Author adherence and experiences with PRISMA-P 2015: a cross-sectional study

OBJECTIVE: To explore systematic review protocol authors' adherence to and experiences with the Preferred Reporting Items for Systematic reviews and Meta-analysis Protocols (PRISMA-P) 2015 reporting guideline. METHODS: We randomly sampled 100 non-overlapping systematic review protocols from May 2021 to May 2024: 50 from PubMed-indexed journals and 50 from Open Science Framework (OSF)/The International Prospective Register of Systematic Reviews (PROSPERO). Two authors independently assessed adherence to the 26 PRISMA-P 2015 items as fully, partially (i.e., when some, but not all, aspects of an item were reported according to predefined criteria), or not reported (or not applicable). We analysed adherence within the two protocol types separately. Forty-three systematic review protocol authors were invited for interviews on their experiences with using PRISMA-P 2015. We used a semi-structured interview guide, involving three predefined themes: level of experience, views on using PRISMA-P 2015, and reflections on the guideline's strengths and challenges. We applied framework analysis to the interview transcripts. RESULTS: Among the 50 PubMed protocols, the median proportion of fully reported applicable PRISMA-P items was 60% (interquartile range 52%-64%). The median proportion for OSF/PROSPERO protocols was 45% (interquartile range 38%-58%). Across both types of protocols, >25% were found not to report six items: protocol amendments, role of funder, criteria for quantitative synthesis, methods for planned summary other than quantitative, meta-bias(es), and confidence in the cumulative evidence. In both types of protocols, >25% were found to partially report 11 items. Fifteen authors participated in the interviews. Through the interviews, suggestions for updating PRISMA-P 2015 emerged. Most suggestions concerned either adding or modifying existing content, e.g., to report conflicts of interest, or clarify guidance on how to report data synthesis when no meta-analysis is planned; other suggestions were more general, e.g., to add links to the Elaboration and Explanation paper. CONCLUSION: Adherence to some PRISMA-P 2015 items was low among 50 PubMed systematic review protocols and even lower among 50 OSF/PROSPERO protocols. Six items were often not reported across both types of protocols. The interviewed authors suggested various additions and modifications of the guideline. Findings from this study provide context for users of the reporting guideline and will inform a forthcoming update of PRISMA-P 2015. PLAIN LANGUAGE SUMMARY: Researchers collect and summarize findings from several studies to better understand a specific health issue. These summaries are called systematic reviews. Healthcare guidelines are often based on results from systematic reviews. This makes systematic reviews very important for healthcare professionals and patients. To make sure that these results can be trusted and used, systematic reviews need to be done in a clear way. It needs to be so clear that other researchers can understand and repeat it. One way to do this is to write in detail how the review is planned before starting it. This is called a protocol. It can be useful with a guide to help authors write protocols with enough details. Such a guideline for writing protocols exists and is called PRISMA-P. It includes a checklist of 26 different items that should be described in a protocol. However, research shows that authors may not follow such guidelines closely. So far, no research has looked at how protocol authors use PRISMA-P or what makes it easier or harder to follow. In our study, we wanted to explore how protocol authors use PRISMA-P and what they think about it. We checked 100 protocols to see how well all 26 PRISMA-P items were described. We also interviewed 15 protocol authors about how they use PRISMA-P and if they had suggestions for improving it. We found that in more than 25% of the protocols, six specific items were not described. We also found that in more than 25% of the protocols, 11 other items were only partially described. The interviewed authors suggested many ways to improve PRISMA-P. For example, they suggested making instructions clearer on how to follow the individual items. They also suggested adding new content to PRISMA-P. Our findings will make it easier to read, write and evaluate protocols for systematic reviews. The findings can also be used for a future update of PRISMA-P.