Support Care Cancer (2015) 23:71–78

DOI 10.1007/s00520-014-2322-0

ORIGINAL ARTICLE

The development and evaluation of an oncological palliative

care deprescribing guideline: the ‘OncPal deprescribing
guideline’
Julian Lindsay & Michael Dooley & Jennifer Martin &
Michael Fay & Alison Kearney & Mohsina Khatun &
Michael Barras

Received: 4 March 2014 / Accepted: 11 June 2014 / Published online: 1 July 2014
# Springer-Verlag Berlin Heidelberg 2014

Abstract concordance. Descriptive data on the incidence of PIMs iden-

Purpose Current data suggests that potentially inappropriate tified were also assessed.
medicines (PIMs) are common in palliative cancer patients; Results A total of 61 patients were recruited. The OncPal
however, there is a lack of criteria to assist clinicians in Deprescribing Guideline matched 94 % of 617 medicines to
identifying PIMs in these patients. The aims of this study were the expert panel with a Kappa value of 0.83 [95 % CI (0.76,
to design and validate a deprescribing guideline for palliative 0.89)] demonstrating an ‘outstanding’ concordance. Forty-
cancer patients and to undertake a descriptive analysis of the three (70 %) patients were taking at least one PIM, with
identified PIMs. 21.4 % of the total medicines assessed identified as PIMs.
Methods This prospective, non-interventional cohort study The medication-associated cost per patient/month was
consisted of four major stages: developing an ‘OncPal AUD$26.71.
Deprescribing Guideline’ from current evidence, the prospec- Conclusion A guideline to assist in the de-escalation of inap-
tive recruitment of consecutive palliative cancer inpatients propriate medications in palliative cancer patients was devel-
with an estimated <6-month prognosis, the assessment of all oped from current literature. The OncPal Deprescribing
medications to identify PIMs using both a panel of medical Guideline was successfully validated, demonstrating statisti-
experts without access to the guideline as well as a Clinical cally significant concordance with an expert panel. We found
Pharmacist independently using the OncPal Deprescribing that the incidence of PIMs was high in our patient group,
Guideline and the evaluation of the guideline by testing demonstrating the potential benefits for the OncPal
Deprescribing Guideline in clinical practice.

J. Lindsay (*) Keywords Deprescribing . De-prescribing . Drug therapy .

Royal North Shore Hospital, New South Wales, Australia, Monash
Potentially inappropriate medications . Medical futility .
University, Victoria, Australia
e-mail: [email protected] Neoplasms . Cancer

M. Dooley
Pharmacy Department, Alfred Health, Monash University, Victoria,
Introduction
Australia

J. Martin Cancer patients who have transitioned from curative intent

Princess Alexandra Hospital, The University of Queensland Medical chemotherapy or radiotherapy to palliative therapy often re-
School, Queensland, Australia
main on medications with potentially harmful effects or no
M. Fay : A. Kearney : M. Barras short-term benefit [1]. With estimated life expectancies com-
Royal Brisbane and Women’s Hospital, The University of monly less than 6 months [1], when these patients transition to
Queensland Medical School, Queensland, Australia palliative therapy, many continue to be prescribed numerous
medications for the secondary prevention of co-morbid dis-
M. Khatun
Queensland Institute of Medical Research (QIMR), Queensland, eases. When this transition is established, the focus of care
Australia should be on the patient’s quality of life, including alleviating
72 Support Care Cancer (2015) 23:71–78

suffering from cancer-related symptoms and treating short- PIMs for concordance. The flow of the study is shown in
term, acute medical illnesses [1]. In addition to a patient’s Fig. 1.
potential adverse drug effects and pill burden, there are also
financial impacts to patients, as well as the healthcare sector, Guideline development
particularly from the rising cost of preventative medicines.
Deprescribing unnecessary or ‘futile’ medications [2] in a The OncPal Deprescribing Guideline is shown in Appendix 1.
palliative cancer patient can benefit the patient by reducing the The guideline was created by investigating the current litera-
associated cost, potential adverse effects and the burden of ture for the de-escalation of medications by systematically
polypharmacy in the last months of life. However, the ratio- reviewing each medication class according to the EphMRA
nalization of drug therapy in this patient group is poorly (European Pharmaceutical Market Research Association) an-
established when compared to geriatric medicine. A re- atomical classification list [4]. As the guideline was developed
cent literature review by our study group identified that as a tool to assist in the identification of medications suitable
potentially inappropriate medicines (PIMs) are common for discontinuation in palliative cancer patients, EphMRA
in palliative cancer patients; however, there is a lack of medication classes not listed in the table have either demon-
criteria to easily identify PIMs in this patient group [3]. strated benefits in this population or the literature is lacking to
Additionally, there are no studies measuring the patient guide a decision-making process.
outcomes of ceasing PIMs in palliative cancer patients. Many guidelines currently in use in the geriatric population
We concluded that further research was warranted to were originally developed using a Delphi consensus model,
establish and implement clear guidelines for the identi- for example Beers Criteria in 1991 [5]. This study used a
fication of PIMs in palliative cancer patients. single-phase consensus model by sending the draft guideline
The primary aim of this study was to develop a specialized to three palliative care consultants, three oncology consultants
oncological palliative deprescribing guideline that assists in and three senior pharmacists for their feedback. Minor
identifying PIMs to aid in the rationalization of medicines in
this patient group. We then aimed to validate the guideline by
assessing the concordance with an expert medical panel and to
undertake a descriptive analysis of identified PIMs in this Develop the "Onc-Pal De-
Prescribing Guideline"
cohort. The purpose of developing and validating the guide-
line was to assist clinicians in targeting PIMs to improve
appropriate prescribing and rationalize medications. This will
reduce adverse drug effects, improve quality of life and reduce
medication costs. The guideline is not intended to replace or Prospective chart review of
substitute the decision to prescribe, which requires complex palliative cancer patients with a
<6-month prognosis
clinical and ethical decisions in regard to each individual
patient, but rather is intended to be a tool to highlight potential
targets for deprescribing by the treating team. It is also imper-
ative that a sensitive discussion should precede any
deprescibing as the discontinuation of a long-standing medi- Pharmacist using "OncPal
Expert Medical Panel
Deprescribing Guideline"
cation may cause patient distress. independently assess all
to independently assess
patients for PIMs
all patients for PIMs

Patients and methods

Study design Compare Panel vs. Guidelines for

highlighted PIMs in all medicines using
This prospective, non-interventional cohort study consisted of concordance statistical methods
four major stages: the development of an ‘OncPal
Deprescribing Guideline’ from current evidence, a prospec-
tive chart review of consecutive palliative cancer inpatients
with an estimated <6-month prognosis, an independent as-
sessment of all medications to identify PIMs using both a Analyse data for
secondary outcomes
panel of medical experts blinded to the guideline as well as a
pharmacist independently using the Onc-Pal Deprescribing
Guideline, and assessing the results of each patient’s identified Fig. 1 Study design
Support Care Cancer (2015) 23:71–78 73

adjustments to the draft were made when a suitable argument measured by identifying each medication as a ‘PIM’ or ‘not
was presented. PIM’ with a >90 % consistency with an expert panel.
Assessments of the individual medicine classes were evaluat-
Patient identification and data collection ed using two methods as follows: Cohen’s Kappa values were
used to make the conclusion about the interrater reliability to
The inpatient medical notes of consecutive cancer patients examine the agreement between expert medical panel and the
admitted to an Australian tertiary, 900-bed hospital from 01 guideline for the medicines associated with PIMs, and
March 2013 to 01 September 2013 were initially screened by McNemar’s chi2 values were used to assess if there was
the ward pharmacist using criteria of predicting factors for a symmetry in discordance, for individual drug classes. In ad-
life expectancy of less than 6 months. This process was based dition, an overall Cohen’s Kappa was calculated for the con-
on the validated method for prognosis assessment, initially cordance of all assessed medicines. Descriptive statistics [11]
screening the inpatients for widespread metastatic disease, were used to measure proportion of palliative cancer patients
cerebral metastases, ascites, recurrent hypercalcaemia and/or taking at least one PIM, medications and medication classes
the treating team’s estimated prognosis. Following the initial most frequently identified as PIMs suitable for discontinua-
screening, a palliative care consultant confirmed an accurate tion, patient demographics associated with PIMs such as
estimated prognosis of 6 months or less of life using current analysis of variance (ANOVA) testing of patient’s age and
validated methods including a Karnofsky performance status, total number of medicines and potential drug avoidance cost
lab results and type of malignancy [6–8]. All data was scanned of PIMs.
as a secure PDF document for the evaluation stage of the
study. The number of participants required for the study was
calculated using a sample size with a 90 % power to detect a
Kappa value significantly different to 0 [9]. We aimed for a Results
90 % power of a two-tailed test for a Kappa value of at least
0.6, estimating the guideline would have greater than 90 % Demographic
concordance with the expert panel. The calculated value was
based on the assessment of 30 medicines. As individual drug There were a total of 61 patients identified (Table 1).
classes needed to be tested, we estimated from the literature There were a total of 617 medicines assessed with a
that the most common drugs would occur in 60 % of patients median (range) of 10 (4–21) medicines per patient. The
[10]. Therefore, greater than 50 participants were required. median (range) age of patients was 66 years old (23–
93), with a relatively even distribution of genders (34
Medication assessment males and 27 females). The median (range) Karnofsky
performance status was 50 (20–80). The primary cancer
Each patient’s data was assessed to highlight PIMs by site was relatively consistent with the general population
an expert panel comprising of a radiation oncology [12, 13] shown in Table 2.
consultant, medical consultant and palliative care con-
sultant who used their knowledge and expertise to Guideline concordance
identify PIMs. Independently, blinded and in parallel,
the patient’s data was assessed by a clinical pharmacist, The proportion of the medicines that were assessed correctly
instructed to systematically use the OncPal using the OncPal Deprescribing Guideline compared to the
Deprescribing Guideline to identify PIMs in accordance expert panel was 94 % (580/617). Using Cohen’s Kappa to
with its considerations only. If there was any ambiguity measure the overall agreement, the concordance was Kappa
or clinical judgement was required, the medication was 0.83 [95 % confidence interval (0.76, 0.89)]. The strength of
deemed ‘not a PIM’.

Analysis and secondary outcomes Table 1 Patient demographics

The results from the two independent assessments were com- Total Min Max Male Female
pared (OncPal Deprescribing Guideline vs. expert panel) for Patients 61 34 27
each patient. To allow the guideline to be accurately assessed, Medicines 617
medicines were divided into 24 drug classes as well as dupli- Median medicines per patient 10 4 21
cates. Duplicates of the same drug class for an individual Median age 66 23 93
patient were documented separately to eliminate bias in the Median Karnofsky score 50 20 80
analysis. The equivalence/concordance of the guideline was
74 Support Care Cancer (2015) 23:71–78

Table 2 Distribution of different cancers among the palliative cancer patients

Primary Cancer Frequency Proportion (%) United States rate 2012 (%) [12] Australia Rate 2010 (%) [13]

Breast 2 3 5 7
Colorectal 8 13 25 10
Endocrine 1 2 0.5 5
Gynae 10 16 5 N/A
Head and neck 9 15 1.5 N/A
Lung 16 26 28 18
Prostate 5 8 5 7
Sarcoma 1 2 0.2 N/A
Skin 6 10 2 5
Urothelial 3 5 5 N/A
Total 61 100.0

this agreement is considered to be ‘outstanding’ (values of Secondary outcomes

Kappa from 0.40 to 0.59 are considered moderate agreement,
0.60 to 0.79 substantial, and >0.80 outstanding agreement) There were 132 PIMs identified by the expert panel, 70 % (43/
[14]. Individual drug classes and the associated results are 61) of patients were found to have at least one PIM, with
shown in Table 3. 21.4 % (132/617) of total medication found to be a PIM by the

Table 3 Concordance of medicines assessed by Guideline vs. Expert Panel

Medicine class Total medicines Guideline Overall Individual Individual McNemar’s

(excluding duplicates) accuracy (%) Kappa Kappa test (p value)

Aspirin/anticoagulants 15 66.7 0.83 0.40 0.063

Dyslipidaemia 19 100.0 N/A N/A
Antihypertensives 30 90.0 0.52 1.0
Oesteoporosis 4 100.0 N/A N/A
Peptic ulcer prophylaxis 38 78.9 0.45 0.008
Oral hypoglycaemics 3 100.0 N/A N/A
CAMs 20 95.0 0.64 1.0
Opioid analgesics 42 100.0
Non-opioid analgesics 39 100.0
Steroids 30 100.0
NSAIDs 5 100.0
Laxatives 35 100.0
Benzodiazepines 21 100.0
Nausea 33 100.0
Antihistamines 2 100.0
Antiepileptics/nerve pain 14 100.0
Anti-infectives 11 100.0
Antiarrhythmics 4 50.0
Insulins 1 100.0
Eye prep 4 100.0
Prostate hyperplasia 5 100.0
Neoplastic/immunomodulators 9 44.4
Psycotropic drugs 25 100.0
Inhaled respiratory 21 95.2
Other 18 77.78

CAMs complementary alternative medicines, NSAIDs non-steroidal anti-inflammatory drugs

Support Care Cancer (2015) 23:71–78 75

expert panel. Medication classes most frequently identified as (PBS) DPMQ section (Dispensed Price for Maximum
PIMs and suitable targets for discontinuation include antihy- Quantity) [15].
pertensives (44 %, 27 patients), dyslipidaemic agents (31 %,
19 patients), and CAMs (complementary alternative medi-
cines) (31 %, 19 patients). Table 4 shows the total and pro-
portion of PIMs in each different drug class with ‘duplicates’ Discussion
excluded for drugs in the same class per patient. For example,
we found that there were 30 patients taking at least one Guideline validity
antihypertensive, and 27 of these were assessed as PIMs.
We did not find any patient-specific factors associated with Deprescribing has potential benefits in patients who are taking
the incidence of PIMs. The results of the ANOVA showed that medicines which have no benefit in regard to their prognosis
there were no significant differences (p value 0.316) in mean [2]. We developed the OncPal Deprescribing Guideline as a
age of the patients who used more than 10 medicines and 10 or tool to improve the rationalization of medicines in palliative
less medicines. cancer patients, with the aim to reduce adverse drug effects,
Potential drug avoidance cost of PIMs was calculated at pill burden and medicine-associated costs. We then tested the
AUD$26.71 per month per patient for the full cost value of guidelines’ validity, measuring its concordance with an expert
medicines sourced from the Pharmaceutical Benefits Scheme panel.
The OncPal Deprescribing Guideline demonstrated its va-
lidity, correctly assessing 94 % of 617 medicines with an
Table 4 Proportion of PIMs per drug class overall Kappa value of 0.83, an ‘outstanding agreement’.
For a more accurate assessment of the guideline, we demon-
Medicine class Total Total PIMs Proportion of
medicines by panel medicines
strated the concordance of each individual medicine class,
(excluding (excluding assessed as using the proportion of accuracy and Kappa values, as well
duplicates) duplicates) PIM (excluding as McNemar values for the symmetry of discordance in med-
duplicates) (%)
icine classes included in the guideline. When the guideline
Aspirin/anticoagulants 15 10 66.7 assessed the medicines with 100 % accuracy, it was not
Dyslipidaemia 19 19 100.0 possible to use these tests.
Antihypertensives 30 27 90.0 All medicine classes in the guideline demonstrated a Kappa
Osteoporosis 4 4 100.0 value of either ‘moderate’ or ‘substantial’ agreement. Aspirin/
Peptic ulcer 38 5 13.2 anticoagulation was the weakest class for the guideline,
prophylaxis
Oral hypoglycaemics 3 3 100.0 assessing medication accuracy 66.7 % of the time, with a
CAMs 20 19 95.0 Kappa value of 0.40. The McNemar values for anticoagulants
Opioid analgesics 42 0 0.0 and antihypertensives of 0.063 and 1.00 respectively, demon-
Non-opioid 39 0 0.0 strate that that there was symmetry in the discordance (>0.5).
analgesics Analyzing this symmetry, we found that when the guideline
Steroids 30 0 0.0
incorrectly assessed this medicine class, the guideline evaluated
NSAIDs 5 0 0.0
Laxatives 35 0 0.0
the medicine as ‘Not a PIM’ and the Expert Panel as a ‘PIM’.
Benzodiazepines 21 0 0.0
Therefore, on the 34 % of occasions that the guideline was
Nausea 33 0 0.0 incorrect, it never recommended ceasing anticoagulation when
Antihistamines 2 0 0.0 the expert panel thought it was necessary. This is a favourable
Antiepileptics/nerve 14 0 0.0 result as it demonstrates a superior safety for the patient.
pain Symmetry was also evident in the CAMs, as the guideline
Anti-infectives 11 0 0.0
recommended to cease one medicine that the expert panel did
Antiarrhythmics 4 2 50.0
not, as the patient was taking zinc for taste disturbances.
Insulins 1 0 0.0
The only medicine class that was assessed incorrectly by
Eye Prep 4 0 0.0
Prostate hyperplasia 5 0 0.0
the guideline in a majority of cases was neoplastic/
Neoplastic/ 9 5 55.6 immunomodulator oral anticancer treatments that were being
immunomodulators administered to 9 patients (15 %). As this class of drug
Psychotropic drugs 25 0 0.0 requires a complex, patient specific decision to evaluate the
Inhaled respiratory 21 1 4.8
benefit of its continuation, it is a class that could not be
Other 18 4 22.0
included in the OncPal Deprescribing Guideline. The decision
CAMs complementary alternative medicines, NSAIDs non-steroidal anti- to de-escalate this medicine class needs to be evaluated on an
inflammatory drugs individual basis by the treating physician.
76 Support Care Cancer (2015) 23:71–78

Incidence of PIMs decision. This challenge indicates a gap in knowledge, which

we believe warrants future research.
Seventy percent of all patients in the study were found to be
taking at least one PIM. This result was consistent with results Limitations
from other studies, which ranged from 22–95 % [3]. This broad
variance appears to be highly dependent on patient specific We have identified three limitations of this study: the population
factors. The study with 95 % of patients, for example, was all size, that it was conducted at a single centre and the theoretical
lung cancer patients who potentially had more co-morbidities nature of the assessment (non-interventional design).
throughout their treatment and more PIMs in the late stages of Due to limited resources, it was only possible to collect
their disease [16]. We identified a high proportion of patients data for 6 months. Although this was sufficient to assess the
taking at least one PIM, which could be attributed to the guideline’s overall validity, a number of individual medication
complexities of patients we identified. As we recruited only classes had too few examples to statistically analyse concor-
inpatients rather than ambulatory care patients seen in some dance. For example, although the guideline correctly assessed
trials, it is possible that our population had more co-morbidities oral hypoglycaemics 100 % of the time, there were only three
that were being treated for secondary prevention. examples, and therefore, we were unable to test statistically.
The proportion of total medications that were identified as As the study was conducted at a single centre, the second-
a PIM has not been previously described in the literature. Our ary outcomes, such as the incidence of PIMs, may differ to
finding of 21.4 % demonstrates the potential for the use of the other centres due to prescribing habits and cancer types treated
guideline in this population, particularly as we found that this by particular specialists.
patient group takes an average of 10 medications, a pill burden Lastly, the study was only able to validate the guideline
figure consistent with the literature [16]. against expert clinical practice; however, further interventional
We found that the medications most frequently identified as studies are required to assess patient outcomes after the guideline
PIMs by the expert panel included antihypertensives, is used to modify prescribing habits in these patients.
dyslipidaemics and CAMs. This is partially consistent with
the literature where ‘statins’ (a member of the dyslipidaemic
drugs) and antihypertensives are commonly described, al-
though proton pump inhibitors (PPIs) and antidiabetic medi- Conclusion
cations have also been described [3]. A possible explanation
for this discrepancy is that our study was at a single site, which A contemporary deprescribing guideline to highlight medica-
showed a trend in some prescribing habits. We found a large tions as potential targets for discontinuation was developed
proportion of patients taking corticosteroids, in which prophy- from current literature and consensus from palliative care
lactic PPIs were necessary. Also, the relatively small number specialists. The ‘OncPal Deprescribing Guideline’ was suc-
of participants may contribute to the discrepancy as well. cessfully validated, demonstrating a statistically significant
Although the primary concern with PIMs is the pill burden ‘outstanding’ concordance with an expert panel. The inci-
and associated potential adverse drug effects the patients may dence of PIMs in palliative cancer patients was high, with
experience, we demonstrated that there is also an associated 70 % taking at least one PIM, demonstrating the potential
cost to the patient and/or healthcare facility, often without benefits for guidelines in clinical practice.
benefit in efficacy, which also needs to be considered. This
is a particularly import finding for healthcare facilities that
Conflict of interest The authors had no financial relationship with the
specialize or have a large cohort of palliative cancer patients.
organization in which the research was conducted and have full control of all
primary data. We agree to allow the journal to review the data if requested.
Challenges

During the assessment stage of the study, the expert panel

encountered one particular challenge where an agreement could Appendix 1 OncPal De-prescribing guideline
not be made on the assessment of particular medication classes.
In this circumstance, the medicines involved were documented This guideline has been developed to assist in highlight-
as ‘not a PIM’, for the purpose of the study. The classes included ing medications with a limited benefit that are suitable
antidepressants, eye preparations and, when multiple opioids targets for discontinuation in palliative cancer patients.
were presented, for example, oxycodone and methadone. To Medication classes not listed below have demonstrated
assist in the decision, the expert panel investigated the literature benefits in this population or the literature is lacking to
as this was their common practice in these scenarios. guide a decision-making process. If the foreseeable bene-
Unfortunately, there was no evidence found to support the fits of any medications do NOT outweigh the adverse
Support Care Cancer (2015) 23:71–78 77

effects and/or associated risks, it is recommended to con-

sider appropriate de-escalation

Medication Medication Considerations Explanation

class for limited
benefit
Blood and Aspirin For primary Long-term benefits at population
blood-forming organs prevention only. level. Little short or intermediate
term risk of stopping (1). Drugs
for primary prevention have, in
general, no place in the treatment
of end-of-life patients since
the time-to-benefit usually exceeds
life expectancy (2).
Cardiovascular system Dyslipidaemia medications All indications. Long-term benefits at population
Statins level. Little short or intermediate
Fibrates term risk of stopping (1).
Ezetimibe
Antihypertensives If sole use is to reduce mild Long-term benefits at population
ACE inhibitors to moderate hypertension level. Ongoing therapy
Sartans for secondary prevention unnecessary in most shortened
Beta blockers of cardiovascular events life expectancy (1).
Calcium channel blockers or as management of
Thiazide stable coronary artery disease.ab
Diuretics
Musculo-skeletal system Osteoporosis medications Except if used for the treatment Except if used for the treatment
Bisphosphonates of hypercalcaemia secondary of hypercalcaemia secondary
Raloxifene to bone metastases. to bone metastases. Long-term
Strontium benefits at population level.
Denosumab Little short or intermediate
term risk of stopping (1).
Alimentary tract Peptic ulcer prophylaxis Lack of any medical history of Ongoing therapy unnecessary
and metabolism Proton pump inhibitors gastrointestinal bleeding, peptic in most shortened life
H2 antagonists ulcer, gastritis, GORD or the expectancy (1).
concomitant use of anti-inflammatory
agents including NSAIDs
and steroids (3).
Oral Hypoglycaemics If sole use is to reduce Potential short-term complications
Metformin mild hyperglycaemia outweigh benefit (1).
Sulfonylureas for secondary prevention
Thiazolidinediones of diabetic associated events.c
DPP-4 inhibitors
GLP-1 analogues
Acarbose
Vitamins If not indicated to treat No evidence for effectiveness (4, 5).d
Minerals a low blood plasma
Complementary—alternative concentration.
medicines
a
Some short-term benefits need consideration—recommended to monitor blood pressure after discontinuation for symptomatic hypertension
b
The use of these agents in symptom management for an underlying disease should be continued. For example, antihypertensives in heart failure or rate control in
irregular heart rhythm (6)
c
Some short-term benefits need consideration—recommended to monitor blood sugar levels infrequently after discontinuation for symptomatic hyperglycaemia.
Aim for blood sugar levels below 20mmols/L (7)
d
Some topical preparations may provide some benefits (5)
1. Stevenson J, Abernethy AP, Miller C, Currow DC. Managing co-morbidities in patients at the end of life. BMJ. 2004;329 (7471):909–12
2. Holmes HM, Hayley DC, Alexander GC, Sachs GA. Reconsidering medication appropriateness for patients late in life. Archives of internal medicine. 2006;166
(6):605–9
3. Fede A, Miranda M, Antonangelo D, Trevizan L, Schaffhausser H, Hamermesz B, et al. Use of unnecessary medications by patients with advanced cancer:
cross-sectional survey. Supportive care in cancer: official journal of the Multinational Association of Supportive Care in Cancer. 2011;19 (9):1313–8
4. Macpherson H, Pipingas A, Pase MP. Multivitamin-multimineral supplementation and mortality: a meta-analysis of randomized controlled trials. The American
journal of clinical nutrition. 2013;97 (2):437–44
5. Kassab S, Cummings M, Berkovitz S, van Haselen R, Fisher P. Homeopathic medicines for adverse effects of cancer treatments. The Cochrane database of
systematic reviews. 2009 (2):CD004845
6. Goodlin SJ, Hauptman PJ, Arnold R, Grady K, Hershberger RE, Kutner J, et al. Consensus statement: palliative and supportive care in advanced heart failure.
Journal of cardiac failure. 2004;10 (3):200–9
7. McCoubrie R, Jeffrey D, Paton C, Dawes L. Managing diabetes mellitus in patients with advanced cancer: a case note audit and guidelines. European journal of
cancer care. 2005;14 (3):244–8
78 Support Care Cancer (2015) 23:71–78

References advanced cancer: prospective cohort study. Bmj 343:d4920. doi:

10.1136/bmj.d4920
8. Tredan O, Ray-Coquard I, Chvetzoff G, Rebattu P, Bajard A,
1. Fede A, Miranda M, Antonangelo D, Trevizan L, Schaffhausser Chabaud S, Perol D, Saba C, Quiblier F, Blay JY, Bachelot T
H, Hamermesz B, Zimmermann C, Del Giglio A, Riechelmann (2011) Validation of prognostic scores for survival in cancer patients
RP (2011) Use of unnecessary medications by patients with beyond first-line therapy. BMC Cancer 11:95. doi:10.1186/1471-
advanced cancer: cross-sectional survey. Supportive care in can- 2407-11-95
cer : official journal of the Multinational Association of 9. Sim J, Wright C (2005) The Kappa statistic in reliability studies: use,
Supportive Care in Cancer 19(9):1313–1318. doi:10.1007/ interpretation, and sample size requirements. Phys Ther 85(3):257–
s00520-010-0947-1 268
2. Jecker NS, Schneiderman LJ (1992) Futility and rationing. The 10. Riechelmann RP, Krzyzanowska MK, Zimmermann C (2009) Futile
American journal of medicine 92(2):189–196 medication use in terminally ill cancer patients. Supportive care in
3. Lindsay J, Dooley M, Martin J, Fay M, Kearney A, Barras M cancer : official journal of the Multinational Association of
(2013) Reducing potentially inappropriate medications in pal- Supportive Care in Cancer 17(6):745–748. doi:10.1007/s00520-
liative cancer patients: evidence to support deprescribing ap- 008-0541-y
proaches. Supportive care in cancer : official journal of the 11. De Muth JE (2009) Overview of biostatistics used in clinical re-
Multinational Association of Supportive Care in Cancer. doi: search. American journal of health-system pharmacy : AJHP : official
10.1007/s00520-013-2098-7 journal of the American Society of Health-System Pharmacists 66(1):
4. Tollier C, Fusier I, Husson MC (2005) ATC and EphMRA classifi- 70–81. doi:10.2146/ajhp070006
cations: evolution from 1996 to 2003 and comparative analysis. 12. Siegel R, Naishadham D, Jemal A (2013) Cancer statistics, 2013.
Therapie 60(1):47–56 CA: a cancer journal for clinicians 63(1):11–30. doi:10.3322/caac.
5. Beers MH, Ouslander JG, Rollingher I, Reuben DB, Brooks J, Beck 21166
JC (1991) Explicit criteria for determining inappropriate medication 13. AIHW & AACR (2012) Cancer in Australia: an overview. Cancer
use in nursing home residents. UCLA Division of Geriatric Medicine series Canberra: AIHW 74 (70)
Archives of internal medicine 151(9):1825–1832 14. Landis JR, Koch GG (1977) The measurement of observer agreement
6. Hwang SS, Scott CB, Chang VT, Cogswell J, Srinivas S, for categorical data. Biometrics 33(1):159–174
Kasimis B (2004) Prediction of survival for advanced cancer 15. Australian Government Department of Health (2013) Pharmaceutical
patients by recursive partitioning analysis: role of Karnofsky benefits scheme. Commonwealth of Australia. http://www.pbs.gov.
performance status, quality of life, and symptom distress. au/pbs/home. Accessed 12/10/2013
Cancer Investig 22(5):678–687 16. Todd A, Williamson S, Husband A, Baqir W, Mahony M (2013)
7. Gwilliam B, Keeley V, Todd C, Gittins M, Roberts C, Kelly L, Patients with advanced lung cancer: is there scope to discontinue
Barclay S, Stone PC (2011) Development of prognosis in palliative inappropriate medication? Int J Clin Pharm 35(2):181–184. doi:10.
care study (PiPS) predictor models to improve prognostication in 1007/s11096-012-9731-2