Background

Systematic reviews are the reference standard for synthesizing evidence in health care because of their methodological rigor. They are used to support the development of clinical practice guidelines and inform clinical decision-making. They are becoming increasingly common; in 2010, 11 new reviews were estimated to be published daily [1]. Ideally, systematic reviews are based on pre-defined eligibility criteria and conducted according to a pre-defined methodological approach as outlined in an associated protocol.

The preparation of a protocol is an essential component of the systematic review process; it ensures that a systematic review is carefully planned and that what is planned is explicitly documented before the review starts, thus promoting consistent conduct by the review team, accountability, research integrity, and transparency of the eventual completed review. A protocol may also reduce arbitrariness in decision-making when extracting and using data from primary research, since planning provides an opportunity for the review team to anticipate potential problems. When clearly reported protocols are made available, they enable readers to identify deviations from planned methods in completed reviews and whether they bias the interpretation of a review results and conclusions. Bias related to the selective reporting of outcomes has been characterized as a serious problem in clinical research, including systematic reviews [27].

Until recently, systematic review protocols were generally available only through select organizations, such as The Cochrane [8] and Campbell Collaborations and the Joanna Briggs Institute, for which the preparation of a protocol is mandatory. Outside of these organizations, the existence of a protocol is infrequently reported in completed reviews [9, 10]. Fewer than half of 300 systematic reviews indexed on MEDLINE in November 2004 (most recent generalizable sample; 2014 update underway) report working from a protocol [10], 80% of which are non-Cochrane affiliated. Of the non-Cochrane therapeutic reviews, only 11% mentioned the existence of a protocol [10]. The majority of reviews in health care are conducted and published outside of Cochrane, however [10]. The paucity of protocols may be due, in part, to the authors’ lack of knowledge about how to write them and what to include. Currently, little succinct guidance is available for those preparing systematic review protocols, although the recent Standards for Systematic Reviews prepared by the Institute of Medicine (IOM) provide some guidance toward addressing this gap [11].

Many groups have called for the widespread preparation and registration of systematic review protocols in order to increase the availability and accessibility of a priori methods for systematic reviews [1214]. Such an effort may reduce the duplication of effort [15] and reduce the publication bias of systematic reviews. This challenge has been taken up by the Centre for Reviews and Dissemination, University of York, which has spearheaded the establishment of an international register—PROSPERO (International Prospective Register of Ongoing Systematic Reviews, http://www.crd.york.ac.uk/prospero) [16, 17]. The register, which enables the permanent documentation of 22 mandatory (and 18 optional) items about the a priori design and conduct of a review, was launched in February 2011. At the time of writing, >5,000 systematic review protocols from over 70 countries have been registered since its inception. Starting in October 2013, new Cochrane protocols were and continue to be automatically added to PROSPERO.

Along with the improved accessibility of protocols through registration comes the need for strengthened transparency, accuracy, and completeness of the reports of protocols intended for dissemination. A template to aid in the preparation of systematic review protocols, such as a reporting guideline, may help achieve this. Furthermore, such guidance will enable authors to create a clear and complete document of their a priori methods, which may facilitate the registration of key information into the PROSPERO database. Building on an established guideline for systematic reviews and meta-analyses of studies evaluating health care interventions—the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA, http://www.prisma-statement.org) [12, 13]—we have developed PRISMA for Protocols (PRISMA-P) 2014. Table 1 summarizes the difference in intentions between PRISMA-P and PROSPERO.

Table 1 PROSPERO and PRISMA-P

The aim of PRISMA-P 2015 is to improve the quality of systematic review protocols, similar to the impact achieved by other reporting guidelines [1820]. By helping authors document an a priori road map of their systematic review, PRISMA-P also has the potential to improve the conduct of systematic reviews, as has been suggested of other reporting guidelines [21]. This Statement paper summarizes the development of the guideline and presents the PRISMA-P checklist.

Terminology

There is no standard definition for a systematic review and meta-analysis protocol, and we note that some terminology contained within these definitions may carry different meanings for different readers (i.e., ‘systematic search’). The terms ‘systematic review’, ‘meta-analysis,’ and ‘protocol’ are defined in Table 2. The former two terms are in accordance with the definitions reported in the PRISMA Statement [13] and are in line with those used by the Agency for Healthcare Research and Quality’s Evidence-based Practice Center (EPC) program [22], The Cochrane Collaboration [23], and the 2011 guidance from the Institute of Medicine [11]. The definition provided is a culmination of the terminology used by the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) 2013 initiative [24], the PROSPERO register, and the IOM Standards (Table 2).

Table 2 PRISMA-P terminology

Scope

The PRISMA-P checklist is intended primarily for the preparation of protocols of systematic reviews and meta-analyses that summarize aggregate data from studies, particularly the evaluations of the effects of interventions. There are many review types that are outside of this scope. As such, given the general lack of protocol guidance for other types of reviews, we encourage reviewers preparing any type of review protocol to make use of PRISMA-P as applicable. Readers can also use the checklist to assess the completeness of the reporting of published protocols. However, it is not recommended to use the checklist as an assessment tool to gauge the appropriateness of the methods of a systematic review protocol; it has not been validated for that purpose.

Development of PRISMA-P 2015

An international steering committee (MC, DG, AL, DM, MP, PS, and LAS) comprising members with wide-ranging experience in systematic review methodology, protocol registry development, and reporting guideline development led the development of PRISMA-P, coordinated by LS. The process proposed by the Enhancing the Quality and Transparency of Health Research (EQUATOR) Network was used to guide PRISMA-P development [27]. The process has 18 step-by-step recommendations grouped into five main stages:

  1. 1.

    Initial steps (determine the need for a reporting guideline);

  2. 2.

    Pre-meeting activities (identify contributors, conduct Delphi exercise, generate a list of potential items, and prepare for face-to-face meeting);

  3. 3.

    Face-to-face consensus meeting (present results of pre-meeting activities and relevant evidence);

  4. 4.

    Post-meeting activities (develop guidance Statement, Explanation and Elaboration document, and a publication strategy);

  5. 5.

    Post-publication activities (encourage uptake of guideline).

The first stage, ‘Initial steps,’ was described above; details of the remaining four steps are below.

Pre-meeting activities

In developing the PRISMA-P checklist, the steering committee compiled a list of items from various tools relating to the preparation of systematic review protocols for discussion at a consensus meeting of experts. Specifically, we mapped items from a Delphi exercise carried out during the development of PROSPERO [28], PROSPERO register items, PRISMA checklist items [13], SPIRIT 2013 checklist items [29], and items of IOM Standard 2.6 [11] against each other to identify unique and overlapping concepts. Lessons learned from the development of the SPIRIT checklist with respect to the concept and content of research protocols were used to guide discussion and debate at the meeting.

PRISMA-P consensus meeting

Twenty-three international experts attended the PRISMA-P consensus meeting on June 23–24, 2011, in Rockville, MD, USA to gain consensus on and reduce the number of potential PRISMA-P items. Delegates included journal editors, systematic review methodologists (including directors and representatives from international Cochrane Centres, Agency for Healthcare Research and Quality’s (AHRQ’s) Evidence-based Practice Centres, and the UK National Institute for Health Research), reporting guideline developers, information specialists, biostatisticians, and health research funders. Through group discussion at the meeting, 38 potential checklist items were reduced to 22.

Post-meeting activities

Following the meeting, the steering committee revised the draft 22-item checklist and refined their wording such that they accurately reflected meeting discussions. The draft checklist was also presented to the PROSPERO group, at a scientific meeting of the Cochrane Collaboration, for input and feedback and to AHRQ’s Learning Network. After each of these reviews, the steering committee made minor amendments to the items. The checklist was then circulated to all meeting invitees for critical input.

The PRISMA-P 2015 checklist

The final PRISMA-P 2015 checklist contains 17 numbered items (26 including sub-items) Items are categorized into three main sections: administrative information, introduction, and methods (Table 3).

Table 3 PRISMA-P 2015 checklist: recommended items to include in a systematic review protocol a

We made a conscious effort to harmonize the PRISMA-P checklist items with the items of the PRISMA checklist to facilitate authors in transitioning their protocol into a report of a systematic review. Thirteen PRISMA-P sub-items have existing PRISMA counterparts. Where PRISMA wording or content did not sufficiently address protocol reporting, checklist items were modified.

Readers familiar with PRISMA will notice that PRISMA-P does not contain a flow diagram documenting the flow of studies throughout the systematic review process. Such documentation is possible only after a review has been carried out and remains an essential component to include in the report of a completed systematic review or meta-analysis; for further guidance, see the PRISMA Explanation and Elaboration document [12].

We strongly recommend that the present document and the accompanying PRISMA-P 2015 Explanation and Elaboration document [30], which includes examples of good reporting, rationale, and evidence (where available), be read together with the PRISMA-P 2015 checklist.

PRISMA-P 2015 explanation and elaboration

Once the steering committee prepared the PRISMA-P 2015 Statement and checklist, they drafted the content of an Explanation and Elaboration document, with assistance from the larger PRISMA-P group. The explanatory text was derived largely from discussions at the PRISMA-P meeting (recorded at the time) as well as the PRISMA Explanation and Elaboration document [12]. Examples of well-reported PRISMA-P items came from protocols registered in the PROSPERO database, AHRQ’s EPC Program, and the Cochrane Database of Systematic Reviews or those published elsewhere. After the entire group had an opportunity to suggest additions, deletions, and changes, the steering committee combined all amendments to create the PRISMA-P 2014 Explanation and Elaboration document [30].

Post-publication activities

The post-publication activities recommended by EQUATOR include seeking and responding to criticism, encouraging the endorsement of and adherence to the guideline from various stakeholders, translating the guideline into other languages, evaluating its impact, ensuring website development, and updating of the guideline. The PRISMA-P 2015 checklist and related publications are freely available on the websites of the PRISMA Group (http://www.prisma-statement.org) and EQUATOR Network (http://www.equator-network.org). The PROSPERO register also contains a link to the guidance to encourage registrants to prepare a complete documentation of their protocol if they have not done so already.

We plan to develop an educational webinar about the rationale, usefulness, and potential impact of PRISMA-P, similar to what was done for PRISMA [31]. In addition, the potential for PRISMA-P 2015 to be used as an educational tool for authors, peer reviewers, and editors will be explored. Targeted implementation activities for PRISMA-P will be developed in a systematic manner together with experts in knowledge translation. The PRISMA website and social media (@PRISMAStatement, http://www.twitter.com/PRISMAStatement) will be used to make announcements about the launch of PRISMA-P and educational initiatives.

Endorsement

We encourage journals publishing systematic review products to modify their ‘Instructions for Authors’ section to endorse PRISMA-P 2015 and to consider publishing systematic review protocols, if they do not do so already. We plan to communicate with known endorsers of PRISMA (http://prisma-statement.org/endorsers.htm) as well as to other, relevant non-endorsing journals, to ask them to consider extending their support to PRISMA-P.

To help ensure optimal uptake by systematic reviewers, we propose a uniform endorsement policy across organizations and journals involved in the development and publication of systematic review protocols, demonstrated by the adoption of the following statement:

[this organization/journal] requires a completed PRISMA-P 2015 checklist as a condition of submission of systematic review protocols. We recommend that, while completing the PRISMA-P 2015 checklist, you ensure your protocol addresses all items. Taking the time to ensure that your protocol adheres to these basic reporting elements will improve your manuscript and potentially enhance its chances of eventual acceptance.’

Such a statement could be included in a journal’s ‘Instructions to Authors,’ or for funding agencies and those commissioning systematic reviews, in their Application Guidelines, recommending that applicants developing the proposals of systematic reviews for funding use PRISMA-P 2014. Peer reviewers and scientific committees can also use the checklist to gauge the extent to which protocols include necessary information.

As has been done for previous reporting guidelines [18, 32] we plan to evaluate whether and to what degree endorsement of PRISMA-P 2015 by journals (and potentially by other organizations) influences the completeness of reported protocols. Such an evaluation will be planned after allowing sufficient time for the wide dissemination of PRISMA-P 2015.

Implementation

The current system of implementing reporting guidelines is not optimal. At present, their primary mechanism of uptake is through endorsement by journals at their discretion, if at all. In journals that do endorse guidelines, language describing their support is often vague, leaving authors unclear on what they are supposed to do with a given reporting guideline during the submission process [33]. Furthermore, policies around how journal editors and peer reviewers should ensure and/or enforce adherence to reporting checklists are even less clear, if they exist at all [34]. Other barriers to implementation may include a lack of awareness of the guideline and perceived burden of using a reporting guideline checklist during the editorial process [35].

Some well-known checklists, such as PRISMA, include a column to the right of the main checklists in which users report the page number on which a specific item is reported. This was initially intended to help authors ensure each checklist item is addressed and to aid peer reviewers in locating reported text for each item within a document. However, this system is not optimal. One major problem is that peer reviewers still have to search within a considerable body of text to locate the exact text describing a checklist item. When multiple items are listed separately but reported together or vice versa, this problem is compounded, because exactly which content pertains to each item may remain unclear.

The lack of implementation and adherence to reporting guidelines is systemic; additional authorities encountered early in the research process should promote a clearer message about author adherence to reporting standards if improvements in reporting are to be made. In targeting protocols of systematic reviews, PRISMA-P has a unique opportunity to not only affect the way in which protocols are reported but to also impact the way in which reviews are eventually conducted, perhaps allowing for a more seamless transition into a completely reported systematic review.

To overcome known challenges with reporting guideline uptake [36, 37], we are developing a prospective implementation strategy for PRISMA-P 2015 using knowledge translation principles involving theoretically derived interventions [37] which have demonstrated effectiveness in the development of implementation interventions for clinical practice guidelines [38, 39]. An initial list of proposed stakeholders who can assist in the implementation of PRISMA-P, along with proposed actions and benefits, is provided in Table 4.

Table 4 Proposed stakeholders, actions, and potential benefits for supporting adherence to PRISMA-P

Discussion

Studies comparing trial protocols to final reports have widely documented both the presence and the extent of reporting biases in publications of randomized trials [2, 40]. Protocols for systematic reviews are rarely available for such comparisons, with the exception of select organizations. Of 288 reviews with available protocols in a 2006/2007 cohort, 64 (22%) were observed to have at least one discrepant outcome with their completed reviews; only 4 described reasons for the change in the completed review [3]. Discrepant outcomes added or upgraded from secondary to primary at the review stage were more likely to be statistically significant than those outcomes that had not changed. This practice (i.e., including, excluding, or changing outcomes in association with the strength or direction of findings) has the potential to bias the findings of any meta-analysis and the review’s conclusions. As review protocols are expected to become increasingly available with the advent of PROSPERO, clear reporting will become essential to facilitate the identification of discrepancies between protocol and review by readers and help them determine whether they need to be cautious in interpreting findings.

Reporting and publishing protocols is an important step in increasing the transparency of the research process and reliability of published papers. For example, some journals require a copy of the protocol as part of the peer review process of randomized trials. As of 1 March 2014, BioMed Central has published 4,158 trial protocols across 66 of its 258 open-access journals, including 1,026 in Trials. Systematic Reviews, a BioMed Central journal launched in February 2012, is committed to publishing systematic review products, including protocols [41], and has published 142 protocols since inception (to 8 June 2014).

Journals, granting agencies, and systematic review organizations are encouraged to endorse PRISMA-P 2015 in their ‘Instructions to Authors’ and guidance for applicants and to implement its use during their peer review process of systematic review proposals. Reviewers are encouraged to use the PRISMA-P checklist and Explanation and Elaboration [30] document to guide them through the documentation of a protocol. Doing so will enhance the completeness of reporting of review protocols, facilitate the assessment of potential in systematic reviews, and hopefully strengthen the methodological quality and reliability of completed systematic reviews.