Methods transparency: show the decisions behind a cohort study

1. Choose guidance that matches the design

STROBE addresses reporting of cohort, case-control and cross-sectional studies. Its authors distinguish reporting guidance from prescriptions for study design and from an instrument for assessing study quality. Use it to find missing explanations, not to score a manuscript as scientifically valid.

For MT24, the teaching design is a retrospective cohort assembled from clinic visits. A review checklist for another design cannot substitute for explaining how this cohort was selected, what was measured and how follow-up was determined.

Reference: STROBE: aims and use

2. Replace a vague sentence with inspectable decisions

The starting sentence is: “We reviewed clinic records and analysed outcomes.” It leaves the source population, time window, unit of observation and outcome undefined. An editor cannot infer those details from the study title.

3. Give each reporting statement an evidence source

A reader needs more than a reassuring reference to “electronic records”. Identify the fields and rules that implement each definition. STROBE's cohort checklist asks authors to explain the setting, selection, variables, data sources and analytical handling. The following mapping is original guidance, not an official checklist translation.

Reference: STROBE: cohort reporting checklist

On small screens, scroll the table horizontally, or focus it and use the arrow keys.

4. Do not turn missing observation into no event

A person whose follow-up record stops early is not automatically a person who had no return visit. In MT24, separate absence of a recorded outcome from evidence that the full 30-day observation window was available. Otherwise the wording may imply information the dataset does not contain.

The last eligible index visit is 30 June 2025. Under the example's calendar-date convention, a full 30-day follow-up reaches 30 July 2025. A data extract ending on 30 June cannot establish that complete window. Specify the observation boundary and seek the methods team's decision on incomplete follow-up; this example prescribes no missing-data or statistical method.

Reference: STROBE: cohort reporting checklist

5. Report four different evidence states

An adequately described procedure can remain unexecuted. A stored script may not run in another environment, and a successful run may not reproduce the reported result. Record the exact inputs, code version and environment when actual execution occurs, then compare the output with a stated target.

Restricted data can still have an explicit access procedure. Describe the restriction and who can assess an access request without promising that every reader may download participant records. An empty repository or an inaccessible link should not be called an available analysis package.

On small screens, scroll the table horizontally, or focus it and use the arrow keys.

6. Leave the reviewer a precise unresolved question

Ask the methods reviewer to examine the specific ambiguity: whether same-clinic records capture the intended outcome, whether observation is complete, or whether the chosen analytical method fits the question. Do not replace that discussion with “STROBE compliant”. Reporting clarity helps appraisal; it does not settle the underlying scientific decision.

Before submission, reconcile the paragraph, data dictionary and analysis account. Keep amendments and departures from the plan visible. MT24 supplies no sample size, approval number or result to copy into a real manuscript. Those statements must come from the real study's records.

Questions for a practical final check

• Is the cohort defined by explicit selection rules?
• Can each variable be traced to its source?
• Does follow-up extend far enough for the stated outcome?
• Are incomplete records distinguished from no event?
• Are description, access, execution and comparison reported separately?