CONSORT, STROBE, PRISMA: why reporting guidelines aren't optional

Three letters, three study designs

Most clinical and epidemiological research is now expected to follow one of a small number of reporting guidelines, depending on the study design. CONSORT applies to randomised controlled trials. STROBE applies to observational studies, including cohort, case-control and cross-sectional designs. PRISMA applies to systematic reviews and meta-analyses. Each provides a checklist, generally between twenty and forty items, specifying what a manuscript should report, from how the sample size was justified to how missing data were handled.

These guidelines are not a proprietary requirement from a single journal. They were developed, and are maintained, by international working groups of methodologists, and the large majority of biomedical journals now require authors to submit a completed checklist alongside the manuscript.

What a checklist actually protects against

It is tempting to treat these checklists as a bureaucratic step, filled in after the manuscript is written, mapping existing sentences to item numbers. Used this way, they add little value. Used properly, they are a specification for what the study report needs to contain, and they exist because incomplete reporting has a long, documented history of misleading readers. A trial that does not report how randomisation was concealed, or how many participants were excluded and why, cannot be properly assessed for bias, regardless of how sound the underlying study was.

In this sense, the guidelines protect the reader as much as the author. A well reported study, even one with limitations, allows those limitations to be judged. A poorly reported one does not.

The mistake of leaving it for the end

The most common problem we see is not disagreement with the checklist items, but timing. When a reporting guideline is consulted only while drafting the manuscript, several of its items are already impossible to satisfy properly, because the information was never collected or recorded during the study. A CONSORT flow diagram, for example, needs accurate counts of participants at every stage, from screening to analysis, which are far easier to track prospectively than to reconstruct afterwards. The same applies to a pre-specified statistical analysis plan, an explicit item in CONSORT, which cannot be produced retrospectively without undermining its own purpose.

Evidence that it matters beyond compliance

The case for reporting guidelines is not only methodological. Adherence has also been associated, in the published literature, with higher citation counts and greater downstream use of a study's results, independent of the underlying findings. This is consistent with a straightforward explanation: a clearly and completely reported study is easier for other researchers to evaluate, replicate and cite with confidence. Reporting quality, in other words, is not separate from the scientific value of a study. It is part of it.

A practical way to use the checklists early

A simple habit helps considerably: at the protocol stage, go through the relevant checklist item by item, and note, for each one, which document or data field will eventually provide the answer. Where the answer is not obvious, that is usually a sign that something needs to be added to the data collection plan, not to the eventual manuscript. Treated this way, CONSORT, STROBE and PRISMA stop being a compliance exercise and become part of the study design itself, which is closer to what they were intended to be.