an introduction to

systematic
reviews
00_Gough et al_Prelims.indd 1 3/3/2017 10:09:02 AM
 Sara Miller McCune founded SAGE Publishing in 1965 to support
the dissemination of usable knowledge and educate a global
community. SAGE publishes more than 1000 journals and over
800 new books each year, spanning a wide range of subject areas.
Our growing selection of library products includes archives, data,
case studies and video. SAGE remains majority owned by our
founder and after her lifetime will become owned by a charitable
trust that secures the company’s continued independence.

Los Angeles | London | New Delhi | Singapore | Washington DC | Melbourne

00_Gough et al_Prelims.indd 2 3/3/2017 10:09:02 AM

 2dnitiodn David Gough
E
Sandy Oliver
James Thomas

an introduction to
systematic
reviews
00_Gough et al_Prelims.indd 3 3/3/2017 10:09:03 AM
 SAGE Publications Ltd Editorial Arrangement  David Gough, Sandy Oliver and
1 Oliver’s Yard James Thomas
55 City Road
London EC1Y 1SP Foreword  Ann Oakley Chapter 7  Jeff Brunton,
Chapter 1  David Gough, James Thomas and
SAGE Publications Inc. Sandy Oliver and James Sergio Graziosi
2455 Teller Road Thomas Chapter 8  James Thomas,
Thousand Oaks, California 91320 Chapter 2  Rebecca Rees Alison O’Mara-Eves, Angela
and Sandy Oliver Harden and Mark Newman
SAGE Publications India Pvt Ltd Chapter 3  David Gough Chapter 9  James Thomas,
B 1/I 1 Mohan Cooperative Industrial Area and James Thomas Alison O’Mara-Eves, Dylan
Mathura Road Chapter 4  Sandy Oliver, Kneale and Ian Shemilt
New Delhi 110 044 Kelly Dickson, Mukdarut Chapter 10  Kristin Liabo,
Bangpan and Mark Newman David Gough and Angela
SAGE Publications Asia-Pacific Pte Ltd Chapter 5  Ginny Brunton, Harden
3 Church Street Claire Stansfield, Jenny Chapter 11  David Gough,
#10-04 Samsung Hub Caird and James Thomas Ruth Stewart and Jan
Singapore 049483 Chapter 6  Katy Sutcliffe, Tripney
Sandy Oliver and Michelle
Richardson

First published 2017

Apart from any fair dealing for the purposes of research or

Editor: Jai Seaman private study, or criticism or review, as permitted under the
Assistant editor: Alysha Owen Copyright, Designs and Patents Act, 1988, this publication
Production editor: Ian Antcliff may be reproduced, stored or transmitted in any form, or
Copyeditor: Christine Bitten by any means, only with the prior permission in writing of
Proofreader: Derek Markham the publishers, or in the case of reprographic reproduction,
Indexer: David Rudeforth in accordance with the terms of licences issued by
Marketing manager: Ben Griffin-Sherwood the Copyright Licensing Agency. Enquiries concerning
Cover design: Shaun Mercier reproduction outside those terms should be sent to the
Typeset by: C&M Digitals (P) Ltd, Chennai, India publishers.
Printed in the UK

Library of Congress Control Number: 2016948864

British Library Cataloguing in Publication data

A catalogue record for this book is available from

the British Library

ISBN 978-1-4739-2942-5
ISBN 978-1-4739-2943-2 (pbk)

At SAGE we take sustainability seriously. Most of our products are printed in the UK using FSC papers and boards.
When we print overseas we ensure sustainable papers are used as measured by the PREPS grading system.
We undertake an annual audit to monitor our sustainability.

00_Gough et al_Prelims.indd 4 3/3/2017 10:09:03 AM

 1
INTRODUCING SYSTEMATIC
REVIEWS
David Gough, Sandy Oliver and James Thomas

Aims of chapter

This chapter:

{{ Introduces the logic and purpose of {{ Introduces some of the current debates
systematic reviews {{ Explains to readers what to expect from
{{ Explains their value for making decisions the rest of the book including what is
{{ Considers what ‘systematic’ means when new compared to the 1st edition
applied to reviewing literature
{{ Explains how systematic review meth-
ods may vary while being systematic

THE ROLE OF RESEARCH REVIEWS

We undertake and read research to find things out. We want to better understand the world.
We develop theories and concepts and gather data to develop insights and answer a vast
breadth of research questions related to a rich array of disciplines, interests and perspectives
of academics, policy makers, professional practitioners, societal groups and individuals.
Often we undertake new ‘primary’ research; but it is also sensible to gather together and
examine what is known from existing research. If we do not assess what has been studied
already then how can we be in a position to know ‘what is known’ or to plan what more
needs to be studied? This book is about how we can go about this – how to bring together
and analyse what we already know from research.

01_Gough et al_Ch-01.indd 1 3/3/2017 10:14:51 AM

 2 A N IN T R ODU CT ION T O SYST E M A T IC R E VIE W S

Research needs to use rigorous and accountable methods to be considered research.

There needs to be an appropriate systematic rigorous method and this must be explicit
so that the results can be interpreted and assessed in the light of how the results were
produced. The same logic applies to reviews of research; the reviews need to use appropri-
ate and explicit methods. In other words, reviews of research are a form of research. They
are a secondary level of analysis (secondary research) that brings together the findings of
primary research. We therefore define a systematic review as ‘a review of existing research
using explicit, accountable rigorous research methods’. There is not just one way to under-
take a systematic review. Research asks all sorts of different questions and uses a wide range
of different methods. Reviewing the literature also requires a variety of methods to bring
findings together systematically. These range from reviews of statistical data to answer
questions about what interventions are effective, to reviews of more qualitative data aiming
to develop new theories and concepts.
Reviewing evidence, and synthesising findings, is something that we do all the time
when going about our daily lives. For example, consider the range of activities involved in
buying a new car. We approach the problem with an overarching question: ‘which car shall I
buy?’ that can be broken down into a series of questions including: ‘what cars are available?’;
‘what type of car do I need?’; ‘which cars can I afford?’; and, if manufacturers’ marketing
departments are doing their job, ‘which car will make me happy?’. We then gather data
together to help us make our decision. We buy car magazines, read online reviews, talk to
people we know and, when we’ve narrowed our options down a little, visit car showrooms
and take some cars out for a test drive. We critically review the evidence we have gathered
(including our personal experience) and identify possible reasons for doubting the veracity
of individual claims. If we’ve decided that we need a small, cheap car, for example, we will
understand that the conclusions of a review written by people who like to drive the latest
sports cars may be less useful than a review written for the ‘thrift supplement’ of a weekend
newspaper. We may prioritise particular characteristics, such as reliability or boot space,
above others, such as fuel economy or safety, and attempt to identify reviews which assess
cars with similar requirements to our own.
The example above, while simple compared to the many very complex decisions made
in life, introduces us to the purpose of reviews and some of the key issues that we need to
grapple with while undertaking a review. Starting our product research by relying first on
what other people have written gives us access to a wide range of ideas about how to judge
cars, more evidence than we could collect ourselves and leaves us with a smaller task when
it comes to visiting showrooms or test driving cars. Our ‘decision question’ drives what we
are doing (‘which car shall I buy?’) and all the other decisions and judgements we make are
based on the need to answer this question. We are faced with many different possible answers
(e.g. the make, model and optional extras of our car) and a mass of evidence that purports
to answer our question. We need to come to an overall understanding of how this hetero-
geneous set of data is able to help us come to a decision and, in order to do this, we need to
understand why the data are heterogeneous. In the example above, reviews of the same cars
come to different conclusions because the people conducting them have different perspec-
tives, priorities and understandings about what they understand the ‘best’ car to be (and,
indeed, what ‘best’ means). In the same way, reviews often depend on judgements, not only
about the methodological quality of research (was it well conducted?), but also its relevance
to answering the question at hand.

01_Gough et al_Ch-01.indd 2 3/3/2017 10:14:51 AM

 INTRODUC ING SYSTEMATIC REVIEWS 3

Our experience as reviewers of research is that there are very many excellent studies
published in the social science and health research literatures. However, there are also
many studies that have obvious methodological or conceptual limitations or do not report
adequate detail for their reliability to be assessed. Even where a study is well conceived,
executed and reported, it may by chance have found and reported atypical findings and so
should not be relied upon alone. For all these reasons, it is wiser to make decisions based on
all the relevant – and reliable – research that has been undertaken rather than an individual
study or limited groups of studies. If there are variations in the quality or relevance in this
previous research, then the review can take this into account when examining its results
and drawing conclusions. If there are variations in research participants, settings or con-
ceptualisations of the phenomena under investigation, these also can be considered and
may add strength to the findings (please also see Chapter 8 for a discussion of these issues).
While primary research is essential for producing much crucial original data and insights,
its findings may receive little attention when research publications are read by only a few.
Reviews can inform us about what is known, how it is known, how this varies across studies,
and thus also what is not known from previous research. It can therefore provide a basis for
planning and interpreting new primary research. It may not be a sensible use of resources
and in some cases it may be unethical to undertake research without being properly informed
about previous research; indeed, without a review of previous research the need for new
primary research is unknown. When a need for new primary research has been established,
having a comprehensive picture of what is already known can help us to understand its
meaning and how it might be used.
In the past, individuals may have been able to keep abreast of all the studies on a topic,
but this is increasingly difficult and (as we shall see in the next section) expert knowledge of
research may produce hidden biases. We therefore need reviews because:

1 Any individual research study may be fallible, either by chance, or because of how it was
designed and conducted or reported. There are even cases of research reports being fabricated.
2 Any individual study may have limited relevance because of its question, scope and context.
3 A review provides a more comprehensive and stronger picture based on many studies and
settings rather than a single study.
4 The task of keeping abreast of all previous and new research is usually too large for an
individual.
5 Findings from a review provide a context for interpreting the results of a new primary study.
6 Undertaking new primary studies without being informed about previous research may result in
unnecessary, inappropriate, irrelevant or unethical research.

SYSTEMATIC, TRADITIONAL AND EXPERT REVIEWS

It has become clear that, when intervening in people’s lives, it is possible to do more harm
than good (Chalmers 2003) (see Box 1.1). Examining existing research is one way of reducing
the chances of doing this. As reviews of such research are increasingly used to inform policy
and practice decisions, the reliability of these reviews is critically important. Methodological
work on reviewing over the past two decades has built up a formidable empirical basis for
reviewing health care evaluations, and systematic reviews have therefore become established
as a key component in evidence-informed decision making. So influential has the use of

01_Gough et al_Ch-01.indd 3 3/3/2017 10:14:51 AM

 4 A N IN T R ODU CT ION T O SYST E M A T IC R E VIE W S

research through systematic reviews become that their development can be considered to be
one of the turning points in the history of science:

This careful analysis of information has revealed huge gaps in our knowledge. It has exposed that
so-called ‘best practices’ were sometimes murderously flawed; and by doing nothing more than sifting
methodically through pre-existing data it has saved more lives than you could possibly imagine.1

The logic of reviewing is thus twofold: first, that as the opening section discussed, looking at
already existing research evidence is a useful thing to do; and second, that as reviews inform
decisions that affect people’s lives, it is important that they be done well. This book uses the
term ‘systematic review’ to indicate that reviews of research are themselves pieces of research
and so need to be undertaken according to some sort of method.

BOX 11

Examples of decisions not informed by research

Expert advice: Dr Benjamin Spock’s advice to parents was to place infants on their fronts to
sleep – advice not supported by research. When this policy was reversed, rates of sudden
infant death dropped dramatically (Chalmers 2001).

Expert panel: In the BSE (mad cow) crisis in the UK in the late twentieth century where
there were many deaths from eating infected meat, ‘... highly problematic policy decisions
were often misrepresented as based on, and only on, sound science’ (van Zwanenberg and
Millstone 2005).

Well-intentioned interventions: In the Scared Straight programme criminals give lectures

to ‘at risk’ youth about the dangers of a life of crime, but this is statistically associated with
higher not lower rates of crime in the at-risk youth (Petrosino et al. 2002, 2013).

Reviewing research systematically involves four key activities: clarifying the question being
asked; identifying and describing the relevant research (‘mapping’ the research); critically
appraising research reports in a systematic manner, bringing together the findings into a
coherent statement, known as synthesis; and establishing what evidence claims can be made
from the research (see Box 1.2 for definitions). As with all pieces of research, there is an expec-
tation that the methods will be explained and justified, which is how we reach our definition
that a systematic review is a review of existing research using explicit, accountable rigorous
research methods.
Most literature reviews that were carried out a decade or more ago were contributions to
academic debates, think pieces, not undertaken in a systematic way. Reviewers did not neces-
sarily attempt to identify all the relevant research, check that it was reliable or write up their
results in an accountable manner.

BBC Radio 4, Moments of Genius: Ben Goldacre on systematic reviews. Available at: www.bbc.co.uk/radio4/features/
1

moments-of-genius/ben-goldacre/index.shtml.

01_Gough et al_Ch-01.indd 4 3/3/2017 10:14:51 AM

 INTRODUC ING SYSTEMATIC REVIEWS 5

Traditional literature reviews typically present research findings relating to a topic of

interest. They summarise what is known on a topic. They tend to provide details on the
studies that they consider without explaining the criteria used to identify and include
those studies or why certain studies are described and discussed while others are not.
Potentially relevant studies may not have been included because the review author was
unaware of them or, being aware of them, decided for reasons unspecified not to include
them. If the process of identifying and including studies is not explicit, it is not possible
to assess the appropriateness of such decisions or whether they were applied in a con-
sistent and rigorous manner. It is thus also not possible to interpret the meaning of the
review findings.

BOX 12

Key terms
Systematic: undertaken according to a fixed plan or system or method

Review: a critical appraisal and analysis

Explicit: a clear, understandable statement of all the relevant details

Accountable: answerable, responsible and justified

Map (systematic): a systematic description and analysis of the research field defined by a
review question

Synthesis: creating something new from separate elements

Systematic review: a review of existing research using explicit, accountable rigorous research
methods

Evidence claim: the statements that can be justified in respect of answering the review
question(s) from the research evidence reviewed

The aim of reviewing systematically is to have such explicit, rigorous and accountable
methods. Just as primary research is expected to report transparent, rigorous methods, the
same standards can apply to systematic reviews. Just as primary research is undertaken
to answer specific questions, reviews of existing research can be productively focused on
answering questions rather than addressing topic areas. The focus on a question drives the
choice of the methods to find the answers.
Individual experts or expert panels are often consulted to answer questions about what is
known from research. Experts may of course have many specialist skills, including knowledge
of research, practical experience of the phenomena being considered, and human insight and
implicit knowledge that have not been formalised in research. However, there can also be
dangers from this richness of knowledge not being explicit.
One danger is that the experts’ ideological and theoretical perspectives, and thus the con-
ceptual framework determining their assessment of the research, will not be explicit; and as

01_Gough et al_Ch-01.indd 5 3/3/2017 10:14:51 AM

 6 A N IN T R ODU CT ION T O SYST E M A T IC R E VIE W S

with everyone, these perspectives may be influenced by personal interests in the issues being
discussed. Second, the boundaries of the experts’ knowledge may not be transparent; that is
the boundaries of studies familiar to them and thus the evidence being considered. A third
danger is that even if the boundaries of the studies are clear, the expert may know some of the
studies within those boundaries better than others, so not all the research will have equal rep-
resentation in the conclusions they draw. Fourth and fifth dangers are the related problems
of how the experts assess the quality and appraise the relevance and then synthesise different
pieces of evidence. Sixth, it may not be clear the extent to which the expert draws on other
forms of knowledge, such as practice knowledge, in forming their overall conclusions. An
expert witness in a court may, for example, provide an opinion that the court believes is
based on research, but is in fact based on a mixture of research and practice wisdom. Seventh,
the manner in which someone is assessed as being expert on a particular area may not be
appropriate. They may not be expert at all on this topic, or they may be expert but their
esteem and credibility is based on practice knowledge. A court may, for example, give high
credibility to research reports from someone who has high esteem as a practitioner rather
than as a researcher.
In many ways an expert review or expert panel is similar to a traditional literature review.
There may be great insight and knowledge but with a lack of transparency about what this
is or how it is being used. With experts and expert panels there may be a lack of clarity
about the:

{{ perspective and conceptual framework, including ideological and theoretical assumptions and
personal interest;
{{ inclusion criteria for evidence;
{{ nature of the search for evidence;
{{ sort of evidence that is thus being considered;
{{ quality and relevance appraisal of that evidence;
{{ method of synthesis of evidence;
{{ use of evidence other than research;
{{ basis for their expertise: (i) how their expertise is assessed, (ii) how its relevance to the topic in
question is assessed, (iii) how its relationship to research skills and knowledge is assessed;
{{ basis of the evidence claims made.

In order to address such issues, systematic reviews often proceed through a number of stages,
as shown in Figure 1.1. We shall see in later chapters that these stages are an oversimplifica-
tion, but they are sufficiently common for them to provide a guide to the part of the review
process being discussed in each chapter (the structure of the book and how the chapters fit
into this is discussed at the end of this chapter).
Though the idea of systematic reviews is simple, and their impact profound, they are
often difficult to carry out in practice, and precisely how they should be done is the subject
of much debate and some empirical research. Identifying the relevant research, checking
that it is reliable and understanding how a set of research studies can help us in addressing
a policy or practice concern is not a straightforward activity, and there are many ways of
going about reviewing the literature systematically. This book takes a careful look at how
approaches differ (while still being systematic) and considers the issues involved in choosing
between these different approaches.

01_Gough et al_Ch-01.indd 6 3/3/2017 10:14:51 AM

 INTRODUC ING SYSTEMATIC REVIEWS 7

QUESTIONS, METHODS AND ANSWERS

Primary research asks many different questions from a variety of standpoints (Gough et al.
2009), and this richness of questions, approaches and research methods is also reflected
in systematic reviews. If, for example, questions are asked about the meaning attached to
different situations by different actors in society, then qualitative exploratory methods are
likely to be appropriate. To find out how societal attitudes vary across the country, then a
survey may be most helpful, whereas knowing how many people receive different state ser-
vices may be found from routine government administrative data. To investigate whether a
particular social intervention has had the required effect, an experimental study may be the
most powerful method.
The idea that different research questions may be answered best by different methods and
by different types of data also applies to reviews. For instance, systematic reviews addressing
questions about the effects of health interventions have widely agreed systematic methods
for setting the scope, judging the quality of studies, and presenting the synthesised find-
ings, often using statistical meta-analysis of the results of randomised controlled trials (see,
for example, Higgins and Green 2011). However, a systematic question-driven approach to
reviews can apply equally to research questions of process or of meaning that are addressed
by more qualitative primary research and by review methods that reflect those qualitative
research approaches (see, for example, Patterson et al. 2001; Pope et al. 2007; Sandelowski
and Barroso 2007).
Reviews and their findings can vary on many different ‘dimensions of difference’ and
these are discussed in detail in Chapter 3. The diversity of methods that are used to bring
together (‘synthesise’) study findings lie on a continuum between approaches that aim to
aggregate or ‘add up’ findings from multiple, similar studies; and those that aim to configure
or ‘organise’ findings of studies (Sandelowski et al. 2006, 2011; Voils et al. 2008). Aggregative
methods of analysis are often used to answer tightly specified questions using quantitative
pre-specified methods to test theory using empirical observations. Configurative approaches
are more likely to address more open questions that explore and explain variation in study
findings. While some configurative methods of analysis bring together qualitative data with
an emphasis on the exploration of meaning and people’s lived experiences, many statistical
methods also configure study findings to understand why a given ‘dependent variable’ might
vary in different situations (see Chapters 8 and 9). Most reviews involve some aggregation
and some configuration.
Reviews need to specify the questions they are asking and the methods used to address
these, and this is often written as a ‘protocol’ prior to undertaking the review (see also
Chapter 4). Writing a protocol or other specification of the review methods at the begin-
ning of a review can be a very useful activity, as it helps the review team to gain a shared
understanding of the scope of the review and the methods that they will use to answer the
review’s questions. It can also form the basis of the review report. Other benefits thought
to accrue from writing a protocol include a reduction of bias (PLoS Medicine Editors 2011),
though this area is somewhat contested, as others would say that protocols are inappropriate
for some types of review where there is a lot of iteration in the method during the process
of the review and iterative methods are required at some stages – for example, searching for
relevant studies (Greenhalgh et al. 2005a, 2005b).

01_Gough et al_Ch-01.indd 7 3/3/2017 10:14:51 AM

 8 A N IN T R ODU CT ION T O SYST E M A T IC R E VIE W S

BOX 13

Key terms
Quantitative research: the systematic empirical investigation of quantitative properties of
phenomena and their relationships. Quantitative research often involves measurement of
some kind.

Qualitative research: in-depth enquiry to understand the meaning of phenomena and their
relationships.

Aggregative synthesis: where the synthesis is predominantly aggregating (adding up) data
to answer the review question. Aggregation commonly uses quantitative data but qualitative
data can also be aggregated.

Configurative synthesis: where the synthesis is predominantly configuring (organising)

data from the included studies to answer the review question. Qualitative and quantitative
data can be configured to understand variation in study findings. Aggregation and con-
figuration fall on a continuum, and all reviews are likely both to aggregate and configure
data to some extent.

Protocol: a statement of the approach and methods to be used in a review made prior to
the review being undertaken.

With these complex choices about review methods, it is important for reviewers to have a
clear understanding of the ‘meaning’ of the question. For example, the starting point for
a review might be, ‘We want to know about the effects of classroom teaching assistants’.
Further discussion may clarify that this question could be more precisely framed to indicate
the more specific question being asked and then the type of synthesis that would be most
appropriate, for example:

{{ ‘Do students in classes where there is a classroom teaching assistant get higher or lower scores
on test scores?’ The most appropriate synthesis method is probably aggregative quantitative,
preferably meta-analysis (if the primary research meets the necessary conditions).
{{ ‘How can we conceptualise the way that the presence of classroom assistants changes
relationships between students and teachers and between teachers in class?’ The most
appropriate synthesis method is probably configurative (if the primary research meets the
necessary conditions).

Once the meaning of the question is clear, the appropriate method can be chosen. Although
it may seem a simple task to match methods to question types, the reality is more complex.
A question may be answered in more than one way. The selection of the appropriate method
for a review depends on the type of question to be answered, but also must consider the use
to which the review will be put, and practical issues, such as the experience and perspective
of the review team (see Chapter 4). Also, reviews vary in how extensive the review question is
(the breadth of question and the depth in which it is examined) and in the time and resources
used to undertake it (see Chapter 3).

01_Gough et al_Ch-01.indd 8 3/3/2017 10:14:52 AM

 INTRODUC ING SYSTEMATIC REVIEWS 9

When thinking about the role of a review, it can be helpful to always keep in mind the
basic logic of undertaking reviews:

{{ What is the issue at hand?

{{ What do we want to know (from research that might help us with this issue – and so what are our
research questions)?
{{ What do we know already (so undertake or use existing systematic reviews)?
{{ And how do we know it (from what methods of systematic review and what type of primary research)?
{{ What more do we want to know (from further reviews or further primary research)?
{{ And how could we know it (how can the systematic reviews inform further research strategies)?

CHALLENGES OF, AND CRITIQUES OF,

SYSTEMATIC REVIEWING
Systematic reviewing has only recently become a major area of methodological development.
Although the idea of being more explicit about reviewing research is not new, it was only in the
1980s that some of the texts on some of the major types of review, such as statistical meta-analysis
(Glass et al. 1981) and meta-ethnography (Noblit and Hare 1988), were published (these types
of review are explained later in the book, particularly in Chapters 3 and 8). Although reviews of
literature have been advocated for very many years (Bohlin 2012; Chalmers et al. 2002; Shadish
and Lecy 2015), systematic reviewing is still a young and rapidly developing field of study and
methods of reviewing have not yet been developed for all areas of science. It is an exciting time
yet there are many challenges to be overcome (Oakley et al. 2005).
First, there are many conceptual and methodological challenges. This book discusses a
wide range of approaches to reviewing from an equally broad range of different epistemo-
logical positions. In it we argue that this range of methods is useful, but we realise that this
diversity raises many complex issues, particularly in relation to mixing results from different
research traditions. Related to the conceptual challenges are more detailed methodological
issues. While there is a strong empirical base underpinning some review methods, many
have been designed and developed based on primary research methods and on the logic of
systematic reviews, with comparatively little methodological study of the impact of different
approaches to particular aspects of reviewing – such as search strategies and data coding. The
methods of evidence-informed policy and practice are not always evidence informed; we
need more empirical data to support the selection of specific review methods.
A second challenge, and related to the methodological issue, is the lack of an agreed termi-
nology to describe, discuss and develop methods (Gough et al. 2012). Some of this linguistic
confusion arises from fundamental debates about the nature of knowledge and the role of
research in this; and some is a lack of clarity about widely used but unclear distinctions, such
as quantitative and qualitative research. There are, however, many further problems in the
terminology used for reviews of research evidence.
The term ‘meta’ is one such confusing word. Meta has many meanings and in relation
to research it is often used to mean ‘about’ or ‘beyond’ and so ‘meta-evaluations’ are
‘evaluations of evaluations’. Such meta-evaluations can be a form of systematic review,
but they can also be simply the evaluation of the quality of one or more evaluations
(Gough et al. 2014). The term ‘meta-analysis’ can mean analysis of analysis and so can be

01_Gough et al_Ch-01.indd 9 3/3/2017 10:14:52 AM

 10 A N IN T R ODU CT ION T O SYST E M A T IC R E VIE W S

another term for systematic review, but meta-analysis has been used so often to refer to
statistical reviews that in some circles it has become synonymous with statistical synthesis. A
‘meta-review’, on the other hand, is a review of (‘about’) reviews, of which there are several
forms (see Chapter 3). Also, many words used in reviewing can give the impression that a
particular type of review is being assumed. Words such as ‘protocol’ (methods of a review)
suggest pre-specified methods and the word ‘synthesis’ may suggest aggregation to some
people and the configuring of findings to others (see Chapter 3 on ‘dimensions of difference’
in reviews). This book does not provide an overarching classification for systematic reviews.
Instead it aims to discuss the nature of reviews and how they vary and includes some of the
terminology used to describe this variation (see Chapter 3).

BOX 14

Key terms
Meta: about, above or beyond something.

Meta-evaluation: evaluation of evaluations. These can be systematic reviews; alternatively,

they can be formative or summative evaluations of evaluations, including standards for
such evaluations.

Meta-analysis: the term has been used in the past to refer to all forms of review, but now
usually refers to the statistical meta-analysis of data from primary research studies.

Review of reviews: a review of previous reviews. This contrasts with reviews that only review
primary research studies.

A third challenge relates to resource constraints. Reviews are major pieces of research and
require time and other resources. The resources available will impact on the type of review
that can be undertaken. More fundamentally, there is the extent of investment by society in
reviews of research. This is not just an issue of overall funding for research, but the balance of
investment between: primary research; reviews of what is known from that primary research;
and systems to assist the interpretation and application of that research (see Chapter 11).
Currently, the funding of research is predominantly on primary production compared to
synthesis and use of research.
The most appropriate balance between primary research and reviews is difficult to specify
but is likely to vary for different funders, producers and users of research. A research insti-
tute or evidence or brokerage centre closely related to policy or practice decision makers, for
example, would be particularly likely to have a high balance of reviews to primary research.
The challenge for all the individuals and organisations involved in research is to consider
whether their needs, roles and responsibilities are best met by the current balance they take
between primary research and reviews of and use of that research.
A fourth challenge is the capacity constraints in terms of individual and organisational skills
and infrastructure to undertake reviews. There are relatively few people with advanced review
skills and so even if funding was available, it would take time to build up the necessary capacity.

01_Gough et al_Ch-01.indd 10 3/3/2017 10:14:52 AM

 INTRODUC ING SYSTEMATIC REVIEWS 11

Fifth are the capacity constraints for using reviews. This not only involves the capacity to
read and understand reviews, but also the capacity to interpret and apply reviews in mean-
ingful and useful ways. Reviews of research are only one part of the research generation and
use cycle. The cycle cannot work effectively without active engagement between research and
the users of research and this may require further intermediary processes and intermediary
organisations (Gough et al. 2011; and see Chapter 11). If formal processes are required to
be explicit about ideological and theoretical perspectives and methods in the production of
knowledge and its synthesis, then maybe similar processes are also required to support the
use of such knowledge.
Sixth are broader political challenges. There are many critics of systematic reviews. One
criticism is the mistaken belief that systematic reviews are only concerned with questions
of studies of effectiveness and so represent an empiricist (or positivist) research paradigm.
In social science there are often strong views about the appropriateness of different research
paradigms (Oakley 2000a, 2000b) and some argue that the empiricist paradigm is deficient,
making systematic reviews deficient too. However, as has already been explained in this
chapter, the logic of reviews can apply to many questions and methods, not only empirical
statistical reviews; meta-ethnography, for example, was introduced as a method in the late
1980s (Noblit and Hare 1988). Moreover, we argue that systematic reviews of effectiveness,
framed with the help of stakeholders, are not deficient but important contributions to
accumulating knowledge.
A related criticism is that the review process is atheoretical and mechanical and ignores
meaning. This is another criticism of the empiricist paradigm where (in both primary
research and reviews) a pre-specified empiricist strategy is used to test hypotheses such
as the effectiveness of interventions. It is a criticism of a particular research paradigm
and of the narrowness of some studies within that paradigm rather than of systematic
reviews. The preference of one research paradigm over another or the existence of some
poor quality primary research studies or systematic reviews is not an argument about the
inherent appropriateness or importance of systematic reviews. Decision makers at various
levels need to have different kinds of research questions addressed in order to inform the
formulation of policy or practice and to implement change, and so the authors of this
book value a plurality of perspectives, research questions and methods and thus also of
review questions and methods (Oakley 2000a, 2000b). We value theory-testing reviews
asking questions of effectiveness that aggregate findings and we also value reviews that
configure and develop and critique concepts and ideas.
Another criticism is that reviews often only consider relatively few studies and thus are
ignoring much relevant research. There are at least two issues here. First, many reviews have
narrow review questions and so narrowly define the boundaries (inclusion criteria) of the
studies they consider and so their conclusions must be limited to research so defined. This
needs to be explicit in the title, introduction and summary of a review to avoid misrepre-
senting the data on which it was reaching its conclusions. The review needs to state what it
is not including and thus what it is not studying. A review that was titled ‘the importance
of X for Y’ which only considered a few aspects of X and Y could rightly be criticised for
misrepresentation or bias.
The criteria for the inclusion and exclusion of studies can include such issues as the topic
focus, the method of primary research and the quality of the research. Researchers with differ-
ent perspectives and working within different research paradigms will have different views of

01_Gough et al_Ch-01.indd 11 3/3/2017 10:14:52 AM

 12 A N IN T R ODU CT ION T O SYST E M A T IC R E VIE W S

what constitutes good quality and relevant evidence. This is the nature of academic discourse
and occurs just as much in primary research as with research reviews. With reviews the argu-
ment is being played out at a meta-level rather than in discussing individual studies. Reviews
allow broader discussions with explicit assumptions and leveraging many studies rather than
debates about individual studies.
The second issue in relation to the low numbers of studies in some reviews is concerned
with the inefficient process of searching for studies on electronic databases. Many irrelevant
studies have to be sifted through to find the few that are on topic. Reports of reviews will
include the number of studies found in the search, which may be very many thousands, of
which only a few may be relevant. Critics use those numbers of discarded studies to argue
that studies are being ignored. What is being ignored here, however, is that electronic search-
ing is imprecise and captures many studies that employ the same terms without sharing the
same focus (see Chapters 5 and 7). These extraneous studies need to be excluded after the
electronic searching is completed. In sum, there are two related issues: what is the focus of a
review and what number of studies will help in addressing that focus?
A broader and potentially more powerful criticism is that systematic reviews appeal to
government because they fit with a new managerialism for controlling research. The state
can specify what research it wants and how this should be reviewed and thus control the
research agenda and the research results. The overt process of setting review questions in
discussion with a range of different users is one way to engage with such concern though
there are of course issues of who has the power and control to decide perspectives and prior-
ities. Researchers and research funders are in a very strong position to determine the nature
and outcome of research. Involving a broader range of users in defining reviews of what we
know, what we don’t know and what more we want to know, can give voice to others in
society and make research more, not less, democratic (Gough and Elbourne 2002; Gough
2007a, 2011; and see Chapter 2). Being more explicit about the personal and political in
research and increasing the potential for the increased involvement of different sections
of society nationally and internationally is an important goal for all research including
systematic reviews.
Systematic reviews have an integral role in the production of research knowledge and
are an essential part of the process of interpreting and applying research findings to bene-
fit society. Systematic reviews play a key part in developing future primary research and in
advancing methods that better achieve their purpose – so-called ‘fit for purpose’ research
methods. They provide a potential means by which all voices in society can engage in deter-
mining research agendas. Whether these voices are heard depends on socio-political forces
within and beyond research.

THE AIMS OF THIS BOOK

This book provides an introduction to the logic of systematic reviews, to the range of current
and developing methods for reviewing, and to the consequences of reviewing systematically
for the production and use of research. There are many excellent books available on different
types of systematic review. This book differs from most others currently available in exam-
ining the nature of the basic components of reviews driven by any research questions and
including any research methods and types of data.

01_Gough et al_Ch-01.indd 12 3/3/2017 10:14:52 AM

 INTRODUC ING SYSTEMATIC REVIEWS 13

It examines formal, explicit and rigorous methods for undertaking reviews of research
knowledge. This idea of gathering research literature together is straightforward; the chal-
lenge is that research questions and methods are very diverse and so we need many different
types of review addressing different questions using different types of research for different
purposes. Understanding the complexities of reviewing research has practical relevance for
using the messages that research offers, but its importance is greater than this.
The book has been designed as a resource for four main audiences. First, it is a resource
for those undertaking reviews. It does not provide a step-by-step guide for carrying out every
stage of every possible type of review. Considering detailed methods of reviewing for all
possible types of research question, ranging from measuring the effects of interventions to
developing theory to understand how things can be best understood, would be too much for
a single volume and sections would quickly become out of date as new insights are being pub-
lished regularly. Instead, it aims to provide an explanation of the main issues encountered at
different stages of different types of systematic reviewing, and thus the thinking behind the
many decisions required in any review. As we shall see later in the book, reviews vary con-
siderably in their questions and methods, and in their scope and purpose. An understanding
of how the aims of a review can be achieved is the most fundamental requirement for being
able to undertake a review or for using review findings appropriately. The book is described
as an introduction because it makes known to readers, in considerable depth, the thinking
underlying the aims and methods of reviews.
Systematic reviews raise issues about how primary research is undertaken, how different
approaches and methods are fit for purpose, and the implications for what more needs to
be known and how the gaps can be filled by primary research. The second audience for the
book therefore consists of those who fund, plan or undertake primary research. Reviews tell
us what we know and don’t know in relation to a question, and how we know this. They also
raise the issue of what future research might be, what we might know and how we might
know it. The review process thus enables a consideration of what would be the appropriate,
fit-for-purpose research strategies and methods to achieve specific research objectives. It also
provides an opportunity for non-researchers to be involved in such processes; to consider the
research to date and to participate in developing future research agendas.
The third audience for the book consists of those who may use reviews to inform decision
making. If decisions are being made on the basis of a review, and yet reviews can vary consid-
erably in terms of focus, question, purpose, method, data and rigour, then an understanding
of these characteristics is important for interpreting the quality and relevance of that review
for the decision makers.
The fourth audience for the book will be those who have a wider interest in the production
and use of research in society. In being question-driven, systematic reviews raise issues about
the purpose of research and thus the drivers producing that research. They raise questions about
whose questions are being asked and the methods being used to provide those answers. Reviews
of research are driven by the needs of the different people asking these different questions,
including all those who may use the research or other stakeholders affected by it. For this rea-
son, reviews and their methods need to be relatively fit for achieving their intended purpose(s).
These questions relate to fundamental issues about the funding and use of research and its role
in society. They also relate to how reviews can be interpreted and applied in practice, thereby
raising a whole range of issues about moving from knowledge to action, described in different
ways as knowledge translation, knowledge mobilisation and exchange (see Chapter 11).

01_Gough et al_Ch-01.indd 13 3/3/2017 10:14:52 AM

 14 A N IN T R ODU CT ION T O SYST E M A T IC R E VIE W S

The book reflects the development of theory and practice at the EPPI-Centre where all of
the authors do or, until recently, did work (Oakley et al. 2005). There is not an EPPI-Centre
method for undertaking reviews; rather there are principles that guide our work. These
include the following:

1 Both primary research and reviews of research are essential to advance knowledge.
2 There is a wide range of review methods just as there is a wide range of primary research
methods.
3 Reviews should follow the research method principles of quality, rigour and accountability that
are expected in primary research.
4 Review methods often reflect the methods, epistemological assumptions and methodological
challenges found in primary research.
5 Reviews should be driven by questions which may vary in many ways, including their implicit or
explicit assumptions.
6 Those asking questions of research are ‘users’ of that research, with particular perspectives
(values with associated priorities and ideological and theoretical assumptions) that inform the
process of undertaking primary and secondary (reviews of) research.

Although the main focus of the work of the EPPI-Centre has been on the development of
methods of systematic review, we are increasingly working on the very related area of the
use of reviews in decision making and ‘research on research use’ (see Chapter 11). This is also
reflected in our teaching which includes both a MSc in Systematic Reviews for Policy and
Practice and a MSc in Social Policy and Social Research.

DEVELOPMENTS IN THIS 2ND EDITION

This 2nd edition of the book has the same overall aims and approach as the 1st edition
(Gough et al. 2012a), but as five years have passed since the 1st edition every chapter has
been updated and revised.
The main changes in structure are:

{{ The expansion of the text on synthesis from one to two chapters (Chapters 8 and 9). There is
simply too much material to include in a single chapter. Chapter 8 discusses the broad range of
approaches to synthesis in terms of the aims of synthesis and how this is achieved; and focuses
on non-statistical methods of synthesis. Chapter 9 provides more detail on technical aspects of
some quantitative approaches to synthesis.
{{ The chapter on critical appraisal of studies included in a review has been changed to be
one aspect of a larger chapter considering the evidence claims made by a review. This is
partly because critical appraisal occurs throughout a review, not just a specific stage. More
fundamentally, it is because critical appraisal is also undertaken on the whole review in
terms of the quality and relevance of its methods, the sufficiency of its data from included
studies, and the extent that it is able to make evidence claims to answer the review
question. The chapter has thus been re-named as ‘Developing Justifiable Evidence Claims’
and moved to Chapter 10. In other words, moved from being before to being after the
chapters on synthesis (Chapters 8 and 9).
{{ The final chapter of the 1st edition, on future developments has been removed. Instead,
individual chapters discuss how methods are developing.

01_Gough et al_Ch-01.indd 14 3/3/2017 10:14:52 AM

 INTRODUC ING SYSTEMATIC REVIEWS 15

There are too many specific changes of detail to list here but there are a few aspects of the 1st
edition that we have been keen to emphasise more in this 2nd edition.
First, we continue to stress that systematic reviews are not limited to any particular
research question or method and that reviews will vary in aim and method just as with pri-
mary research. Systematic reviews differ in some ways from primary research but for us the
best way to consider them is as a level of analysis rather than being seen as a very different
form of undertaking to primary research.
Second, we continue to emphasise the way that the rich diversity of reviews varies on a
number of key dimensions (Gough et al. 2012b). There is an increasing growth of brands of
types of reviews which we welcome in terms of the growing rich diversity of reviews. However,
without specifying where these review types sit on the key dimensions there can be a lack of
clarity about the distinctive characteristics of these ‘brands’; and even when a review might be
given a particular brand ‘label’, there is such wide variation in the range of approaches covered
by a label, that it does not necessarily describe the approach taken very precisely.
Third, we have made some adjustments to the ‘dimensions of difference’ we use to describe
variation between reviews (Gough et al. 2012a, 2012b). We have adjusted the language of the
terms we use and introduced new dimensions such as the extent to which methods adopt
widely agreed procedures (see Chapter 3). Also, in the 1st edition, we highlighted the dis-
tinction between aggregative and configuring methods of analysis in reviews. In this second
edition, we have increased our discussion of configurative approaches and the way in which
configuration and aggregation are often both present in different ways in many types of
research synthesis.
Fourth, we have put even more emphasis on considering the extent of the ‘research prob-
lem’ that a review attempts to address. Both primary research and reviews of that research vary
in their ambition and resources and how much they aim to achieve. We therefore frame reviews
as a process that takes place between: (i) there being an issue that somebody wants to know
more about from research; and (ii) what the review achieves in answering that question. We
consider reviews in relation to the user perspectives, aims of a review, and the role that reviews
play in the research production and use process. This approach is shown in the diagrams at the
start of each chapter showing its position as prior to, after, or stages within a review.
Fifth, there are some general changes in research that have parallel developments in reviews.
One example is the advances in information technology that is changing the relationship
between individual level, primary research level and synthesis level of data analysis. Another
related example is the increased use of large data sets including routine administrative data
and the blurring between experimental and observational research. Another example is the
increased awareness of the need to consider theory and empirical data together with more
complex primary and consequently more complex secondary forms of research analysis (see
Chapters 8 and 9). In addition, developments in information technology have also supported
the advance of tools and technologies of information management in reviews (see Chapter 7).

STRUCTURE OF THE BOOK

The book is organised around the stages prior to, during or after a review as illustrated in
Figure 1.1 which is used at the very start of most of the chapters of the book (indicating
which stage is being considered in each chapter).

01_Gough et al_Ch-01.indd 15 3/3/2017 10:14:52 AM

 16 A N IN T R ODU CT ION T O SYST E M A T IC R E VIE W S

Clarifying the problem and question

Engaging stakeholders Forming review team
Exploring the problem and how research might help
Developing review question and scope
Deciding work to be done and developing conceptual framework

Finding Describing Synthesising Appraising

studies in terms of using relevance and
within the conceptual conceptual quality of
scope framework framework the evidence
and to
manage
the review

Next steps
Engaging stakeholders to interpret and make use of the evidence

Figure 1.1  Stages of the review process

The current chapter (Chapter 1) provides an introduction to the book. Chapters 2, 3

and 4 are concerned with the initial stages of clarifying the problem and question, and
how this will be addressed in the review.
Chapter 2 (Stakeholder Perspectives and Participation in Reviews) focuses on the multi-
ple perspectives that can be involved in a review reflecting different priorities, values and
theoretical assumptions. It discusses how these effect a review and how such different per-
spectives can be brought into and participate in the review process. Chapter 3 (Commonality
and Diversity in Reviews) considers the range of different types of review and how they
vary. It examines the nature of this variation on such ‘dimensions of difference’ as structure,
resources and approach of the review. Chapter 4 (Getting Started with a Review) then exam-
ines how to begin the process of a systematic review. It discusses in more detail the choice of
aim and method of review, the team to undertake the review, quality assurance of the review
process and product, and the management of all this work.
Chapters 5 and 6 follow on with the next stages of a review. Chapter 5 (Finding Relevant
Studies) explores the work required to find the research that will be included in a review.
This includes the overall approach and specific strategies and resources and ‘screening’ to
check that the studies do meet the needs of the review. Chapter 6 (Describing and Analysing
Studies) follows by explaining the different aims and methods of identifying and recording
(coding) relevant information from the studies included in a review. This includes the prin-
ciples and processes of good coding and how these vary between different types of review.
A review requires the management and analysis of many pieces of detailed information.
Chapter 7 (Tools and Technologies for Information Management) examines the ways that
information technology can assist the review process in terms of supporting individual stages

01_Gough et al_Ch-01.indd 16 3/3/2017 10:14:52 AM

 INTRODUC ING SYSTEMATIC REVIEWS 17

such as data coding and analysis as well as the management of the whole review, and the rela-
tionship between different levels of analysis in research (from raw data, to analysis in primary
research and reviews of such research).
Chapters 8 and 9 then examine the methods used to synthesise the findings of included
studies. Chapter 8 (Synthesis Methods for Combining and Configuring Textual or Mixed
Methods Data) develops the ideas introduced earlier in Chapters 3 and 4 on the variation
between reviews to consider how reviews can vary in their relation to theory and concep-
tual development. The chapter also introduces a range of methods of non-statistical and
mixed methods synthesis. Chapter 9 (Synthesis Methods for Combining and Configuring
Quantitative Data) then concentrates on quantitative methods of analysis such as statistical
meta-analysis.
The final stage of the review process is considered in Chapter 10 (Developing Justifiable
Evidence Claims). The chapter considers the three main components for assessing the extent
that a review has addressed the question that it aimed to answer (the methods of the review,
the studies included in the review and the totality of evidence produced by that method
using that included evidence).
The final chapter of the book, Chapter 11 (Using Research Findings) focuses on how the
findings of a review might be used. It provides a discussion of the nature of research use, how
this occurs in practice, how this use might be developed and changed and how we can study
these processes of research use. The chapter is relevant to the whole book as it provides some
background as to how reviews might be used and so inform the development of methods to
assist such use of research.

CONCLUSION
The starting point for the book is that research as systematic enquiry is an important form
of knowledge and that we should balance the investment of resources and energy in new
research with the reviewing of what we know from previous research and processes for inter-
preting and applying that knowledge. Reviews should be one of the first things that we do
before embarking on any form of important decision including academics planning new
primary research. Traditionally, reviews have been undertaken without clear, formal, explicit
and systematic methods, which undermines their status and utility as research, and similar
arguments can be made about expert opinion (even if such opinion may have great uses in
other circumstances). Systematic reviews have common principles and similar processes, but
can vary as much as primary research in terms of their extent, breadth and depth, and in
types of question, data and method. Systematic reviews, like any form of research, can be
undertaken well and badly, and so need appropriate quality assurance processes to evaluate
them. To progress, systematic reviewers need to be aware of the many practical, methodolog-
ical and political challenges involved in this work and their wider role in the production and
use of research in society.

01_Gough et al_Ch-01.indd 17 3/3/2017 10:14:52 AM