in depressive symptoms at long-term follow-up in people
living with dementia offered animal-assisted therapy versus usual care, but could not detect a
difference post-intervention, or in quality of life at any time point.
13.2.4.16 Robotic pet therapy
High-quality evidence from 1 RCT containing 61 participants found a clinically meaningful

improvement in depressive symptoms in people living with dementia offered robotic pet
therapy versus usual care.
13.2.4.17 Adapted mindfulness program
Very low- to low-quality evidence from 1 RCT containing 28 participants found a clinically
meaningful improvement in quality of life, but could not detect a difference post-intervention
in cognition or depressive symptoms in people living with dementia who took part in an
adapted mindfulness program plus treatment as usual versus treatment as usual alone.
13.2.4.18 Home Safety Toolkit
Moderate-quality evidence from 1 RCT containing 108 people could not differentiate
caregiver self-efficacy, caregiver strain, home safety or risky behaviour scales between
people offered a home safety toolkit intervention versus usual care.

Relative value of different The committee noted that there were four primary outcomes of
outcomes relevance to the questions considered in this review: cognition,
functional ability, independence and wellbeing (most commonly
measured through quality of life). They noted that different interventions
often primarily targeted different domains and would therefore be most
likely to have an impact on those domains, but it would be appropriate
291
to measure the impact of all interventions across all domains, as there

are considerable levels of interaction between each of the different
outcomes.
The committee also noted that it is important to separate out the short
term benefits of the interventions (i.e. those that only persist for as long
as the intervention is delivered) from any longer term benefits that
remain after the intervention is stopped. The evidence was therefore
divided in to two time points; outcomes measured at the end of an
intervention, and any long-term follow-up measured after the
intervention was stopped.
Trade-off between Cognitive stimulation therapy and reminiscence therapy
benefits and harms The committee noted there were improvements in cognition in people
offered cognitive stimulation therapy, and the average improvement was
around the level that would be considered meaningful for an individual
(approximately 1.4 points on the MMSE). Only a small number of trials
reported follow-up after the intervention stopped, but there was no
evidence the effects went away again after the intervention, with very
similar (and still statistically significant) effect sizes. The effects of group
interventions seemed to be larger than those of individual interventions,
and no evidence of benefit was found on outcomes other than cognition.
Similar short-term results were found for reminiscence therapy, with
again an improvement in cognition, though in this case with no evidence
of those effects persisting after the end of the intervention, and the
number of included studies was smaller. Again, group interventions
appeared to provide more benefits on average than those delivered
individually. The committee noted that in practice the two interventions
were not mutually exclusive, with cognitive stimulation therapy often
including elements of reminiscence.
The committee agreed there was evidence of benefits from both
interventions, but the evidence was stronger for cognitive stimulation
than for reminiscence, and therefore agreed it was appropriate to make
a strong (‘offer’) recommendation for cognitive stimulation therapy, and
a weaker (‘consider’) recommendation for reminiscence therapy.
Cognitive training
The committee noted there was now a considerable body of evidence
on cognitive training providing no evidence of value in any of the
relevant domains, and therefore agreed it was appropriate to make a
recommendation that cognitive training not be offered. The majority of
the evidence base was in people with mild to moderate Alzheimer’s
disease, and therefore the committee agreed to restrict the
recommendation to this population, as they noted there were other
types of dementia (e.g. semantic dementia) where cognitive training
might have benefits, and where it has not yet been tested.
Cognitive rehabilitation and occupational therapy
The committee noted that both the large studies of cognitive
rehabilitation showed improvements in activities of daily living, with
particularly large benefits shown in the recent UK HTA study. Benefits
for depression and quality of life were also shown with occupational
therapy, although a difference in activities of daily living could not be
demonstrated here, as a number of the trials did not measure this as an
outcome. However, the committee agreed that since the primary focus
of occupational therapy interventions was on ADL, the impairments in
quality of life seen were highly likely to be mediated through
improvements in ADL.
The committee agreed the evidence was sufficient to make ’consider’
recommendations for both cognitive rehabilitation and occupational
therapy, and also agreed it was important to stress these interventions
were designed to target functional ability, as they would be unlikely to
be effective for people who do not have concerns about functional
limitations.
292
Individualised activities
The committee noted there was evidence of some benefits across a
range of domains from exercise, aromatherapy, music therapy,
mindfulness and animal assisted therapy. However, the magnitudes of
these benefits were uncertain, and there was considerable variability in
both the structure and intensity of the interventions tested. The
committee also noted that taking one particular activity and offering that
to all people living with dementia is unlikely to be the most effective
approach, nor is this what is done in practice.
The committee agreed the more appropriate recommendation was that
people should be offered access to a range of activities that should be
tailored to their individual preferences. They agreed such an approach
was likely to be more effective than the blanket provision of a specific
activity (such has music therapy being provided to everyone), and was
justified by the evidence showing evidence of benefits across a range of
different activities. They also agreed that it was appropriate not to list
any specific activities under this heading, as the important part of the
recommendation was about the activities being individualised, rather
than what they actually are.
The individual activity with the strongest evidence of benefits individually
was exercise, but the committee were aware of the soon to be
published DAPA (Dementia and Physical Activity) study looking at the
effectiveness and cost-effectiveness of structured exercise provision in
the UK, and felt that in the absence of those results it was not
appropriate to make a specific positive recommendation for exercise.
Nutrition
The committee noted there was no evidence of benefits from a range of
supplements and nutritional interventions, including omega-3 fatty acids,
souvenaid, ginseng and various vitamin and herbal supplements. The
committee therefore agreed it was appropriate to make a ‘do not offer’
recommendation for these for the purposes of treating dementia. Some
evidence was identified showing potential benefits of huperzine A, but
this evidence was of low quality and conducted in populations often not
directly comparable to the UK (in particular, it was unclear if people
were taking cholinesterase inhibitors in the study, and how robust the
diagnosis of Alzheimer’s disease in many of the studies was). The
committee agreed the evidence was therefore not sufficient to make
either a positive or a negative recommendation.
The studies did not suggest any evidence of benefits from ginkgo biloba
supplements in a population of people with Alzheimer’s disease alone.
In a mixed population with Alzheimer’s disease, vascular dementia, or
comorbid Alzheimer’s disease and vascular dementia, there was
evidence of benefits, but the majority of the evidence in these studies
came from studies where people were only recruited if they had
behavioural symptoms at baseline (defined as being above a certain
threshold on the NPI). The committee therefore agreed this evidence
was best considered in the section of the guideline on managing non-
cognitive symptoms in people living with dementia (section 14).
Other interventions
The committee noted that there were no meaningful benefits found in
trials of psychotherapy (specifically interpersonal therapy), acupuncture
or non-invasive brain stimulation, and therefore it was appropriate to
make ‘do not offer’ recommendations for these interventions even in the
absence of proven clinical harm, as money spent on these interventions
would be better used on interventions where there is evidence of
benefits. However, they noted that non-invasive brain stimulation is still
an active area of research, and therefore added a caveat to that
recommendation to allow it still to be used within the context of clinical
trials. No positive evidence was found in this section for light therapy
either, but since this was already recommended within the section of the
293
guideline on managing sleep problems, the committee agreed it was

appropriate to make no comment on this within this section.
Finally, the committee considered the evidence on self-management
groups was insufficient to make either a positive or a negative
recommendation. In particular, self-management interventions were
agreed to comprise such a wide range of possible interventions that the
literature currently available did not cover the full range of possible
interventions adequately to be able to make recommendations, and
therefore the committee agreed the appropriate action was to make a
research recommendation around the effectiveness of self-management
for people living with dementia and their carers.
Trade-off between net For these review questions, the systematic literature reviews identified
health benefits and one economic study each for the interventions of cognitive rehabilitation,
resource use maintenance cognitive stimulation therapy, joint reminiscence group
therapy and exercise.
The committee noted that a commonly cited paper assessing the cost
effectiveness of cognitive stimulation therapy, Knapp et al. (2006) had
not been included as evidence for this guideline, but understood this
was because it did not meet the NICE reference case, as health
outcomes were not measured using QALYs. The committee were
satisfied that the trial data underpinning Knapp et al. (2006) was
considered in the original economic modelling using quantitative
synthesis techniques.
Although cognitive rehabilitation as per the Clare et al (in press) trial
was found to be more expensive than usual care, and was unable to
demonstrate a statistically significant benefit in terms of QALYs, the
committee were not convinced that the clinical benefit that this treatment
provided was adequately captured. The committee agreed that a similar
situation existed for joint reminiscence group therapy as per the Woods
et al. (2016) study and exercise as per the Sopina et al. (2017) study.
Although cognitive rehabilitation, maintenance cognitive stimulation
therapy and joint reminiscence group therapy and exercise were not
found to be cost effective at conventional thresholds in the primary
analyses, the guideline committee questioned the robustness of these
studies in the light of a systematic review and meta-analyses that
indicated positive benefits on clinically important domains. These
interventions were therefore prioritised for original economic modelling.
All interventions selected for original economic modelling were
associated with QALY gains in the region of 0.03 QALYs with the
exceptions of cognitive training (0.003 QALYS) and reminiscence
therapy (-0.004).
However, interventions varied in terms of costs and resultant ICERs,
with most of the high-intensity interventions modelled never likely to be
a cost-effective option even if QALYs are valued at £50,000 or more
each. The committee noted that the interventions tested in the trials
(and therefore on which the costs in this analysis were based) were
considerably more intensive than those currently offered in UK practice.
Therefore, the evidence was agreed to demonstrate that high-intensity,
long-term interventions were unlikely to be an effective use of
resources.
However, the committee noted that there was no clear pattern in the
evidence that high intensity interventions were more effective than lower
intensity ones. In particular, a number of recent trials of intensive
interventions (e.g. reminiscence therapy - Woods 2016) did not show
larger effects than trials of less intensive interventions. Therefore, the
committee was confident that these interventions could be delivered in a
much cheaper way than those measured in the more intensive trials,
whilst still providing benefits for people living with dementia.
When uncertainty in model parameters was explored, some of these
intensive interventions had the potential to be cost-effective.
294
In the base case, group cognitive stimulation therapy was found to be

marginally cost effective at the £20,000/QALY threshold. However, the
one-way sensitivity analysis conducted for group cognitive stimulation
therapy found that the model was extremely sensitive to many
parameters in the model, and varying any parameter within plausible
range generates results lying on either side of a £20,000/QALY
threshold. The case for group cognitive stimulation therapy was
strengthened by a scenario analysis where costs relating to staff
training, staff travel, venue and administration were set to zero –that is,
only staff time and participant travel expenses were accounted for. This
too resulted in group cognitive stimulation therapy being associated with
an ICER of less than £20,000/QALY. Furthermore, the committee
reported that many CCGs did not pay for participant travel and used
staff at band 2 and band 4 on the agenda for change pay scale, rather
than the staff at band 4 and band 6 as assumed in the original model.
Assuming that the effectiveness of the intervention remains the same,
these further modifications would only reduce the cost and increase the
cost effectiveness of group cognitive stimulation therapy. Furthermore,
the committee noted that one of the cost–utility analyses from the
included study by D’Amico (2015) found that maintenance CST was
associated with an ICER of between £20,000 and £30,000/QALY
(health and social cares perspective; effects measured using proxy-EQ-
5D). The committee therefore felt it appropriate to create a strong
(‘offer’) recommendation for group cognitive stimulation therapy.
The committee also noted that the original models were likely to be
somewhat conservative, as there were potential benefits they did not
capture such as impact on carers, and potential cost savings to other
services that may result from providing people with appropriate support.
Although reminiscence therapy was associated with a negative QALY
gain, and was technically dominated, this was driven by a loss in the
behavioural domain larger than we would expect, which was only
partially offset by a benefit in the cognitive domain larger than we would
expect. The committee also took into account the very small loss of
QALYs associated with the model (-0.004) and did not believe that
reminiscence therapy would result in any harm, and considered it
misleading to think of it as a dominated strategy.
The committee also considered the original economic evidence for
cognitive training in detail and found the high ICER of the base-case
model to be an unacceptable use of NHS resources. The committee
considered the univariable sensitivity analysis, which only took into
account effects in the cognitive domain, and produced an ICER well
below traditionally acceptable cost-effectiveness thresholds. However,
the committee was not convinced that this evidence was reliable, as the
meta-analysis on which it was based was disproportionately influenced
by a single, small RCT (Bergamaschi et al., 2013), which showed a very
large positive effect on the MMSE scale of 6.3 points in favour of
cognitive training. The committee did not believe that the intervention
could feasibly have an effect of this magnitude, and noted that the
apparent cost effectiveness of cognitive training, in this sensitivity
analysis, disappeared when this datapoint was excluded from analysis.
Therefore, the committee was confident in making a strong (‘do not’)
recommendation against cognitive training.
The committee also considered the original evidence for cognitive
rehabilitation and occupational therapy and agreed that, although the
base-case analyses resulted in ICERs higher than £30,000/QALY, use
of alternate model structures could result in more favourable ICERs.
The committee noted that, using the univariable ADL model (that is,
focusing only on the domain in which cognitive rehabilitation aims to
achieve its effect), cognitive rehabilitation resulted in an ICER of
£26,006/QALY compared with standard care. Furthermore, the
committee agreed that, while large ADL benefits had been shown in the
295
recent UK HTA study, these had not translated into quality of life
benefits as measured by the instrument used in that RCT (DEMQOL-U).
As ADL measurements had been specifically targeted at factors that
affected each participant's day-to-day life, committee members
expressed a belief that a more sensitive quality of life instrument would
have reflected the benefits in ADL shown. The committee also agreed
that the available analyses did not adequately capture the benefits of
occupational therapy. The committee noted that the role of occupational
therapy is well established for a broad range of people who have
restricted ADL. It noted that, in some analogous conditions (for
example, Parkinson's disease) NICE committees had reviewed
evidence that enabled them to make strong (“offer”) recommendations
for occupational therapy. The committee concluded that this was
supportive evidence that patients with dementia are likely to obtain cost-
effective benefit from this intervention, too. Overall, the committee
believed that both these interventions are most likely to benefit people
with mild or moderate dementia. As a result of this, the committee felt a
‘consider’ recommendation was most appropriate for cognitive
rehabilitation and occupational therapy.
The committee also noted that resources were already spent in the
current system on providing both group activities and support with
activities of daily living for people living with dementia. The
recommendations were therefore agreed not to be likely to be
associated with a substantial resource impact associated with new
investment, but were rather advising people on the most effective ways
to use the resources already allocated to supportive interventions for
people living with dementia. They also noted there would be potential
savings from the number of ‘do not’ recommendations made for
ineffective interventions.
Quality of evidence The committee noted that the evidence around cognitive stimulation,
reminiscence therapy, occupational therapy and cognitive rehabilitation
was predominantly in a population of people with mild to moderate
dementia, and agreed it was likely that interventions would need to be
delivered in different ways for people with more severe dementia.
Therefore, they agreed it was appropriate to focus those
recommendations on people with mild to moderate dementia. The
evidence on individual activities (particular music and exercise) came
from a more varied population and did include studies in people with
severe dementia, and therefore the committee were comfortable for that
recommendation to apply to the whole population.
The committee noted that, for the interventions that were not found to
be effective, some of the trials were conducted solely or predominantly
in people with Alzheimer’s disease (cognitive training, interpersonal
therapy and non-invasive brain stimulation), and therefore the
recommendations made should be specific to that population. The trials
on acupuncture and nutrition tended to recruit a more general
population of people living with dementia, and therefore the committee
were confident to apply those recommendations to the broader group.
Other considerations The committee noted that one possible interpretation of the evidence
base identified for this review question was that the benefits of many of
these interventions were driven less by the specific content of the
interventions, and more by the benefits from support groups more
generally. Therefore, the committee agreed it was appropriate to make
a research recommendation around the effectiveness of unstructured
activities as an intervention, to test this hypothesis.
296
82. Offer a range of activities to promote wellbeing that are tailored to the person’s
individual preferences.
83. Offer group cognitive stimulation therapy to people living with mild to moderate
dementia.
84. Consider group reminiscence therapy for people living with mild to moderate
dementia.
85. Consider cognitive rehabilitation or occupational therapy to support functional

ability in people living with mild to moderate dementia.
86. Do not offer acupuncture to treat dementia.
87. Do not offer ginseng, vitamin E supplements or herbal formulations to treat

dementia.
88. Do not offer cognitive training to treat mild to moderate Alzheimer’s disease.
89. Do not offer interpersonal therapy to treat the cognitive symptoms of mild to
moderate Alzheimer’s disease.
90. Do not offer non-invasive brain stimulation (including transcranial magnetic

stimulation) to treat mild to moderate Alzheimer’s disease, except as part of a
randomised controlled trial.
10. What are the most effective psychosocial interventions for improving cognition,
independence, activities of daily living and wellbeing in people living with
dementia?
11. What is the effectiveness of unstructured community activities on wellbeing for

12. What is the effectiveness and cost-effectiveness of self-management training for

people living with dementia and their carers?
see appendix L.
297
Managing non-cognitive symptoms
14 Managing non-cognitive symptoms

The cognitive problems associated with dementia are well established, but up to 90% of
people with dementia may also be affected by non-cognitive symptoms of dementia. These
symptoms can lead to significant changes in behaviour that include increased aggression,
anxiety, apathy, agitation, depression, delusions, hallucinations and sleep disturbances.
These behaviours often reflect a high level of distress being felt by the person with dementia
that they may not be able to communicate or understand, but these behaviours can also
have significant adverse effects on the people involved in caring for the person with
dementia. For example, the occurrence of sleep problems and wandering by people with
dementia will disrupt their carer’s sleep patterns and is likely to have a severe effect on their
carer’s mental state and ability to cope over time. Carer responses to these non-cognitive
symptoms can also increase distress for the person with dementia and may lead to their
being prescribed medication to try to control these behaviours. The inability of these
treatments to successfully manage the non-cognitive behavioural symptoms may ultimately
result in the institutionalisation of the person with dementia when the carer becomes unable
to cope. However, staff within these care homes may also struggle to manage these
behavioural issues and may in turn refer the person with dementia to specialist nursing care.
Thus, the non-cognitive symptoms associated with dementia have severe adverse effects on
the person with dementia, family and both paid and unpaid carers. If these could be treated
successfully this would have a big impact on the quality of life for everyone concerned and
could lead to people with dementia being able to remain in their homes or with family.
There are a range of potential treatments for the non-cognitive symptoms of dementia which
can be divided into two groups: pharmacological and non-pharmacological interventions.
Pharmacological interventions are targeted to the problematic behaviour of the person with
dementia and include the use of antidepressants, antipsychotics, mood stabilisers and drugs
to modify sleep patterns. In contrast, non-pharmacological interventions take a wider view
and may include approaches aimed at: resetting sleep patterns using bright light therapy or
by increasing the activity levels of the person with dementia; calming and distracting an
agitated person; and altering the carer’s behaviour to better cope with and manage the
person with dementia. In addition, anxiety and depression may be treated using cognitive
behavioural therapy, multisensory stimulation, relaxation and animal-assisted therapies.
This chapter focuses on the range of pharmacological and non-pharmacological

interventions to address non-cognitive symptoms for which rigorous evidence from
randomised controlled trials is available. The non-cognitive symptoms examined include
depression, which is the subject of specific NICE guidance in its own right. However, it was
decided that there were additional factors that needed to be taken into consideration for
people with dementia and as a result depression was addressed in this specific context.
298
14.1 Interventions for treating illness emergent non-cognitive

symptoms in people living with dementia
Review questions
 What are the most effective pharmacological interventions for managing illness emergent
non-cognitive symptoms, such as psychosis, depression, behavioural changes in people
living with dementia?
 What are the most effective non-pharmacological interventions for managing illness
emergent non-cognitive symptoms, such as psychosis, depression, behavioural changes
in people living with dementia?
14.1.1 Introduction
The aim of these review questions was to determine the effectiveness of different
pharmacological and non-pharmacological interventions for treating illness emergent non
cognitive symptoms in people living with dementia. The review identified studies that fulfilled
the conditions specified in Table 77 and Table 78. For full details of the review protocols, see
appendix C.
Table 77: Review summary: pharmacological interventions

Interventions Pharmacological interventions for treating illness emergent non-
cognitive symptoms, which may include:
 Antipsychotics
 Cholinesterase inhibitors
 Memantine
 Carbamazepine
 Valproate (mood stabilisers)
 Antidepressants
 Anxiolytics
 Propranolol
 Hypnotics
 Standard care
Outcomes  Change in/resolution of non-cognitive symptoms
ability
 Adverse events
Table 78: Review summary: non-pharmacological interventions

Interventions  Non-pharmacological interventions for treating illness emergent non-
cognitive symptoms
 Standard care
299
Outcomes  Change in/resolution of non-cognitive symptoms

ability
 Adverse events
This review was conducted as an update from the previous dementia guideline (CG42). All
included RCTs and systematic reviews from the previous guideline, together with all RCTs in
included systematic reviews, were screened at title and abstract level. RCTs from included
systematic reviews were excluded if they did not meet the criteria of enrolling patients with
dementia and an illness-emergent non-cognitive symptom at baseline.
In addition, a systematic literature search for randomised controlled trials since the time of
the last guideline identified 2,645 references. These were screened at title and abstract level,
with 103 papers ordered as potentially relevant systematic reviews, 196 ordered as
potentially relevant RCTs; 40 systematic reviews and 104 RCTs were ordered for
pharmacological interventions; 63 systematic reviews and 92 RCTs were ordered as
potentially relevant for non-pharmacological interventions. Finally, 21 papers identified
through the literature search for the question on the management of pre-existing mental
health conditions were also included as part of this question. Summaries of the included
studies are provided below, with details of all excluded studies given in Appendix F.
For the full evidence tables and full GRADE profiles for included studies, please see
Appendix E and Appendix G. References for the included studies are given in appendix I.
14.1.2.1.1 Anxiety, depression, antidepressants and antipsychotics
Twenty-seven studies were included in the evidence review for this question, 16 on
depression and/or anxiety (including 5 systematic reviews containing an additional 32 RCTs),
1 on antidepressants for other behavioural symptoms (a systematic review containing 9
RCTs), and 9 on the use of antipsychotics (including 3 systematic reviews containing an
additional 33 RCTs). A systematic review of memantine for mild Alzheimer’s disease was
also identified, containing 3 RCTs. A summary of the included studies is given in Table 79.
14.1.2.1.2 Sleep problems
Nine studies were included in the evidence review for this question, consisting of 7 RCTs and
2 systematic reviews containing 5 additional RCTs. A summary of the included studies is
given in Table 80. One additional study was identified during the rerun process, but this was
excluded at full text screening.
14.1.2.1.3 Agitation and aggression
Twenty-eight studies were included in the evidence review for this question, 22 RCTs and 6
systematic reviews containing 42 additional RCTs. A summary of the included studies is
given in Table 81.
300
14.1.2.2 1 Summary tables for included studies
4.1.2.2.1 2 Anxiety, depression, antidepressants and antipsychotic
3 Table 79: Included studies for anxiety, depression, antidepressants and antipsychotics
Ballard (2008) RCT 165 people with Intervention: Antipsychotic Cognition Location: UK
dementia and taking withdrawal Functional ability Follow up: 12
antipsychotics Comparator: Continuation months
Ballard (2009) RCT 165 people with Intervention: Antipsychotic Mortality Location: UK
dementia and taking withdrawal Follow up: 24-54
antipsychotics Comparator: Continuation months
Ballard (2015) RCT 199 people in care Intervention: Memantine Cognition Location:
homes taking Comparator: Antipsychotics Behavioural symptoms UK/Norway
antipsychotics Anxiety Follow up: 24
Adverse events weeks
Mortality
Banerjee (2011) RCT 326 people with Intervention: Sertraline or Depression Location: UK
Alzheimer’s disease mirtazapine Cognition Follow up: 35
Comparator: Placebo Quality of life weeks
Adverse events
Boström (2015) RCT 186 people with Intervention: High-Intensity Depression Location: Sweden
dementia Functional Exercise program Follow up: 7
Comparator: Non-exercise months
activity program
Brodaty (2003) RCT 86 people with dementia Intervention: Psychogeriatric Depression Location: Australia
management Follow up: 12
Comparator: Usual care weeks
Cooke (2010) RCT 47 people with Intervention: Music therapy Quality of life Location: Australia
confirmed or probable Comparator: Reading therapy Depression Follow up: 16
dementia weeks
301
Fossey (2006) RCT 349 people with Intervention: Psychosocial Neuroleptic use Location: UK
dementia intervention Agitation Follow up: 12
Comparator: Usual care Aggression months
Hickman (2007) RCT 66 people with dementia Intervention: Ambient bright Depression Location: USA
lighting Follow up: 3 weeks
Comparator: Standard lighting
Holmes (2007) RCT 27 people with Intervention: Risperidone Anxiety Location: UK
Alzheimer’s disease Comparator: Rivastigmine Follow up: 6 weeks
Ing-Randolph (2015) Systematic review 9 studies in people with Intervention Group music Anxiety Location: N/A
dementia-associated Comparator: Usual care Follow up: 5
anxiety weeks-6 months
Kiosses (2015) RCT 74 people with cognitive Intervention: PATH Depression Location: USA
impairment (39 with Comparator: ST:CI Disability Follow up: 12
dementia) weeks
Leong (2014) Systematic review 10 RCTs in people with Intervention: Antidepressants Depression Location: N/A
depression and Comparator: Placebo Follow up: 6-24
dementia weeks
Leontjevas (2013) RCT 403 residents of Intervention: Multidisciplinary Depression Location:
dementia units care programme Quality of life Netherlands
Comparator: Usual care Follow up: 20
months
Lyketsos (2003) RCT 44 people with Intervention: Sertraline Depression Location: USA
Alzheimer’s disease Comparator: Placebo Cognition Follow up: 12
Ma (2014) Systematic review 19 studies in people Intervention: Antipsychotics Behavioural symptoms Location: N/A
with dementia Comparator: Placebo Anxiety Follow up: 6
Adverse events weeks-26 weeks
Mortality
Moulton (2014) Systematic review 11 studies in people Intervention: Pharmacological Depression Location: N/A
with Huntington’s treatment Follow up: 4
disease Comparator: Placebo weeks-1 year
302
Ortega (2015) Systematic review 6 RCTs in people with Intervention: Psychological Anxiety Location: N/A
anxiety/ depression and treatment Depression Follow up: 6
dementia Comparator: Usual care weeks-1 year
Pan (2014) Systematic review 10 RCTs in people with Intervention: Antipsychotic Behavioural symptoms Location: N/A
dementia and taking withdrawal Mortality Follow up: 4
antipsychotics Comparator: Continuation weeks-54 months
Petrovsky (2015) Systematic review 10 studies in people Intervention: Active music Anxiety Location: N/A
with anxiety/depression interventions Depression Follow up: <24
and dementia Comparator: Usual care weeks
Porsteinsson (2014) RCT 186 people with Intervention: Citalopram Anxiety Location:
probably Alzheimer’s Comparator: Usual care Agitation US/Canada
disease Follow up: 9 weeks
Richter (2012) Systematic review 4 RCTs in care homes Interventions: Psychosocial Antipsychotic use Location: N/A
interventions Behavioural symptoms Follow up: 30
Comparator: Usual care days-12 months
Schneider (2011) Systematic review 3 RCTs in mild Intervention: Memantine Cognition Location: N/A
Alzheimer’s disease Comparator: Placebo Behavioural symptoms Follow up: 24
weeks
Seitz (2011) Systematic review 9 RCTs in people with Interventions: Antidepressants Psychosis Location: N/A
psychosis/agitation Comparator: Usual care Agitation Follow up: 4
weeks-12 weeks
Sung (2010) RCT 52 people with dementia Intervention: Preferred music Anxiety Location: Taiwan
listening Follow up: 6 weeks
Comparator: Usual care
Verhey (2006) RCT 58 people with dementia Intervention: Olanzapine Anxiety Location:
and agitation Comparator: Haloperidol Netherlands
Follow up: 5 weeks
Weintraub (2010) RCT 131 people with Intervention: Sertraline Depression Location: USA
Alzheimer’s disease Comparator: Placebo Cognition Follow up: 24
303
4.1.2.2.2 1 Sleep problems
2 Table 80: Included studies for sleep problems

Study reference Study design Study population Included studies Outcomes of interest
Alessi (2005) RCT Nursing home residents with abnormal Intervention: Multicomponent Acitgraph measurements
sleep/wake patters nonpharmacological
intervention
Comparator: Usual care
Chong (2005) RCT Community-dwelling patients with mild to Intervention: continuous Epworth sleepiness scale
moderate Alzheimer’s disease and sleep positive airway pressure
disordered breathing. (CPAP) for 6 weeks
Comparator: sham CPAP for
3 weeks then CPAP for 3
weeks
Forbes (2009) Systematic Dementia patients treated with light therapy to 10 RCTs included in final Acitgraph measurements (or
Review manage sleep, cognitive, functional, psychiatric review equivalent systems)
and behavioural disturbances Sleep relevant RCTs = 6
Intervention: light therapy
Comparator:
Harris (2012) RCT Nursing home patients with dementia and sleep Intervention: slow-stroke back Acitgraph measurements
disturbances massage
Comparator: usual care
Larsson (2010) RCT Parkinson’s Disease and dementia with Lewy Intervention: Memantine Stavanger sleep questionnaire
bodies patients suffering from sleep disorders Comparator: Placebo Epworth sleepiness scale
enrolled in psychiatric and neurological out-
patient clinics.
McCleery (2016) Systematic Pharmacological interventions in patients with 5 RCTs included in final Acitgraph measurements
Review Alzheimer’s Disease and sleeping problems in review:
nursing homes, long-term care facilities and Melatonin n=4
hospital geriatric centres. Trazadone n=1
McCurry (2005) RCT AD patients living at home with carers and Intervention: Night-time Epworth sleepiness scale
suffering from sleep disturbances. insomnia treatment and PSQI
education (NITE-AD) Actigraph measurements:
304
Comparator: general Night awakenings
dementia education and carer Time in bed (hrs)
support Daytime sleep (hrs)
Total sleep/night (hrs)
McCurry (2011) RCT Patients in an independent community living Intervention(s): walking, light Actigraph measurements:
setting suffering from AD and sleep problems. or NITE-AD Total wake time/night
Comparator: contact control Sleep %
(non-directive dementia care Total sleep time
support)
Number of awakenings
Sleep disorders inventory (SDI)
Richards (2005) RCT Patients in a nursing home setting suffering from Intervention: individualised Actigraph measurements:
dementia and sleep problems social activities Day/night sleep ratio
Comparator: usual care Night-time sleep efficiency,%
Night-time minutes to sleep onset
Night-time minutes slept
Night-time awake
Daytime minutes slept
4.1.2.2.3 1 Agitation and aggression
2 Table 81: Included studies for agitation and aggression

Brown (2015) Systematic RCTs comparing opioids to placebo for agitation 0 RCTs included in final N/A
review in dementia review
Forrester (2014) Systematic RCTs of aromatherapy for people living with 5 RCTs in review Agitation (CMAI; NPI; PAS
Review dementia Aggression (NPI)
Behavioural symptoms (NPI)
Quality of life (Blau QOL scale)
Activities of Daily living (Barthel
scale of ADL)
305
Jutkowitz (2016) Systematic Non-pharmacological care delivery interventions 19 RCTs included in final Frequency of agitation and
review to reduce and manage agitation and aggression review: aggression (measured by CMAI)
in people living with dementia in nursing homes Dementia care mapping
and assisted living (DCM) n=3;
Person centred care
(PCC) n=3;
Clinical protocols for
antipsychotic use n=3;
Emotion oriented care n=2;
Unique interventions n=11
Kong (2009) Systematic RCTs of non-pharmacological interventions for 14 RCTs included in final Agitation (measured by CMAI and
review agitation in people living with dementia review: short CMAI; Pittsburgh Agitation
Sensory interventions (n=3); Scale PAS; Agitation Behaviour
Social contact activities (n=5); Mapping Instrument ABMI; Agitated
Behaviour Inventory for Dementia
Environmental modification
ABID; Agitation Visual Analogue
carer training (n=3)
Scale; AVAS; Observed Agitation
Behavioural interventions Scale ; OAS Scale for the
(n=2) Observation of Agitation in
Combination therapy (n=1) Dementia SOAD)
Aggression (measured by the
Ryden Aggression Scale (RAS)
Behavioural symptoms (measured
by Burke Dementia Behavioral
Rating Scale BDBRS; Behavioral
Pathology in Alzheimer’s Disease
Rating Scale BEHAVE-AD;
Disruptive Behavior Scale DBS;
Need Driven Compromised
Behavior Model NDB
Von Gunten Systematic RCTs comparing Ginkgo biloba extract EGb 761 4 RCTs included in final Cognition (based on SKT short
(2015) review to placebo for people living with dementia and review cognitive performance test)
behavioural and psychological symptoms BPSD (based on NPI)
306
Activities of daily living (based on
Gottfries Brane Steen scale (GBS)
and Activities of Daily living-
International scale (ADL-IS)
Clinical Global Impression of
change (based on GBS total score
and Alzheimer’s disease
Cooperative Study-Clinical Global
Impression of Change (ADCS-
CIGIC)
Quality of life – DEMQOL=-
PROXY)
Xiao (2010) Systematic RCTs comparing use of mood stabilisers for 5 RCTs included in final Agitation
review agitation in Alzheimer’s disease review Functional Ability
Neuropsychiatric profile
Cognition
Adverse events
1
Study Study
reference design Study population Intervention Outcomes of interest Other
Burns 2009 RCT 48 people with dementia and agitation Bright light therapy Agitation (CMAI) Study location UK
versus normal light MMSE Follow up 3 weeks
Cornell scale for
depression
Behavioural
psychopathology
(MOUSEPAD)
Cohen RCT 167 people with dementia and TREA – individualised Overall agitation Study location USA
Mansfield exhibiting agitation several times a day interventions for unmet Follow up / duration 10 days
2007 needs versus usual care
Cohen RCT 125 people with dementia and TREA – individualised Overall Agitation (CMAI) Study location USA
Mansfield exhibiting agitation at least several interventions for unmet Overall BPSD (LMBS) Follow up 2 weeks
2012 times a day needs versus usual care
307
Study Study
Cummings RCT 194 [people with Alzheimer’s disease Dextromethorphan Agitation (NPI) Study location USA
2015 and behavioural symptoms interfering quinidine versus NPI total score Follow up 10 weeks
with daytime routine placebo Cornell scale for
depression
Global assessment
(CGIC)
Deudon 2009 RCT 306 people with dementia presenting Staff training NPI Study location France
with at least one BPSD once a week programme versus Follow up 10 weeks
usual care
Fox 2012 RCT 153 people with Alzheimer’s disease Memantine versus Agitation (CMAI) Study location UK
with significant agitation matched placebo NPI Follow up 12 weeks
Cognition (MMSE)
Global assessment
(CIGIC)
Frakey 2012 RCT 22 people with Alzheimer’s disease and Modafinil versus Apathy (FrSBe) Study location USA
displaying apathetic outcomes matched placebo Activities of Daily Living Follow up 8 weeks
(ADLQ)
Functional status
(DAFS)
Holmes 2004 RCT 96 people with Alzheimer’s disease and Donepezil versus NPI Study location UK
showing neuropsychiatric symptoms matched placebo NPI distress scale Follow up 24 weeks
Adverse events
Howard 2007 RCT 249 people with Alzheimer’s disease Donepezil versus Agitation (CMAI) Study location UK
and agitation matched placebo NPI Follow up 8 weeks
CGIC
MMSE
Lin 2011 RCT 100 people with dementia and agitated Group music versus Agitation (CMAI) Study location Taiwan
behaviours usual care Follow up 6 weeks
308
Study Study
Mahlberg RCT 20 people with Alzheimer’s disease and Rivastigmine versus NPI agitation Study location Germany
2007 showing agitated behaviour matched placebo NPI Follow up 2 weeks
NOSGER
McCabe 2015 RCT 189 people with dementia and agitation Staff training protocol Agitation (CMAI ) Study location Australia
and/or aggression versus usual care Follow up 12 weeks
Porsteinsson RCT 56 people with dementia and agitated Divalproex sodium BPRS Study location USA
2001 symptoms versus usual care CMAI Follow up 6 weeks
PSMS
MMSE
Rapp 2013 RCT 304 people with dementia and agitation Training support and Agitation (CMAI) Study location Germany
activity therapy Follow up 10 weeks
Usual care
Rea 2015 RCT 113 people with Alzheimer’s disease Donepezil plus choline Apathy Study location Italy
showing apathetic signs alphoscerate versus NPI Follow up 2 years
donepezil Frontal Assessment
Battery
Ridder 2013 RCT 21 people diagnosed with dementia Music therapy versus Anxiety (CMAI) Study location Netherlands
and symptoms of agitation standard care Activities of daily living Follow up 6 weeks
(ADRQL)
Sung 2006 RCT 36 people with dementia and presence Group music and Agitation (CMAI) Study location Taiwan
of agitated behaviours movement versus Follow up 4 weeks
standard care
Van der Ploeg RCT 44 people with dementia showing Individualised activity Agitation Study location Australia
(2013) physical agitated behaviour several session versus non Positive affect Follow up 4 weeks
times a day personalise intervention Negative affect
Constructive
engagement
Negative engagement
309
Study Study
Van den RCT 50 people with dementia and showing Tetrahydrocannabinol Agitation (CMAI) Study location Netherlands
Elsen (2015) agitated, aggressive or aberrant versus placebo Study NPI total Follow up 3 weeks
behaviour location CGIC
QoL-AD
Wang 2008 RCT 22 people with Alzheimer’s disease and Prazosin versus GCIC Study location USA
exhibiting agitation or aggression at matched placebo BPRS Follow up 8 weeks
least twice a week NPI
Yang (2015) RCT 186 people with dementia and Aroma acupressure Agitation (CMAI) Study location Taiwan
symptoms of agitation versus control Follow up 4 weeks
Aromatherapy versus
control
Zwijsen RCT 659 people with dementia and showing Staff protocol versus Agitation (CMAI) Study location Netherlands
(2014) challenging behaviours (specifically usual care NPI Follow up 20 months
agitation)
310
14.1.3.1 Systematic review of published economic evaluations

(CUAs) evaluating the most effective pharmacological and non-pharmacological
interventions for managing illness emergence non-cognitive symptoms in people living with
dementia. In total, 2,385 articles were returned. In total, 4 publications were judged to be at
least partially applicable to the review questions and were therefore included. Details of the
literature search are provided in Appendix D.
14.1.3.2 Antidepressants
Banerjee et al. (2013) compared the cost effectiveness of sertraline and of mirtazapine with
placebo, for the treatment of depression in people with Alzheimer’s disease referred to old-
age psychiatry services. Romeo et al. (2013) present the same study and analysis in a
separate publication. The authors conducted a cost–utility analysis alongside the HTA-SADD
RCT of 326 participants, collecting primary service-use and EQ-5D data. The primary
analysis was a cost-effectiveness analysis with change in the Cornell Scale for Depression
and Dementia as the health outcome. A secondary analysis considered incremental QALYs.
A 39-week time horizon was taken for this secondary analysis, matching the trial duration.
There was no extrapolation beyond the trial duration. For further details, please see the
economic evidence profile in Appendix M.
Service-use data included direct costs associated with hospital-based care, community-
based care and day services recorded during the follow-up period. Informal care data
(unpaid carer costs) were also collected. QALYs were estimated using data obtained from
the EQ-5D questionnaire. The authors performed non-parametric bootstrapping to generate
additional pairs of incremental cost and QALY outcomes in order to present a cost-
effectiveness acceptability analysis. The mean results with informal care costs excluded are
presented in Table 82.
Table 82: Base-case cost–utility results – Banerjee (2013) and Romeo (2013)
Effects Costs Effects ICER
Treatment Costs (QALYs) (£)[95% CI] (QALYs) [95% CI] (£/QALY)
Placebo £2,146 0.55
QALYs
Mirtazapine £2,550 0.60 £404 [-972, 1,626] 0.05 QALYs [-0.10, £8,080
QALYs 0.10] /QALY
Sertraline £2,839 0.57 £289 [-1,545, 1,151] -0.02 QALYs [-0.07, Dominated
QALYs 0.03]
The base-case analysis produces an ICER of £8,080 per QALY for mirtazapine compared
with placebo, and shows sertraline to be dominated by mirtazapine. There is large
uncertainty around the incremental costs and QALYs, with all 95% confidence intervals
crossing zero. In addition, some estimates are conspicuously skewed, especially estimated
incremental QALYs for mirtazapine compared with placebo: the 95% confidence interval
ranges from −0.1 to +0.1, but the mean is +0.05 QALYs. Despite its low mean ICER
compared with placebo, mirtazapine is shown to have only approximately 20% probability of
being cost-effective at a threshold value of £20,000/QALY. Mirtazapine is shown to have a
probability in excess of 90% of being cost-effective compared with sertraline at all threshold
values. Deterministic sensitivity analysis was not presented for analyses which excluded
311
informal care costs. The authors conclude that their analysis provides no support for the use
of antidepressants as first-line therapy for depression in people with Alzheimer’s disease
referred to old-age psychiatry services.
14.1.3.3 Antipsychotics
Rosenheck et al. (2007) compared the cost effectiveness of olanzapine, quetiapine and
risperidone with placebo, for the treatment of psychosis and aggression in people with
Alzheimer’s disease in ambulatory outpatients living at home or in assisted living. The
authors conducted a cost–benefit analysis alongside Schneider et al. (2006) (n=421),
assessing quality-adjusted life-years (QALYs) using the Health Utilities Index Mark. This was
supplemented by the Alzheimer’s Disease Related Quality of Life Scale, the Alzheimer’s
disease Cooperative Study Activities of Daily Living Scale (ADCS-ADL) and the AD
Dependence Scale. A secondary analysis excluded observations after the first medication
change. For further details, please see the economic evidence profile in Appendix M.
Service-use data included direct costs associated with experimental medication costs.
Concomitant medication cost and monthly health service costs were also collected. The
authors performed an analyses of net health benefits when the value of health benefits was
valued at $50,000 per QALY and $100,000 per QALY. The mean results with concomitant
medication cost and monthly health service costs excluded are presented in Table 83.
Table 83: Base-case cost–utility results – Rosenheck et al., (2007)

Treatment Costs Effects Costs Effects ICER
0.14
Placebo $4,923 QALYs
-0.02
0.12 QALYs
Olanzapine $6,480 QALYs $1,557 Dominated
0.15 0.01
Quetiapine $7,839 QALYs $2,916 QALYs ext. dom.
0.16 0.02
Risperidone $10,215 QALYs $5,292 QALYs $264,600 /QALY
The base-case analysis suggests that olanzapine is dominated by placebo as placebo is

cheaper and produces more health benefits. Quetiapine and risperidone both produced a
very small incremental health benefit (of +0.01 and +0.02 QALY, respectively), but also result
in relatively large costs, with the result that risperidone produces an ICER of $264,600/QALY
compared with placebo, and quetiapine is extendedly dominated by placebo and risperidone.
At the cost-effectiveness thresholds of $50,000 and $100,000 per QALY considered by the
authors, no intervention was found to be cost-effective.
Kirbach et al. (2008) compared the cost effectiveness of olanzapine with no treatment, for
the treatment of agitation and psychosis in people with Alzheimer’s disease in the USA. The
authors created a Markov model, with a 6-month cycle length and a 13-year treatment
horizon Transition probabilities for Alzheimer’s disease progression were taken from the
Consortium to Establish a Registry for Alzheimer’s disease (CERAD) cohort, in which there
was a 70% prevalence for agitation and up to 36% prevalence for varying psychoses. In
common with Rosenheck et al. (2007), Kirbach et al. (2008) used results from Schneider et
al. (2006) to estimate olanzapine cost-effectiveness. Economic evidence profiles for both
studies are available in Appendix M.
312
The study considered both ‘direct’ and ‘indirect’ costs, though did not specify what each
comprised. The results presented here are from analyses of ‘direct’ costs only, as these were
judged less likely to include items that are beyond the NICE reference case. Utility weights
used to estimate QALYs were provided by Murman and Colenda (2005).
Table 84: Base-case cost–utility results – Kirbach et al (2008)

No olanzapine $34,215 a NR
0.15
Olanzapine $39,781 NR $5,566 a QALYs $37,104 /QALY
a Not reported in paper, but may be inferred from ICER, incremental QALYs and absolute costs of olanzapine
The base-case analysis (Table 84) suggests that treatment with olanzapine incurs additional
costs (primarily due to cost of the drug itself) but also provides QALY gains, with an ICER of
$37,104 per QALY. Several 1-way, 2-way and 3-way deterministic sensitivity analyses
produced ICERs ranging from $31,336 to $42,039.
However, this finding is at odds with Rosenheck et al. (2007) (see above), which is
noteworthy, given that both models relied on the same evidence to estimate the
effectiveness of olanzapine.
14.1.3.4 Non-pharmacological interventions
Livingston et al. (2014) compared the cost effectiveness of a 6-month, multimodal, non-
pharmacological intervention with usual care, for reducing agitation in people with dementia.
The intervention comprised music-based group therapy, structured teaching with a therapist,
psychoeducational staff training by a psychologist and intensive family member–staff
communication comprising provision of basic information, everyday availability of
professional carers to answer family members’ questions, and a 1-hour session of
psychoeducational counselling by a psychologist to a close family member of each
participant. For further details, please see the economic evidence profile in Appendix M.
Service-use data included direct costs associated with providing the treatment including staff
time. Data from a longitudinal epidemiological study (LASER-AD) were used to quantify the
relationship between agitation and health and social care costs and agitation and utility. The
analysis adopted a 12-month time horizon.
Table 85: Base-case cost–utility results – Livingston et al. (2015)

Multimodal intervention NR NR
-0.00583
Usual care NR NR £716 QALYs dominated
Base-case results (Table 85) suggest that, although the treatment cost £406 to provide to
each patient, the net cost impact was a saving of £716 compared with usual care. This was
due to a reduction in the costs of managing agitation. The intervention was also more
effective than the comparator, so is considered dominant. PSA suggested that the
intervention had an 82.2% probability of being cost effective, if QALYs are valued at £20,000
each.
313
This analysis, however, had some serious limitations. Critically, data for treatment effects
were from a paper published by Fischer-Terworth and Probst (2011), a non-randomised
study with a small number of participants (n=49).
Zwijsen et al. (2016) compared the cost effectiveness of a non-pharmacological intervention

(‘Grip on Challenging Behaviour’; GRIP), for the management of challenging behaviour in
dementia special care units in comparison with usual care in the Netherlands. The economic
evaluation was performed from a societal perspective alongside a cluster-randomised
controlled trial (Zwijsen et al., 2011; see 14.1.2, above). QALYs were estimated using the
EQ-5D. Challenging behaviour and quality of life was assessed on 5 different occasions,
each 4 months apart. For further details, please see the economic evidence profile in
Appendix M.
Costs were estimated using standard Dutch sources. Staff time was estimated using
prospective 1-monthy diaries.
Table 86: Base-case cost–utility results – Zwijsen et al. (2016)

483
Usual care EUR€ NR
931 276
GRIP EUR€ NR EUR€ -0.02 QALYs dominated
The base-case analysis (Table 86) suggests that GRIP is associated with increased costs
and less QALYs compared with usual care. Probabilistic analysis showed that the probability
of GRIP being cost-effective in comparison with usual care was zero regardless of the value
placed on a QALY. The authors concluded that GRIP was not considered cost-effective in
comparison with usual care.
14.1.3.5 Exercise
D’Amico et al. (2016) conducted a cost–utility analysis alongside EVIDEM-E RCT,

evaluating a 12-week trial of a dyadic exercise regimen (tailored walking) for people living
with dementia and their main carer as therapy for behavioural and psychological symptoms
of dementia. For further details, please see the economic evidence profile in Appendix M.
Data on care and support service use were collected using an adapted version of the Client
Service Receipt Inventory. Unit costs were taken from standard sources (PSSRU, BNF), if
possible, and estimated from market sources if not. All costs were expressed in 2011 UK
pounds. QALYs were calculated for participants with dementia only, using the DEMQOL-
Proxy, completed by the carer, with societal weights.
Table 87: Base-case cost–utility results from a health and social care perspective for
exercise compared with control for D’Amico et al. (2016)
Strategy Cost Effect Cost a [95%CI] Effect a [95%CI] ICER
Control £1,984 NR
£-169.70 0.0055 QALYs
Exercise £2,122 NR [−1240.0, 900.5] [-0.0031, 0.0140] Dominant
Base-case results from a health and social care perspective (Table 61) show that exercise
was associated with lower costs and higher QALYs than control, resulting in the being a
314
dominant. However, findings were very uncertain in both cost and effect dimensions. The
cost savings of the exercise group compared with the control group appear to be driven by
less usage of hospital services by those in the exercise group; however, the authors note
that neither the costs or QALY difference were statistically significant.
14.1.4.1 Anxiety and depression
14.1.4.1.1 Pharmacological treatment
Sertraline vs placebo
Low-quality evidence from up to 3 RCTs containing 348 people could not differentiate levels
of depressive symptoms (Cornell Scale for Depression in Dementia/Hamilton Depression
rating Scale) between people taking sertraline and placebo at any time between 12 and 39
weeks.
Low- to moderate-quality evidence from up to 2 RCTs containing 217 people could not
differentiate global impression of change scores (mADCS-CGIC), cognition (MMSE),
activities of daily living, neuropsychiatric symptoms (NPI) or quality of life (self- or carer-
reported) between people taking sertraline and placebo at any time between 12 and 39
weeks.
Moderate- to high-quality evidence from 1 RCT containing 173 people found worse levels of
carer mental health (GHQ/SF-12) in carers of people taking sertraline compared with placebo
at 13 weeks, but could not differentiate levels at 39 weeks, or carer burden or physical health
at 13 or 39 weeks.
Moderate-quality evidence from 3 RCTs containing 385 people found higher levels of
adverse events in people taking sertraline compared with placebo, but very low-quality
evidence from 2 RCTs containing 348 people could not differentiate levels of serious adverse
events.
Mirtazapine vs placebo
depressive symptoms (Cornell Scale for Depression in Dementia) between people taking
mirtazapine and placebo at 13 or 39 weeks.
Low- to moderate-quality evidence from 1 RCT containing 158 people could not differentiate
cognition (MMSE), activities of daily living, neuropsychiatric symptoms (NPI) or quality of life
(self- or carer-reported) between people taking mirtazapine and placebo at 13 or 39 weeks.
Moderate-quality evidence from 1 RCT containing 158 could not differentiate levels of carer
burden, physical or mental health between carers of people taking mirtazapine and placebo
at 13 or 39 weeks.
Moderate-quality evidence from 1 RCT containing 215 people found higher levels of adverse
events in people taking mirtazapine compared with placebo, but low-quality evidence from
the same study could not differentiate levels of serious adverse events.
315
Antidepressants vs placebo
Low-quality evidence from a systematic review containing 10 RCTs could not identify
evidence of significant benefit with antidepressants compared with placebo for the
management of depressive symptoms in people with dementia.
Low-quality evidence from a systematic review containing 11 studies (5 RCTs) could not
identify evidence of significant benefit with pharmacological treatment compared with
placebo for the management of depressive symptoms in people with Huntington’s disease.
14.1.4.1.2 Psychological treatment vs usual care
Low-quality evidence from 6 RCTs containing 439 people found lower levels of depressive
symptoms in people offered psychological treatment compared with usual care.
Low- to moderate-quality evidence from up to 2 RCTs containing 65 people found lower

levels of anxiety (RAID) in people offered psychological treatment compared with usual care,
but could not differentiate levels of anxiety as measured by self-rating or the NPI-A.
Very-low to low-quality evidence from up to 4 RCTs containing 381 people could not
differentiate levels of quality of life (self-report or proxy), activities of daily living,
neuropsychiatric symptoms, cognition (MMSE) or carer depressive symptoms between
people offered psychological treatment compared with usual care.
14.1.4.1.3 PATH vs ST-CI
Low- to moderate-quality evidence from 1 RCT containing 74 people with either dementia
(n=39) or mild cognitive impairment (n=35) found lower levels of depressive symptoms and
disability in people offered the PATH intervention compared with the ST-CI intervention for
depression.
14.1.4.1.4 Structured depression management vs usual care (in nursing homes)
Moderate-quality evidence from 1 RCT containing 393 people could not differentiate levels of
depressive symptoms (Cornell Scale for Depression in Dementia/Geriatric Depression Scale)
or severe depression between people offered structured depression management compared
with usual care.
High-quality evidence from 1 RCT containing 393 people found higher levels of quality of life
(EQ-VAS) in people offered structured depression management compared with usual care.
14.1.4.1.5 Psychogeriatric management vs usual care
depressive symptoms or psychosis between people offered psychogeriatric case
management, psychogeriatric consultation or usual care.
14.1.4.1.6 Ambient bright lighting
Very low- to low-quality evidence from 1 RCT containing 66 people found higher levels of
depressive symptoms (Cornell Scale for Depression in Dementia) in men exposed to bright
morning light compared with standard lighting, but could not differentiate levels in those
exposed to bright evening or all-day light compared with standard lighting, or in women in
any lighting conditions.
316
14.1.4.1.7 Music therapy
Group music therapy
Moderate-quality evidence from a systematic review containing 8 studies (5 RCTs) could not
identify evidence of significant benefit of group music therapy compared with standard
therapy or non-music interventions for the management of anxiety in people with dementia.
Active music therapy
Low-quality evidence from a systematic review containing 10 studies (3 RCTs) could not
identify evidence of significant benefit with active music therapy compared with usual care or
non-music interventions for the management of anxiety and depressive symptoms in people
with dementia.
Active music therapy vs reading
quality of life (DQOL) or depressive symptoms (Geriatric Depression Scale) between people
offered active music therapy compared with reading therapy.
Preferred music listening vs usual care
Very-low quality evidence from 1 RCT containing 52 people could not differentiate levels of
anxiety (RAID) between people offered preferred music listening compared with usual care.
14.1.4.1.8 High-intensity exercise vs non-exercise activity program
levels of anxiety (RAID) or depressive symptoms (Geriatric Depression Scale/MADRS)
between people offered high-intensity exercise compared with a non-exercise activity
program.
14.1.4.2 Antidepressants for other non-cognitive symptoms
Very low- to moderate-quality evidence from up to 4 RCTs containing 419 people found
improvements in scores on the Cohen-Mansfield Agitation Inventory with SSRIs versus
placebo, but could not differentiate total neuropsychiatric symptoms, behavioural symptoms
or adverse events.
Very low- to moderate-quality evidence from up to 2 RCTs containing 103 people could not
differentiate any outcome measures between SSRIs and atypical antipsychotics, SSRIs and
typical antipsychotics, trazodone and placebo, or trazadone and typical antipsychotics.
14.1.4.3 Antipsychotics
14.1.4.3.1 Atypical antipsychotics versus placebo
Moderate- to high-quality evidence from up to 17 RCTs containing 5,028 people found

improvements in the NPI, Brief Psychiatric Rating Scale, Cohen-Mansfield Agitation
Inventory and Clinical Global Impression of Change with atypical antipsychotics versus
placebo, but higher rates of mortality, somnolence, and extrapyramidal and cerebrovascular
adverse events.
317
14.1.4.3.2 Olanzapine vs haloperidol
Low-quality evidence from 1 RCT containing 58 people could not differentiate cognition
(MMSE), anxiety (CMAI) or neuropsychiatric symptoms (NPI) between people taking
olanzapine and haloperidol.
14.1.4.3.3 Risperidone vs rivastigmine
Moderate-quality evidence from 1 RCT containing 27 people found lower levels of anxiety in
people taking risperidone versus rivastigmine.
14.1.4.3.4 Antipsychotic withdrawal
High-quality evidence from 7 RCTs containing 366 people found a higher proportion of
people who discontinued antipsychotics had a worsening of behavioural and psychological
symptoms of dementia compared with those who continued, but low- to moderate-quality
evidence from up to 6 RCTs containing 462 people could not differentiate overall levels of
behavioural and psychological symptoms of dementia, or rates of early study termination or
mortality.
cognition (SIB/MMSE), neuropsychiatric symptoms (NPI), Parkinsonism (modified UPDRS),
activities of daily living (BADL) or language difficulties (STALD/FAS) between people who
continued antipsychotic medication compared with those who discontinued.
High-quality evidence from 1 RCT containing 109 people found higher levels of
neuropsychiatric symptoms (NPI) in people who discontinued antipsychotic medication
compared with those who continued, but moderate-quality evidence from the same study
could not differentiate levels of cognition (SIB)
High-quality evidence from 1 RCT containing 165 people found lower levels of mortality in
people who discontinued antipsychotic medication compared with those who continued.
14.1.4.3.5 Antipsychotic switch to memantine
agitation (CMAI), neuropsychiatric symptoms (NPI), cognitions (MMSE), activities of daily
living (BADL), serious adverse events or mortality between people on antipsychotics at
baseline who either continued on antipsychotics or were switched to memantine.
14.1.4.3.6 Enhanced psychosocial care
Moderate-quality evidence from 1 RCT containing 338 people found a lower proportion of
people taking antipsychotics in homes that offered an enhanced psychosocial care
intervention compared with usual care, but very low- to low-quality evidence from the same
study could not differentiate rates of falls or levels of aggression and wellbeing.
14.1.4.4 Memantine vs placebo
Low-quality evidence from 3 RCTs containing 427 people could not differentiate cognition
(ADAS-cog), activities of daily living (ADCS-ADL) or neuropsychiatric symptoms (NPI)
between people with mild Alzheimer’s disease taking memantine versus placebo.
318
14.1.4.5 Sleep problems
14.1.4.5.1 Melatonin vs placebo
detect a difference in total night-time sleep time, ratio of daytime to night-time sleep, sleep
efficiency, nocturnal time awake, number of night-time awakenings, carer-rated sleep activity,
activities of daily living, sleep latency or numbers of adverse events between people with
sleep problems taking melatonin versus placebo.
14.1.4.5.2 Trazadone vs placebo
Moderate- to high-quality evidence from 1 RCT containing 30 people found higher levels of
total night-time sleep and better sleep efficiency in people with sleep problems taking
trazadone versus placebo over a two week period, but could not differentiate numbers of
night-time awakenings, total daytime sleep, number of daytime naps or activities of daily
living.
14.1.4.5.3 Memantine vs placebo
Low- to moderate-quality evidence from 1 RCT containing 60 people found a reduction in

REM sleep behaviour disorder (measured using the Stavanger Sleep Questionnaire) in
people with sleep problems taking memantine versus placebo, but could not differentiate
scores on the Epworth Sleepiness Scale.
14.1.4.5.4 Light therapy
Low-quality evidence from 2 RCTs containing 48 people could not detect a difference in total
sleep duration, sleep latency, night-time activity counts or the number of night-time
awakenings in people with dementia and sleep problems exposed to bright light therapy.
14.1.4.5.5 Slow-stroke back massage
Moderate-quality evidence from 1 RCT containing 40 people could not detect a difference in
total night-time sleep time or sleep efficiency in people with dementia and sleep problems
exposed to massage therapy for 2 nights.
14.1.4.5.6 Multicomponent non-pharmacological interventions vs usual care
Moderate- to high-quality evidence from up to 3 RCTs containing 207 people found

improvements in total night-time sleep, total night-time awake time and scores on the Sleep
Disorders Inventory in people offered a multicomponent non-pharmacological intervention
including light exposure, exercise, environmental modification and sleep hygiene advice
versus usual care, but could not differentiate number of night-time awakenings, total daytime
sleep or depressive symptoms.
14.1.4.5.7 Individualised activities
Low- to moderate-quality evidence from 1 RCT containing 50 people found a reduction in the
amount of daytime sleep in people with dementia and a sleep efficiency of <50% at baseline
exposed to individualised social activities for 21 days, but could not differentiate the day/night
sleep ratio or the number of night-time minutes to sleep onset, night-time minutes slept,
night-time sleep efficiency or night-time minutes awake.
319
14.1.4.5.8 Continuous positive air pressure
Moderate-quality evidence from 1 RCT containing 39 people did not detect a difference in the
Epworth Sleepiness Scores in people with dementia and sleep disordered breathing treated
with continuous positive air pressure compared with sham intervention for 3 weeks.
14.1.4.6 Non-pharmacological management of agitation, aggression and apathy
14.1.4.6.1 Sensory interventions
Moderate-quality evidence from up to 5 RCTs containing 446 people could not differentiate
levels of agitation, positive affect, negative affect, depressive symptoms, quality of life,
behavioural pathology or cognition between people offered sensory interventions or usual
care.
14.1.4.6.2 Social contact
Very low-quality evidence from 1 RCT containing 164 people could not differentiate levels of
agitation between people offered social contact interventions (simulated presence or pet
therapy) or usual care.
14.1.4.6.3 Activities
Low- to moderate-quality evidence from up to 6 RCTs containing 465 people found reduced
levels of negative affect and increased levels of pleasurable and interested affect in people
offered activity based intervention versus usual care, but very low- to low-quality evidence
could not differentiate levels of agitation, constructive engagement or negative engagement.
14.1.4.6.4 Care delivery interventions
Very low-to moderate quality evidence from up to 5 RCTs containing 71 people could not
differentiate levels of agitation, aggressive behaviours, or rates of psychotropic prescription
in centres offered or not offered a care delivery intervention, but found higher rates of
antidepressant and cholinesterase inhibitor prescriptions in centres that offered the
intervention.
14.1.4.6.5 Staff training
Moderate- to high-quality evidence from 1 RCT containing 272 people found reduced levels
of anxiety and verbally aggressive behaviours in centres offered a staff training intervention,
but could not differentiate levels of physically aggressive behaviours.
14.1.4.6.6 Herbal extracts
Low- to high-quality evidence from up to 4 RCTs containing 1,596 people found improved
NPI scores, activities of daily living, quality of life and cognition in people offered ginkgo
biloba extract versus placebo.
14.1.4.7 Pharmacological management of agitation, aggression and apathy
14.1.4.7.1 Mood stabilisers vs placebo
Very low- to high-quality evidence from up to 4 RCTs containing 254 people found NPI
scores, cognition (MMSE) and adverse events were significantly worse in people offered
mood stabilisers versus placebo, but could not differentiate levels of agitation (CMAI; NPI
BPRS subscale), total neuropsychiatric symptoms or physical self-maintenance.
320
14.1.4.7.2 Cholinesterase inhibitors versus usual care
Low- to high-quality evidence from up to 3 RCTs containing 317 people found global
assessment, agitation (NPI subscale) and cognition were significantly better in people offered
cholinesterase inhibitors versus placebo, but could not differentiate levels of agitation (CMAI)
or total neuropsychiatric symptoms (NPI).
Moderate- to high-quality evidence from 1 RCT containing 149 people found significantly
lower NPI scores in people offered memantine versus placebo but could not differentiate
levels of agitation, global assessment, clinician assessment or cognition.
14.1.4.7.4 Tetrahydrocannabinol versus placebo
agitation (CMAI; NPI agitation aggression subscale), neuropsychiatric profile, aberrant
behaviours, clinician global assessment, activities of daily living or quality of life in people
offered tetrahydrocannabinol versus placebo.
14.1.4.7.5 Prazosin versus placebo
Very low- to low-quality evidence from 1 RCT containing 13 people found improvements in
psychiatric assessment and global assessment in people offered prazosin versus placebo,
but could not differentiate neuropsychiatric profile (NPI).
14.1.4.7.6 Dextromethorphan-quinidine vs placebo
Moderate- to high-quality evidence from 1 RCT containing 279 people found improved
neuropsychiatric symptoms (NPI), agitation/aggression and depressive symptoms (Cornell
scale) but higher levels of adverse events in people offered dextromethorphan-quinidine
versus placebo. The evidence could not differentiate global assessment, cognition, quality of
life, serious adverse events or mortality.
14.1.4.7.7 Modafinil vs placebo
apathy, function, activities of daily living or carer burden between people offered modafinil
versus placebo.
14.1.4.7.8 Donepezil and choline alphoscerate vs donepezil
High-quality evidence from 1 RCT containing 113 people found improved levels of apathy,
neuropsychiatric symptoms (NPI) and cognition (MMSE, ADAScog) in people offered
donepezil and choline alphoscerate versus donepezil alone.
14.1.4.8.1 Antidepressants
One directly applicable cost–utility analysis with very serious limitations found that
mirtazapine has a probability in excess of 90% of being superior to sertraline, regardless of
the assumed value of a QALY, but it only had a probability of approximately 20% of being
associated with an ICER of £20,000 per QALY or better compared with placebo. The study
undertook only limited exploration of uncertainty and had a short time horizon.
321
14.1.4.8.2 Antipsychotics
Two partially applicable cost–utility analyses from the USA explored the cost effectiveness of
second-generation antipsychotics compared with each other and/or usual care.
 One RCT-based analysis with potentially serious limitations found that there were no
significant differences in QALYs across the treatment groups, with olanzapine being
worse than placebo. Both risperidone and sertraline produced very small increases in
QALYs which were insufficient to outweigh additional costs even when QALYs are valued
at $100,000 each.
 One analysis with very serious limitations, based on an observational study, found that,
assuming QALYs are valued at $50,000 each or more, olanzapine is likely to be
considered cost-effective, compared with usual care, for the treatment of agitation and
psychosis in Alzheimer’s disease. However, effect data were drawn from a study that
found that olanzapine resulted in a loss of QALYs.
14.1.4.8.3 Non-pharmacological interventions
One directly applicable cost–utility analysis with very serious limitations explored a
multimodal, non-pharmacological intervention consisting of music therapy, sensory
interventions and training. The intervention was found to have a 82% probability of being cost
effective if QALYs are valued at £20,000 each. However, the analysis relied on a non-
randomised study with a small number of participants (n=49) for effects.
One partially applicable economic evaluation with potentially serious limitations, which was
performed alongside a cluster RCT in the Netherlands, found that the Grip on Challenging
Behaviours care programme (GRIP) was less effective and more expensive than usual care.
14.1.4.8.4 Exercise
One partially applicable cost–utility analysis with minor limitations conducted alongside an
RCT explored the cost effectiveness of a dyadic exercise regimen for people living with
dementia and their main carer as therapy for behavioural and psychological symptoms of
dementia. Exercise was found to result in lower costs and higher QALYs than control,
resulting in the exercise intervention being dominant. The authors noted that neither the
costs nor QALY difference were statistically significant.
14.1.5.1 Pharmacological interventions

Relative value of different The committee agreed that evaluations of most interventions depended
outcomes on 3 components; whether the intervention improves the specific
symptom(s) it is targeting; whether there are any broader impacts on
cognition, function or wellbeing; and adverse events. Many
pharmacological treatments are known to have worse adverse event
profiles in people with cognitive decline than in people without, and
therefore the committee agreed this trade-off was always an important
one to consider.
Sleep disturbances
The committee agreed that whilst daytime sleep was a relevant
outcome, the most important individual measure would be of nocturnal
time awake, as this is a particular issue that can impact on carer quality
of life. It also agreed that it would be important to consider participant
and carer reported outcomes alongside the actigraph measurements
which are commonly reported in trials of sleep disturbances.
322
Trade-off between Depression and anxiety

benefits and harms The committee agreed the evidence did not show significant benefits
from using antidepressants to treat mild to moderate depression in
people with mild to moderate dementia. In particular, the large DIADS-2
and HTA-SADD studies did not produce any positive findings. Whilst
only 2 specific antidepressants were tested in these trials, the
committee agreed this was likely to represent a class effect, and
therefore any recommendation made should apply to all
antidepressants. There were, however, specific caveats that the
committee agreed were important. First, it agreed that it would not be
appropriate to extrapolate these findings to people with either severe
depression (where there may be an urgent need for treatment) or
severe dementia. Second, it agreed that if someone had previously
responded well to antidepressant treatment, then it would be
appropriate to use the same treatment if the person later develops
dementia.
Antipsychotics
The committee agreed there was a clear pattern in the evidence for
antipsychotics. They showed clear evidence of efficacy (reductions in
agitation and NPI scores), but also evidence of significant harms, with
increase in rates of all types of adverse events, and mortality. The
committee agreed that the significant risks of treatment meant their use
should be restricted as much as possible, and limited only to situations
either where there is an urgent need for treatment to prevent harm to
the person living with dementia or others, or where the use is for the
treatment of an underlying psychosis, and would be equally appropriate
in a person who does not have dementia. The committee also agreed
that a specific discussion is necessary with the person living with
dementia and their carers/family members about the benefits and harms
of treatment. It agreed that treatment should be restricted to the lowest
effective does and the shortest possible time, in order to reduce adverse
events as far as possible.
The committee agreed that it was necessary to regularly review people
taking antipsychotics to ensure the treatment is still necessary, and to
encourage a discussion about discontinuation wherever this is possible.
It also agreed that the use of an antipsychotic was not a reason to
discontinue non-pharmacological treatment, and that people either
taking or being discontinued from antipsychotics should have access to
the same range of non-pharmacological options as people not being
treated with antipsychotics.
It was noted that the majority of the included studies were for non-
cognitive symptoms such as agitation or similar behavioural symptoms,
rather than as treatments for psychosis.
It was also agreed to be appropriate to add a specific recommendation
in the guideline around the risk of worsening motor features and
antipsychotic sensitivity reactions in people with Parkinson’s disease
dementia or dementia with Lewy bodies taking antipsychotics. This
recommendation links to the NICE guideline on Parkinson’s disease,
which contains additional recommendations on this topic.
Sleep disturbances
The committee agreed the evidence did not show significant benefit
from using melatonin to treat insomnia in people with Alzheimer’s
disease, and that the evidence base contained a sufficiently large
sample size that any meaningful benefit would have been detected. As
a result of this and the known adverse events associated with melatonin
treatment (including headaches and dizziness) the committee agreed it
should not be used to manage insomnia. The committee agreed it was
not appropriate to extrapolate this evidence to other subtypes of
dementia. For example, melatonin is used to treat REM sleep disorder
in people with Parkinson’s disease, and the committee did not want to
323
restrict its use for people with Parkinson’s disease dementia or

dementia with Lewy bodies without robust evidence.
The committee agreed that the evidence for trazadone was promising,
but noted that since the trial only had a 2 week duration it was unclear
whether the increase in night-time sleep time would be sustained long-
term. The committee agreed that the full effects of trazodone could take
longer to be detected and that the increase in night-time sleep detected
here could be due to the sedative effect of the drug rather than an effect
on sleeping per se. In addition it agreed that the lack of improvement in
carer reported outcomes was an issue and that the side-effects of
trazadone treatment should also be considered before prescription. It
therefore decided against recommending the use of trazadone, but
included a research recommendation to examine the effectiveness of
pharmacological interventions to treat sleep problems in people with
dementia who had failed to respond to non-pharmacological
interventions.
Agitation, aggression and apathy
The committee agreed the evidence relating to the use of valproate was
demonstrably robust. It noted the significant presence of adverse events
in people living with dementia receiving the intervention, but recognised
that this was contrary to the beneficial effect mood stabilisers can have
in non-dementia specific populations. As a result, the committee agreed
it would be appropriate to recommend limiting the use of valproate in
people living with dementia to those with pre-existing mood disorders
only (specifically, in situations where they had already shown
effectiveness before the onset of cognitive decline). The committee
noted there was only very limited evidence on other mood stabilisers,
and therefore it was appropriate to restrict the recommendation solely to
valproate.
In addition, the committee noted that preliminary findings regarding the
use of dextromethorphan-quinidine for agitation in people living with
dementia, and donepezil plus choline alphoscerate for the management
of apathy in people living with dementia did show some positive
preliminary results. Although these results were limited the committee
agreed it would be beneficial to pursue these lines of research and
suggested these would be appropriate topics in which to make research
recommendations.
Trade-off between net Depression and anxiety
health benefits and The committee noted that the findings of the economic evaluation
resource use conducted alongside the HTA-SADD trial were also negative, with
mirtazapine only having a 20% probability of being cost-effective versus
placebo at £20,000/QALY, and sertraline an even lower probability. The
committee agreed this added further justification to their conclusion that
antidepressants not be routinely used in this population.
Antipsychotics
The committee agreed that the most robust economic evidence
available in this area came from the Rosenheck et al publication
conducted alongside the CATIE-AD trial. This study concluded that
antipsychotic treatment was not cost-effective for the trial population,
and the committee agreed this further supported their conclusion that
treatment should be restricted to those cases where it is urgently
needed.
Sleep disturbances
No economic evidence was identified for sleep disturbances, but the
implementation of the negative recommendation for melatonin was
agreed to be likely to be cost-saving.
324
No positive recommendations were made for pharmacological treatment

for agitation, aggression and apathy, and therefore the committee was
not concerned by the lack of economic evidence identified.
Quality of evidence Depression and anxiety
The committee agreed that both the DIADS-2 and HTA-SADD trials
were of good quality, with sufficiently large sample sizes that it would be
reasonable to expect an effect to have been detected, if a meaningful
one was present at the population level. It agreed that whilst there may
be individual people who respond well to antidepressants, it is not
usually possible to identify these people prospectively, and therefore no
specific recommendations could be made for this group.
Antipsychotics
The committee agreed that there were good quality studies with large
sample sizes looking at both antipsychotic efficacy and the effects of
antipsychotic discontinuation. There were also long-term studies looking
at the effects of antipsychotics on mortality, and therefore the committee
agreed there was a robust evidence base behind the recommendation
made.
Sleep disturbances
The committee noted that since the trazadone trial only ran for 2 weeks
and involved a small number of participants, care was needed in
interpreting the positive findings presented.
The committee were concerned about the use of the Epworth
Sleepiness Scale and Stavanger Sleep Questionnaire as primary
outcome measures in the Larsson study. Combined with the lack of
actigraph data and carer outcomes this led it to rate the study quality as
poor and as a result they lacked confidence in the reported
improvement in REM sleep disorder presented.
The committee observed that aside from the evidence relating to mood
stabilisers, the evidence from all other pharmacological interventions
associated with the management of agitation, aggression or apathy
came from a limited number of RCTs or single studies only. For this
reason the committee were cautious and agreed it would be
inappropriate to make broad based recommendations on these
interventions without a more comprehensive evidence base.
Other considerations A common exclusion criteria across many of the trials in this review was
either people who were deemed to need treatment sufficiently urgently
that they could not be included in the study, or had sufficiently severe
symptoms that randomisation was not considered appropriate.
Therefore, the majority of the evidence base consists of people not
considered at urgent need of treatment, and it is unlikely that RCTs
would be conducted in this very severe population. Therefore, the
committee agreed it was important to note that the recommendations
made focus on this non-urgent population, and individual clinician
judgement would be important in those people where it was felt there
was an urgent need for intervention to prevent harm to the individual.
14.1.5.2 Non-pharmacological interventions

Relative value of different The committee agreed that evaluations of most interventions depended
outcomes on 3 components; whether the intervention improves the specific
symptom(s) it is targeting; whether there are any broader impacts on
cognition, function or wellbeing; and adverse events.
Trade-off between Depression and anxiety
benefits and harms The committee agreed there was evidence of efficacy for psychological
treatments in the management of depression and anxiety. There were
significant reductions in overall levels of depression across the studies,
as well as reductions in the primary anxiety measure in the 2 studies
325
where people were required to have elevated levels of anxiety at study

entry. The population in the studies was composed of people with mild
to moderate dementia and mild to moderate depression, and therefore
the committee agreed it was appropriate to make a recommendation for
this group.
The only other intervention which showed evidence of benefit was the
PATH intervention, but this was conducted in a setting where only
around 50% of the people had dementia (the rest having mild cognitive
impairment), and since the results were not reported separately for the 2
populations, the committee were not confident to conclude the
intervention was specifically effective for people living with dementia.
Managing distress
The committee agreed that reactions which are classified as
behavioural symptoms of dementia were often responses to other
underlying problems in the context of difficulty in communicating needs
effectively. For example, people with pain or delirium or who are
responding to inappropriate care may be labelled as having behavioural
problems when in fact there is a need to treat the underlying pain or
delirium, and/or to improve the environment. The committee therefore
agreed that, before any interventions for distress are considered, it is
important that a thorough structured assessment of the person and their
environment be conducted to try and identify and address the
underlying causes of distress.
If this assessment is unsuccessful in identifying approaches that can
resolve the problem, then in view of the clearly established harms of
antipsychotics, the committee agreed it was appropriate that non-
pharmacological management (both environmental and psychosocial)
be offered before any thought is given to the use of antipsychotics. The
committee also noted that the evidence showed that staff training in
appropriate use of non-pharmacological methods showed the use of
antipsychotics could be significantly reduced without any subsequent
increase in neuropsychiatric symptoms, and therefore it was agreed this
would form an appropriate part of the training staff should receive in
managing non-cognitive symptoms (this is included as part of a
recommendation in section 16 on staff training).
The committee also noted the evidence on two somewhat different
approaches to managing symptoms; more structured psychosocial
interventions and less structured interventions based around offering
enjoyable and personalised activities. The committee agreed it was
appropriate for both to be mentioned in the recommendations as they
may each be useful in different situations.
Sleep disturbances
The committee agreed there was a lack of evidence in support of the
use of light therapy for people with dementia and sleep problems.
The benefits of multicomponent interventions, including NITE-AD, were
considered and the committee agreed that although the participants
showed an increase in sleep at night-time and a reduction in awake-
time at night, the importance of these improvements was hard to judge
in the absence of data on the effect on participant or carer quality of life.
Since the evidence suggested that light therapy alone was ineffective at
increasing sleep the committee attributed the changes seen to the other
interventions included in the programme, namely improved sleep
hygiene and daily exercise. The discussion expanded to include the
reduction in day-time sleep associated with individualised social
activities, which despite not being associated with a significant increase
in night-time sleep could have an important effect on people’s quality of
life by enriching their environment. The committee commented on the
value of improved sleep hygiene, exposing people to daylight, physical
and other pleasurable activities on general health and wellbeing of
people with dementia. The committee agreed the positive results of the
326
3 trials looking at multicomponent interventions justified a

recommendation in favour of an intervention involving sleep hygiene,
daylight exposure and exercise, but in the absence of robust data on
quality of life felt this recommendation should be kept at the “consider”
level.
The committee recognised there was some overlap in symptoms
experienced by people living with dementia; managing symptoms is not
necessarily discrete.
Overall, the committee agreed that staff training studies demonstrated
the most positive findings. The committee considered the relevance of
training implementation. It noted that the health technology assessment
presented for Livingston (2014) had demonstrated similar findings and
consequently the committee concluded that the focus for
implementation should rest upon care providers where group training
sessions could be offered. It recognised that the evidence reported in
Deudon (2009) suggested that supporting and mentoring staff could
have positive effects. For this reason the committee agreed it was
appropriate to incorporate specific aspects of that reported intervention
within a recommendation. Specifically, the committee recognised the
benefit of face to face interventions rather than e-learning interventions.
It agreed that it was important for initial sessions to be followed by on
the job support sessions, focusing upon specific content.
In addition, the committee considered the evidence for ginkgo biloba in
treating symptoms of agitation in people with dementia. It acknowledged
that the pooled outcomes from 4 trials demonstrated positive outcomes
for treating behavioural and psychological symptoms of dementia but
was also mindful that evidence from non-dementia specific populations
had observed effects from drug interactions. The committee also noted
that ginkgo biloba is on a list of items that currently cannot be
prescribed in primary care, and therefore agreed it was not appropriate
to make any recommendations on its use.
Trade-off between net Depression and anxiety
health benefits and No economic evidence was identified for non-pharmacological
resource use management of depression and anxiety. However, the committee
agreed that the recommendation made was broadly similar to that in the
NICE guideline for depression, and therefore it was unlikely there would
be significantly higher resource use in this group.
Managing distress
The committee noted there was no specific evidence available on the
cost-effectiveness of non-pharmacological interventions for managing
distress. However, it did note the evidence from the 2014 Livingston
HTA report which demonstrated that successful non-pharmacological
interventions for managing non-cognitive symptoms could be cost-
saving, due to the reductions in subsequent treatment costs for those
receiving early interventions. The committee agreed that this recent
HTA report represented the best quality economic evidence available,
and supported the recommendation for the first line use of non-
pharmacological management.
Agitation and aggression
The committee noted that the Grip on Challenging Care programme
(GRIP) (trialled by Zwijsen in 2011), a non-pharmacological treatment
strategy for patients with dementia who display symptoms of agitation
and aggression, resulted in a loss of QALYs, despite incurring a small
but modest cost for the treatment. GRIP is therefore a dominated
strategy. The committee also took note of the EVIDEM-E study and
concluded that evidence for exercise for behavioural and psychological
symptoms of dementia did not show a significant difference in terms of
costs or benefits and were therefore unable to recommend it.
327
Sleep problems
The potential cost of a programme similar to NITE-AD was discussed by
the committee with concern being raised about the cost of multiple
sessions with a geropsychologist. It was suggested that another suitably
trained person could deliver these sessions instead and at less cost to
the health system. In addition, the committee noted that the study by
Alessi 2005, which contained a similar multicomponent intervention but
without the same extensive input from a geropsychologist, appeared to
be equally effective, and was therefore confident that such an
intervention could be delivered without imposing a large additional
resource burden.
Quality of evidence Depression and anxiety
The committee noted that the trials included in the meta-analysis of
psychological interventions were heterogeneous, both in terms of the
entry criteria into the trial and the interventions studied. Unfortunately,
the sample sizes of the individual trials were small, so it was not
possible to identify which individual interventions (e.g. CBT, counselling,
psychodynamic interpersonal therapy) were likely to have the most
robust effects, and therefore the committee agreed that only a general
recommendation for psychological treatments was appropriate. The
committee agreed it was appropriate to make a research
recommendation about the most effective psychological treatments for
anxiety and depression, in order to identify which of this class of
interventions is most appropriate to offer to people living with dementia.
Managing distress
The committee noted that the evidence on using enhanced
psychosocial care to reduce antipsychotic prescribing rates came from a
single study. Therefore, whilst it was confident to recommend that this
should form part of the training given to staff to manage anxiety, it did
not feel that it was appropriate to recommend any specific form these
interventions should take based on this single study.
Sleep problems
The committee commented that the short time frame of 2 days used in
the slow-stroke back massage pilot study (Harris 2012) was probably
insufficient to determine effects on sleep problems. The committee
agreed that it was difficult to fully disentangle the effects of the individual
components of the NITE-AD intervention, and therefore it was not
possible to make recommendations about which parts of the
intervention were the most important.
The committee was concerned that patients with sleep disordered
breathing in the Chong 2005 study were not generally representative of
people with dementia and a sleep problem, and that the use of a mask
in people with dementia would be both problematic and potentially
distressing.
Other considerations The committee agreed that one of the factors leading to sleep problems
for people with dementia is likely to be a lack of stimulation during the
day. This can lead to day-time sleepiness and therefore affect sleep
patterns at night. The committee therefore agreed that the non-
pharmacological interventions considered elsewhere in this guideline for
the improvement of function and wellbeing in people which dementia
would be likely to also have an impact on sleep problems for some
people.
The committee noted that only very limited evidence as found on
managing non-cognitive symptoms in people with Parkinson’s disease
dementia or dementia with Lewy bodies, and therefore agreed it was
appropriate to cross-refer to the advice in the NICE Parkinson’s disease
guideline. However, the committee noted these interventions may need
to be modified to be appropriate or a population of people living with
328
dementia, and agreed it was important to highlight this within the

recommendation.
Agitation, aggression, distress and psychosis
91. Before starting non-pharmacological or pharmacological treatment for distress in

people living with dementia, conduct a structured assessment to:
 explore possible reasons for the person’s distress and
 check for and address clinical or environmental causes (for example
pain, delirium or inappropriate care).
92. As initial and ongoing management, offer psychosocial and environmental

interventions to reduce distress in people living with dementia.
93. Only offer antipsychoticsh,i for people living with dementia who are either:
 at risk of harming themselves or others or
 experiencing agitation, hallucinations or delusions that are causing them
severe distress.
94. Be aware that for people with dementia with Lewy bodies or Parkinson’s disease
dementia, antipsychotics can worsen the motor features of the condition, and in
some cases cause severe antipsychotic sensitivity reactions. For more guidance,
see the advice on managing delusions and hallucinations in the NICE guideline on
Parkinson’s disease. Be aware that interventions may need to be modified for
95. Before starting antipsychotics, discuss the benefits and harms with the person
and their family members or carers (as appropriate). Consider using a decision aid
to support this discussion.
96. When using antipsychotics:

 use the lowest effective dose and use them for the shortest possible
time
 reassess the person at least every 6 weeks, to check whether they still
need medication.
97. Stop treatment with antipsychotics:

 the person is not getting a clear ongoing benefit from taking them and
h The MHRA (2012) has given advice for health and social care professionals on prescribing antipsychotics to
people living with dementia to treat the behavioural and psychological symptoms of dementia.
i At the time of publication (June 2018), the only antipsychotics with a UK marketing authorisation for this
indication were risperidone and haloperidol. The marketing authorisation for risperidone only covers short-term
treatment (up to 6 weeks) of persistent aggression in people with moderate to severe Alzheimer's disease
unresponsive to non-pharmacological approaches and when there is a risk of harm to self or others. The
marketing authorisation for haloperidol only covers treatment of persistent aggression and psychotic
symptoms in people with moderate to severe Alzheimer's dementia and vascular dementia when non-
pharmacological treatments have failed and when there is a risk of harm to self or others. The prescriber
should follow relevant professional guidance, taking full responsibility for the decision. Informed consent
should be obtained and documented. See the General Medical Council’s Prescribing guidance: prescribing
unlicensed medicines for further information.
329
 after discussion with the person taking them and their family members or
carers (as appropriate).
98. Ensure that people living with dementia can continue to access psychosocial and
environmental interventions for distress while they are taking antipsychotics and
after they have stopped taking them.
99. For people living with dementia who experience agitation or aggression, offer
personalised activities to promote engagement, pleasure and interest.
100. Do not offer valproate to manage agitation or aggression in people living with
dementia, unless it is indicated for another condition.j
Depression and anxiety
101. For people living with mild to moderate dementia who have mild to moderate
depression and/or anxiety, consider psychological treatments.
102. Do not routinely offer antidepressants to manage mild to moderate depression

in people living with mild to moderate dementia, unless they are indicated for a
pre-existing severe mental health condition.
Sleep problems
103. Do not offer melatonin to manage insomnia in people living with Alzheimer’s
disease.
104. For people living with dementia who have sleep problems, consider a
personalised multicomponent sleep management approach that includes sleep
hygiene education, exposure to daylight, exercise and personalised activities.
Parkinson’s disease dementia and dementia with Lewy bodies
105. For guidance on the management of Parkinson’s disease symptoms in people

with Parkinson’s disease dementia and Dementia with Lewy bodies, see the NICE
guideline on Parkinson’s disease. Be aware these interventions may need to be
modified for people living with dementia.
13. What are the most effective psychological treatments for managing depression or
anxiety in people living with dementia at each stage of the condition?
14. What is the effectiveness and cost-effectiveness of dextromethorphan-quinidine

for managing agitation in people living with dementia?
15. What is the effectiveness and cost-effectiveness of choline alphoscerate for

managing apathy in people living with dementia?
jIf relevant, follow MHRA advice that valproate medicines are contraindicated in women and girls of childbearing
potential unless a Pregnancy Prevention Programme is in place.
330
16. What is the effectiveness of pharmacological treatments for sleep problems in

people who have not responded to non-pharmacological management?
see appendix L.
331
Supporting informal carers
15 Supporting informal carers

Informal carers are the wives, husbands, partners, children, other family members, and
friends who provide vital unpaid care and support for people with dementia living in the
community. In the UK, informal carers provide 1.34 billion hours of unpaid care to people with
dementia each year, equating to a cost of £11.6 billion per year, or 44% of the total cost of
dementia (Dementia UK, 2nd Edition, Alzheimer’s Society, 2014). People with dementia have
particular needs for care which are greater and more complex than those of other users of
long-term care (World Alzheimer Report 2013), but provision of home care for people with
dementia is not keeping pace with rising need (Carers UK, 2015). This places significant
demands on informal carers.
For many carers, providing care has positive aspects, such as feeling closer to the person
with dementia and fulfilling commitments made in lasting relationships. However, caring also
brings many challenges (Merrilees, 2016). Carers experience loss and grief at the changes in
the person they care for and in their relationship. Many feel isolated and lonely, and lack
confidence in their ability to cope. Caregiving can be mentally and physically exhausting,
especially where the person’s behaviour causes concern or personal care needs are high.
Many carers also have other responsibilities, such as caring for children or grandchildren, or
experience practical problems such as lack of financial resources or unsuitable
accommodation. Some carers, especially older carers, have health and mobility problems of
their own.
Consequently, carers of people with dementia experience high levels of stress, and this
impacts on their own physical and mental well-being. It is essential that carers have good
support to enable them to manage the stresses and demands of caregiving and enable them
to fulfil their role. This includes both support for the specific challenges of caregiving and
support to address their own needs. Effective support will be tailored to the carer’s personal
circumstances and needs as well as the characteristics of the person with dementia for
whom they are providing care; for example support needs will differ at different stages of
dementia progression, and will continue after the person with dementia has moved into
residential care. Carers are entitled under the Care Act 2014 to an assessment of their own
needs and to have these needs addressed. The key questions for this review therefore focus
firstly on how carers’ needs can best be assessed and secondly on what kinds of
interventions and services are effective in supporting carers’ well-being and hence enabling
them, if they so choose, to continue providing care..
332
15.1 Supporting informal carers of people living with dementia

Review questions
 How effective are carers’ assessments in identifying the needs of informal carers of
 What interventions/services are most effective for supporting the wellbeing of informal
carers of people living with dementia?
15.1.1 Introduction
The aim of these review questions was to determine the effectiveness of different methods of
assessing the needs of carers, and the effectiveness of interventions and services for
supporting informal carers of people living with dementia. The 2014 Care Act gives carers a
legal right to an assessment if they request one, and therefore the focus of the assessments
question was on how and by whom these assessments should be conducted. The review
identified studies that fulfilled the conditions specified in Table 88 and Table 89. For full
details of the review protocols, see appendix C.
Table 88: Review summary: informal carers’ assessments

Population Carers of people (aged 40 years and over) living with dementia
Interventions Formal assessments of the needs of carers of people living with
dementia
Comparator  Alternative assessment methods
 No formal assessment
Outcomes  Access to health and social care support
 Carer burden and stress
 Carer experience and satisfaction
 Carer health-related quality of life
Table 89: Review summary: interventions/services for informal carers

Population Carers of people (aged 40 years and over) living with dementia
Interventions Interventions or services designed to improve the wellbeing of informal
carers of people living with dementia, which may include:
 Peer support groups
 Training/information courses
 Information
 Psychosocial support
 Cognitive behavioural therapy
 Respite breaks
 Standard care
Outcomes  Carer burden and stress
 Carer experience and satisfaction
 Carer health-related quality of life
333
This review was conducted as an update from the previous dementia guideline (CG42). All
included RCTs and systematic reviews from the previous guideline were considered.
Relevant RCTs, together with identified RCTs in the included systematic reviews, were
screened at title and abstract level, with 112 potentially relevant references ordered for full-
text review. The previous guideline included interventions for all carers, whilst this update
only considered interventions for carers of people living with dementia, and therefore not all
papers included in the previous guideline were eligible for inclusion in this update.
In addition, a systematic literature search for RCTs published since the last guideline
identified 6,738 references. These were screened at title and abstract level, with 63 papers
ordered as potentially relevant systematic reviews, 247 ordered as potentially relevant RCTs
of interventions/services for carers, and 3 ordered as potentially providing relevant data on
carers’ assessments. Finally, because of the limited RCT data available on carers’
assessments, a separate search was conducted to look for comparative observational
studies on carers’ assessments, identified from the full set of observational studies related to
carers of people living with dementia using a keyword search for variations on the terms
assess/assessment. In total, 1,477 references were identified, of which 7 were judged to be
potentially relevant and ordered for full-text review.
15.1.2.1 Interventions/services for informal carers
Interventions were classified into one of 7 different types, based on a classification by

Sörensen et al 2002 used in the previous guideline, but updated to take into account the
changes in types of interventions available, in particular the increasing using of technology
and more formal models of case management. These intervention types were defined as
follows:
 Psychoeducational interventions – Structured programmes providing information about
how to effectively respond to dementia-related difficulties, such as memory or behavioural
problems.
 Skills training – Training and information on specific practical aspects of care for a person
living with dementia.
 Psychoeducation and skills training – Interventions containing both of the components
listed above.
 Supportive interventions – Professionally or peer-led unstructured support groups
 Respite care – In-home or site-specific assistance with care, designed to give the carer
time off from providing care.
 Psychotherapy – Formal therapeutic interaction between a trained professional and the
carer. This category includes both cognitive-behavioural therapy and interventions
modelled on a cognitive-behavioural approach.
 Case management – A specified case manager is responsible for the assessment,
planning, facilitation, coordination and evaluation of an individual’s care, to ensure this
care is optimised.
 Multicomponent intervention – Any intervention containing elements of at least two of the
categories above (with the exception of the combined psychoeducation and skills training
category). Interventions containing multiple components from the same category (e.g. an
intervention containing both group and individual psychoeducation) would be classified
under that rather than as a multicomponent intervention. Interventions were sub-
categorised according to whether all components of the interventions were aimed at the
carer, or whether some components were aimed at the carer and some the care receiver.
A table summarising what components were contained in multicomponent studies is given
in Appendix E.
334
Where appropriate, these types were further sub-categorised accordingly to the way they
were delivered (individual versus group, in person versus telephone versus online).
The most recent, high quality systematic reviews in each of these areas were included as
part of the analysis, with older systematic reviews being excluded if they did not provide
additional relevant data. All RCTs contained within those systematic reviews were retrieved
to check for additional relevant data, which was also included in any analyses undertaken.
Any additional identified RCTs not contained in any included systematic reviews were
included separately. Some interventions did not fit anywhere within the classification, and
were therefore kept separate. These interventions were:
 Exercise
 Attendance at a memory clinic
 Meditation/mindfulness
 Cranial electrotherapy stimulation
In total, 8 systematic reviews and 38 additional RCTs were included, with 112 studies from
the old guideline excluded at full-text review, and 266 studies (55 systematic reviews and
210 RCTs) from the new search excluded at full-text review. Studies excluded from both the
old dementia guideline and the new search are listed, with reasons for exclusion, in appendix
F, and evidence tables for both systematic reviews and RCTs are available in appendix E.
Evidence was further summarised in GRADE profiles, which are given in appendix G.
15.1.2.2 Carers’ assessments
The 10 identified papers (from both the RCT and observational searches) were screened at
full-text level, with 10 excluded due to not matching the study protocol, and therefore no
studies were included in the final review. Studies excluded are listed, with reasons for
exclusion, in appendix F.
15.1.2.3 Analyses
For both questions, data from different studies were meta-analysed when possible, with
GRADE tables and meta-analysis results given in appendices G and H, respectively.
15.1.2.3.1 Meta-regression
A meta-regression analysis was additionally carried out for the question on interventions for
informal carers. In this analysis, rather than the separate categories of intervention being
compared to usual care (or another active intervention) independently, all the studies are
combined into a single network meta-analysis, with a regression model being fitted to the
effect sizes from each trial to estimate the contribution of different intervention components to
the overall effect size. This analysis has the advantage of being able to decompose
multicomponent interventions, and estimate the effect of individual components within those
interventions, rather than treating multicomponent interventions as a single homogeneous
class. Sufficient data were only available to undertake this analysis for the outcome of carer
depression, with the following analyses being conducted:
 An analysis containing the full list of all components included within any intervention.
 A restricted analysis, only including those intervention components included in enough
studies to provide reliable estimates of effect.
 An analysis of the effect of the mode of intervention delivery (individual, group, telephone,
technology-based).
335
 An analysis of the effect of the target of an intervention (carer only, or dyadic for the
person living with dementia and their carer).
 An analysis including both intervention components and mode of delivery.
 An analysis including both intervention components and the target of an intervention.
 Two analyses looking at intervention components, and also including specific parameters
for the numbers of components in an intervention, to test for either an additive effect of
multiple interventions, or for a diminution of effect, where the overall effectiveness of a
multicomponent intervention is less than the sum of its components.
Further details on and result of this analysis are given in appendix H.
15.1.2.4.1 Carers’ assessments
No evidence was identified evaluating the effectiveness of carers’ assessments. Many of the
multicomponent interventions identified did include a structured assessment, but it was not
possible to isolate the effect of one particular component of the interventions.
15.1.2.4.2 Interventions/services for informal carers
Systematic reviews

Study
referenc study Outcomes of
e design Study population Included studies interest
Jensen Systema RCTs comparing educational 7 RCTs comparing Carer burden
(2015) tic interventions with usual care educational Carer depression
review for informal carers of people interventions for Carer quality of life
living with dementia carers of people
Admissions to long
living with
stay facility
dementia versus
usual care
Laver Systema RCTs comparing the efficacy of 17 RCTs included Carer depressive
(2016) tic multicomponent interventions for carer only symptoms
review for the carer and dyadic multicomponent Carer quality of life
interventions for the carer and interventions Carer reaction to
person living with dementia to versus usual care behavioural and
usual care 23 RCTs included psychological
for dyadic symptoms of
multicomponent dementia
interventions
versus usual care
Lins Systema RCTs comparing the efficacy of 9 RCTs included in Carer depression
(2014) tic telephone counselling for the review Carer burden
review informal carers of people living Distress
with dementia to usual care
Anxiety
Quality of life
Care-giving self-
efficacy
Maayan Systema RCTs comparing respite care 3 RCTs compared Carer Burden (ZBI)
(2014) tic with a control intervention for respite care for Carer psychological
review carers of people living with carers of people stress and health
dementia living with
336
Study
referenc study Outcomes of
e design Study population Included studies interest
dementia versus a
control intervention
Parker Systema Studies assessing the 13 studies Psychological
(2008) tic effectiveness of interventions considered psycho morbidity
review that assist carers to provide educational Self-reported
support for people living with interventions; perceptions of
dementia in the community 7 studies knowledge
considered Quality of life
support; Health service
12 studies utilisation (including
reported carer satisfaction
multicomponent
interventions
Scott Systema Studies considering trials of 4 studies included Carer depression
(2016) tic pure technology based CBT in the review
review interventions for carers of
Thomps Systema RCTs evaluating the efficacy of 4 studies classed Quality of life
on tic individual or group based as technology Physical and mental
(2007) review technology interventions based; health
13 studies classed Burden or
as group based; satisfaction
27 studies classed Time spent on
as individual based caring activities
Vernooij- Systema Studies using cognitive 11 RCTs included Depression
Dassen tic reframing to reduce carer’s in the review Anxiety
(2011) review problems by changing their Quality of life
beliefs and interpretations.
Self-efficacy
Burden
337

Randomised controlled trials
A total of 111 unique RCTs containing relevant data were identified, of which 107 compared
the following interventions with either standard care or a less intensive intervention:
 9 RCTs of psychoeducational interventions (4 group, 2 individual, 2 technology-based, 1
telephone-based)
 12 RCTs of skills training interventions (4 group, 5 individual, 2 technology-based, 1
telephone-based)
 20 RCTs of combined psychoeducation and skills training interventions (6 group, 9
individual, 2 technology-based, 3 telephone-based)
 7 RCTs of supportive interventions (3 group, 1 individual, 3 telephone-based)
 3 RCTs of respite care
 15 RCTs of psychotherapy interventions (8 group, 3 individual, 4 technology-based)
 3 RCTs of case management
 30 RCTs of multicomponent interventions (14 carer only, 16 in carer and care receiver
dyads)
 2 RCTs of exercise interventions (1 individual, 1 telephone-based)
 1 RCT of attendance at a memory clinic
 5 RCTs of meditation/mindfulness interventions
 1 RCT of cranial electrotherapy stimulation
Additionally, 4 RCTs compared different intervention types at the same intensity of

intervention:
 3 RCTs compared psychotherapy with a psychoeducational intervention
 1 RCT compared cognitive behavioural therapy with acceptance and commitment therapy
At the re-runs stage 1 RCT of a spiritual care educational program was identified.

evaluating interventions targeted on informal carers of people living with dementia or carers’
assessments that have been published since the literature reviews in CG42. In total, 2,454
articles were returned, of which 22 were selected as potentially relevant and retrieved for full-
text review. Additionally, 2 studies included in CG42 were deemed to be suitable for full-text
review against the current protocol. In total, 8 publications, reporting on 5 studies, were
judged to be at least partially applicable to the review question regarding
interventions/services for supporting the wellbeing of informal carers, and were therefore
included. Of these studies, 2 evaluated psychoeducation and skills training interventions, 2
evaluated multicomponent carer interventions and 1 evaluated a supportive intervention. No
relevant CUAs were identified evaluating carers’ assessments.
Details of the literature search are provided in Appendix D.
338

15.1.3.1.1 Supportive interventions
Charlesworth et al. (2008) compared the befriending of carers by trained lay workers with
usual care, which was all normal available care without the befriending intervention. The
trained lay workers offered emotional support and a degree of informational support. The
economic evaluation was conducted alongside a UK RCT (BECCA; n=236), over a time
horizon of 15 months. Utilities were elicited from carers using the EQ-5D. Carers also
provided resource use data, with unit costs from standard UK sources (price year 2005). For
further details, please see the economic evidence profile in Appendix M.
Trained lay workers befriending carers was associated with 0.017 additional carer QALYs,
and £2,003 of additional costs, compared with usual care over 15 months. The resulting
ICER was £117,039. Cost-effectiveness acceptability analysis, conducted only from a
societal perspective (including the cost of informal carer time), showed the intervention to
have a 29.4% probably of achieving an ICER of £30,000 per QALY or better. Including QALY
gains for the person living with dementia appeared to increase its likelihood of being cost-
effective; however this analysis was only presented with societal costs included, such as
informal carer time forgone.
15.1.3.1.2 Multicomponent interventions
Martikainen et al. (2004) compared a programme of short psychoeducation courses with

counselling support for carers, including physical and recreational training for their relatives
with Alzheimer’s disease, with existing community services. A 5-year Markov model based
on nursing home placement used effectiveness data from 1 US RCT. Resource use and
costs were from Finnish sources. Utilities were informed by a published HUI-2 values. Cost-
effectiveness results focused on the PWD’s outcomes; however, carer QALYs were reported
such that an ICER including them can be estimated. For further details, please see the
economic evidence profile in Appendix M.
The analysis found that family meetings reduce carer QALYs by 0.01 and reduces costs by
€2992, meaning €299,200 is saved for every 1 QALY lost. Combining QALYs from carers
and people living with dementia, this analysis suggests that the intervention dominates usual
care.
Drummond et al. (1991) conducted an economic evaluation of a multicomponent support

programme for carers of people with dementia. The intervention included weekly visits by
carer support nurses, weekly periods of respite care, monthly family meetings, and education
about dementia and caregiving. The programme was compared with conventional nursing in
a Canadian RCT (n=60) with 6 months of follow-up. The CUA, conducted concurrently,
collected resource use data from carer interviews and health records. Unit costs were
obtained from Canadian national sources (price year 1988). Utilities were elicited from RCT
participants using the Caregiver Quality of Life Index by time trade-off. For further details,
The carer support intervention was found to be associated with an ICER of $20,036 per
QALY gained compared with conventional care. No sensitivity analysis or cost-effectiveness
acceptability analysis was presented.
339

15.1.3.1.3 Psychoeducation and skill training
Livingston et al. (2014) reported on the cost-effectiveness of a programme of manual-based

coping strategy sessions for carers of PWD, compared with usual care. The CUA was
conducted alongside a UK RCT (START; n=260), over a time horizon of 24 months. Utilities
were elicited from carers using the EQ-5D questionnaire. Carers also provided resource use
data, with unit costs from standard UK sources (price year 2009). For further details, please
see the economic evidence profile in Appendix M.
The intervention was associated with 0.03 additional QALYs per carer, and an additional
£336 of carer-related costs, resulting in an ICER of £11,200 compared with usual care. In
probabilistic sensitivity analysis, the intervention was found to have a 65% probably of
achieving an ICER of £20,000 per QALY or better. Methodological sensitivity analyses
conducted, varying the time horizon and approach to missing data, did not materially impact
upon results.
Joling et al. (2013) compared a programme of 6 counselling sessions held every 2–3
months, 4 of which could include the carer’s family and friends, with usual care. The CUA
was conducted alongside a 12-month Dutch RCT (n=192). Resource use was collected using
cost diaries, with unit costs from standard Dutch tariffs (price year 2009). A societal
perspective was taken; however, the value of productivity losses can be subtracted from total
costs to estimate a health and social care perspective. Utilities for carers and people living
with dementia were informed by the SF-12 for the estimation of QALYs. For further details,
Using carer outcomes only, Joling et al. suggest that the family meetings intervention
dominates usual care. Including carer-PWD dyad outcomes, Joling et al. found the
intervention to have an ICER of €1,875.
No relevant CUAs were identified evaluating carers’ assessments, therefore there is no

economic evidence for this review question.
15.1.4.1 Psychoeducational interventions
differentiate levels of burden (carer), depressive symptoms (carer), anxiety (carer), stress
(carer), quality of life (carer), self-efficacy (carer), social support (carer), severity of memory,
behavioural and psychological symptoms (person living with dementia), reactions to memory,
behavioural and psychological symptoms of dementia (carer), activities of daily living (person
living with dementia) or the proportion of people entering long stay care (person living with
dementia) between people offered psychoeducational interventions or usual care.
15.1.4.2 Skills training
Low- to moderate-quality evidence from up to 6 RCTs containing 360 people found people
offered skills training had improvements in burden (carer) and quality of life (carer) compared
with people offered usual care.
340

differentiate levels of depressive symptoms (carer), anxiety (carer), stress (carer), self-
efficacy (carer), social support (carer), severity of memory, behavioural and psychological
symptoms (person living with dementia) or reactions to memory behavioural and
psychological symptoms (carer) between people offered skills training or usual care.
15.1.4.3 Psychoeducation and skills training
Low- to moderate-quality evidence from up to 14 RCTs containing 2,031 people found

people offered psychoeducation and skills training had improvements in burden (carer),
depressive symptoms (carer), anxiety (carer), stress (carer) and severity of memory,
behavioural and psychological symptoms (person living with dementia) compared with
people offered usual care.
Very low- to moderate quality evidence from up to 7 RCTs containing 973 people could not
differentiate levels of quality of life (carer), self-efficacy (carer), reactions to memory,
behavioural and psychological symptoms (carer), activities of daily living (person living with
dementia) or the proportion of people entering long stay care (person living with dementia)
between people offered psychoeducation and skills training or usual care
15.1.4.4 Supportive interventions
differentiate burden (carer), depressive symptoms (carer), anxiety (carer), stress (carer),
quality of life (carer), social support (carer) or severity of memory, behavioural and
psychological symptoms (person living with dementia) between people offered supportive
interventions or usual care.
15.1.4.5 Respite care
Moderate-quality evidence from 1 RCT of 55 people could not differentiate burden (carer),
depressive symptoms (carer) or anxiety (carer) between people offered respite care or usual
care.
Moderate-quality evidence from 1 RCT of 38 people found people offered polarity therapy
had less severe depressive symptoms (carer), and stress (carer) compared with people
offered respite care.
Low- to high-quality evidence from up to 14 RCTs containing 1,034 people found people
offered psychotherapy had improvements in burden (carer), depressive symptoms (carer),
anxiety (carer), quality of life (carer), self-efficacy (carer) and reactions to memory,
behavioural and psychological symptoms (carer) compared with people offered usual care.
Low- to moderate-quality evidence of up to 3 RCTs containing 298 people could not

differentiate stress (carer) or severity of memory, behavioural and psychological symptoms
(person living with dementia) between people offered psychotherapy or usual care.
341

15.1.4.7 Case management
High-quality evidence from 1 RCT of 61 people found people offered case management had
improvements in anxiety (carers) and severity of memory, behavioural and psychological
symptoms (person living with dementia) compared with people offered usual care.
Low- to moderate-quality evidence from up to 3 RCTs of 229 people could not differentiate
levels of burden (carers), depressive symptoms (carers), quality of life (carers), self-efficacy
(carers) or proportions of people entering long stay care (person living with dementia)
between people offered case management or usual care.
15.1.4.8 Multicomponent interventions
Low- to moderate-quality evidence from up to 20 RCTs containing 5,220 people found

people offered multicomponent interventions had improvements in burden (carer),
depressive symptoms (carer), quality of life (carer), social support (carer), severity of
memory, behavioural and psychological symptoms (person living with dementia) and
reactions to memory, behavioural and psychological symptoms (carer) compared with people
offered usual care.
Low- to moderate-quality evidence of up to 7 RCTs containing 992 people could not

differentiate levels of anxiety (carer), self-efficacy (carer), activities of daily living (person
living with dementia) or the proportion of people entering long stay care (person living with
dementia) between people offered multicomponent interventions or usual care.
15.1.4.9 Exercise
differentiate burden (carer), depressive symptoms (carer) or stress (carer) between people
offered exercise or usual care.
15.1.4.10 Memory clinic
Moderate-quality evidence from 1 RCT of 30 people could not differentiate burden (carer) or
reactions to memory, behavioural and psychological symptoms (carer) between people
offered access to a memory clinic or usual care.
15.1.4.11 Mindfulness/meditation
High-quality evidence from up to 5 RCTs containing 192 people found people offered
meditation/mindfulness interventions had improvements in depressive symptoms (carer) but
worsening in the severity of memory, behavioural and psychological symptoms (person living
with dementia) compared with people offered usual care.
Low-to moderate-quality evidence from up to 3 RCTs containing 133 people could not
differentiate burden (carer), anxiety (carer), stress (carer), self-efficacy (carer), social support
(carer) or reactions to memory, behavioural and psychological symptoms between people
offered meditation/mindfulness interventions or usual care.
342

15.1.4.12 Cranial electrotherapy stimulation
Very low-quality evidence from 1 RCT of 38 people could not differentiate burden (carer) or
depressive symptoms (carer) between people offered cranial electrotherapy stimulation or
usual care.
15.1.4.13 Psychotherapy versus psychoeducational interventions
differentiate burden (carer), depressive symptoms (carer), anxiety (carer), self-efficacy
(carer), severity of memory, behavioural and psychological symptoms (person living with
dementia) or reactions to memory, behavioural and psychological symptoms (carer) between
people offered cognitive behavioural therapy or psychoeducational interventions
15.1.4.14 Cognitive behavioural therapy versus acceptance and commitment therapy
Low-quality evidence from 1 RCT of 87 people could not differentiate depressive symptoms
(carer) or anxiety (carer) between people offered cognitive behavioural therapy or
acceptance and commitment therapy.
15.1.4.15 Spiritual care
Low-quality evidence from 1 quasi experimental RCT of 54 people found carer self-efficacy
improved for carers of people with Alzheimer’s disease who took part in a spiritual care
education program compared with those who did not receive the intervention.
15.1.4.16 Meta-regression (carer depression)
Moderate-quality evidence from a meta-regression 73 RCTs found that skills training,

mindfulness and psychotherapy significantly reduced carer depressive symptoms compared
to usual care, with the largest effects found in group interventions offered to carers alone
(rather than in a dyad with the person living with dementia also present).
15.1.4.17.1 Psychoeducational and skill training
One directly applicable cost-utility analysis with minor limitations compared a manual-based
programme of coping strategy sessions for carers with usual care. The intervention was
associated with 0.03 additional QALYs per carer, and an additional £336 of carer-related
costs. The ICER was £11,200. The intervention had an ICER of no more than £20,000 per
QALY in 65% of analysis replications (designed to capture uncertainty in the sample), and no
more than £30,000 per QALY in 75%.
One partially applicable cost–utility analyses with very serious limitations compared
programmes of family meetings for carers of PWD with usual care. One found that family
meetings provide 0.02 additional carer QALYs at a lower cost (-€845) compared with usual
care over 12 months.
343

15.1.4.17.2 Supportive interventions
One directly applicable cost-utility analysis with potentially serious limitations compared lay
workers befriending carers with usual care. Befriending carers generated 0.017 additional
carer QALYs, and incurred an additional £2,003, over 15 months. The ICER was £117,039.
15.1.4.17.3 Multicomponent interventions
One partially applicable cost-utility analyses with very serious limitations compared
programmes of family meetings for carers of PWD with usual care. It found that family
meetings cause a loss of 0.01 carer QALYs, with a cost saving of €299,200 per lost QALY.
Both interventions were found to be dominant when QALYs of the PWD were included.
One partially applicable cost-utility analysis with very serious limitations compared a
multicomponent intervention of support for carers with usual care. The intervention was
associated with a gain of 0.11 carer QALYs, estimated using the Caregiver Quality of Life
Index, and an additional cost of Can$2204 (price year 1988). The ICER was Can$20,036.

Relative value of different The committee acknowledged the variety of measures used to assess
outcomes each outcome and agreed the use of a standardised mean difference
was the appropriate method of interpretation. A value of 0.2 was used
as the cut-off for a meaningful effect size throughout the discussions.
The committee recognised there were some limitations in linking the
outcomes with clinical experience. Some interventions may be more
fitted to specific situations: for example, the interpretation of depression
as an outcome may have to be qualified within the context of the
recruited population; in mixed carer populations, both depressed and
non-depressed people may have been included in the sample
(combined with the general underdiagnosing of depression in carers of
people living with dementia). In other instances, disaggregation of data
meant it was possible to interpret outcomes at the subgroup level,
whereby the use of skills training demonstrated significant benefits for
carer burden at the group and individual level, but did not differentiate
for the use of technology based interventions.
Of the outcomes available, the committee agreed that carer burden and
carer quality of life would be the most directly applicable for decision
making, as they are both overall measures of the impact that caring for
someone living with dementia has on the life of the carer. They also
agreed that measures of behavioural and psychological symptoms
would be valuable, as these would identify if any changes interventions
introduced in the way carers interact with the person they are caring for
would impact on the person living with dementia themselves (with
positively or negatively), as well as on the carer.
Trade-off between Psychoeducation and skills training
benefits and harms The committee noted that skills training was closely linked to
psychoeducation and agreed that the stratification of evidence into
individual, group based or technology based interventions was the most
appropriate means of presentation. However, the committee
acknowledged that individual, group based and technological
interventions involve different approaches and interactions. The
committee noted the choice of intervention technique may reflect the
344

carer’s individual or personal preference and specific approaches may

be more considerate of different cultural needs.
The committee agreed that whilst there was no evidence of benefits
from psychoeducation alone, and only limited evidence of benefits from
skills training alone, interventions combining psychoeducation and skills
training showed meaningful benefits across a range of domains, and
included benefits for both the carer and the person living with dementia.
Respite care
The committee acknowledged the limited evidence on respite care
meant it was difficult to come to a conclusion about its overall benefit. It
noted that respite can be very difficult to measure because it does not
benefit all carers in exactly the same way. Some people may find
respite a relief from stress, whereas, for others, stress levels may
remain high. The committee noted there were caveats on the
interpretation of the results and observed that if carers do not find
respite care helpful they might be unlikely to continue in a trial. In real
world examples, respite care packages will probably evolve based on
an individual’s needs or personal preference. The overall consensus
from the committee was that although it was hard to measure and
capture real world outcomes, the lack of evidence does not mean we
should conclude that respite care does not work. Further, it noted that
the trials only included a very specific type of respite care (small respite
periods of short duration, as opposed to an extended single period of
respite). Finally, it noted that many of the other interventions examined
will necessarily have involved a level of respite care being provided in
order for the carer to be able to attend the intervention.
Psychotherapy and counselling
The committee recognised that psychotherapy and counselling were
distinct interventions and therefore should be classified discretely. The
committee recognised there is a distinction between psychotherapy for
specific mental health issues and post-diagnostic counselling aimed at
supporting adjustment to the diagnosis, and acknowledged that the
evidence that had been presented was mostly related to cognitive
behavioural therapy (CBT). The committee had concerns about the
applicability of the trials to the current UK context; the trials often
included people without problems at baseline, whilst in the UK, people
often only receive these interventions when demonstrating and showing
signs of a specific need which prompts a referral (a situation where
larger gains may well be expected, as the carer has more potential to
benefit). The committee queried whether these interventions would be
most effective for selective groups of carers or everyone. In particular,
they felt that carers with a diagnosis of depression would receive
standard psychological treatments, and therefore the relevant question
for this guideline was whether a specific group of people with sub-
threshold depression would benefit from earlier psychological
interventions, and the committee felt it was not possible to answer this
from the current evidence due to the heterogeneous nature of the
populations involved in the trials.
The committee discussed the applicability of technology based
interventions and raised the need to be mindful of selection of
participants, as people with skills in using web based technologies may
experience interventions differently to others. On the other hand, the
committee noted anecdotal evidence that, in general, web based CBT
may be as effective as face to face delivery – this may especially be the
case for younger people with dementia and their carers who are seen to
be increasingly using web based resources.
345

However, the committee noted there was evidence that group CBT may
provide benefits, particularly in levels of depression, and therefore
agreed it was appropriate to make a research recommendation, looking
at the effectiveness of these interventions in a more targeted group of
people at risk of depression.
Case management
The committee agreed there was limited evidence available on case
management interventions. It was agreed that the effectiveness of case
management was best assessed as part of the questions on non-
pharmacological treatments for people living with dementia, as there is
a larger volume of evidence available in this context.
Multicomponent interventions
Multi-component interventions were described as encompassing at least
2 or more elements to the intervention, both aimed at the carer. The
committee noted that an intervention containing only a course of
psychoeducation and skills training would not normally be classified as
multicomponent, therefore this was considered in its own category as
part of the review.
The committee agreed that many multicomponent interventions
demonstrated moderate significant effects, yet when each of the
individual components was considered, they were only able to
demonstrate a small or negligible effect. One possible explanation for
this is that some of the multicomponent intervention trials may be better
designed than some of the individual component studies, where a
structured package has been developed (and tailored to individual
needs), rather than simply offering a single individual intervention. It
may also be the case either that the effects of the interventions are
additive, or that the higher levels of contact time usually seen with
multicomponent interventions lead to better results.
Psychoeducation and skills training interventions
The committee agreed that there was clear evidence of benefits from
interventions containing both psychoeducation and skills training, and
the findings were broadly similar to those for more general
multicomponent interventions. The committee agreed that it was
therefore appropriate to recommend psychoeducation and skills training
interventions, and the content of these should include at least those
features which were commonly present in the clinical data showing the
effectiveness of these interventions. It also agreed that recommending a
specific psychoeducation and skills training intervention was more
appropriate than making a more generic recommendation around
multicomponent interventions, both because of the economic evidence
described in the “trade-off between net health benefits and resource
use” section below, and because it was a clearly defined intervention
type that should be practical to implement in practice.
Mindfulness and meditation
The committee agreed with the evidence seen on mindfulness although
noted that the high quality status would need to be qualified alongside
the small to moderate effect that was observed. It did not believe the
evidence was sufficiently robust to suggest recommending these
interventions over the psychoeducation and skills training interventions
discussed above.
Supportive interventions; exercise; memory clinics; cranial
electrical stimulation
The committee noted there were non-significant effects across a range
of interventions including exercise, cranial electrical stimulation, memory
346

clinics and supportive interventions, and therefore did not feel it

appropriate to recommend any of these approaches.
Spiritual care
The committee agreed that, although a significant result was found in
the one study looking at spiritual care, the fact this came from only one
small study meant the evidence was insufficient for any positive
recommendations to be made.
Trade-off between net Supportive interventions
health benefits and The committee discussed the health economic evidence for supportive
resource use interventions, which was based on one trial of a carer befriending
intervention. The committee discussed the importance of capturing the
resource impact of interventions preventing carer breakdown and
delaying or avoiding the need for residential care for the person living
with dementia. The committee recognised that the mean ICER was
significantly higher than would usually be considered an effective use of
NHS resources, and agreed that the expected additional resource use
required for the intervention would be likely to cause a net loss of health
within the wider health care system.
The committee discussed the importance of critiquing health economic
evaluations of very specific interventions, and agreed that is it important
to determine the degree to which any single intervention is
representative of that wider category of interventions.
Multicomponent interventions
The committee discussed the two CUAs evaluating multicomponent
interventions. The committee discussed the extent to which the studies
could be used to inform present decision making, for example given the
publication date (1991) of the Drummond et al. study, and ultimately felt
that they could not rely on the economic evidence to support a
recommendation. The committee understood that some estimation of
ICERs by the NICE technical team had been necessary in order to
present results that satisfied the reference case, and that this added to
the overall uncertainty.
Psychoeducation and skill training
The committee discussed the Livingston et al. evaluation of individual,
manual-based carer support. The results were robust to probabilistic
sensitivity analysis, with the intervention having a 65% probability of
being cost effective at a threshold of £20,000/QALY and a 75%
probability of being cost effective at £30,000/QALY. The committee
discussed whether this was sufficient to determine whether an
intervention should be recommended. The committee agreed that the
study provided evidence to recommend a psychoeducation and skills
training intervention for carers, but not necessarily the specific
intervention evaluated (START), as it was acknowledged that a
substantial number of possible alternative interventions were not
captured in the clinical or economic evidence. The committee agreed
that the topics covered by START are a good representation of the
topics that should be covered in this type of intervention.
Quality of evidence The committee noted that the evidence in relation to respite care
consisted of fairly poor examples of respite interventions; the trials may
not have been long enough to actually allow carers to use the time as
they desire. Furthermore, polarity therapy was agreed to be not directly
relevant as an intervention for the UK health and social care setting.
With the exception of respite care, no other real overarching concerns
with the overall quality of the evidence were raised.
The committee recognised that the vast array of interventions and
outcomes meant it was very hard to present outcomes for interventions
347

in a truly meaningful way, and recommendations could only really be

made at the class level.
The committee discussed the potential impact of the placebo effect on
outcome effect sizes, as the evidence consisted of studies where the
intervention recipient and provider were not blinded. However, they
noted that in the case of the types of supportive interventions
considered, the benefits gained simply from taking part in an activity
regardless of the content of it, could reasonably be considered a part of
the intervention itself, rather than a confounding factor.
The committee noted that the fact the results of the meta-regression
analysis were consistent with the pairwise analyses was a useful
validation of the findings, and gave further justification to the
recommendations made around skills training and group interventions,
and the research recommendation on group CBT.
Other considerations The committee agreed that carer interventions were generally
considered to be clinically beneficial. They were confident that the
evidence they had seen on multicomponent interventions was sufficient
to enable an ‘offer’ recommendation to be made. This could be
achieved by taking into consideration specific elements drawn from
individual components which had been found to be effective on their
own, together with the common elements from the psychoeducation and
skills training interventions evaluated, and the elements found in
psychoeducation and skills training interventions shown to be cost-
effective. The committee agreed it was not appropriate to provide a
prescriptive list of requirements, but felt comfortable providing a
minimum set of domains which should be covered in any
psychoeducation and skills training intervention package.
The committee acknowledged that specific attention should be paid to
carers’ individual and personal preferences. For example different
people will prefer face to face, web based, or group based support. A
personalised approach may identify issues which had not been
observed in the evaluated interventions. The committee therefore
agreed the recommendation should consider personalised strategies.
However, they also noted the evidence suggested the largest benefits
would be found in group interventions, and therefore agreed it was
appropriate to make an awareness raising recommendation on this
point.
The committee discussed wording around the use of ‘pleasant activities’
and agreed it was best applied jointly as a collaborative statement
focusing on both the carer and the person living with dementia. The
committee also discussed the relevance of psychotherapy. They agreed
it was important for carers to have access to a range of interventions to
manage any serious issues, rather than automatically being referred for
psychotherapy. The use of Improving Access to Psychological
Therapies (IAPT) was highlighted as it is often accessed via self-referral
if people have knowledge of it.
It was noted there is a legal necessity to place carers at the centre of
the decision making process. The committee were mindful that any
intervention offered should be easily accessible although noted that
intervention choice may be influenced by mode and/or setting.
The committee also agreed it was appropriate to cross-refer to the NICE
guideline depression in this section, due to the established higher risk of
depression in carers of people living with dementia.
348


Relative value of different The committee recognised the Care Act 2014 sets out best practice but
outcomes noted that different local authorities may interpret the Act differently,
with statutory guidance used to help support implementation. Guideline
recommendations can highlight or support specific parts of the Act.
Trade-off between The committee acknowledged that carers’ assessments are a legal
benefits and harms requirement and noted that assessing carers and identifying appropriate
interventions were two separate issues. For the former, reading the
Care Act may help in identifying appropriate wording. However, the
committee were reminded that the purpose of any recommendation
should not simply be to explain legislative requirements, but should be
used to enable a practical application of carers’ assessments.
The committee noted there are very practical implications which inhibit
any standardisation in the application of carers’ assessments across the
board. In some areas there has been the expectation that numbers of
carer’s assessments would increase, given the requirements of the
Care Act, but in some cases carers are not coming forward or making
use of their entitlement to receive an assessment. This may be a
consequence of accessibility or awareness issues. The committee
noted that some carers may be unaware of their rights and of the
services that are available.
The committee highlighted issues with continuity of care that may arise
from ‘artificial’ divisions between under and over 65s in care
organisations. Changing demographics and funding arrangements may
have an impact upon considerations regarding repeat reviews.
Trade-off between net No economic evidence was identified for this review question and
health benefits and economic modelling was not prioritised. Since the recommendations
resource use made were only about ensuring people were aware of their legal rights,
the committee agreed it was not necessary to consider the resource
implications of those rights being accessed.
Quality of evidence The committee noted the lack of evidence from research in this area but
received a contextual presentation placing carers’ assessments within
the framework of the Care Act 2014, delivered by one of the Local
Authority Commissioners on the committee. This covered both the legal
rights people have and the extent to which the legislation has been fully
implemented in practice.
Other considerations The committee agreed that recommendations within this area needed to
be considered within an equality impact framework. It was noted that
young carers in particular often do not access the support to which they
are entitled, and that they often feel that support is not appropriate for
their age group. Culturally appropriate approaches may be needed and
the committee raised awareness of issues facing ethnic groups; for
example people from black African and Caribbean family origin often do
not access the levels of support to which they are entitled. The
committee recognised there is variation/inequality around the country
regarding who does the assessment which needs to be highlighted.
106. Offer carers of people living with dementia a psychoeducation and skills
training intervention that includes:
349

 education about dementia, its symptoms and the changes to expect as

the condition progresses
 developing personalised strategies and building carer skills
 training to help them provide care, including how to understand and
respond to changes in behaviour
 training to help them adapt their communication styles to improve
interactions with the person living with dementia
 advice on how to look after their own physical and mental health, and
their emotional and spiritual wellbeing
 advice on planning enjoyable and meaningful activities to do with the
person they care for
 information about relevant services (including support services and
psychological therapies for carers) and how to access them
 advice on planning for the future.
107. Ensure that the support offered to carers is:

 tailored to their needs and preferences and to what they want it to
achieve (for example, providing information on carer’s employment
rights for carers who work or want to work)
 designed to help them support people living with dementia
 available at a location they can get to easily
 provided in a format suitable for them (for example individual or group
sessions, or online training and support)
 available from diagnosis and as needed after this.
108. Be aware that carer interventions are likely to be most effective when provided
as group sessions.
109. Advise carers about their right to the following and how to get them:
 a formal assessment of their own needs (known as a 'Carer's
Assessment'), including their physical and mental health
 an assessment of their need for short breaks and other respite care .
110. Be aware that carers of people living with dementia are at an increased risk of
depression. For guidance on identifying and managing depression, see the NICE
guideline on depression in adults.
17. What is the effectiveness and cost-effectiveness of group-based cognitive

behavioural therapy for carers of people living with dementia who are at high risk
of developing depression?
appendix L.
350

Staff training
Staff training
16 Staff training
The need for improvement in dementia care has increasingly become a focus of research,
development and policy initiatives. The demand for high quality dementia care through a
skilled workforce will increase as both dementia prevalence and diagnosis rates rise.
The focus of the Prime Minister’s Challenge on Dementia 2020 (DH, 2015) and the 2015–16
Mandate from Government to Health Education England (HEE) is the requirement for an
informed and effective workforce for people living with dementia. This means all health and
social care staff involved in the care of people who may have dementia should have the
necessary dementia core skills, education and training to provide the best quality care in the
roles and settings where they work.
The Prime Minister’s Challenge on Dementia 2020 Implementation Plan (DH, 2016) presents
the progress that has been made in ensuring that the dementia workforce is fully equipped
through the development of such initiatives as the Core Skills, Education and Training
Framework led by Skills for Health and HEE. One of the questions often raised, however, is
what difference does a well-developed workforce make to the experiences of people living
with dementia and their carers?
The Implementation Plan, supported by other reports such as the “Fix Dementia Care”
(Alzheimer’s Society, 2016), demonstrate a direct correlation between poor skills, education
and training to poor outcomes, poor effectiveness and poor experience.
A skilled educated and trained workforce can improve the experience of the person living
with dementia and their carers throughout the dementia pathway, from diagnosis, care and
treatment to living well and the end of life. The experience can potentially be improved at:
 First contact, as primary care has an increased focus and understanding of dementia
leading to a timely diagnosis and increased diagnosis rates.
 Assessment and treatment, as memory assessment services improve their response
times, discharge plans and ongoing care and support plans and coordination.
 Acute care, as staff understand the alternative options to hospital admission and, if
admission is necessary, ensure people with dementia and their carers receive
personalised care helping to reduce length of stay, hospital incidents, improved
satisfaction, co-ordinated discharge and reduced readmissions.
 Post-diagnostic care, as staff working in communities support people living with dementia
to remain independent and active citizens for as long as possible delaying the possible
need for residential care and improving the quality of life both of the person living with
dementia and their carers/families.
 Care homes, as care home staff deliver personalised co-ordinated and dignified care and
support reducing the possible need for hospital admission, reducing incidents and
improving satisfaction.
 End of Life, as palliative care and hospice teams include the care and support of people
living with dementia as part of their commissioned services and service offer increasing
choice and control for people with dementia at the end of life.
351

Staff training
Staff training
16.1 Staff training

Review question
What effect does training for staff working with people living with dementia have upon the
experiences of people living with dementia in their care?
16.1.1 Introduction
This question considered both quantitative and qualitative evidence on effective models of
staff training for improving the care and experiences of people living with dementia. The
quantitative part of this review included a collaboration between the NICE Guideline Updates
Team and the Cochrane Dementia and Cognitive Impairment Group.

Phenomena of interest Aspects of training programmes for staff caring for people living with
dementia, which may include training on:
 The natural history of dementia, different subtypes, prognosis etc.
 Communication skills
 Principles of person-centred care
 Roles of different health and social care professionals, and how care
should be co-ordinated between these different services
 Adult protection policies and procedures
 Awareness of abuse and neglect
 Principles of palliative care
 Appropriate prescribing (antipsychotics)
 Avoiding unnecessary hospital admissions
 Managing behaviour and non-cognitive symptoms
 Enablement and reablement
 Nutrition and swallowing difficulties
 Legislative rights

Interventions  Training interventions for formal, paid staff working with people living
with dementia
Comparator  No specific training intervention
Outcomes  Appropriate use of procedures/medicines
Qualitative studies needed to report the views of either people living with dementia or their
carers, and match the criteria given in Table 91. The aims of this review were to establish the
352

Staff training
Staff training
most effective ways of managing the transition between different settings for people living
with dementia, and their carers. The review focused on identifying studies that fulfilled the
conditions specified in Table 40. For full details of the review protocol, see Appendix C.
Randomised controlled trials were included if they explored the effectiveness of staff training
interventions for improving the experiences of people living with dementia and meet the
criteria given in Table 92. Papers were excluded if they:
 did not include the views of people living with dementia or their carers
 were not in the English language
 were abstracts, conference proceedings or other unpublished studies.
For the purposes of this question, a care provider is defined as an organisation that delivers
health and/or social care. This includes all providers registered with the Care Quality
Commission as well as unregistered providers such as Community Based Support Services
which may be delivering health and social care through a commissioning agreement.
and abstracts, and the full texts of 11 references that were potentially relevant to the review
question were requested. All of these studies were excluded on full text review, with reasons
for exclusion presented in Appendix F.
The RCT data included in this review primarily came from an ongoing Cochrane review on
‘Educational interventions for improving clinical competencies of medical practitioners to
detect, diagnose, and manage people with cognitive impairment and dementia’. Whilst this
review was not published at the time this question was considered in the guideline, the list of
included studies was provided by the Cochrane Dementia and Cognitive Impairment Group,
and these studies were screened at full text level to identify relevant studies. In addition, the
studies included in 4 other recent, high-quality systematic reviews were also screened to
identify any additional studies not included in the Cochrane review, particularly studies
primarily targeted at social care rather than healthcare staff. The systematic reviews used as
sources for RCT data are summarised in Table 93. For the full evidence tables and full
GRADE profiles of included RCTs, please see Appendix E and Appendix G. References for
the included studies are given in appendix I.
353

Staff training
Staff training
1 Table 93: Systematic reviews used to identify primary studies

Study details Study population Interventions
Belisario (2013) – Protocol for an ongoing Generalist clinicians working in either primary care or Educational interventions whose primary
review secondary care settings, and specialist consultants in any objective is to improve clinicians' skills in
Educational interventions for improving the related medical field and who are working in any clinical evaluating, diagnosing, managing (or a
skills of medical practitioners to detect, setting. combination of these) people with cognitive
diagnose, and manage people with cognitive impairment or dementia.
impairment and dementia
Bird (2016) Staff working in residential dementia care Interventions in long-term facilities helping
Do interventions with staff in long-term staff develop their capacity to provide better
residential facilities improve quality of care or care and/or QOL for residents living with
quality for life people with dementia? A dementia
systematic review of the evidence
Machiels (2017) Nursing staff working with people living with dementia Interventions that aim to improve
Interventions to improve communication communication (verbal and/ or non-verbal)
between people with dementia and nursing between nursing staff and people living with
staff during daily nursing care: A systematic dementia
review
Scerri (2016) Staff working in a ‘general hospital’, defined as ‘a hospital not Any dementia training programmes directed to
Dementia training programmes for staff specialising in the treatment of a particular illness or of staff working in general hospital settings
working in general hospital settings - a patients of a particular sex or age group’
systematic review of the literature
Spector (2013) Paid care staff in nursing or residential care homes Training interventions designed to help staff
A systematic review of staff training manage behavioural and psychological
interventions to reduce the behavioural and symptoms of dementia
psychological symptoms of dementia
2 A total of 88 unique studies were identified from these reviews, and these studies were themselves screened at full text level. Twenty-five RCTs
3 met the criteria for inclusion (reported in 26 papers), with the remaining 62 studies excluded, with reasons for exclusion given in Appendix F. The
4 included studies are summarised in Table 94.
354

Staff training
Staff training
1 Table 94: Summary of included studies – RCT evidence
Beer (2011) 351 people living with dementia. Intervention: Training to meet the perceived Quality of life, behavioural and psychological
Living in residential care. need of GPs and residential care staff. Main symptoms of dementia, pain, use of physical
topics of the educational programs were: restraint.
communication, personal care and activities,
positive values, behaviours of concern, pain
management, dementia, depression and
delirium, and effective working between GPs
and residential care facility staff.
Comparison: No training.
Burgio (2002) 88 people living in residential Intervention: Behaviour management skills Resident agitation observations.
care who displayed behavioural training of residential care staff.
disturbances. Their mean MMSE Comparison: Conventional staff
was approximately 7. 106 management.
nursing assistants.
Chang (2005) 20 people living with dementia in Intervention: Feeding skills training program. Food intake.
residential care identified as Comparison: No training.
having eating problems.
Chenoweth (2009) 289 people living with dementia Intervention 1: Dementia care mapping. Staff Agitation, measured with the Cohen-
in residential care. carers received training. Mansfield agitation inventory. Psychiatric
Intervention 2: Person-centred care. Staff symptoms, neuropsychological status, quality
carers received training. of life, falls.
Comparison: Usual care.
Chenoweth (2014) 601 people living with dementia Intervention 1: Person-centred care. Nurses Quality of life (DEMQOL), agitation,
in residential care. received training. emotional responses, depression scores.
Intervention 2: Person-centred dementia
environment.
Clare (2013) 65 people living with dementia in Intervention: Care staff training to observe Quality of life, well-being, behaviour and
residential care. 65 care staff. and identify signs of awareness in cognition.
participants with advanced dementia
Davison (2007) 113 people living with dementia Intervention 1: Care staff received training to Frequency of challenging behaviours.
in residential care. 90 care staff. manage dementia-related challenging
355

Staff training
Staff training

behaviours. In addition, there was a peer-
support group.
Intervention 2: Care staff received training to
manage dementia-related challenging
behaviours.
Comparison: Wait-list control.
Deudon (2009) 306 people living with dementia Intervention: Care staff education to manage Cohen-Mansfield Agitation Inventory (CMAI)
in residential care. the behavioural and psychological symptoms and an Observation Scale (OS) score.
of dementia.
Comparator: No training.
Döpp (2015) 71 people living with dementia. Intervention: A training package for Daily functioning, carers’ sense of
71 informal carers. occupational therapists including the usual competence, quality of life, and self-
postgraduate course, outreach visits, perceived performance of daily activities of
regional meetings, and access to a reporting both people living with dementia and carers.
system.
Physicians and managers received
newsletters, had access to a website.
Comparison: A postgraduate course for
occupational therapists only.
Finnema (2005) 146 people living with dementia Intervention: Care staff training to provide Behaviour and mood related to adaptation to
in psychogeriatric wards in integrated emotion-orientated care. the illness and the institutionalisation.
residential care. 99 nursing Comparison: No training.
assistants.
Fossey (2006) 246 people living with dementia Intervention: Training and support delivered Proportion of residents in each home who
in residential care. to residential care staff, focusing on were prescribed neuroleptics and mean
alternatives to drugs for the management of levels of agitated and disruptive behaviour
agitated behaviour in dementia. (Cohen-Mansfield agitation inventory).
Huizing (2006) 167 people living with dementia. Intervention: Training care staff to reduce Restraint use.
Living in psychogeriatric their use of physical restraints.
residential care wards. Comparison: No training.
Leone (2013) 230 people living with dementia Intervention: Care staff education to manage Katz ADL Scale, neuropsychiatric symptoms,
with apathy. apathy. The Apathy Inventory-Clinician to measure
Comparison: No training. apathy, behavioural disturbance.
356

Staff training
Staff training

Magai (2002) 91 people living with dementia. Intervention: Training staff carers in Symptoms (depression, agitation,
Living in residential care. sensitivity to nonverbal communication. behavioural symptoms), facial expressions of
Comparison 1: “Training” on cognitive and emotion.
behavioural aspects.
Comparison 2: No training group.
McCallion (1999) 105 people living with dementia. Intervention: Nursing assistants underwent a Penn State Mental Health Questionnaire,
Living in residential care. Nursing Assistant Communication Skills turnover rates, signs and symptoms of
Program. depression and aggressive behaviours,
Comparison: No training. disorientation, irritability, agitation,
psychotropic medication and restraint use.
Pellfolk (2010) 353 people living with dementia Intervention: Nursing staff underwent a Use of physical restraints, number of falls,
in group dwelling units for restraint minimisation education program. use of psychoactive medication.
people living with dementia. Comparison: No training.
Robison (2007) 388 family members of people Intervention: Nursing staff received training Ease of communicating with staff, staff
living with dementia in that was designed to improve communication behaviours, and care involvement of the
residential care. and cooperation between staff and families family.
of people living with dementia.
Sloane (2004) 69 people living with dementia, Intervention 1: Care staff were educated to Agitation, and aggression, discomfort, bath
who had agitation during deliver person-centred showering. completeness, skin condition, skin microbial
bathing, living in residential care. Intervention 2: Care staff were educated to flora.
deliver towel bathing.
Testad (2005) 142 people living with dementia. Intervention: Care staff training on restraint Number of restraints used and agitation.
Living in residential care. use and alternatives.
Testad (2010) 90 people living with dementia. Intervention: Care staff were trained using Use of restraints, agitation and use of
Living in residential care. the Relation-Related Care course to reduce antipsychotics.
agitation and restraint use.
van de Ven (2013) 192 people living with dementia. Intervention: Staff were trained to use Agitation, neuropsychiatric symptoms, and
Living in residential care. dementia care mapping. quality of life.
357

Staff training
Staff training

van Weert (2005) 121 people living with dementia. Intervention: Care staff were trained in verbal Smiling, gaze, negative verbal behaviours
Living in residential care. and non-verbal communication, and in and verbal expressed autonomy.
multisensory stimulation.
Verkaik (2011) 97 people living with dementia. Intervention: Nurses were trained to apply a Depression, observed mood.
Living in residential care. guideline to their residents who had
depression. The aim of the guideline was to
individualise pleasant activities and decrease
unpleasant events
Visser (2008) 76 people living with dementia. Intervention 1: Care staff attended a Agitation, quality of life.
Living in residential care. behaviourally-based education programme.
Intervention 2: Care staff attended a
behaviourally-based education programme.
They also participated in a peer support
group.
Wenborn (2013) 210 people living with dementia. Intervention: Adjusting care homes to Quality of life, anxiety, depression and
Living in residential care. improve residents’ engagement in activities, challenging behaviour.
care staff training to promote residents’
engagement in activities.
358

Staff training
Staff training
total of 1,414 citations was returned. Following review of titles and abstracts, no full text
studies were retrieved for detailed consideration. Therefore, no relevant cost–utility analyses
were identified for this question
16.1.4.1 Residential care staff training
16.1.4.1.1 Flexible education
quality of life, pain, behavioural and psychological symptoms of dementia or the use of
physical restraint between people living with dementia in residential care where staff were
offered a flexible training package, and people living with dementia in residential care where
no specific additional training was offered.
16.1.4.1.2 Activity provision
Moderate-quality evidence from 1 RCT containing 159 people could not differentiate quality
of life, cognition, challenging behaviours, depression, anxiety or the total number of
medicines prescribed between people living with dementia in residential care where staff
were offered training in activity provision, and people living with dementia in residential care
where no specific additional training was offered.
16.1.4.1.3 Multisensory stimulation
Moderate-quality evidence from 1 RCT containing 121 people found improvements in verbal
communication and increases in the duration of morning care for people living with dementia
in residential care where staff were offered training in multisensory stimulation, compared
with people living with dementia in residential care where no specific additional training was
offered.
16.1.4.1.4 Behaviour management
Very low-quality evidence form 1 RCT containing 79 people could not differentiate between
levels of agitation in people living with dementia in residential care where staff were offered
behavioural management training, and people living with dementia in residential care where
no specific additional training was offered.
16.1.4.1.5 Feeding skills
Very low-quality evidence from 1 RCT containing 20 people found higher levels of feeding
difficulties in people living with dementia in residential care where staff were offered feeding
skills training, compared with people living with dementia in residential care where no specific
additional training was offered.
359

Staff training
Staff training
16.1.4.1.6 Dementia care mapping
Moderate- to high-quality evidence from 1 RCT containing 159 people found reductions in
agitation and the number of falls in people living with dementia in residential care where staff
were offered training in dementia care mapping, compared with people living with dementia
in residential care where no specific additional training was offered, but could not differentiate
quality of life or behavioural and psychological symptoms of dementia.
16.1.4.1.7 Person-centred care
Moderate- to high-quality evidence from up to 2 RCTs containing 269 people found less
agitation and improvements in quality of life, behavioural and psychological symptoms of
dementia and the number of falls in people living with dementia in residential care where staff
were offered training in person-centred care, compared with people living with dementia in
residential care where no specific additional training was offered.
16.1.4.1.8 Awareness and communication
Low- to moderate-quality evidence from 1 RCT containing 65 people found improvements in

quality of life in people living with advanced dementia in residential care who had little or no
verbal communication where staff were offered training in identifying signs of awareness in
people with advanced dementia and improving their communication skills, compared with
people living with dementia in residential care where no specific additional training was
offered, but could not differentiate wellbeing, cognition or behavioural symptoms.
16.1.4.1.9 Challenging behaviours
Very low- to moderate-quality evidence from up to 2 RCTs containing 350 people found less
agitation in people living with dementia in residential care where staff were offered training in
managing challenging behaviours, compared with people living with dementia in residential
care where no specific additional training was offered, but could not differentiate aggressive
behaviours, quality of life, numbers of hospitalisations or numbers of psychotropic medicines
prescribed.
Very low- to low-quality evidence from 1 RCT containing up to 67 people could not
differentiate agitation, aggressive behaviours or quality of life between people living with
dementia in residential care where staff were offered training in managing challenging
behaviours and additional peer support, and people living with dementia in residential care
16.1.4.1.10 Communication skills
depression and verbally aggressive behaviours and reduced use of mechanical restraints in
people living with dementia in residential care where staff were offered training in
communication skills, compared with people living with dementia in residential care where no
specific additional training was offered, but increased levels of disorientation. The evidence
could not differentiate physically aggressive behaviours, use of chemical restraints or levels
of irritability or withdrawal.
360

Staff training
Staff training
16.1.4.1.11 Emotion-oriented care
Moderate- to high-quality evidence from 1 RCT containing 146 people could not differentiate
cognition, agitation, affect or satisfaction between people living with dementia in residential
care where staff were offered training in emotion-oriented care, and people living with
dementia in residential care where no specific additional training was offered.
16.1.4.1.12 Reducing antipsychotic drug use
Very low- to moderate-quality evidence from 1 RCT containing 338 people found a lower
proportion of people taking antipsychotics in residential care homes where staff were offered
psychosocial care training, compared with people living with dementia in residential care
where no specific additional training was offered, but could not differentiate rates of falls or
levels of aggression and wellbeing.
16.1.4.1.13 Towel bathing and person-centred showering

levels of aggression and discomfort in people living with dementia in residential care where
staff were offered training in either towel bathing or person-centred showering, compared
with people living with dementia in residential care where no specific additional training was
offered, but could not differentiate agitation or specific types of aggression.
16.1.4.1.14 Apathy management
Low- to moderate-quality evidence from 1 RCT containing 230 people found reduced levels
of apathy in people living with dementia in residential care where staff were offered training in
managing apathy, compared with people living with dementia in residential care where no
specific additional training was offered, but could not differentiate other measures of apathy,
activities of daily living or behavioural and psychological symptoms of dementia.
16.1.4.1.15 Non-verbal emotion signals
Very low-quality evidence from 1 RCT containing up to 68 people could not differentiate
dementia symptoms or emotions between people living with dementia in residential care
where staff were offered training in sensitivity to non-verbal emotion signals, and people
living with dementia in residential care where no specific additional training was offered.
16.1.4.2 Residential care staff and nurse training
16.1.4.2.1 Communication, empathy and conflict resolution
communication and interaction between people living with dementia in residential care where
nurses and other staff were offered training in communication, empathy development and
conflict resolution, compared with people living with dementia in residential care where no
specific additional training was offered, but could not differentiate the level of involvement of
other family members in care.
361

Staff training
Staff training
16.1.4.3 Restraint use reduction
Very low- to moderate-quality evidence from up to 2 RCTs containing 288 people found
reductions in the use of physical restraints on people living with dementia in residential care
where nurses and other staff were offered training in restraint use reduction, compared with
people living with dementia in residential care where no specific additional training was
offered, but could not differentiate numbers of medicines prescribed, functional ability, falls,
agitation or aggressive behaviours.
16.1.4.4 Residential care nurse training
16.1.4.4.1 Managing depression
depression between people living with dementia in residential care where nurses were
offered training in managing depression, and people living with dementia in residential care
16.1.4.4.2 Restraint use reduction
restraint use between people living with dementia in residential care where nurses were
offered training in restraint use reduction, and people living with dementia in residential care
16.1.4.4.3 Dementia care mapping
behavioural and psychological symptoms in people living with dementia in residential care
where staff were offered training in dementia care mapping, compared with people living with
dementia in residential care where no specific additional training was offered, but could not
differentiate agitation or quality of life.
16.1.4.5 Occupational therapist training
16.1.4.5.1 Interdisciplinary training
activities of daily living or quality of life between people living with dementia offered
occupational therapy by therapists who had been given specific additional training in
dementia, and people living with dementia offered occupational therapy by therapists who
had been given no specific additional training.
16.1.4.6 GP training
16.1.4.6.1 Flexible education
quality of life, pain, behavioural and psychological symptoms of dementia or the use of
physical restraint between people living with dementia in residential care where GPs were
offered a flexible training package, and people living with dementia in residential care where
no specific additional GP training was offered.
362

Staff training
Staff training
16.1.4.7 Pooled analysis: person-centred care versus control
Moderate- to high-quality evidence from up to 5 RCTs containing 941 people found

improvements in quality of life and levels of agitation in people living with dementia in
residential care where staff were offered training falling under the broad category of person-
centred care, compared with people living with dementia in residential care where no specific
additional training was offered, but low-quality evidence could not differentiate levels of
behavioural and psychological symptoms of dementia.
16.1.5 Evidence to recommendation

Relative value of different The committee agreed that, since the aim of this review question was
outcomes to identify staff training interventions that improve the experience of
the person living with dementia, outcomes for that individual (such as
quality of life or behavioural and psychological symptoms) would be
most relevant. They noted that the review was not focused on
identifying interventions that improve the experience of staff and,
therefore, trials that only reported on outcomes for staff would not be
relevant to include within the review.
Trade-off between Person-centred and outcome-focussed care
benefits and harms The committee agreed there were a number of trials (Chenoweth
2009 and 2014, Davison 2007, Deudon 2009, Finnema 2005, Fossey
2006 and van de Ven 2013) that, whilst including quite disparate
training interventions, could be grouped under the general heading of
person-centred care. Whilst the committee acknowledged that there
were difficulties in combining the data from these studies, it was
agreed that is was an appropriate thing to do, as the studies reported
a range of positive and negative findings without obvious differences
in study design which could explain the different outcomes.
Therefore, it was agreed to be appropriate to calculate the average
effectiveness across this group of studies, to ensure biased
recommendations were not made by focusing only on the positive
studies. The meta-analysis found significant improvements in both
agitation and quality of life in people treated by staff offered person-
centred training interventions, and the committee agreed it was
therefore appropriate to recommend such interventions.
No clear evidence was identified for any individual training
programme (such as dementia care mapping) being more effective
than another, and therefore the committee agreed it was appropriate
to make a more general recommendation which highlighted the key
elements of the interventions, rather than being more prescriptive on
exactly how an intervention should be structured or delivered.
The committee agreed that concerns had been raised by the LGBT
community that their needs are not being addressed. Therefore, the
bullet point “Respecting the person’s identity, sexuality and culture”
was included.
In the recommendation on training for care providers, the committee
agreed it was appropriate to specify those components which were
consistently included as part of the trials in the evidence base. This
included general education about dementia, assessing and
responding to individual’s symptoms and needs and understanding
and managing non-cognitive symptoms such as agitation, aggression
363

Staff training
Staff training
and pain. These trials also all included some follow-up sessions to
provide feedback to staff, and give advice on specific complex cases.
This recommendation also included items on antipsychotics and
restraint (explained below), and a specific recommendation for
younger people living with dementia, based on evidence from and
explained in section 17.
Antipsychotics and restraint use
Trials which focused primarily on managing agitation and/or
aggression whilst reducing the use of ether antipsychotics medicines
or physical restraint were also identified. The aim of these trials was
somewhat different, in that rather than trying to improve symptoms,
they focused on reducing the use of potentially harmful medicines or
procedures, without an increase in symptoms over a defined time
period. The committee noted there was clear evidence from these
studies that an approximately 50% reduction could be achieved in the
use of either antipsychotics or physical restraint without any
significant increase in behavioural or other symptoms, and the
committee therefore agreed it was appropriate to include this in the
recommendation for training interventions.
Multi-sensory stimulation
The committee agreed there was some evidence of benefits from a
multisensory stimulation intervention in people with moderate to
severe dementia. It noted that in practice these interventions are
sometimes used across a wider range of individuals, but agreed the
evidence was not sufficient to extrapolate beyond this more limited
population (particularly, since it was only based on evidence from a
single trial). Since the same quantity of evidence was not available as
for person-centred care, they agreed that it was appropriate to restrict
this recommendation to a ‘consider’ recommendation.
Other interventions
The evidence base also contained a number of more specific
targeted interventions. These again were often presented under the
broad heading of person-centred care, but only focused on a specific
subset of care rather than a whole person-centred approach. The
committee agreed these trials did not demonstrate the same positive
results as the more inclusive training programmes and agreed that it
was not possible to establish whether this was because these
interventions are less effective, or because the trials were too small
to detect an effect. It was therefore agreed that the evidence was not
sufficient to make any recommendations based on this evidence.
Consideration of health Person-centred and outcome-focussed care
benefits and resource The committee noted there was a lack of cost-effectiveness evidence
use available to support recommendations on this topic, and therefore
were conscious that it was important not to impose substantial
additional costs. The committee therefore agreed that it was
appropriate to subdivide the recommendation into two specific target
groups.
The first group is comprised of staff directly providing care and
support to people living with dementia. The committee agreed it was
this group for whom training would have the highest impact, and
therefore it was appropriate, in line with the interventions shown to be
effective in the trials, to recommend this training be face-to-face and
include the option for mentoring or additional support after the initial
intervention is delivered. It noted that there were a number of ways in
which training was delivered within the context of the trials whereby in
some studies all staff were trained by an external provider, whilst in
others the provider only trained a small number of staff, who then
364

Staff training
Staff training
passed on that knowledge to their colleagues. The committee agreed

that both of these approaches were appropriate to consider in
practice.
The second group of individuals considered in the recommendations
consists of care and support providers more generally. This may
include staff working in care environments or with people living with
dementia, but not directly involved in providing care and support
themselves. The committee agreed this group would also benefit
from training, but it was not possible to justify the costs associated
with providing face-to-face training in this larger group. The
committee noted that in practice this was often provided as online
training, and agreed that for this broader group this was an
appropriate approach to take.
Antipsychotics and restraint use
The committee agreed that, even though there would be additional
costs associated with delivering this training, this would be offset by
considerable reductions in the costs of antipsychotic prescribing,
provided that reductions similar to those found in the studies could be
achieved in practice.
Multi-sensory stimulation
The committee noted that the primary cost associated with
multisensory stimulation is the initial cost of purchasing the
equipment, and staff training costs. It therefore agreed it was
important the recommendation focus on training staff in the use of
such equipment and techniques (which would only be relevant if the
equipment was available at their site), rather than recommending
additional sites purchase that equipment as there was not sufficient
evidence to justify the purchase of such equipment as part of the
recommendation.
Quality of evidence The committee agreed the evidence underpinning the
recommendations on person-centred care and multisensory
stimulation was of moderate to high quality. However, it noted that
the evidence base was entirely composed of studies conducted in
care homes, and not in other clinical or community settings. The
committee agreed that the principles of good training would be similar
across these different settings, and therefore it was appropriate to
extrapolate the evidence base from care homes to being applicable
to all care and support providers. However, it also agreed that there
may be other interventions that are more effective in these other
settings, and therefore made research recommendations to look at
the most effective training interventions for community staff and acute
hospital staff.
Other considerations The committee noted that in a number of the included studies, carers
were also invited to attend the training alongside staff. Whilst there
was no direct evidence that this led to improved outcomes, the
committee agreed that it was a positive thing to encourage, both
because of the gains that carers could make themselves, and
because they provided a valuable additional perspective at any
training. The committee agreed the evidence did not justify
recommending additional resources be devoted to training
specifically organised for carers but agreed it was appropriate to
consider inviting carers along to training sessions that were already
being run, provided sufficient capacity is available.
365

Staff training
Staff training
111. Care and support providers should provide all staff with training in person-
centred and outcome-focused care for people living with dementia, which should
include:
 understanding the signs and symptoms of dementia, and the changes to
expect as the condition progresses
 understanding the person as an individual, and their life story
 respecting the person's individual identity, sexuality and culture
 understanding the needs of the person and their family members or
carers
 the principles of the Mental Capacity Act 2005 and the Care Act 2014.
112. Care providers should provide additional face-to-face training and mentoring to
staff who deliver care and support to people living with dementia. This should
include:
 understanding the organisation’s model of dementia care and how it
provides care
 how to monitor and respond to the lived experience of people living with
dementia, including adapting communication styles
 initial training on understanding, reacting to and helping people living
with dementia who experience agitation, aggression, pain, or other
behaviours indicating distress
 follow-up sessions where staff can receive additional feedback and
discuss particular situations
 advice on interventions that reduce the need for antipsychotics and allow
doses to be safely reduced
 promoting freedom of movement and minimising the use of restraint
 if relevant to staff, the specific needs of younger people living with
dementia and people who are working or looking for work.
113. Consider giving carers and/or family members the opportunity to attend and
take part in staff dementia training sessions.
114. Consider training staff to provide multi-sensory stimulation for people with
moderate to severe dementia and communication difficulties.
18. What is the cost effectiveness of using a dementia-specific addition to the Care
Certificate for community staff, including dementia-specific elements on
managing anxiety, communication, nutritional status and personal care?
19. What is the effectiveness of training acute hospital staff in managing behaviours
that challenge in people living with dementia on improving outcomes for people
and their carers?
366

Staff training
Staff training
appendix L.
367

Needs of younger people living with dementia
17 Needs of younger people living with

dementia
It is estimated that at least 40,000 people in the UK are living with young-onset dementia,
defined as symptom onset before the age of 65 (Dementia UK, 2nd edition). Young-onset
dementia differs from dementia in later life in several important respects. Less typical clinical
syndromes (such as behavioural, language-led, dyspraxic or visuospatial presentations) and
underlying pathologies (such as frontotemporal dementia) contribute a higher proportion of
cases. A family history of young-onset dementia, or related conditions such as motor neuron
disease or Parkinson's disease, may be relevant in this age group, as dementias caused by
single gene mutations typically have a lower age of onset than sporadic forms.
The onset of dementia at an earlier age has a range of consequences that require particular
consideration from health and social care professionals and systems. These include, but are
not limited to, loss of employment (and hence income and status), reduced ability to care for
elderly parents or dependent children, a need to reconsider plans for retirement, and stigma
from having an illness more typically associated with later life. Carers of people living with
young-onset dementia also have particular needs. They are most commonly spouses of a
similar age, so may also experience difficulty maintaining employment and other
responsibilities. But elderly parents and children can also become carers and face particular
challenges as a result. It is incumbent on professionals working with people affected by
young-onset dementia to be mindful of the additional challenges faced by this group, and to
respond to these in a personalised way. For example, some people living with young onset
dementia may be able to continue working if appropriately supported.
368

17.1 The specific needs of younger people living with dementia

Review question
 What are the specific needs of younger people living with dementia?
17.1.1 Introduction
Table 95: Review summary: needs of younger people living with dementia
Population  People (aged between 40 years and 65 years) living with dementia
 Carers of people (aged between 40 years 65 years) living with
dementia
Phenomena of interest Any factors which either uniquely impact on younger people living with
dementia or have a disproportionate impact on this group, which may
include:
 Being in work at time of diagnosis
 Having a partner who still works
 Dependent children
 Caring for older relatives
 Large financial commitments (e.g. mortgage)
Qualitative studies and qualitative evidence syntheses were included if they explored the
specific needs of younger people living with dementia and focussed upon improving
outcomes for people with dementia and their carers. Studies needed to contain participants
from the UK, report the views of either people living with dementia or their carers, and match
the criteria given in either Table 95. Full details of the review protocol are given in appendix
C. Papers were excluded if they:
 did not include the views people living with dementia or their carers in the UK
 included only quantitative analysis of the collected information
 were not in English
and abstracts, and the full texts of 25 references that were potentially relevant to these
review questions were requested. Seven qualitative studies exploring care coordination were
included in the review. The included studies are summarised in Table 96. For the full
evidence tables and full CERQual profiles please see Appendix E and Appendix G.
References for the included studies are given in appendix I. The 18 excluded papers, with
reasons for exclusion, are presented in Appendix F.
369


Study details Study population Subject of study Outcomes
Chaplin 2016 5 younger people Topic: the experiences of The opinions of younger
living with people with dementia in people living with dementia
dementia employment
Method of data collection:
interviews
Clayton-Turner 28 younger people Topic: standard care The opinions of younger
2015 living with Method of data collection: people living with dementia
dementia and 15 Interviews and their carers
carers
Clemerson 8 younger people Topic: comparing a The opinions of younger
2014 living with memory service and a people living with dementia
dementia young onset dementia
service
semi-structured interviews
Hegarty 2014 4 men and 2 Topic: a walking group for The opinions of younger
women who were younger people living with women living with
younger people dementia dementia and their carers
living with Method of data collection:
dementia, and their focus group interview for
carers younger people living with
dementia. A questionnaire
for their spouses.
Higgins 2010 5 younger people Topic: a club for younger The opinions of younger
living with people with dementia people living with dementia
dementia and 6 Method of data collection: and their carers
carers interviews
Johnson 2008 16 younger women Topic: a service for The opinions of younger
living with younger women living with women living with
dementia and their dementia dementia and their carers
carers Method of data collection:
written and verbal
feedback
Pipon-young 8 younger people Topic: the experiences of The opinions of younger
2011 living with younger people with people living with dementia
dementia dementia
interviews and group
discussions
370

17.1.4.1 Experiences and coping in employment
The following themes were identified for ‘experiences and coping in employment’ for people
living with dementia and their carers:
 People needed information about their rights in the workplace, and employers also
needed to be educated about the same (low confidence)
 People living with dementia had an awareness of changes in their functioning in the work
place as they developed dementia. (low confidence)
 People living with dementia experienced a shock at losing their expected future. (low
confidence)
 A reluctance from people living with dementia to acknowledge the signs of cognitive
decline (low confidence)
 Attempting to self-manage – developing coping strategies, and spending more time and
effort in planning and organising tasks (low confidence)
 Feeling under scrutiny by managers and colleagues (low confidence)
 A lack of consultation about management decisions – not feeling they were offered the
reasonable adjustments they were entitled to (low confidence)
 Feeling abandoned by the workplace and consequent feelings of resentment towards the
workplace (low confidence)
 Financial hardship and consequent worry (low confidence)
17.1.4.2 General experiences and coping
The following themes identified for ‘general experiences and coping’ for people living with
dementia and their carers:
 Feelings of shock and a sense of loss at receiving the diagnosis, but also relief at having
the diagnosis confirmed (low confidence)
 Experiences of feeling ‘too young’ – assuming dementia was something that affected
older people (high confidence)
 Sense of pressure at still having responsibility for children, a mortgage or a business to
run (low confidence)
 Coping by normalising the situation - creating an identity as an older person, even
transiently, allowed people to make sense of developing Alzheimer’s disease by
normalising the life-cycle (very low confidence)
 Loss of adult competency - emerged through people’s experience of either feeling more
‘childlike’ due to a loss of skills or being treated this way by others (very low confidence)
 Negative impact of other’s perceptions (low confidence)
 A reduced sense of self-worth (very low confidence)
 Trying to hold on to their existing self-concept - the importance of acknowledging that
although they have dementia, there were many aspects of their lives that remained the
same (high confidence)
 A fear of disclosing the diagnosis and a desire to hide it from others (low confidence)
 The importance of remaining independent, active and involved (low confidence)
371

 The importance of knowing other people with dementia and being able to share
understandings through similar experiences (low confidence)
 Lack of age-appropriate services (very low confidence)
 The intention to regain control emerged as a common coping strategy in response to the
experience of loss of agency (very low confidence)
 People may well still be driving, and this should be discussed (low confidence)
17.1.4.3 Group activities (walking group, day service and lunchtime social group)
The following themes identified for ‘group activities designed specifically for young people
living with dementia’ for people living with dementia and their carers:
 Benefits of building supportive and positive relationships, and a social network (low
confidence)
 Provided a sense of belonging, purpose and achievement (low confidence)
 Improved self-confidence by being able to interact with a group of people similar to
themselves (low confidence)

Relative value of different The committee agreed the key finding of the review would be any
outcomes specific needs identified for younger people living with dementia that
were different to those of the main dementia population. This would
then be used to ensure the pathway was equally relevant to and
accessible for, younger people living with dementia (and their carers).
Trade-off between Chaplin (2016) and Clayton-Turner (2015) provided evidence that at
benefits and harms diagnosis, the person and their family members or carers should be
offered oral and written information that explains the person’s rights
and needs for reasonable adjustments if they are in work or looking
for work. In the committee’s experience, with reasonable
adjustments, some younger people living with dementia are able to
continue to work for many years. The committee noted that in their
experience Disability Employment Advisors have little experience in
this area and it agreed that a recommendation in this area would be
useful. The committee agreed that the phrase ‘needs for reasonable
adjustments’ should be included in the recommendation as this infers
that the person’s needs should be assessed. The committee agreed
that the ‘employment’ being undertaken could be voluntary and
therefore, the word ‘work’ was used rather than ‘employment’ to
reflect this. The committee agreed that care providers should provide
face-to-face training and mentoring to staff who deliver care and
support to people living with dementia. This should include the
specific needs of younger people living with dementia (where this is
relevant to their role).
Pipon-Young (2012) and Clayton-Turner (2015) discussed the issues
around younger people living with dementia and financial
commitments such as mortgages they may still have. The committee
agreed that this adds further importance to the recommendations
372

which have already been drafted to offer information regarding

financial and legal advice services to all people living with dementia.
Some of the studies either involved peer support groups (Clayton-
Turner 2015, Hegarty 2014, Higgins 2010, Johnson 2008) or
discussed a need for them (Pipon-Young 2012). The committee
agreed that the recommendations already include offering information
regarding local support groups.
The committee agreed that people living with dementia who work
during office hours may find it difficult to access services. Therefore,
the committee agreed that service providers should design services
to be accessible to as many people living with dementia as possible,
including people who are in work.
The committee noted that the carers of younger people living with
dementia might still be in employment. Therefore, the committee
agreed that the support offered to carers should be tailored to their
needs and preferences and to what they want it to achieve, for
example, carer’s employment rights.
Consideration of health The committee discussed the potential impact of the
benefits and resource recommendations and agreed that they should not result in additional
use expenditure.
Quality of evidence The committee agreed that the overall quality of the evidence was
low or very low, because of the paucity of studies in this area and the
low numbers of study participants. However, the committee agreed
that the findings of the review did match their experience and
therefore were confident they would be replicated in larger studies.
They agreed that, due to the relatively low quality of the evidence
available, it was more appropriate to make modifications to
recommendations made in other sections with stronger evidence
bases to ensure they appropriately considered the needs of younger
people, rather than writing a separate set of recommendations to
cover younger people.
The committee agreed that the work-related issues raised in the
study Chaplin 2016 were very important to people living with
dementia and very common for younger people. Therefore, this study
had a high value.
Other considerations No specific equality issues were identified for this review question.
Additional comments around the needs of younger people living with dementia were added
to the sections on ‘involving people living with dementia in decisions about their care’
(section 6), ‘staff training’ section (section 16) and ‘care planning, review and co-ordination’
(section 7).
373

Assessing and managing comorbidities
18 Assessing and managing comorbidities

Dementia is primarily a condition of old age and as such many people with dementia will
have additional long term illnesses also associated with increasing age.
People with dementia often have several additional long term conditions; a UK based study
found that, on average, people with dementia had 4.6 chronic illnesses in addition to their
dementia (Guthrie 2012) and other geriatric conditions such as delirium, falls and
incontinence are also more frequent.
The most common chronic illnesses in people with dementia include diabetes, hypertension,
cardiovascular disease and age-related musculoskeletal disorders.
The increasing multi-morbidity associated with older age leads to a higher level of clinical
complexity which health care professionals may find challenging especially within health care
systems where clinical guidelines and service organisation are often focused on a single
illness. In addition, research to date has often considered dementia in isolation with little, if
any, regard as to how other complex health needs might impact on the person living with
dementia and their family’s needs and experiences and service use and provision. Also
certain comorbid medical conditions may exacerbate the progression of dementia; for
example, there is evidence that cognitive decline may be accelerated in older people with
type 2 diabetes. The presence of dementia may also adversely affect the clinical care of
other conditions by limiting their ability to self-care, take medication, attend specialist care
appointments and engage in health promotion. There has been little research investigating
the effect of co-existent multi-morbidity on a person living with dementia's health, well-being
and clinical care and what 'good practice' would look like to both service users and providers.
Healthcare professionals who do not work in mental health often have little understanding of
the needs and experiences of people with dementia and consequently their care needs are
frequently not met with evidence of service duplication, delays in the identification of
problems and unnecessary interventions. Whilst improving quality of care for people with
dementia remains a key government target, there is a growing concern that current health
care service organisation is not meeting the needs of an increasing ageing population. Key
questions to be addressed include the optimal ways to manage both co-existing long term
illnesses after a diagnosis of dementia, and how best to manage intercurrent illness in people
living with dementia.
374

18.1 Assessing and treating intercurrent illness in people living

with dementia
Review questions
 Are there effective methods for assessing intercurrent illness in people living with
dementia that are different from those already in use for people who do not have
dementia?
 Are there effective methods for treating intercurrent illness in people living with dementia
that are different from those already in use for people who do not have dementia?
18.1.1 Introduction
The aim of these review questions was to identify the most effective methods for assessing
the symptoms and severity of intercurrent illness and to identify the most effective
interventions and strategies for treating intercurrent illness in people living with dementia.
Both review questions considered whether the methods used for assessing and treating
intercurrent illness in people living with dementia are different from those used for people
who do not have dementia. They sought to explore the methods of assessing and treating an
unrelated acute condition presenting in people living with dementia and specifically focused
on people showing symptoms of the following.
 Pain
 Falls (& loss of mobility)
 Delirium
 Urinary tract infections
The review identified studies that fulfilled the conditions specified in Table 97 or Table 98.
For full details of the review protocols, see Appendix C.
Table 97: Review summary: assessing intercurrent illness in people living with
dementia
Population  Studies containing people (aged 40 years and over) with and without
a diagnosis of dementia, and showing symptoms of an intercurrent
illness
Interventions  Standardised observations, assessments, scales or tools used to
assess the presentation and severity of an acute condition
specifically for people living with dementia
Comparator  Standardised observations, assessment scales or tools used to
assess the presentation and severity of an acute condition for
people with an intercurrent illness who do not have dementia
 Usual care
Outcomes  Rates of accurately identified intercurrent illness in people living with
dementia
 Diagnostic test accuracy (including sensitivity, specificity, PPV, NPV
etc.)
ability
 Health-related quality of life of people living with dementia
375

Table 98: Review summary: managing intercurrent illness in people living with
dementia
Population  People (aged 40 years and over) with a diagnosis of dementia and
showing symptoms of an intercurrent illness
Interventions  Pharmacological interventions/self-care strategies/monitoring or
observational strategies specifically designed for people living with
dementia and an intercurrent illness
Comparator  Pharmacological interventions/self-care strategies/monitoring or
observational strategies for people with an intercurrent illness but
not specific to people living with dementia
 Usual care
Outcomes  Symptom resolution/reduction of intercurrent illness
ability
 Change in appropriate polypharmacy
 Intervention-related problems such as potentially avoidable hospital
admissions and re-admissions, errors, poor adherence and
potentially avoidable adverse effects (e.g. pressure sores)
 Intervention related outcomes including concordance, compliance
satisfaction of person living with dementia and their informal carers
 Health related quality of life of person living with dementia and
his/her informal carers
and abstracts and 285 references were ordered for full text across both review questions.
258 papers were subsequently excluded because they did not fit the inclusion criteria (see
Appendix F for a detailed list of excluded studies and reasons for their exclusion).
Seven studies were included in the evidence review for the question considering
assessments for an intercurrent illness. Five studies were included for assessing pain, 1
study for assessing falls and 1 study for assessing delirium. No studies were identified as
relevant to consider assessments for urinary tract infections in people living with dementia
compared with those who do not have dementia.
Fourteen studies were included in the evidence review for the question considering
management of an intercurrent illness. Eight studies were included for falls, 3 for pain, 2 for
hip fractures and 1 for delirium.
A summary of the characteristics of the included studies for assessment is provided in Table
99 and for management in Table 100. For the full evidence tables and full GRADE profiles
376

please see Appendix E and Appendix G. References for the included studies are given in
appendix I.
377

2 Table 99: Included studies - assessment of intercurrent illness

Study Study design Study population Intervention & Assessment scales Relevant Comments
reference comparator (including outcomes
study aim)
Pain assessment
Mosele et Prospective Participants with To assess the psychometric Pain assessment in  Presence of pain All participants were
al (2012) cohort MMSE≥24 (n=290) properties of PAINAD scale advanced dementia as assessed by admitted to the acute
versus participants with compared with the NRS in (PAINAD) versus PAINAD and geriatric section of
MMSE<24=310) people with different stages Numerical rating scale NRS Padua University.
of cognitive impairment (NRS) and verbal Study location Italy
descriptor scale
(VDS)
(observational pain
assessment versus
self-report
assessment)
Horgas et Cross Cognitively intact To compare NOPPAIN Non Communicative  Pain verification All participants were
al (2007) sectional participants with a ratings with self-report in Patients Pain – correlation selected as a
study mean MMSE score of participants with and Assessment between subsample from a
27 (n=20) without cognitive (NOPPAIN) versus observational larger parent study.
Versus impairment numerical rating scale and self-reported Participants were
Cognitively impaired (NRS and verbal pain scores enrolled in assisted
participants with mean descriptor scale living facilities, nursing
MMSE score of 17 (VDS) facilities or retirement
(n=20) (observer reported apartments.
versus self- report Study location USA
assessment)
De Waters Correlational Cognitively intact To psychometrically Pain assessment in  Pain verification - All participants
et al (2008) design participants with a evaluate the PAINAD advanced dementia correlation hospitalised for a hip
mean MMSE score of alongside the NRS in (PAINAD) versus between fracture.
26 (n=13) participants with and Numerical rating scale observational Study location USA
Versus (NRS)
378

study aim)
Cognitively impaired without cognitive (observational pain and self-reported
participants with a impairment assessment versus pain scores
mean MMSE score self-report
of14 (n=12) assessment)
Van Herk Multi centre Participants without To assess reliability and Rotterdam Elderly  Pain verification Study location
et al (2009) case control cognitive impairment validity of a newly Pain Observation  Comparison of Netherlands
study with MMSE ≥18 (n=50) developed observational Scale (REPOS) pain scores
versus participants with pain assessment tool for Versus NRS and
cognitive impairment people whom self-report is PAINAD
with MMSE< 18 impossible in people with (observational pain
(n=124) and without cognitive assessment versus
impairment observational and
self-report
assessment)
Lukas et al Retrospective Participants with To determine the Abbey pain scale  Pain verification: Study location
(2013) cohort MMSE ≥24 (n=60) comparative ability of Pain Assessment in level of Australia
versus participants with observer-rated pain Advanced Dementia agreement
MMSE<20 (n=65) assessment tools to identify (PAINAD) regarding
the presence or intensity of Non communicative presence of pain
pain in cognitively intact Patients Pain  Correlation
older people or people with Assessment regarding pain
moderate to severe (NOPPAIN) intensity
cognitive impairment
(observational pain
assessment versus
observational pain
assessment versus
self-report
assessment)
Falls assessment
379

study aim)
Kato-Narita Case control Participants without To analyse the correlation Berg Balance Scale Number of falls Alzheimer’s disease
et al (2011) study cognitive impairment between falls and loss of (BBS) Performance on diagnosed by CDR
(based on Mayo Older functional capacity in Disability Assessment scale rating.
American Normative people with Alzheimer’s for Dementia (DAD) Participants recruited
Studies Criteria ;n=40) disease and those without from an out-patients
versus cognitive impairment service at a university
participants with hospital.
Alzheimer’s disease Study location
(based on Clinical Brazil
Dementia Rating;
n=48)
Delirium assessment
Sepulveda Cross Participants without To assess Delirium rating DRS-R98  DRS-R98 scores Participants with
et al (2015) sectional cognitive impairment scale- revised 98 against versus  ROC analyses dementia were
analysis (diagnostic criteria not other assessment scales in ICD-10 identified as a
reported; n=40) people with dementia and subsample of whole
DSM-III-R
versus those without cognitive population based on a
impairment DSM-IV classification by
participants with
DSM-5 Spanish Informant
possible Dementia
based on IQCD score questionnaire on
>85 (n= 85) cognitive decline
(score >85).
Study location Spain
1 Table 100 Included studies for management of intercurrent illness

Intervention & comparator (including
Study reference Study design Study population study aim) Relevant outcomes
Pain management
Fuchs-Lacelle et al Cluster RCT 173 people over 65 with dementia and Intervention:  Medication level
(2008) severe communication impairment  Completion of the Pain Assessment  Nursing stress scale
Checklist for Seniors with Limited
380


Ability to Communicate (PACSLAC)
every other day for 3 months
Comparator:
 Completion of an Activity Log every
other day for 3 months
Husebo et al Cluster RCT 352 people with dementia (MMSE < 20) Intervention:  Neuropsychiatric symptoms
(2014)  Stepwise protocol for treating pain
Comparator:
 Usual care
Sandvik et al Cluster RCT 352 people with dementia (MMSE < 20) Intervention:  Pain
(2014),  Stepwise protocol for treating pain  Adverse events
Comparator:
 Usual care
Delirium
Kolanowski et al RCT 16 people with delirium superimposed Intervention:  Delirium symptoms
(2011) on dementia  Standard nursing care and prescribed  Activities of daily living
rehabilitation therapies, plus 30  Cognition
minutes per day of cognitively
stimulating recreational activities for
30 days.
Comparator:
 Standard nursing care and prescribed
rehabilitation therapies
Hip fracture rehabilitation
Smith et al (2015) Systematic RCTs evaluating the effectiveness for Intervention:  Mortality
review people with dementia of enhanced care  Enhanced models of care and/or  Activities of daily living
and rehabilitation following hip fracture rehabilitation:  Adverse events
surgery compared with usual care. Comparator:  Hospitalisation
 Standard nursing, medical and
therapy intervention
381


Stenvall et al Cluster RCT 199 people post neck of femur fracture Intervention:  Falls
(2007) (including 64 people with dementia)  Comprehensive geriatric
assessments, management and
rehabilitation
 Active prevention, detection and
treatment of postoperative
complications such as falls, delirium,
pain and decubitus ulcers
Comparator:
 Specialist orthopaedic unit following
conventional postoperative routines
Falls
Chan et al (2015) Systematic RCTs that compared the efficacy of Intervention:  Falls
review physical exercise with routine medical  Group or home-based exercise  Fractures
care or other controlled activities in Comparator:
preventing falls in older people with
cognitive impairment  Routine care or less intensive
interventions
Oliver et al (2006) Systematic Trials, case-control or observational Multiple intervention types:  Association of dementia
review cohort studies of patients in hospitals or  In hospital multifactorial interventions prevalence in study with effect
care homes that reported the number of size
 In care home multifactorial
rate of falls or fractures or people who
interventions
fell.
 Hip protectors in care homes
 Removal of physical restraint
 Fall alarm devices
 Exercise
 Changes or differences in physical
environment
 Calcium and vitamin D in care homes
 Medication review
Pitkälä et al (2013) RCT 210 home-dwelling patients with Home-based exercise:  Falls
Alzheimer’s disease and their carers  Hospital admissions
382


 Physiotherapist led individually  Cost
tailored training
Group-based exercise:
 Physiotherapist led group exercise
consisting of endurance, balance and
strength training, and exercise for
improving executive functioning
Control group:
 Usual care provided by the Finnish
healthcare system, plus oral and
written advice on nutrition and
exercise methods
Shaw et al (2003) RCT 274 cognitively impaired older people Intervention:  Falls
presenting to the accident and  Multifactorial assessment and  Fractures
emergency department after a fall intervention  A&E attendance
Control:  Hospital attendance
 Usual care from all health  Mortality
professional who were involved in
their management
Suttanon et al RCT 40 people with mild to moderate Home-based exercise programme:  Falls
(2013) Alzheimer’s disease  Six-month individualised home-based  Quality of life
exercise programme supervised by a  Carer burden
physiotherapist
 Based on the Otago home-exercise
programme
Control (education) programme:
 Education and information
programme delivered by an
occupational therapist
 Designed to provide the same
number of home visits and phone
calls as the exercise programme
383


Tchalla et al RCT 96 people with Alzheimer’s disease Intervention:  Falls
(2013)  Fall reduction program following an
initial Comprehensive Gerontological
Assessment. Participants were
equipped with an HBTec-TS system
Control:
 Fall reduction program following an
initial Comprehensive Gerontological
Assessment. No HBTec-TS system
was implemented
Toulotte et al RCT 20 elderly dementia patients with a Training group:  Falls
(2003) history of falling  Two supervised 1 hour exercise
sessions per week for 16 weeks
 Exercises to develop muscular
strength, proprioception, static and
dynamic balance and flexibility
Control (education) programme:
 Usual care
Wesson et al RCT 22 person living with dementia and Home-based exercise program:  Falls
(2013) carer dyads  Strength and balance training  Carer burden
exercises
Home hazard reduction:
 Six occupational therapist and five
physiotherapist visits over 12 weeks
Control group:
 Usual care
384

A single search was undertaken for review questions 20 and 21. A total of 2,565 citations
was returned. Following review of titles and abstracts, the full text of 1 study was retrieved for
detailed consideration, but it did not meet inclusion criteria. Therefore, no relevant cost–utility
analyses were identified for these questions.
18.1.4.1 Assessment
18.1.4.1.1 Pain assessment
Low- to moderate-quality evidence from 1 observational study with 600 participants found
that rates of those assessed as having pain increased significantly for people with cognitive
impairment when pain was assessed by an observational rating scale (PAINAD) compared
with those who did not have cognitive impairment. When pain was assessed by a self-report
scale (NRS), there were no significant differences between people with cognitive impairment
and those who did not have cognitive impairment.
Low-quality evidence from 1 observational study with 40 participants found that pain intensity
ratings from an observational rating scale (NOPPAIN) did not correlate with pain intensity
ratings from self-report scales (VDS and NRS), for people with cognitive impairment;
however, for people who did not have cognitive impairment, there was a significant positive
correlation between pain intensity ratings from the observational rating and self-report scales.
The same study found there was a significant positive correlation between pain intensity
ratings from an observational rating scale (NOPPAIN) and total number of pain indicators
observed for both people with cognitive impairment and those who did not have cognitive
impairment.
Very low-quality evidence from 1 observational study with 25 participants found a significant
positive correlation between pain ratings on an observational rating scale (PAINAD) and self-
report scale (NRS) for both people with cognitive impairment and those who did not have
cognitive impairment.
Very low-quality evidence from 1 observational study with 174 participants found ratings from
two observation scales (REPOS versus PAINAD) were significantly positively correlated but
when self-report observations were obtained by a nurse, pain ratings were not positively
correlated with an observational rating scale (REPOS) for people with cognitive impairment.
Significant positive correlations were found for both the observational rating scale and nurse
led self-report rating scale for people who did not have cognitive impairment. The same study
also found that pain scores recorded on the observational rating scales were significantly
higher for people with cognitive impairment compared with those who did not have cognitive
impairment.
Moderate-quality evidence from 1 observational study with 108 participants found higher
levels of agreement between the presence of self-reported pain and the presence of
observer-rated pain in people without cognitive impairment compared with people with
cognitive impairment.
385

Moderate-quality evidence from 1 observational study with 108 participants found significant
positive correlations between observational rating scales (PAINAD, NOPPAIN and Abbey
pain scale) for people with cognitive impairment. The relationship between observational
rating scales and self-report ratings was not significantly correlated for people who did not
have cognitive impairment.
18.1.4.1.2 Falls assessment
Low-quality evidence from 1 observational study with 88 participants found people with
cognitive impairment scored significantly lower on a balance assessment than those who did
not have cognitive impairment. The same study also found there was a significant negative
correlation between the number of falls recorded and scores on the balance scale for people
with cognitive impairment and a significant positive correlation between the number of falls
recorded and scores on the balance scale for those who did not have cognitive impairment.
18.1.4.1.3 Delirium assessment
Low-quality evidence from 1 observational study with 125 participants found the difference in
scores assessed by the Delirium rating scale (DRS) compared with the ICD-10, DSM-II-R,
DSMIV and DSM-5 were significantly higher for people who did not have cognitive
impairment compared with those with cognitive impairment.
18.1.4.2 Management
18.1.4.2.1 Pain management
Low-quality evidence from 1 cluster-randomised RCT of 173 people followed up for 3 months
showed that people monitored using the PACSLAC had a significantly higher increase in the
amount of pain medications used than those assessed using an activity log.
Low-quality evidence from 1 cluster-randomised RCT of 173 people followed up for 3 months
showed that nurses monitoring people using the PACSLAC reported lower levels of stress
(as measured by the Nursing stress scale) than those monitoring people using an activity log.
Moderate-quality evidence from 1 cluster-randomised RCT of 327 people followed up for 8

weeks showed that people treated using a stepwise treatment protocol had significantly
lower overall pain intensity scores, measured using the MOBID-2, than those receiving usual
care. This included reductions in both musculoskeletal and internal organ, head and skin
pain.
Moderate-quality evidence from 1 cluster-randomised RCT of 298 people followed up for 8

weeks showed that people treated using a stepwise treatment protocol had significantly
fewer behavioural symptoms (as measured by the NPI-NH) than those receiving usual care.
This included reductions in mood symptoms, depressive symptoms and apathy, but no
significant differences in anxiety or irritability.
18.1.4.3 Delirium
Very low-quality evidence from 1 RCT of 16 people followed up for 30 days showed that
people randomised to cognitively stimulating activities had significantly higher rates of
improvement in activities of daily living (BI) and cognition (MMSE), but could not distinguish a
difference in rates of improvement in delirium symptoms (CAM and DRS).
386

18.1.4.3.1 Hip fracture rehabilitation
Moderate-quality evidence from 1 cluster RCT of 199 could not distinguish the effectiveness
of a neck of femur fracture rehabilitation programme on reducing the incidence of falls
between people with and without dementia.
Very low- to low-quality evidence from 1 RCT of 47 people could not distinguish the odds of
mortality or activities of daily living independence at 12 months between people offered
enhanced or conventional inpatient care. There was very low-quality evidence from the same
study of lower incidence of:
 urinary tract infections
 nutritional problems
 postoperative delirium
 recurrent falls
Very low-quality evidence from the same study could not distinguish:
 rates of pneumonia
 rates of decubital ulcers
 rates of postoperative fracture
 length of stay
 number of drugs prescribed on discharge
Very low-quality evidence from 2 RCTs of 177 people could not distinguish the odds of
mortality at 12 months between people offered enhanced or conventional home and inpatient
care.
Very low-quality evidence from 1 RCT of 47 people found higher activities of daily living
scores for people offered enhanced rather than conventional home and inpatient care, but
could not distinguish the odds of falls at 12 months.
Very low-quality evidence from the same study could not distinguish:
 Frequency of hospital admissions
 Attendance at accident and emergency
Very low-quality evidence from 1 RCT of 126 people could not distinguish the odds of
delirium incidence between people offered geriatrician-led vs orthopaedic-led inpatient
management.
18.1.4.3.2 Falls
Low- to moderate-quality evidence from 2 RCTs of 148 people found home-based exercise
programmes reduced the proportion of people falling and the mean number of falls in a
population of community-dwelling people with dementia, but could not differentiate levels of
carer burden.
Moderate-quality evidence from 1 RCT of 123 people found group-based exercise programs
reduced the proportion of people falling in a population of community-dwelling people with
dementia, but could not differentiate the mean number of falls.
387

Low- to moderate-quality evidence from a systematic review of 7 RCTs found exercise

programmes reduced the proportion of people falling by a similar amount to equivalent
interventions in a population without dementia, but could not differentiate the proportion of
people with subsequent fractures.

Relative value of different The committee noted that both the evidence presented and members’
outcomes own experience pointed to the under-diagnosis of pain in people living
with dementia, and therefore any evidence that showed either an
increase in the number of people living with dementia correctly identified
as being in pain, or evidence of effective treatment protocols for this
group would be highly relevant. The committee recognised the
relevance of the evidence associated with pain and agreed there was
appropriate evidence to support a recommendation for the use of
structured observational tools alongside self-reported pain. This was
supported by the evidence presented, which demonstrated a lack of
correlation between self-reported and observer-reported pain in people
living with dementia, compared with people without dementia, and that
whwn observational tools were used, a similar proportion of people with
and without dementia were found to be in pain, which was not the case
when self-report alone was used. The committee also agreed that
specific mention should be made of people who are unable to self-report
pain, where observational tools may present the only viable method of
evaluating pain.
The committee noted that there were already published NICE guidelines
in a number of the areas under consideration, such as falls and delirium,
and these guidelines did not explicitly exclude people living with
dementia from their scope. Therefore, evidence that merely confirmed
what was already in those guidelines was considered but did not justify
the need for specific recommendations. The committee agreed that
further recommendations would only be relevant were the evidence
pointed towards the need for differences in the management between
people living with dementia and people without dementia.
Trade-off between The committee noted that, for pain management, there was a key trade-
benefits and harms off between the risk of overtreatment and what was believed to be the
larger current problem of people being prescribed insufficient pain relief.
The committee agreed there was appropriate evidence to support a
recommendation for the use of structured observational tools alongside
self-reported pain. This was supported by the evidence presented,
which demonstrated a lack of correlation between self-reported and
observer-reported pain in people living with dementia, compared with
people without dementia. The committee also agreed that specific
mention should be made of people who are unable to self-report pain,
where observational tools may present the only viable method of
evaluating pain. The committee observed that pain management was
undertaken with a stepped approach which balances the need for
analgesia alongside a change in a person’s behaviour or any signs of
distress attributed to pain. In practice, treating pain in all people should
be done with a holistic approach, which takes into consideration a need
to assess and intervene. The committee noted that RCT evidence
indicated the effectiveness of a stepwise protocol to manage pain,
388

consistent with the way pain would be managed in people without

dementia.
The committee agreed that it was appropriate to make a
recommendation to use such a protocol, noting that it was unlikely that
there would be major differences in the way pain was treated in people
with and without dementia, once the pain had been correctly identified.
It was also noted that pain assessment and management was an
iterative process, and there is a need to repeat assessments both when
people show signs of pain and display behaviours indicative of being in
pain, and during treatment to ensure healthcare staff are following the
appropriate step of the protocol.
The committee agreed that the principles of assessment and treatment
of falls in people living with dementia should not necessarily differ to
those applied to people without dementia, and the evidence presented
was broadly supportive of this. It was agreed therefore that it would be
appropriate to cross refer to the recommendations in the NICE guideline
on Falls in Older People (CG161). However, it also noted that many of
the trials of falls interventions in people living with dementia had specific
modifications to ensure they worked appropriately in that group (e.g.
involvement of carers in delivering or monitoring interventions). The
committee therefore agreed that it would be appropriate to add a
recommendation for healthcare professionals to consider the specific
needs of people living with dementia when referring them for falls
intervention programmes or to consider how interventions may need to
be modified to ensure adequate participation.
The committee noted that there was a specific area where the evidence
on falls management in people living with dementia did not align with
that of people without dementia, which was the effectiveness of multi-
factorial interventions. These interventions are recommended in the falls
guideline (CG61) but the evidence presented did not show a significant
effect in a population of people living with dementia. The committee
noted that the population in this RCT contained a significant proportion
of people identified as having severe dementia and this raised concerns
that the interventions may be less effective in this group, and that this
may be attributable to the intensity of the interventions and the large
number of tests involved, which may cause additional distress to
someone living with dementia and in particular severe dementia. These
negative factors would outweigh any benefits of the intervention. The
committee discussed the evidence noting that it was strong enough to
recommend that such interventions should not be used in people with
severe dementia without consideration, on a case-by-case basis, to
ensure that the benefits of the intervention were expected to be greater
than the possible harms.
Trade-off between net No health economic evidence was identified for inclusion when
health benefits and addressing these review questions. The committee noted that the
resource use recommendations made for pain and falls were unlikely to be more
intensive than managing the same intercurrent illnesses for people
without dementia, as the recommended approaches for treating pain
and falls were similar to those for people without dementia. Therefore, it
would be unlikely to add a significant additional cost to the NHS.
Quality of evidence The committee acknowledged that the lack of longitudinal evidence in
the studies meant that it was difficult to provide specific
recommendations on the frequency of pain assessment. All the included
studies involved regular monitoring using a prescribed protocol, but this
may not be either practical or appropriate for all people with dementia in
routine practice. It was acknowledged that there was a need to balance
harms and benefits and, for this reason, the committee agreed it would
389

not be appropriate to give specific recommendations about assessment

frequency. Generally the committee agreed that reassessment would be
appropriate where there was a concern that a person may continue to
be in pain or to be experiencing a new episode of pain, as this would be
the situation where a change of pain management could be relevant.
Due to the lack of relevant evidence related to intercurrent delirium or
urinary tract infection, the committee agreed that it was unable to draft
recommendations in these areas.
Other considerations The committee recognised that, in general, there was a lack of evidence
appropriate to address these review questions. However, it noted that
there were several NIHR-funded trials currently in development that
might help mitigate this lack of evidence in the future. It therefore
agreed that it was not necessary to address the current evidence gap by
making specific recommendations for future research above that
already being undertaken and/or considered. An exception to this was in
the area of long term recovery from delirium superimposed on
dementia, where the committee agreed it was appropriate to
recommend future research, as there is currently a lack of evidence on
how best to help people with dementia to recover to their baseline
cognitive status after an acute episode of delirium.
Pain
115. Consider using a structured observational pain assessment tool:

 alongside self-reported pain and standard clinical assessment for people
living with moderate to severe dementia
 alongside standard clinical assessment for people living with dementia
who are unable to self-report pain.
116. For people living with dementia who are in pain, consider using a stepwise
treatment protocol that balances pain management and potential adverse events.
117. Repeat pain assessments for people living with dementia:

 who seem to be in pain
 who show signs of behavioural changes that may be caused by pain
 after any pain management intervention.
Falls
118. For guidance on managing the risk of falling for people living with dementia (in
community and inpatient settings), see the NICE guideline on falls in older people.
When using this guideline:
 take account of the additional support people living with dementia may
need to participate effectively
 be aware that multifactorial falls interventions may not be suitable for a
person living with severe dementia.
390

20. What are the most clinically and cost-effective non-pharmacological interventions
for helping the long-term recovery of people with delirium superimposed on
dementia?
appendix L.
391

18.2 Management strategies for people living with dementia and

co-existing physical long term conditions
Review question
 What are the optimal management strategies (including treatments) for people living with
dementia with co-existing physical long term conditions?
18.2.1 Introduction
The aim of this review question was to identify the most effective interventions/ strategies to
manage medical comorbidities (for example diabetes, cardiovascular disease etc.) in people
living with dementia and to consider if the most effective interventions and strategies used for
treating medical comorbidities in people living with dementia are different from the
interventions and strategies used for people with medical comorbidities who do not have
dementia.
The focus of the question was to consider strategies that reduce the progression of co-
existing conditions and to specifically focus on people living with dementia and the following
co-existing long term conditions:
 Continence
 Recurrent falls (rehabilitation)
 Hypertension
 Diabetes
 Risk of Cardiovascular disease (anticoagulation)
 Sensory impairment
of the review protocols, see Appendix C.
Table 101: Review summary: management of physical health comorbidities

Population  Studies including people (aged 40 years and over) with a diagnosis
of dementia and living with a co-existing long term condition
Interventions  Pharmacological interventions/ self-care strategies/ observational or
monitoring strategies specific to people living with dementia and a
coexisting long term condition
Comparator  Pharmacological interventions/ self- care strategies, monitoring or
observational strategies for people living with a coexisting long term
condition but not specific to people living with dementia
 Standard care
Outcomes  Clinical progression of comorbidity and associated symptoms.
ability
 Intervention related problems such as potentially avoidable hospital
potentially avoidable adverse effects
satisfaction of person living with dementia and their carers
392

A systematic search identified 9,306 references filtered by randomised controlled trials. The
references were screened on their titles and abstracts and 100 references were ordered for
full text. The included studies and references of any eligible systematic reviews were also
screened and full text copies of any appropriate studies were ordered, giving a total of 119
full-text studies. 112 papers were subsequently excluded because they did not fit the
inclusion criteria (see Appendix F for a detailed list of excluded studies and reasons for their
exclusion). Seven randomised controlled trials were included in the evidence review.
Four studies were included for strategies to treat risk of cardiovascular disease (2 studies
focused upon strategies to treat hypertension, 1 of these trials compared pharmacological
treatment with a PPAR-ˠ agonist (telmisartan) to pharmacological treatment with a CCB
(amlodipine), the other compared the use of relative-measured blood pressure and
automated blood pressure measurement). One study was included for cerebrovascular
lesions (comparing a standardised protocol approach, involving both pharmacological and
behavioural strategies with a standard care approach which was non-specific to vascular
care) and 1 study for diabetes (comparing pharmacological treatment with a PPAR-ˠ agonist-
pioglitazone to no treatment).
Three studies were included for treating incontinence. All 3 trials compared behavioural
strategies.
One additional study was included from rerun searches conducted at the end of the guideline
an was on sensory impairment, comparing active hearing aids to placebo hearing aids in
people living with Alzheimer’s disease and moderate age related hearing loss.
No studies were identified as relevant for falls rehabilitation.
A summary of the characteristics of the included studies is provided in Table 101. Data from
the included studies were extracted into evidence tables. See Appendix E for the full
evidence tables, and for the full GRADE profiles see Appendix G. References for the
393


Study reference Study design Study population Intervention & Relevant outcomes Comments
comparator
Dementia and cardiovascular risk (hypertension, CV risk factors, diabetes)
Hypertension
Kume (2011) Randomised 20 patients with mild dementia (CDR=1) Participants received  Mean difference in Location: Japan
open-label trial and hypertension (NINCDS-ADRDA either telmisartan or Blood pressure Follow up: 6 months
criteria) amlodipine measurements
Plichart (2013) Randomised 66 outpatients diagnosed with dementia Participants received  Mean difference in Location: France
open (based on DSM-IV criteria) and either relative measured Blood pressure Follow up 3 days
comparative hypertension (BP ≥140/90mmHg at two or home blood pressure measurements
study more occasions) measurement or  Correlation
automated blood pressure between ABPM or
measurement r-HBPM
Vascular disease
Richard (2009) Randomised 130 patients with probable Alzheimer’s Intervention:  Change in Location:
multi centre disease (according to CEMDE and white Multi component vascular comorbidity Netherlands
controlled trial matter lesions of vascular origin care (for outcome 2 year follow up
hypercholesterolemia and  Clinical outcomes
hypertension) including cognitive,
involving pharma functional
treatment 8 to 100 mg behavioural ability
acetylsalicylic acid;  Adverse events
50 mg pyridoxine; 0.5 mg
folic acid per day and a
stepped protocol of
dietary, exercise and
lifestyle changes
Control: Standard care
Diabetes
Sato (2011) Randomised 42 patients with mild Alzheimer’s disease Intervention: 15-30mg  rCBF change from Location Japan
open label (CDR 0.5 or 1) and Type II Diabetes pioglitazone baseline 6 month follow up
controlled trial Control:  Adverse events
394

Study reference Study design Study population Intervention & Relevant outcomes Comments
comparator
(diagnosis based on NINCDS-ADRDA No treatment
criteria)
Continence
Ostaskiewicz Systematic RCTs and quasi RCTs of timed voiding 3 (1 RCT, 1 mixed design  Incontinence rates Location:
(2010) review interventions for management of and 1 controlled cross Multinational
incontinence in people living with dementia over study)
Jirovec 2001 Randomised Carers of 118 patients with cognitive Intervention:  Mean change in Follow up 6 months
controlled trial impairment (based on SPMSQ) and Carers taught an number of
urinary incontinence individualised toileting incontinent
schedule program episodes
Control:  Correlation
No precise details between baseline
and follow up
incontinence
frequency
Engberg (2002) Randomised 19 cognitively impaired older adults Intervention:  Change in daytime After 8 weeks
controlled cross- (MMSE<24) with urinary incontinence Prompted voiding initially incontinent participants in
over study every 2 hours, episodes control group were
Control:  Carer satisfaction crossed over to
with prompted treatment group
No precise details
voiding intervention
Age related hearing loss
Adrait (2017) Randomised 48 people living with probable Alzheimer’s Intervention: Active  Clinical outcomes Location : France
controlled semi- disease (NINCDS-ADRDA) and age- hearing aids including cognitive, After 6 months the
crossover study related hearing loss Control: Placebo hearing functional, participants in
aids behavioural ability control group had
hearing aids
activated
Abbreviations NINCDS and ADRDA = National Institute of communicative disorders and Stroke and the Alzheimer’s disease and related disorders association criteria;
BS = Barthel Index score; DSM- IV Diagnostic & statistical manual 4th ed; MMSE= mini mental state examination; CDR= clinical dementia rating; CEMDA= Cambridge
Examination for Mental Disorders in Elderly; SPMSQ= short portable mental status questionnaire
395

A total of 3,078 citations was returned from the search for this question. Following review of
titles and abstracts, the full text of 1 study was retrieved for detailed consideration, but it did
not meet the inclusion criteria. Therefore, no relevant cost–utility analyses were identified for
this question. For full details of the literature review, please see Appendix D.
18.2.4.1 Hypertension
Very low-quality evidence from 1 RCT of 20 participants with Alzheimer’s disease and
hypertension found the mean difference in systolic blood pressure, diastolic blood pressure
and pulse rate at 6 months was not significantly different for participants receiving
telmisartan, compared with participants receiving amlodipine. The mean difference in
cognitive and functional outcomes (MMSE scores, ADAS-Cog and WMS-R [logical memory
scores]) at 6 months were also not significantly different for participants receiving telmisartan,
compared with participants receiving amlodipine.
Low to moderate-quality evidence from 1 randomised comparative crossover study of 60

participants with cognitive impairment and hypertension found the mean systolic blood
pressure readings were significantly higher when blood pressure readings were taken by a
relative compared with 24 hour readings or daytime only readings taken from an ambulatory
blood pressure measurement device. The mean diastolic blood pressure readings were not
significantly different for readings taken by a relative compared with 24 hour or daytime
readings taken from an ambulatory blood pressure measurement device.
18.2.4.2 Cardiovascular risk
Moderate -to high-quality evidence from 1 RCT with 94 participants with Alzheimer’s disease
and cerebrovascular lesions found the mean difference in change over 2 years for systolic
blood pressure readings, diastolic blood pressure readings, HBA1c, and HDL cholesterol did
not differ significantly for participants receiving a multicomponent vascular care programme
to participants receiving standard care. However, after 2 years, the total cholesterol readings
and LDL cholesterol readings had significantly reduced for participants receiving the vascular
care programme compared with those receiving standard care.
The change in cognitive outcomes (as measured by MMSE, IDDAD and revised MBPC) did
not significantly differ after 2 years for participants receiving a multicomponent vascular care
programme to participants receiving standard care.
18.2.4.3 Diabetes
Low-quality evidence from 1 RCT with 42 participants with mild Alzheimer’s disease and type
II diabetes mellitus found no difference in fasting plasma glucose, HBA1c, or fasting insulin
levels at 6 months between participants receiving pioglitazone compared with those who did
not receive a drug therapy. The clinical outcomes (as measured by MMSE, ADAS-Cog and
WMS-R logical memory) at 6 months did not differ for participants receiving pioglitazone
compared with those who did not receive a drug therapy.
396

18.2.4.4 Incontinence
Low-quality evidence from 1 RCT of 74 participants living at home with cognitive impairment
and incontinence found the number of participants showing decreased incontinence at 6
months did not differ significantly for those that received an individual scheduled toileting
programme compared with those who did not receive the individualised programme. After 6
months, there was no significant difference in mean incontinence frequency for those
participants that received an individual scheduled toileting programme compared with those
who did not receive the individualised programme. There was no difference in cognitive
ability (as measured by MMSE) or mobility (as measured by a composite mobility score) at 6
months for those that received an individual scheduled toileting programme compared with
those who did not receive the individualised programme.
Very low-quality from 1 RCT of 19 participants living at home with cognitive impairment and
incontinence found no significant difference in the reduction of incontinent episodes,
reduction in daytime incontinent episodes per day after 8 weeks for participants receiving
prompted voiding compared with a control group who did not receive prompted voiding. The
mean percentage reduction in daytime wet and day and night time wet after 8 weeks was not
significantly different for participants receiving prompted voiding compared with a control
group who did not receive prompted voiding.
Very low- to moderate-quality evidence from 1 RCT of 174 participants with urinary
incontinence and dementia found that after 2 months, the number of participants showing
reductions in incidence of daytime incontinence did not differ significantly for participants
receiving timed voiding compared with those receiving usual care. The number of
participants with reductions in incidence of night time incontinence was significantly greater
for participants receiving timed voiding compared with usual care. After 2 months the number
of participants whose pad test indicated reductions in the volume of incontinence did not
significantly differ for participants receiving timed voiding compared with usual care.
18.2.4.5 Age related hearing loss
Low- to moderate-quality evidence from 1 RCT of 36 participants with age related hearing
loss and Alzheimer’s disease could not differentiate activities of daily living (ADCS-ADL);
behavioural and neuropsychological symptoms (NPI); Quality of life (ADRQL) or carer
burden (ZBI) at 6 months follow up for people using an active hearing aid, compared with
those using a placebo hearing aid, but did find an improvement in quality of life at 12 months.

Relative value of different The committee agreed that the evaluation of any intervention for this
outcomes question relied on 2 key outcomes. First, how well the intervention
manages the comorbidity itself (where the primary outcome measures
would be the same as for evaluations of that condition in people without
dementia), and secondly whether this treatment leads to any
improvement or worsening of the person’s dementia.
It was noted that there was a considerable body of evidence about the
impact of managing cardiovascular risk factors on the progression of
dementia (which is covered elsewhere in this guideline through a
397

question on modifying risk factors), but most of this evidence was not
eligible for inclusion here, as only outcomes relating to the person’s
dementia were reported, without any data on the impact on the co-
morbidity itself.
Trade-off between Diabetes
benefits and harms The committee agreed that intensive interventions to improve diabetic
control may not always be appropriate in people with severe dementia,
as the benefits may not be sufficient to justify the distress/other harms
these interventions can cause. It was noted that the diabetes guideline
has a recommendation that it may be appropriate to relax the target
HbA1c level for people with significant comorbidities in whom intensive
management would not be appropriate. The committee agreed that
severe dementia was a condition which would often meet this threshold,
and therefore it was felt appropriate to cross-refer to this
recommendation.
It was agreed, however, that whilst this recommendation established the
principle that it is appropriate to reduce the intensity of treatment in
some individuals, there was no evidence to establish which individuals
were likely to benefit from the withdrawal of treatment, and what the
most appropriate point in the person’s dementia trajectory to withdraw
treatment would be. Therefore, the committee agreed there would be
value in additional research specifically looking at what the impact of the
withdrawal of intensive treatments for diabetic control would be in
people with severe dementia, and this would include attempting to
identify subgroups of people in whom withdrawal is or is not beneficial.
Cardiovascular disease
The committee agreed that none of the evidence identified was
sufficient on which to make recommendations. However, as with
diabetes above, it was felt to be likely there were people with severe
dementia in whom it was appropriate to withdraw treatments as the
potential benefits would not be sufficient to justify the distress/other
harms cause. In the absence of evidence no recommendations were
made on this topic, but a research recommendation was made that trials
should be conducted looking at the impact of the withdrawal of
preventative vascular interventions for people with severe dementia.
Falls
No evidence was identified looking at the most appropriate interventions
for rehabilitation following recurrent falls. Therefore, the committee
decided not to add to the recommendations around falls made in the
review question around the management of intercurrent illness.
Incontinence
The committee discussed the NICE guideline on urinary incontinence in
neurological disease. This guideline recommends the use of
antimuscarinic drugs to treat overactive bladder, but also mention they
may be associated with confusion. The committee agreed it was
important to highlight this is a particular concern in people with
dementia, and that therefore antimuscarinic drugs should be avoided if
possible in this population.
There is additionally a NICE technology approval on mirabegron, which
says it is an appropriate treatment for overactive bladder in people for
whom the side effects of antimuscarinics would be unacceptable. The
committee highlighted that people with dementia would fulfil this criteria,
and therefore should be eligible for mirabegron treatment.
Finally, the NICE guideline on faecal incontinence contains a specific
recommendation around management in people with cognitive issues,
and again it was felt appropriate to cross-refer to this recommendation.
398

In the absence of any high-quality evidence around interventions to

reduce incontinence levels in people living with dementia, the committee
felt it appropriate to make a research recommendation around this topic.
Sensory impairment
No evidence was identified looking at the most appropriate interventions
for managing sensory impairment in people living with dementia. The
committee noted the NICE guideline on hearing loss contained
recommendations on assessments for hearing loss in people with
suspected or diagnosed dementia, and agreed it was appropriate to
cross-refer to these recommendations. The committee also agreed it
was appropriate to include an equivalent recommendation around eye
tests for people living with dementia, as this was also an important issue
where the presences of visual impairment is higher in people living with
dementia than those without..
Trade-off between net No health economic evidence was identified for this review question and
health benefits and it was not prioritised for economic modelling. The committee agreed that
resource use since the recommendations made either referred to recommendations
from technology appraisals and other NICE guidance (and therefore any
economic considerations should have been considered during those
earlier evaluations) or were around potential reductions in the intensity
of treatment (and therefore likely to be cost-saving) and that it was thus
unlikely any of the recommendations made would lead to a substantial
increase in resource use.
Quality of evidence The committee agreed that the evidence presented was of a generally
low quality, and did not provide sufficiently robust evidence to make
recommendations. It was noted that the evidence search for this
question was limited to trials in a population of people with dementia,
and that another source of evidence was in published guidelines looking
at populations with comorbidities of interest, where people with
dementia/cognitive impairment may have been considered as a
subgroup. Specific consideration was given to NICE guidance around
diabetes and incontinence, and potentially relevant recommendations
from these guidelines which could be cross-referred to were identified.
The recommendations arising from these considerations are discussed
in the trade-off between benefits and harms section above.
Other considerations The committee noted there was often concern that people living with
dementia were often not offered equitable access to treatment for
comorbidities they may have, and agreed it was important to make a
statement that people with dementia should be offered equivalent
access to treatments for comorbidities. As part of this, it agreed that it
was appropriate to cross-refer to the NICE guidance on manging
multimorbidity, and on older people with social care needs and multiple
long-term conditions.
119. Ensure that people living with dementia have equivalent access to diagnosis,
treatment and care services for comorbidities to people who do not have
dementia. For more guidance on assessing and managing multimorbidity, see the
NICE guidelines on multimorbidity and older people with social care needs and
multiple long-term conditions.
120. For more guidance on providing support for older adults with learning
disabilities, see the NICE guideline on care and support of people growing older
with learning disabilities.
399

121. For guidance on setting HbA1c targets for people living with severe dementia
who have type 2 diabetes, see recommendation 1.6.9 in the NICE guideline on type
2 diabetes in adults.
122. For guidance on pharmacological treatment of overactive bladder, see the NICE
technology appraisal on mirabegron for treating symptoms of overactive bladder.
123. For guidance on treating faecal incontinence, see recommendations 1.7.2 and
1.7.8 in the NICE guideline on faecal incontinence.
Sensory impairment (such as sight loss, hearing loss, or both)
124. For guidance on hearing assessments for people with suspected or diagnosed
dementia, see adults with suspected dementia in the NICE guideline on hearing
loss
125. Encourage people living with dementia to have eye tests every 2 years.
Consider referring people who cannot organise appointments themselves.
21. What is the effectiveness of interventions to improve faecal and urinary

continence in people living with dementia?
treatments for the primary and secondary prevention of vascular outcomes in
people with severe dementia?
intensive treatments for diabetic control in people with severe dementia?
see appendix L.
400

18.3 Managing mental health conditions alongside dementia

Review question
 What are the optimal management strategies (including treatments) for people with
dementia and an enduring mental health condition?
18.3.1 Introduction
The aim of these review questions was to identify the most effective methods for managing
pre-existing mental health comorbidities in people with dementia. This is distinct from the
question of managing dementia-emergent mental health problems, which is addressed
elsewhere in this guideline. All mental health conditions were eligible for inclusion but
particular emphasis was placed on the following:
 Anxiety
 Depression
 Psychotic disorders
 Substance use disorders
 Personality disorder
The review identified studies that fulfilled the conditions specified in Table 102. For full
details of the review protocol, see Appendix C.
Table 102: Review summary: management of mental health comorbidities

Population  People (aged 40 years and over) with a diagnosis of dementia and
living with a psychiatric comorbidity/multimorbidity
Interventions  Pharmacological interventions/non pharmacological
interventions/self-care strategies/observational or monitoring
strategies specific to people living with dementia and a comorbid
psychiatric illness.
Comparator  Pharmacological interventions/ non pharmacological/self- care
strategies/monitoring or observational strategies for people living
with a comorbid psychiatric illness but not specific to people living
with dementia
 Standard care
Outcomes  Clinical progression of mental health condition and associated
symptoms.
ability
 Intervention related problems such as potentially avoidable hospital
potentially avoidable adverse effects
satisfaction of person living with dementia and their carers
 Health related quality of life of person living with dementia and their
informal carers
401

and abstracts and 80 references were ordered for full text review, with an additional 5 papers
identified through reference screening of included articles. All 85 papers were subsequently
excluded because they did not fit the inclusion criteria (see Appendix F for a detailed list of
excluded studies and reasons for their exclusion). In particular, 21 papers did provide
information on people with dementia and one or more of depression, anxiety and psychosis,
but in all these papers the onset of these symptoms came after the diagnosis of dementia,
and therefore these papers were included in the section on illness-emergent non-cognitive
symptoms.
No relevant studies were identified for this review question.
No evidence was identified for this review question.

Relative value of different The committee agreed that the evaluation of any intervention for this
outcomes question relied on 2 key outcomes. First, how well the intervention
manages the mental health comorbidity itself (where the primary
outcome measures would be the same as for evaluations of that
condition in people without dementia), and secondly whether this
treatment leads to any improvement or worsening of the person’s
dementia. Some of the treatments for specified mental health
comorbidities (e.g. antipsychotics) are known to cause harms in people
with dementia, and therefore identified trials of the effectiveness of
these medicines in managing mental health comorbidities in people with
dementia would be valuable, as they would enable to trade-offs
between these benefits and harms to be appropriately quantified.
Trade-off between The committee agreed that the people in this group were likely to be
benefits and harms complex to manage, because of the two way interaction between the
two conditions. First, the presence of many mental health comorbidities
has the potential to make dementia more difficult both to identify and
manage and, conversely, the development/progression of dementia is
likely to make management of the underlying mental health comorbidity
more difficult.
The committee discussed whether, in the absence of evidence, there
were any useful consensus recommendations that could be made for
this group. However, it agreed that any such recommendations would
either need to be vague (to capture the entire spectrum) and therefore
402

not provide useful guidance, or to be very specific and risk being

inappropriate for many individuals within that group. The committee
therefore agreed that no recommendations about management of
people with dementia and long-term mental health comorbidities could
be made.
The committee agreed however, that future research in this area would
be valuable. It has been recognised for a long time that this is a
complex patient group to manage, and yet there are still no randomised
controlled trials in this population. It therefore recommended that trials
should be conducted, looking at the optimum management strategies
for people with an enduring mental health problem who go on to
develop dementia.
Trade-off between net No health economic evidence was identified for this review question and
health benefits and it was not prioritised for economic modelling. The committee agreed that
resource use the people specified in this review question were likely to be complex
and therefore potentially expensive to manage. This meant that
randomised controlled trials in this area are likely to be valuable, as they
have the potential not only to improve patient care, but also to provide
cost savings if more efficient and co-ordinated ways can be found to
manage care for these individuals.
Quality of evidence No evidence was identified for this review question. The committee
agreed this is likely to be because of the relatively small numbers of
people available for such trials (whilst a considerable number of people
do have dementia and a mental health comorbidity, the number of
different mental health comorbidities means there is not a large number
in any individual group), and the complexity in conducting such trials. It
also agreed that the population of people who develop a mental health
comorbidity after a diagnosis of dementia is a larger group, but this is
covered in a separate question in this guideline.
Other considerations The committee noted that the prevalence of mental health problems is
higher in people with learning disabilities than in the general population,
and that there are particular challenges associated with managing
mental health problems in people with learning disabilities and mental
health issues who develop dementia. It was agreed that it would be
highly desirable to be able to provide advice on how to manage
dementia in this population, but the lack of relevant evidence did not
make this possible. It was reinforced that this was an important
subgroup for consideration throughout the guideline, and any evidence
found in other questions relating to this population should be given a
high priority for consideration.
No recommendations were made
24. What are the optimal management strategies for people with enduring mental
health problems (including schizophrenia) who subsequently develop dementia?
appendix L.
403
Palliative care: care towards and at the end of life
19 Palliative care: care towards and at the end

of life
The aim of palliative care is to achieve ‘the best quality of life for patients and their families’
(WHO 1990). Dementia is a life limiting illness that will often exist alongside other chronic
conditions, and therefore most people with dementia will benefit from palliative care. The
number of people who die with dementia is high and likely to substantially increase in the
future. Dementia care is therefore an increasingly important strategic consideration for
palliative and end of life care providers, and policies at national and international level
recognise this (DoH End of life Strategy 2008)
Palliative care for people with dementia is compassionate, person centred and holistic;
encompassing physical and psychological issues and social and spiritual aspects of their life.
It requires attention to mental capacity and advance care planning to help people plan for the
future, in order that they can live well and die in their usual place of residence. It recognises
the central role of informal carers; also attending to their needs to enable them to play the
vital role of supporting the person living with dementia. Care is coordinated and draws on
support from professionals and carers who are well trained and highly skilled (Hospice
enabled dementia care 2015).
Quality of care and access to palliative care for people with dementia is variable. This may in
part be due to the challenges around recognition of end of life and knowing when to adopt a
palliative approach. The GMC defines End of life care as “Patients are ‘approaching the end
of life’ when they are likely to die within the next 12 months”. One of the difficulties caring for
people with a diagnosis of dementia is identifying the last 12 months of life. This means
conversations about end of life do not always happen early enough.
The previous NICE dementia guidance 2006 highlighted the lack of evidence to support
arguments concerning what constitutes good quality palliative care in dementia and
highlighted the need for further research (Bayer 2006). In 2008, the NHS End of Life Strategy
became the building blocks for the development of palliative care in the UK. It identified that
we needed to prepare for larger numbers of people dying and that not everybody received
high quality care. The NICE Quality Standards for End of Life Care in Adults (published 2011
and modified in 2013) set clear standards for quality of care.
Variation in quality of care across care settings and for different diagnoses remained a point
of national debate and led to further reports and guidance including: The Leadership
Alliances “One Chance to get it right” report (2014), Ambitions for End of life care Document
(2015), Choice in End Of Life Care (2015), Health Select committee inquiry into end of life
care (2015). These documents were a response to the variations and an attempt to build
upon the national strategy by introducing frameworks for service development at a local
level.
The European Association for Palliative Care (EAPC) aimed to produce the first definition of
palliative care in dementia based on evidence and consensus from a range of experts (EAPC
White Paper 2013). It proposed the differing trajectory of dementia from other life limiting
illnesses created the need for a dementia specific definition and model of palliative care.

404
19.1 Palliative care

Review question
 What models of palliative care are effective for people with dementia?
19.1.1 Introduction
This question considered both quantitative and qualitative evidence on effective models of
palliative care for people with dementia. The quantitative part of this review was undertaken
as a collaboration between the NICE Internal Clinical Guidelines Team and the Cochrane
Dementia and Cognitive Impairment Group.

Phenomena of interest Aspects of palliative care approaches impacting on people living with
dementia, which may include:
 Meeting physical care needs
 Psychological, social and spiritual care needs
 Planning
 Communication

Interventions  Defined models of palliative care
 Enteral tube feeding interventions
Comparator  Alternative models of palliative care
 No enteral tube feeding
 Standard care
Outcomes  Improvements in care
 Nutritional status
 Pain
 Patient satisfaction and quality of life
 Carer burden, satisfaction and quality of life
 Adverse events
Randomised and non-randomised controlled trials, controlled before-and-after studies,

interrupted time series, qualitative studies and systematic review of either randomised
controlled trials or qualitative studies were included if they explored the effectiveness of
different models of palliative care for people living with dementia. Qualitative studies needed
to report the views of either people living with dementia or their carers, and match the criteria
given in Table 104. For full details of the review protocol, see Appendix C.
Quantitative studies needed to match the criteria given in Table 103. Papers were excluded if
they:
 did not include the views people living with dementia or their carers

405
and abstracts, and the full texts of 34 references that were potentially relevant to the review
question were requested. Eight qualitative studies exploring the preferences of people living
with dementia and their carers about palliative care were included in the review. The included
studies are summarised in Table 105. For the full evidence tables and full CERQual profiles
please see Appendix E and Appendix G. The 26 excluded papers, with reasons for
exclusion, are presented in Appendix F. References for the included studies are given in
appendix I.
Table 105: Summary of included studies – qualitiative evidence

Study
details Study population Topic/methods Outcomes
Crowther Bereaved carers Unstructured Exploration of the experiences of
(2013) interviews bereaved carers of people living with
dementia in the last year of life and time
surrounding death and how the
presence and lack of compassion,
kindness and humanity influenced the
experience of care.
Davies Professionals Semi-structured This study explored perceived barriers
(2014) involved in delivering interviews to the delivery of high-quality palliative
palliative care care for people living with dementia.
Dening Informal carers and Semi-structured Perceived and real barriers that prevent
(2012) staff interviews and people living with dementia and their
focus groups carers receiving end-of-life care of
acceptable quality
Grisaffi GPs Semi-structured GP’s views and experiences of end-of-
(2010) interviews life care for their patients living with
dementia.
Lamahewa Carers of people Topic: decision The opinions of carers of people living
(2017) living with dementia making at the end with dementia
of life for people
with dementia
Method of data
collection: focus
groups and semi-
structured
interviews
Lawrence Care professionals, Structured An exploration of how effective good-
(2011) bereaved family interviews quality end-of-life care might be
carers. delivered for people living with dementia
across care settings
Moore Family carers of Topic: end-of-life The opinions of familial carers of people
(2017) people living with care living with dementia
dementia Method of data
collection:
interviews

406
Study
details Study population Topic/methods Outcomes
Trelar Bereaved carers Mixed Retrospective evaluation of palliative
(2009) methodology care given to the person living with
dementia
19.1.2.2 Quantitative evidence (palliative care interventions)
The studies included in this review were identified through a collaboration with a Cochrane
review on palliative care in people living with advanced dementia. A total of 1,535 unique
citations were identified through a systematic search, of which 21 were retrieved for full-text
appraisal. Two of these studies ultimately met the criteria for inclusion, with the remaining 19
studies excluded, with reasons for exclusion given in Appendix F. An additional RCT was
identified after the final search date of the Cochrane review which was also included in the
analysis, giving 3 included RCTs in total. The included studies are summarised in Table 106.
For the full evidence tables and full GRADE profiles please see Appendix E and Appendix G.
Table 106: Summary of included studies – RCT evidence

Ahronheim (2000) 99 participants with Specialist palliative Palliative care plans,
advanced dementia, care team versus deaths in hospital,
staged as FAST 6d or usual care hospital admissions, use
greater, hospitalised for an of procedures, decisions
acute illness to forgo procedures
Hanson (2011) 256 dyads of a resident Decision aid on Decisional conflict,
with advanced dementia feeding options frequency of feeding
and feeding problems and versus no decision discussions, assisted
their surrogate aid feeding treatments
Hanson (2017) 302 dyads of a resident Goals of Care Quality of
with advanced dementia palliative care communication,
and feeding problems and intervention versus concordance of decision
their family decision maker usual care making, satisfaction with
care
19.1.2.3 Quantitative evidence (enteral tube feeding)
A total of 452 unique citations were identified through a systematic search, of which 16 were
retrieved for full-text appraisal. Seven of these studies ultimately met the criteria for inclusion,
with the remaining 9 studies excluded, with reasons for exclusion given in Appendix F. The
seven included studies were all non-randomised observational studies that compared
outcomes between people who did and did not receive enteral tube feeding.
Goldfeld et al. (2013) compared the cost-effectiveness of no do-not-hospitalise (DNH) order

(meaning that patients can be hospitalised, as per current practice) with a DNH order. This
analysis also compared hospitalisation for suspected pneumonia with no hospitalisation for
suspected pneumonia in patients with advanced dementia who were nursing home (NH)
residents. The authors conducted a cost–utility analysis alongside the Choices, Attitudes,
and Strategies for Care of Advanced Dementia at the End of Life (CASCADE) study (n=268),
a prospective cohort study conducted between 2003 and 2006, and collected Medicare
expenditure and Health Utility Index Mark 2 (HUI2) data. Primary outcome measures were

407
quality-adjusted life days (QALD) and Medicare expenditures over 15 months. For further
Service-use data included direct costs associated with hospital admissions, emergency
department visits, physician and other professional visits in the NH, hospice enrolment, and
skilled nursing facility admissions after hospitalisation. To estimate QALYs, the study authors
developed and validated a method that mapped the Symptom Management at the End-of-
Life in Dementia Scale and Comfort Assessment in Dying with Dementia Scale to the Health
Utility Index Mark 2 (HUI2). The authors employed bootstrap methods to explore the
uncertainty of the incremental expenditure and quality-adjusted survival estimates. The mean
results with informal care costs excluded are presented in Table 107.
Table 107: Base-case cost–utility results – Goldfeld et al (2013)

Incremental
Treatment Costs Effects ICER
Usual hospitalisation practice vs DNH $5,972 +3.7 QALD $1,614 /QALD
Order $589,130 /QALY
Hospitalisation for Suspected $3,697 -9.7 QALD Dominated
Pneumonia vs No Hospitalisation for
Suspected Pneumonia
Do-not-hospitalise -v- usual practice
DNH order NR NR
$5,972 0.010 $589,130
Usual practice (no DNH order) NR NR QALYs /QALY
Do-not-hospitalise for pneumonia –v- usual practice
No hospitalisation for suspected pneumonia NR NR
$3,697 0.027 Dominated
Usual practice (hospitalisation) NR NR QALYs
Current hospitalisation practice, compared with a DNH order resulted in additional average
costs of $5,972 and 3.7 quality-adjusted life-days per patient, with a resulting ICER of
$589,130 per QALY. Probabilistic analysis showed that, if QALYs are valued at less than
$300,000 each, the proportion of incremental net benefits that are positive – indicating usual
hospitalisation practice would be the preferred option – is no more than 20%.
Hospitalisation for suspected pneumonia compared with no hospitalisation for suspected

pneumonia resulted in an additional average cost of $3,697 and a loss of 9.7 quality-adjusted
life-days per patient, meaning hospitalisation was dominated. Probabilistic analysis showed
that, at all assumed QALY values up to $300,000, the proportion of incremental net benefits
that are positive – indicating hospitalisation would be the preferred option – was less than
10%.

408
19.1.4.1.1 Carer identified issues
The following issues around palliative care for people living with dementia were identified by
bereaved carers:
o Ensuring adequate food and fluid intake was considered paramount, but care homes
were occasionally evaluated negatively in this respect (moderate confidence)
 Going beyond task-focused care
o End-of-life care was evaluated positively if it was felt that the professionals cared about
their dying relative (moderate confidence)
o Getting to know individual’s interests, sensitivities and preferences (including food
preferences) was considered important (moderate confidence)
o Knowing the person well and having a sense of their personal and social identity was
said to enable carers and health-care professionals to make better informed best
interests decisions on behalf of a person living with dementia (high confidence)
o When healthcare professionals do not communicate with carers because of poor
communication or lack of time to involve the family, this can complicate decision
making (high confidence)
o Family carers reported often having to retell the same narrative to different health-care
professionals (high confidence)
o Carers sometimes have doubts making decisions, particularly if there was not an up-to-
date living will (high confidence)
o Carers valued continuity and receiving regular feedback about their relative’s health
condition and the progression of dementia. (moderate confidence)
o Carers were rarely informed about the dementia from diagnosis onwards through to the
palliative stages (moderate confidence)
o The unpredictable course of dementia made it very challenging for carers to prepare for
the end of life (moderate confidence)
o Carers valued timely and sensitive information provided by a knowledgeable
professional and that was reinforced in writing (moderate confidence)
o End of life (EOL) plans were not started early enough (moderate confidence)
o Some carers were satisfied with EOL care if they felt adequately informed and
involved, even when EOL care was not in accordance with advance care plans
o Carers often grieve for their relative before the person dies (moderate confidence)
o Participants discussed the failure of services to acknowledge their grief or to provide
information about obtaining support (moderate confidence)
o Despite high levels of grief, many carers felt they did not need formal support or
counselling and did not seek it. (moderate confidence)
o Carers who felt well informed about how dementia progressed, were regularly updated
on their relative’s health condition and felt involved appeared more satisfied with EOL
care. (moderate confidence)
 Planning
o The importance of advance directives and advance statements (moderate confidence)
o The importance of discussing treatment planning with families and the wider care team

409
o Family carers described how little happened routinely; they had to initiate and then
“push” for services to be provided, these were unpredictable and fragmented
 Impact of hospitalisation
o Not liking the hospital environment, and finding it an uncomfortable experience and
place to be (moderate confidence)
o Carers described how acute hospital staff struggled to provide basic care. Carers
perceived a lack of understanding, little compassion and low staffing levels (moderate
confidence)
o Lack of dignity associated with dying on an open ward (moderate confidence)
19.1.4.2 Professional identified issues
The following issues around palliative care for people living with dementia were identified by
professionals working with these people:
o Identifying and responding to the physical care needs of the person living with
dementia – often these needs were not complex, but basic needs were still not being
met (moderate confidence)
o Difficulties of pain management (moderate confidence)
o Palliative care nurses were considered skilled in identifying and managing pain in
patients with complex needs and were also sensitive to nausea and hallucinations in
people with dementia at the end of life (moderate confidence)
 Complex pathways of care
o People with advanced dementia had complex medical and social needs requiring input
from a number of agencies, but the coordination was poor (moderate confidence)
o Out of hours staff often felt unsupported and lacking in access to key information
 Going beyond task-focused care
o Risk of becoming entirely task-focused with little empathy (moderate confidence)
o Difficulties in getting to know individual’s interests, sensitivities and preferences
 Planning
o People with dementia should be given the opportunity to plan for the future (moderate
confidence)
o Whether individuals should be transferred to hospital during the final stages of their life.
Hospitalisation was a frequent occurrence despite agreement among care
professionals that this was often inappropriate (moderate confidence)
o Palliative care staff noted that professionals across care settings could be reluctant to
withdraw active treatment in the absence of explicit planning or a clear consensus
among the care team (moderate confidence)
o Problems with discontinuity of care (low confidence)
 Flexibility
o The growing number of guidelines, standards, rules and regulations placed upon
professionals in health and social care makes palliative care standardised leaving no
room for ﬂexibility (moderate confidence)
o GP’s prior knowledge of the person living with dementia is important in informing
decisions. To help the person overcome the communication and capacity issues,
relatives and carers are seen as an expert source of information regarding the person’s
wishes (low confidence)

410
o NHS Primary Care Trusts have no duty of care for people who are self-funding their
care home (moderate confidence)
 Systemisation
o Some routines are useful, such as certain meetings, pain assessment, when to stop
pursuing certain treatments (moderate confidence)
 Staff training to reduce the need to call for specialist help
o Need for training on appropriate interventions (moderate confidence)
o Many, particularly hospice, ambulance staff and district nurses acknowledged they had
received little or no training in dementia, in particular concerning communication and
managing behavioural problems (moderate confidence)
 Roles of generalists and specialists
o Some district nurses and GPs feel that palliative care should be left to specialists
 Lack of trust, fear of litigation, fear of blame and threats to speciality
o Managing both real and perceived risks can be a difficult challenge (moderate
confidence)
 Difficulties in deciding when to start end-of-life care
o The typically slow erratic decline and the indicators for starting the pathway could lead
to either a person being on it for a long time or ‘yo-yoing’ on and off as their state
fluctuated (low confidence)
19.1.4.3.1 Specialist palliative care team versus usual care
Low- to moderate-quality evidence from 1 RCT of 99 people found that people living with
dementia and hospitalised for an acute illness who were offered support by a specialist
palliative care team were more likely to have palliative care plans developed at any time and
on discharge than people offered usual care, but could not differentiate proportions with
palliative care plans during hospitalisation.
Low-quality evidence from 1 RCT of 99 people could not differentiate deaths in hospital,
hospital admissions, use of procedures or decisions to forgo procedures between people
living with dementia and hospitalised for an acute illness who were offered support by a
specialist palliative care team compared with people offered usual care.
19.1.4.4 Use of decision aid on feeding options
Low- to moderate quality evidence from 1 RCT of 90 people could not differentiate decisional
conflict, frequency of feeding discussions and assisted feeding treatments between people
living with dementia in nursing homes using a structured decision aid on feeding options and
people in nursing homes providing usual care.
19.1.4.5 Goals of care intervention versus usual care
Low- to moderate-quality evidence from 1 cluster RCT of 299 people found improved levels
of quality of communication (end of life), concordance on primary care goals, and number of
palliative care plans addressed in people offered a palliative care intervention (Goals of
Care) versus usual care, but could not differentiate levels of quality of communication
(overall), quality of communication (general), advanced care planning problems, symptom
management and satisfaction with care.

411
19.1.4.5.1 Enteral tube feeding
A high-quality systematic review identified low-quality evidence from seven observational

studies containing 1,813 people, and could not detect any differences in mortality, nutritional
status or the prevalence of pressure ulcers between people given and not given enteral tube
feeding.
One partially applicable cost–utility analysis with potentially serious limitations, based on an
observational study, found that, compared with approaches that avoid hospitalisation, more
aggressive treatment strategies leading to hospitalisation are not cost effective – and, in the
case of hospitalisation for suspected pneumonia, do more harm than good – for nursing
home residents with advanced dementia.

Relative value of different The committee agreed that, in order to recommend specific palliative
outcomes care interventions, data from quantitative studies (particularly
randomised controlled trials) would be necessary. Data from
qualitative studies would be likely to lead to more general
recommendations around important principles of care.
The committee also discussed the existing NICE guidelines on
palliative care and the last days of life, to assess their applicability for
Trade-off between The committee agreed that the general principles recommendations
benefits and harms contained in the NICE guidelines on palliative care and the last days
of life were relevant, and agreed it would be appropriate to cross-
refer to both these pieces of guidance. It agreed that all further
recommendations should then focus on areas of palliative care
particularly relevant to people living with dementia, over and above
standard palliative care. The main distinct features of people who live
with advanced dementia were identified as being:
 the difficulty in assessing pain and distress
 the extreme unpredictability of when death will occur
 people lacking capacity to make palliative and end-of-life care
decisions for themselves.
As a consequence of the extreme unpredictability of when death will
occur and associated difficulty in recognising when end-of-life care is
appropriate, the committee agreed that it was common for palliative
care and end-of-life care to be applied and then unapplied and then
re-applied for people who live with advanced dementia. As a
consequence, the commonly used definition of palliative care being
something that should be in the last year of life is often not applicable
to people who live with dementia. Instead, a definition is needed that
is more flexible and takes into account the needs of people who live
with dementia. The committee therefore made a recommendation
that palliative care should be both flexible and needs-based rather
than time based, taking in to account the disease trajectory, and in
principle be available from the point of diagnosis.
It was further noted that the person living with dementia may also
have comorbidities, and that dementia may mean that these
comorbidities can be exacerbated in an unpredictable way.
The committee noted that the Gold Standards Framework can be
very helpful as an aid to palliative care, but agreed that in the
absence of evidence it was not possible to make a direct reference to
it. The committee also agreed that a palliative care approach is the
responsibility of every health and social care professional working

412
with people living with dementia, and therefore the terms ‘non-
specialist palliative care’ or ‘specialist palliative care’ were avoided.
The committee agreed that anticipatory healthcare planning is
important because for people who live with dementia, their ‘voice’
with regards to knowing their preferences is lost much earlier.
Proactive planning can make the transition much easier, with the
principles of best interest decision making becoming important when
it is no longer possible for the person living with dementia to be
involved in decision making. It was also agreed to be appropriate to
include a recommendation on when it is appropriate to hold a best
interested discussion, and in particular the need for capacity
assessments to be conducted in all but emergency situations.
The committee agreed that one of the distinguishing features of
dementia is the increasing difficulty in communication and ‘trying to
find the person inside’. The committee agreed that structured
observational tools may help to continue person centred care. They
agreed it was important to personalise care despite any
communication issues, and therefore these tools should be used to
assess people’s “likes and dislikes, routines and personal history”.
The committee also agreed that alternative methods of
communication may be necessary to engage people. The most
common example of this would be visual communication tools, but
other methods may also be valuable (e.g. some people use touch as
a way to augment communication). The recommendations stress the
importance of “two-way communication”, as the aim should be to
involve the person rather than simply communicate to them.
The committee reflected on examples of available structured
observational tools (e.g. the ‘This is me’ tool, which is published by
the Alzheimer’s Society). However in the absence of any evidence
meeting the criteria for this review, the committee agreed that it was
not appropriate to recommend any one specific tool above the others.
Eating and drinking
The committee agreed that the evidence available on enteral feeding
did not show any benefits in people living with severe dementia, and
there was also a lack of evidence on the potential harms of these
interventions. The committee agreed it was therefore appropriate to
make a recommendation that enteral feeding should not routinely be
initiated in people living with severe dementia (though it may be used
in situations where enteral feeding is necessary for a pre-existing
condition and has been working satisfactorily). An important caveat to
this recommendation was agreed to be in people where the enteral
feeding is indicated for a potentially reversible co-morbidity. This is
particularly relevant for the rarer types of dementia, including for
younger patients who might otherwise be denied treatment (e.g.
people with frontotemporal dementia who develop bulbar problems
and consequently have difficulties with swallowing). In these
examples, the cognitive aspects of dementia are generally mild when
bulbar problems occur. In contrast, in cases of Alzheimer’s or
vascular dementia, by the time bulbar problems occur, the cognitive
aspects of dementia are usually very severe.
The committee agreed this recommendation should be restricted to
people with severe dementia as enteral feeding may well be
appropriate for people living with mild or moderate dementia if they
had a good quality of life.
The committee agreed it was important to make a positive
counterpart to this recommendation, about supporting and
encouraging oral eating and drinking for as long as possible. The
committee also noted that in situations where there are concerns
around the safety of eating and drinking, it is appropriate to consider
involving a speech and language therapist as part of this process, in

413
a similar way to how they would be involved if these concerns existed

in a person without dementia.
Hospitalisation
The committee agreed that there are clear benefits in maintaining a
familiar environment for a person living with dementia and
acknowledged the possible harms from hospitalisation. The
committee agreed it was therefore important, when considering
hospitalisation for a person living with dementia to conduct a risk
assessment that takes into account the specific difficulties people
living with dementia may experience in a hospital environment. This
assessment should ensure that people living with dementia are only
hospitalised when there are clear benefits which justify the harms that
may be caused.
The committee agreed this recommendation was necessary because
acute hospitals are not set up to accommodate behavioural and other
needs of people living with advanced dementia. In addition, people
who live with advanced dementia, once hospitalised, are more likely
to stay longer in hospital, become stuck in hospital and are more
likely to die there. People who live with dementia could be admitted
into hospital for treatment of an acute illness but maybe discharged
with a significantly worse health status, e.g. their infection may be
treated but they may no longer be able to walk.
Consideration of health The committee noted that there were often high costs associated with
benefits and resource not providing appropriate palliative care and support (in particular,
use inappropriate and expensive hospitalisations). The committee agreed
the potential increased costs associated with a palliative care
approach initiated at the point of diagnosis should be offset by a
reduction in the number of unnecessary hospital admissions.
The committee did not use the terms ‘non-specialist palliative care’ or
‘specialist palliative care’ in any recommendations. This was because
there are large cost implications to the use of specialist palliative care
teams and for which there was not strong enough evidence to
support. Further, the committee agreed a palliative care approach is
the responsibility of all health and social care professionals working
with people living with dementia.
Goldfeld (2013) provided economic evidence that hospitalising
people living with dementia who had pneumonia resulted in a
reduction of QALYs compared with no hospitalisation. The study was
from the USA and therefore not directly applicable, but was agreed to
add weight to the suggestion that hospitalisation should be avoided
where possible.
Quality of evidence The committee agreed that none of the RCTs on specific palliative
care interventions provided strong enough evidence to make
recommendations. However, they noted that a number of larger
RCTs on palliative care in dementia are currently underway, and
therefore it is likely a higher quality evidence base will be available to
support these decision in the future.
Confidence in the quality of the qualitative evidence was moderate
overall. The amount of studies and participants was fairly small, but
the themes identified all matched with the professional experience of
members of the committee.
Other considerations The committee agreed that staff training can be an important
component in the delivery of good quality palliative care, but noted
that these issues are covered in the separate question on staff
training in this guideline.
The committee agreed there was a lack of quantitative evidence on
specific palliative care interventions for people living with dementia,
and therefore agreed it was appropriate to make research

414
recommendations around the most effective models and of palliative

care, and specific palliative care interventions at the end of life.
126. From diagnosis, offer people living with dementia flexible, needs-based
palliative care that takes into account how unpredictable dementia progression
can be.
127. For people living with dementia who are approaching the end of life, use an
anticipatory healthcare planning process (see recommendation 41 on advance
care planning). Involve the person and their family members or carers (as
appropriate) as far as possible, and use the principles of best-interest decision-
making if the person cannot make decisions about their own care.
128. For standards and measures on palliative care, see the NICE quality standard
on end of life care for adults.
129. For guidance on care for people in the last days of life, see the NICE guideline
on care of dying adults.
130. For guidance, on best interests decision-making, see the NICE guideline on
decision-making and mental capacity.
131. Encourage and support people living with dementia to eat and drink, taking into
account their nutritional needs.
132. Consider involving a speech and language therapist if there are concerns about
a person’s safety when eating and drinking.
133. Do not routinely use enteral feeding in people living with severe dementia,
unless indicated for a potentially reversible comorbidity.
134. When thinking about admission to hospital for a person living with severe
dementia, carry out an assessment that balances their current medical needs with
the additional harms they may face in hospital, for example:
 disorientation
 a longer length of stay
 increased mortality
 increased morbidity on discharge
 delirium
 the effects of being in an impersonal or institutional environment.
135. When thinking about admission to hospital for a person living with dementia,
take into account:
 any advance care and support plans
 the value of keeping them in a familiar environment.
136. Consider using a structured tool to assess the likes and dislikes, routines and
personal history of a person living with dementia..

415
25. What are the most effective models of general and specialist palliative care
support to meet the needs of people with advanced dementia?
26. What are the most effective interventions to support staff to recognise advanced
dementia and develop appropriate escalation/end of life plans to facilitate care to
remain at home?
see appendix L.

416
Glossary
20 Glossary
Abbreviations used in this guideline
AAS Anticholinergic Activity Scale
ABC Anticholinergic Burden Classification
ACE Angiotensin converting enzyme
ACER Addenbrooke’s Cognitive Exam-Revised
AChE Acetyl(cholinesterase) inhibitor
ACL Anticholinergic Loading Scale
AD Alzheimer’s disease
ADAS-cog Alzheimer’s Disease Assessment Cognitive Subscale
ADCS-ADL Alzheimer’s Diseases Cooperative Study – Activities Of Daily
Living
ADCS-CGIC Alzheimer’s Disease Cooperative Study- Clinician’s Global
Impression Of Change
ADL Activities Of daily living
Apo E Apolipoprotein E
BBS Berg Balance Scale
BNT Boston Naming Test
BS Barthel Index Score
BTA Brief Test Of Attention
CADASIL Cerebral autosomal dominant arteriopathy with subcortical infarcts
and leukoencephalopathy
CAM Confusion Assessment Method
CBD Corticobasal degeneration
CDR Clinical Dementia Rating Scale
CEMDA Cambridge Examination for Mental Disorders In Elderly
CERAD Consortium to Establish a Registry for Alzheimer’s Disease
CIBIC+ Clinician’s Interview Based Impression of Change Plus Caregivers
Assessment
CJD Creutzfeldt-Jakob disease
CLP Contrastophore-linker-pharmacophore
CrAS Clinician’s Rated Anticholinergic
CSF Cerebrospinal fluid
CT Computed tomography
CTD Cognitive Test for Delirium
CVLT California Verbal Learning Test
DAD Disability Assessment for Dementia
DAD Disability Assessment Daily
DEMQOL Dementia-Specific Quality of Life
DI Delirium Index
D-KEFS Delis-Kaplan Executive Functions System Verbal Fluency Test
DLB Dementia With Lewy bodies
DNAR Do not attempt resuscitation
DRS Delirium Rating Scale
DSM Diagnostic and Statistical Manual of Mental Disorders
EDSSS Expanded Disability Status Scale

417
Glossary

EEG Electroencephalography
FAQ Functional Activities Questionnaire
FBI Frontal Behavioural Inventory
FDG-PET Fluorodeoxyglucose positron emission tomography
FST Faces Symbol Test
FTD Frontotemporal dementia
GDS Global Deterioration Scale
GHQ General Health Questionnaire
GMC General Medical Council
GPCOG General Practitioner Assessment of Cognition
HAND HIV-associated neurocognitive disorder
HDL High-density lipoprotein
HIS Hachinski Ischemic Scale
ICD-10 International Statistical Classification of Disease and Related
Health Problems
ICER Incremental cost-effectiveness ratio
IQCODE Informant Questionnaire on Cognitive Decline in the Elderly
LBCR Lewy Body Composite Risk Score
MAI Medication Appropriateness Index
MCI Mild cognitive impairment
MDRS Mattis Dementia Rating Scale
MIBG Metaiodobenzylguanidine
Mini-ACE Mini-Addenbrooke’s Cognitive Exam
MMSE Mini-Mental State Examination
MoCA Montreal Cognitive Assessment
MRI Magnetic resonance imaging
MUSIC Multiple Sclerosis Inventarium Cognition Score
NMB Net monetary benefit
NOPPAIN Non Communicative Patients Pain Assessment
NRS Numerical Rating Scale
NPV Negative predictive values
NSAIDs Nonsteroidal anti-inflammatory drugs
NPI Neuropsychiatric Inventory
OSLA Observational Scale of Level of Arousal
PACSLAC Pain Assessment Checklist for Seniors with Limited Ability to
Communicate
PAINAD Pain Assessment in Advanced Dementia
PASAT Paced Auditory Serial Addition Test
PDD Parkinson’s disease dementia
PET Positron emission tomography
PIB Pittsburgh compound B
PPA Primary progressive aphasia
PPV Positive Predictive Values
QALY Quality-adjusted life-year
QoL Quality of life

418
Glossary

RBANS Repeatable Battery for The Assessment of Neuropsychological
Status
REPOS Rotterdam Elderly Pain Observation Scale
RFFT Ruff Figure & Fluency Test
RUDAS Rowland Universal Dementia Assessment Scale
SAA Serum Anticholinergic Activity
SCEAM Sheffield Care Environment Assessment Matrix
SDMT Symbol Digit Modalities Test
SMC Subjective Memory Complaints
SMMSE Standardised Mini-Mental State Examination
SPECT Single-photon emission computed tomography
SPMSQ Short Portable Mental Status Questionnaire
SRT Spatial Recall Test
TFLS Texas Functional Living Scale
TOL Tower of London Test of Learning and Memory
TYM Test Your Memory
UPDRS Unified Parkinson’s Disease Rating Scale
WMS Wechsler Memory Scale
VaD Vascular dementia
ZBI Zarit Burden Interview

419

Demencia NICE

Uploaded by

Copyright:

Available Formats

Demencia NICE

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Demencia NICE

Uploaded by

Copyright:

Available Formats

National Institute for Health and Care

Commissioned by the National Institute for

© NICE 2018. All rights reserved. See Notice of rights.

© NICE 2018. All rights reserved. See Notice of rights.

5.1.7 Research recommendations .................................................................. 113

© NICE 2018. All rights reserved. See Notice of rights.

7.2.5 Evidence to recommendations ............................................................... 164

© NICE 2018. All rights reserved. See Notice of rights.

11.2.3 Health economic evidence ..................................................................... 206

© NICE 2018. All rights reserved. See Notice of rights.

13.2.3 Health economic evidence ..................................................................... 269

© NICE 2018. All rights reserved. See Notice of rights.

17.1.6 Recommendations ................................................................................. 373

© NICE 2018. All rights reserved. See Notice of rights.

This guideline makes evidence-based recommendations aiming to support these areas of

© NICE 2018. All rights reserved. See Notice of rights.

What does it cover?

How has it been developed?

How does it relate to statutory and non-statutory

© NICE 2018. All rights reserved. See Notice of rights.

Relevant legislation and statutory guidance

Relevant policies and non-statutory guidance

© NICE 2018. All rights reserved. See Notice of rights.

Consultant Liaison Psychiatrist, Greater Manchester Mental Health Foundation Trust

Professor of Geriatric Medicine, University of Exeter

Joanne Brady (Co-opted member)

Professor of Clinical Psychology of Ageing and Dementia, University of Exeter

Consultant Occupational Therapist, Older Adults Specialist Mental Health Services,

Paul Dunnery (Resigned November 2016)

Lay member – organisation (Alzheimer’s Society)

Consultant Psychiatrist and Lecturer in Psychiatry, St Bartholomew’s Hospital, London

Wayne Goddard (From March 2017)

Head of Strategy and Delivery Integrated Dementia Lead, Doncaster CCG

Kim Grosvenor (From March 2017)

Senior Manager – Dementia Transformation, HWLH CCG

Head of Research & Publications, Admiral Nurse, Dementia UK

Jeremy Isaacs (Co-opted member)

Kate Mitchell (Co-opted member) (Resigned November 2016)

Lay member – carer for a person living with dementia

Professor of Old Age Psychiatry, University of Cambridge

Care Home Manager, Somerset Redstone Trust

Catherine Pascoe (Co-opted member) (Resigned October 2016)

Commissioning Manager Adult Services, Hampshire County Council

Sarah Partington (Co-opted member)

Lay member – person living with dementia

GP and Professor of Primary Care and Ageing, Newcastle University

Directorate Manager – (Social Worker) Directorate Manager HMR Integrated Teams,

(Social Worker) Directorate Manager HMR Integrated Teams, Pennine Acute

Chief Executive, Sheffcare

Practice Manager, East Sussex County Council

Lay member – carer for a person living with dementia

Ruth O’Dea (Guideline committee member)

Care Home Manager, Somerset Redstone Trust

Practice Excellence Advisor, Genesis Housing Association

1.3 NICE Centre for Guidelines Team

Consultant Clinical Adviser

Sohaib Ashraf (From November 2016)

Jean Bennie (From June 2017)

Sue Ellerby (From December 2015 until March 2017)

Consultant Clinical Adviser