Extended definition
Reporting guidelines are standardized checklists that specify the minimum set of information an article of a given study design must report so that the reader can judge its validity and, ideally, reproduce it. They do not say how to conduct the research, but what must appear in the report. Each design has its own. CONSORT, updated by Schulz and colleagues (2010), covers parallel-group randomized controlled trials, with a checklist and a participant flow diagram that have become standard. STROBE, by von Elm and colleagues (2007), does the same for observational cohort, case-control, and cross-sectional studies, in twenty-two items. To these are added PRISMA, for systematic reviews, ARRIVE, for animal studies, and dozens of others. Simera and colleagues (2010) describe the EQUATOR Network, which catalogs and disseminates these guidelines, showing that transparent and complete reporting increases the reliability and utility of research. The common principle is that a study can only be assessed if it is reported in full.
When it applies
Reporting guidelines apply when writing a manuscript, ideally from the start, as a roadmap for what each section must contain. CONSORT applies to randomized trials, STROBE to observational studies, PRISMA to systematic reviews, each according to the design. They apply to submission: many journals require the completed checklist and the page number of each item. They apply to peer review, as an objective reference for pointing out what is missing from the report. They apply to critical reading and evidence synthesis, where the absence of key items signals incomplete reporting. And they apply to teaching scientific writing, offering a verifiable standard instead of tacit convention. Consulting the EQUATOR Network is the way to find the right guideline for each type of study.
When it does not apply
Reporting guidelines do not apply as an assessment of study quality: a poorly designed trial can be reported completely, and following CONSORT does not make a flawed method valid. Reporting and rigor are distinct things, and confusing them is the most common error. They do not apply as a methodological straitjacket: they say what to report, not what to do. The wrong guideline does not apply to the wrong design; using STROBE for a randomized trial, for example, leaves out essential items. They do not apply as a substitute for risk-of-bias assessment, which is an analysis separate from the completeness of the report. And they do not apply as a guarantee of reproducibility on their own: reporting well is a necessary, not sufficient, condition for another researcher to redo the study.
Applications by field
- Clinical trials: CONSORT, with checklist and participant flow diagram required by much of the medical journal landscape.
- Epidemiology and observational studies: STROBE, for cohort, case-control, and cross-sectional designs.
- Evidence synthesis: PRISMA, the natural counterpart of systematic reviews and meta-analyses.
- Preclinical animal research: ARRIVE, aimed at the transparency of in vivo experiments.
Common pitfalls
The first pitfall is confusing complete reporting with methodological quality: following the guideline does not fix a bad design. The second is applying the wrong guideline to the study type, missing items specific to the correct design. The third is treating the checklist as a submission formality, filled in without the items actually being in the text. The fourth is assuming that reporting substitutes for risk-of-bias assessment, when they are complementary and distinct analyses. The fifth is ignoring the EQUATOR Network and improvising the report, when a consolidated guideline exists for nearly every design, ready to guide the writing from the first draft.