research methods

& reporting

BMJ: first published as 10.1136/bmj.b2700 on 21 July 2009. Downloaded from http://www.bmj.com/ on 29 July 2019 by guest. Protected by copyright.
The PRISMA statement for reporting systematic reviews
and meta-analyses of studies that evaluate healthcare
interventions: explanation and elaboration
Alessandro Liberati,1 2 Douglas G Altman,3 Jennifer Tetzlaff,4 Cynthia Mulrow,5 Peter C Gøtzsche,6 John P A
Ioannidis,7 Mike Clarke,8 9 P J Devereaux,10 Jos Kleijnen,11 12 David Moher4 13

Introduction guidance for authors reporting a meta-analysis of ran-

1
Università di Modena e Reggio
Emilia, Modena, Italy Systematic reviews and meta-analyses are essential tools domised trials. Since then, much has happened. First,
2
Centro Cochrane Italiano, Istituto for summarising evidence accurately and reliably. They knowledge about the conduct and reporting of system-
Ricerche Farmacologiche Mario help clinicians keep up to date; provide evidence for atic reviews has expanded considerably. For example,
Negri, Milan, Italy
policy makers to judge risks, benefits, and harms of the Cochrane Library’s Methodology Register (which
3
Centre for Statistics in Medicine,
University of Oxford, Oxford healthcare behaviours and interventions; gather together includes reports of studies relevant to the methods for
4
Ottawa Methods Centre, Ottawa and summarise related research for patients and their systematic reviews) now contains more than 11 000
Hospital Research Institute, carers; provide a starting point for clinical practice entries (March 2009). Second, there have been many
Ottawa, Ontario, Canada
guideline developers; provide summaries of previous conceptual advances, such as “outcome-level” assess-
5
Annals of Internal Medicine,
Philadelphia, Pennsylvania, USA research for funders wishing to support new research;1 ments of the risk of bias,10 11 that apply to systematic
6
Nordic Cochrane Centre, and help editors judge the merits of publishing reports reviews. Third, authors have increasingly used system-
Copenhagen, Denmark of new studies.2 Recent data suggest that at least 2500 atic reviews to summarise evidence other than that pro-
7
Department of Hygiene and new systematic reviews reported in English are indexed vided by randomised trials.
Epidemiology, University of
Ioannina School of Medicine, in Medline annually.3 However, despite advances, the quality of the con-
Ioannina, Greece Unfortunately, there is considerable evidence that duct and reporting of systematic reviews remains well
8
UK Cochrane Centre, Oxford key information is often poorly reported in systematic short of ideal.3‑6 All of these issues prompted the need
9
School of Nursing and Midwifery, reviews, thus diminishing their potential usefulness.3‑6 for an update and expansion of the QUOROM state-
Trinity College, Dublin, Republic
of Ireland
As is true for all research, systematic reviews should ment. Of note, recognising that the updated statement
10
Departments of Medicine, Clinical be reported fully and transparently to allow readers to now addresses the above conceptual and methodo-
Epidemiology and Biostatistics, assess the strengths and weaknesses of the investigation.7 logical issues and may also have broader applicability
McMaster University, Hamilton, That rationale led to the development of the QUOROM than the original QUOROM statement, we changed
Ontario, Canada
11
Kleijnen Systematic Reviews, York
(quality of reporting of meta-analysis) statement; those the name of the reporting guidance to PRISMA (pre-
12
School for Public Health and detailed reporting recommendations were published in ferred reporting items for systematic reviews and meta-
Primary Care (CAPHRI), University 1999.8 In this paper we describe the updating of that analyses).
of Maastricht, Maastricht, guidance. Our aim is to ensure clear presentation of what
Netherlands
was planned, done, and found in a systematic review. Development of PRISMA
13
Department of Epidemiology and
Community Medicine, Faculty of Terminology used to describe systematic reviews and The PRISMA statement was developed by a group
Medicine, Ottawa, Ontario, Canada meta-analyses has evolved over time and varies across of 29 review authors, methodologists, clinicians, medi-
Correspondence to: alesslib@ different groups of researchers and authors (see box 1 at cal editors, and consumers.12 They attended a three
mailbase.it
end of document). In this document we adopt the defini- day meeting in 2005 and participated in extensive
Accepted: 5 June 2009
tions used by the Cochrane Collaboration.9 A systematic post-meeting electronic correspondence. A consensus
Cite this as: BMJ 2009;339:b2700 review attempts to collate all empirical evidence that process that was informed by evidence, whenever pos-
doi: 10.1136/bmj.b2700 fits pre-specified eligibility criteria to answer a specific sible, was used to develop a 27-item checklist (table 1)
research question. It uses explicit, systematic methods and a four-phase flow diagram (fig 1) (also available
that are selected to minimise bias, thus providing reliable as extra items on bmj.com for researchers to down-
findings from which conclusions can be drawn and deci- load and re-use). Items deemed essential for transpar-
sions made. Meta-analysis is the use of statistical methods ent reporting of a systematic review were included in
to summarise and combine the results of independent the checklist. The flow diagram originally proposed
studies. Many systematic reviews contain meta-analyses, by QUOROM was also modified to show numbers of
but not all. identified records, excluded articles, and included stud-
ies. After 11 revisions the group approved the checklist,
The QUOROM statement and its evolution into PRISMA flow diagram, and this explanatory paper.
The QUOROM statement, developed in 1996 and The PRISMA statement itself provides further details
published in 1999,8 was conceived as a reporting regarding its background and development.12 This

BMJ | online