Prisma 2020 checklist for meta analysis: A practical guide

The prisma 2020 checklist for meta analysis helps you report a systematic review clearly, from its rationale and methods to its findings. Used throughout the review rather than as a final form, it can help you capture decisions and analysis details while they are still easy to verify. PRISMA 2020 is a reporting guideline—not a manual for conducting reviews, a statistical-analysis standard, or a tool for judging methodological quality. [1][5]

This guide walks through using the checklist from protocol and search documentation to synthesis results and the flow diagram. The goal is to make your reporting traceable and complete, while keeping methodological decisions under human judgment. A completed checklist can show what you reported; it cannot, by itself, establish that the review was well conducted or free from bias. [1][5]

What do you need before starting?

Have the review protocol or current plan, search records, screening decisions, extraction files, risk-of-bias assessments, and analysis outputs available. You do not need to have completed the meta-analysis before beginning the checklist: PRISMA is most useful when it helps you record reporting details as the work proceeds. [1][3]

The main checklist has 27 items across seven sections. Supporting resources include a checklist for abstracts, an expanded checklist with explanations, and flow-diagram templates for original and updated reviews. Choose the relevant materials for your review, and use the expanded checklist when you need item-level guidance. [1][2]

Step 1: How should you set up the checklist?

Start by mapping each checklist item to the person, file, or workflow step that will supply the information. For example, search documentation should capture the sources searched, dates, and complete strategies; screening records should capture who reviewed records and reports and how disagreements were resolved. Also plan where you will document extraction procedures, risk-of-bias methods, and any automation tools used. [1][3]

This setup makes reporting part of the review process rather than a reconstruction exercise at the end. It is especially useful when several people or systems contribute to the work: the report needs to explain what was done, not merely name the software or provide a final result. Keep the protocol and any later changes accessible so you can distinguish planned methods from decisions made during the review. [1][3]

Common mistake to avoid: treating the checklist as proof of methodological quality. A report can address every item and still describe methods with limitations. Conversely, an omitted detail can make a sound decision difficult for readers to assess. Use PRISMA to improve transparency, not to certify conduct. [1][5]

Step 2: How should you document eligibility, searching, and selection?

Report the eligibility criteria and the information sources you searched. For every database, register, website, or other source, preserve the complete search strategy and the date searched. A brief description of search concepts is not a substitute for the full strategy: readers need enough detail to understand and, where applicable, reproduce the search. [1][3]

For selection, record how many reviewers assessed records and reports, how they worked independently or together, and how disagreements were handled. If automation tools supported searching, screening, or another review task, state how they were used. These details let readers see where human decisions entered the workflow and what role tools played. [1][3]

Keep the units of selection distinct. A record is an entry in a source, a report is a document about a study, and a single study may have more than one report. Reconcile counts at the level where they were generated rather than treating the three terms as interchangeable. [1][2]

Common mistake to avoid: reporting only the total number of “papers” found or included. That can obscure duplicate records, multiple reports from one study, and the path from search results to included studies. Use the appropriate unit at each stage and make the transitions clear in the flow diagram. [1][2]

Step 3: How should you report data collection and risk of bias?

Describe how data were collected from reports, including who extracted the data, whether reviewers worked independently, and how disagreements were resolved. Explain how you handled multiple reports of the same study and how you selected outcomes and results for reporting. Record the methods for assessing risk of bias, including who made assessments and how differences were addressed. [1][3]

Capture these decisions in the extraction and assessment workflow, not just in the manuscript draft. A traceable record can help distinguish information stated in a report from information inferred or obtained elsewhere, and it makes later checking more manageable. The checklist asks for transparent reporting of methods; it does not prescribe one extraction or risk-of-bias procedure for every review. [1][3][5]

Common mistake to avoid: naming an extraction tool or risk-of-bias instrument without explaining how it was applied. Readers also need to know who performed the work and how the process handled disagreement. A tool name alone does not describe the review method. [1][3]

Step 4: How should you describe each synthesis and meta-analysis?

For each synthesis, identify the contributing studies and explain how data were prepared and displayed. State the statistical method and why it was appropriate. Report planned and unplanned analyses, so readers can distinguish analyses specified in advance from those added or changed during the review. [3]

For a meta-analysis, report the model, estimation method, software and version, and methods used to assess heterogeneity. If you used a random-effects analysis, also identify the between-study variance estimator and the confidence-interval method. These details matter because “random effects” alone does not fully describe how the pooled result and its uncertainty were calculated. [3]

Make the effect measure and its rationale clear, and explain any interpretive thresholds used. Report the summary estimate and its precision, relevant heterogeneity measures, and the direction of effect when comparing groups. PRISMA does not set universal numerical cutoffs for heterogeneity or clinical importance, so do not imply that a single threshold is mandated by the guideline. [3]

Common mistake to avoid: reporting only the model name and pooled estimate. A reader also needs the estimation and interval methods, heterogeneity approach, contributing studies, and rationale for the effect measure and interpretation. Report the analytic choices that generated the result, not just the result itself. [3]

Step 5: How should you present study-level and analysis results?

Present individual-study results as well as synthesis results. A pooled estimate or forest plot alone is not enough: readers need to see the contributing study results and how they relate to the synthesis. Where applicable, report the estimate and its precision for each study in a way that supports comparison with the combined result. [3]

Report results from all subgroup analyses, meta-regressions, and sensitivity analyses that you conducted, including those that were not prespecified. Identify which analyses were planned and which were exploratory or added later. If an analysis was not conducted as planned, explain the change rather than presenting the final analysis as though it had always been specified. [3]

Also report the certainty of evidence and the review’s limitations. Keep these judgments distinct from the statistical estimate: the pooled result and its precision describe the synthesis, while certainty and limitations address how confidently readers should interpret the body of evidence and the review process. [1][4]

Common mistake to avoid: selectively presenting the most favorable or most familiar analysis. Omitting unplanned analyses or study-level results makes it harder to assess how robust the synthesis is and whether the reported interpretation reflects the full analysis record. [3]

Step 6: How do you complete the flow diagram accurately?

Use the flow diagram to show how records moved through identification, screening, retrieval, and inclusion, while preserving the distinctions among records, reports, and studies. One study may be represented by multiple reports, so the report count and study count can differ. Explain exclusions at the appropriate stage and ensure that the diagram agrees with the narrative and the underlying screening records. [1][2]

For an updated review, use the relevant flow-diagram template and make clear which information pertains to the update. Treat the diagram as a structured account of selection, not a substitute for describing eligibility criteria and selection methods in the text. [1][2]

Common mistake to avoid: forcing every count into a single “study” total. That can create apparent inconsistencies when multiple reports describe one study or when records are removed before reports are assessed. Reconcile counts using their correct units and retain an auditable record of how they were derived. [1][2]

Step 7: What else should you report before submission?

Check that the report covers registration and protocol amendments, competing interests, and access to data and other materials. Compare the methods stated in the report with the protocol and analysis records. Explain amendments and deviations clearly, including whether they affected eligibility, outcome selection, synthesis, or planned analyses. [1][4]

PRISMA 2020 replaced PRISMA 2009 and expanded reporting guidance in areas including abstracts, full search strategies, reviewer roles and automation, outcome and result selection, synthesis methods and results, certainty, exclusions, protocol amendments, competing interests, and availability of data and materials. Use the current checklist and supporting materials rather than relying on an older checklist or memory. [1]

A review platform such as Arakis can help organize decisions, files, and analysis records so that reporting details are easier to trace. Automation can support workflow, but it does not replace human oversight or methodological judgment, and completing checklist items does not guarantee review quality. [1][5]

Common mistake to avoid: treating a completed checklist as a final validation step. Cross-check the manuscript against the protocol, source records, and analysis outputs. The purpose is to make the review’s methods and findings transparent, including changes and limitations—not to imply that reporting completeness proves the underlying choices were correct. [1][5]

Key takeaways

• PRISMA 2020 is a reporting guideline, not a review-conduct manual, statistical standard, or methodological-quality appraisal tool. [1][5]
• Use the checklist during the review to capture search, selection, extraction, risk-of-bias, and automation details. [1][3]
• For each synthesis, report contributing studies, data preparation, statistical rationale, and planned and unplanned analyses. [3]
• A meta-analysis report should include the model, estimation and interval methods, software and version, heterogeneity methods, and study-level results. [3]
• Keep records, reports, and studies distinct when reconciling selection counts and completing the flow diagram. [1][2]
• A complete PRISMA report improves transparency but does not guarantee sound conduct or freedom from bias. [1][5]

FAQ

What is the PRISMA 2020 checklist for systematic reviews?

It is a reporting checklist for transparent, complete accounts of why a systematic review was conducted, how it was conducted, and what it found. It applies to reviews with or without meta-analysis, but it does not prescribe a complete review method or assess methodological quality. [1][5]

How do I use the PRISMA 2020 checklist in my systematic review?

Map checklist items to the people and records that will supply the information, then capture details as searching, screening, extraction, and analysis proceed. Before submission, check the report against the protocol and underlying records, including amendments and unplanned analyses. [1][3]

What are the key updates in the PRISMA 2020 guidelines?

PRISMA 2020 replaced PRISMA 2009 and expanded guidance on areas such as full search strategies, reviewer and automation reporting, outcome and result selection, synthesis, certainty, exclusions, protocol amendments, competing interests, and data and materials availability. It also provides an abstract checklist and flow-diagram templates. [1][2]

How does PRISMA 2020 improve reporting standards in meta-analysis?

It prompts authors to report how each synthesis was constructed, including contributing studies, statistical methods, model and estimation details, uncertainty, and heterogeneity. It also calls for individual-study results and results from conducted subgroup, meta-regression, and sensitivity analyses, rather than relying on a pooled estimate alone. [3]