A cohort report may describe eligible people, enrolled participants, completed follow-up and an analysis population. Those groups need not have the same size. Before repeating a percentage, identify which group supplies its denominator.
The STROBE cohort checklist asks authors to report participant numbers at successive study stages and reasons for non-participation. It also asks for missing-data counts for variables of interest and a summary of follow-up time. These prompts help a reader locate information; they do not substitute for evaluating the study’s methods.
Build a stage table
| Stage | Reading note |
|---|---|
| Eligibility and inclusion | Who could enter and who actually entered? |
| Follow-up | Who was observed for the relevant period? |
| Analysis | Who contributed to this particular result? |
Retain the report’s exact labels and page locations. If a count is absent, mark it absent rather than deriving it from a different table without justification.
Attach missingness to the variable
A person can contribute information on one variable while lacking another. Avoid describing all missingness with one unexplained dropout number. Inspect the variable and analysis the report is discussing, and preserve its stated approach to incomplete data.
If a denominator differs between tables, first check outcome, time point and eligibility for that analysis. A difference deserves an explanation; it is not by itself evidence of an error or misconduct.
Write a proportion with its base
A useful source note records the numerator, denominator, group and period together. When the report’s basis remains unclear, flag the specific percentage for further review instead of presenting it with false precision.
No cohort results were extracted here. The article provides a documentary reading method, not a causal judgement, disease-risk estimate or clinical recommendation. Its aim is to make the population behind a reported number visible to the next reader.
Primary reporting guidance read 21 October 2026. Suggested note structures and explicitly fictional examples are editorial. No clinical result is evaluated.
Source: STROBE cohort checklist items13/14. Related discussion: health information source quality.
Leave a Reply