PRISMA 2020 Checklist and Compliance Tracker

A working audit for reporting completeness

A PRISMA checklist is useful only when every reporting claim can be located

The practical difficulty with PRISMA 2020 is not obtaining the checklist. It is deciding, item by item, whether the information requested by the checklist is actually present, whether all relevant parts of a multi-part item have been reported, and where another person can find the evidence. This tracker turns that problem into a documented audit rather than a retrospective tick-box exercise.

PRISMA 2020 contains 27 numbered items, many divided into sub-items and supported by a substantially more granular expanded checklist. The expanded checklist distinguishes elements considered essential from additional elements that may improve completeness and usability, and it identifies conditional elements that apply only in specified circumstances.¹² That structure matters because a broad item can appear to be “reported” while one of its important components is still absent.

This is not a theoretical concern. Hamilton and colleagues reviewed 24 tools that had been used to assess adherence to PRISMA 2020 across 2,766 systematic reviews. The assessment methods differed markedly, and only one tool was reported in enough detail to be independently implemented by researchers outside the original team.³ Ivaldi and colleagues reached a related finding from another direction: when 67 systematic reviews that reported using PRISMA 2020 were examined with the expanded checklist, none completely adhered under that study’s operational definitions, and mean adherence was 42.64%.⁴ That percentage is not a universal benchmark, but it illustrates how a granular audit can expose reporting omissions hidden by a high-level checklist.

How this tracker makes a reporting judgment

The official PRISMA checklist does not provide a universal scoring algorithm for deciding that an item is “complete,” “partial,” or “missing.” Published adherence studies have created their own response systems, denominators and aggregation rules, which is one reason reported adherence percentages are difficult to compare across studies.³ The definitions below are therefore explicit operational rules for this resource. They exist to make the audit repeatable, not to redefine PRISMA.

Reporting status and workflow status are intentionally separate. An item may appear complete but still require verification by a co-author. A partial item may already have a revision drafted but not yet checked. Conflating these states creates an audit record that is difficult to interpret.

How to use the audit without turning it into a score

Begin with the manuscript version you actually intend to submit, not an earlier draft. Work from the Methods and Results outward rather than simply ticking items in numerical order. For each row, locate the relevant reporting evidence, compare it with the official checklist and expanded guidance, then record a semantic location such as Methods > Search strategy, Figure 1, Supplement S2, or Results > Risk of bias. Page numbers can be added, but they are fragile because pagination changes during revision and typesetting.

Use the external-location field for registries, protocols, repositories and other research objects outside the manuscript. A URL by itself is not evidence that an item is satisfactorily reported. Open the destination and verify that the information described in the manuscript is genuinely present. This distinction is especially important for registration records, supplementary search strategies and availability statements.

When several authors are working on the report, one person can perform the first audit and another can verify complex rows. This is particularly useful for multi-part synthesis, risk-of-bias, reporting-bias and certainty items, where a short sentence may mention a method without supplying enough information for readers to reconstruct what was done.

Do not use “Not applicable” as a shortcut for a methodological activity that was omitted. For example, Items 24a and 24b explicitly allow the report to state that the review was not registered or that a protocol was not prepared. Those items therefore remain reportable even when the underlying research object does not exist. By contrast, results of sensitivity analyses are genuinely conditional on such analyses having been conducted.

After peer review, create or record the new manuscript version and repeat the rows affected by revision. A location that was accurate in the submission PDF can become wrong when sections, tables and supplements move. The audit should describe the current manuscript, not the manuscript that existed when the checklist was first completed.

A reporting audit should end with a manuscript change, not a completed form

The useful output of this tracker is not the filled table itself. It is the corrected manuscript and a recoverable record showing why each checklist row was judged complete, partial, missing or legitimately inapplicable. If a row is partial, revise the report. If it is missing, add the information or state transparently that the activity did not occur where that is the appropriate reporting response. If a location depends on a supplement or repository, verify the destination.

This approach also changes the role of the checklist at submission. Instead of asking an editor or reviewer to trust a declaration that the manuscript “followed PRISMA,” the authors can provide specific locations that can be checked. Journal implementation practices vary, and mandatory checklist submission does not itself prove that every reported location has been independently verified. A before-and-after study in one surgical journal did, however, observe an increase in measured PRISMA compliance from 48% to 76% after completed reporting-guideline checklists became mandatory, illustrating that implementation can influence reporting behavior while not establishing a universal causal effect across journals.⁵

The tracker therefore has a deliberately narrow job. It helps you control reporting completeness. It does not teach the statistical method, redesign the review, decide whether a risk-of-bias tool was appropriate, or write the manuscript for you. Those are separate methodological tasks.

Continue from the audit to the reporting decision

Frequently asked questions