How to complete the STROBE checklist for an observational study

What is the STROBE checklist?

STROBE stands for STrengthening the Reporting of OBservational studies in Epidemiology. The STROBE Statement was published in late 2007 in several journals at once, including The Lancet, BMJ, PLoS Medicine and Annals of Internal Medicine, with a long Explanation and Elaboration paper that gives the reasoning and a published example for every item. The EQUATOR Network lists STROBE as the main reporting guideline for observational studies.

The checklist has 22 numbered items, several with lettered sub-items. It covers three designs: cohort, case-control and cross-sectional studies. It tells you what information a reader needs to judge your study. It does not tell you how to design or analyse it, and it is not a quality score. A badly designed study can still report every item. A sound one can still be rejected for leaving items out.

If you are not sure that STROBE is the right guideline for your design, start with our guide to which reporting guideline your study needs.

Which STROBE checklist should I use: cohort, case-control or cross-sectional?

The STROBE website offers a combined checklist and three design-specific versions, plus a short checklist for conference abstracts. All of them have the same 22 items. Only a few items change wording by design:

Items 8, 13, 14 and 15 are marked with an asterisk in the checklist. For these, give the information separately for cases and controls in a case-control study, and for exposed and unexposed groups in a cohort or cross-sectional study where that applies.

Use the design-specific version that matches your study. A retrospective chart review that compares outcomes between two exposure groups is usually a cohort study, even though the data already existed. Say so in the title or abstract.

What does each STROBE item ask for, and where does it go?

The table summarises each item in plain language and shows where reviewers expect to find it. The wording is ours. Download the official checklist from the STROBE website for the exact text.

Items 12 and 16 are where most observational manuscripts lose points. For item 16, report both the crude and the adjusted estimate with a 95% confidence interval, and name every variable in the model. If you report an odds ratio, consider adding an absolute risk difference as well. Our guide to odds ratios, relative risks and hazard ratios explains which measure fits which design, and the guide to reporting P values and confidence intervals covers how to present precision.

How do I fill in a STROBE checklist?

1. Download the version for your design from the STROBE website (Word or PDF).
2. Finish the manuscript first. Turn on line numbers in your word processor so the page and line references will not change.
3. Go through the 22 items in order. For each item and sub-item, find where the manuscript reports it. Write the page and line numbers (for example, "p. 5, lines 98–104") or the section, table or figure.
4. Fix the gaps in the manuscript, not in the checklist. If an item is missing, add it to the text. A checklist entry that points to nothing is a red flag for the editor.
5. Mark items that do not apply as "not applicable", with a short reason (for example, "6b: no matching was used"). Do not leave rows blank.
6. Check the asterisked items (8, 13, 14, 15). Give the information separately by group.
7. Say in the Methods that you followed STROBE, for example: "This study is reported following the STROBE guideline for cohort studies."
8. Update the numbers after every revision. Line numbers move when you revise. Recheck the checklist before you resubmit.

Worked example: filling in four STROBE items (synthetic data)

All data below are synthetic and made up for illustration. They do not describe a real study.

A team reports a retrospective cohort of 4,500 singleton pregnancies at one hospital. The exposure is gestational diabetes and the outcome is preterm birth. Their checklist rows might read:

While checking item 13, the team notices that the abstract says 4,520 participants but Figure 1 and Table 1 say 4,500. The difference is 20 pregnancies excluded after a data-cleaning step that was never reported. Reviewers often catch this kind of mismatch. Fix it by adding the step to the flow diagram and making the abstract match.

Which STROBE extension applies to my study?

Extensions add items for specific methods or fields. You use them together with the main checklist, not instead of it. The EQUATOR Network's STROBE page lists the current set. These are the most common:

The EQUATOR page also lists newer or narrower extensions, such as STROBE-AMS (antimicrobial stewardship), STROBE-SIIS (sports injury surveillance) and STROBE-EQUITY (health equity). A retrospective database or chart-review study usually needs both STROBE and RECORD. Some journals require STROBE-MR for Mendelian randomization papers. As of October 2026, for example, PLOS ONE asks for a completed STROBE-MR checklist with page numbers and the relevant text.

How do reviewers and editors check STROBE?

Journals use the checklist in different ways. Some ask for it at submission and send it back if it is incomplete. Some pass it to peer reviewers as a map of the manuscript. Some use it only to check the Methods during editorial screening. The ICMJE Recommendations encourage journals to ask authors to follow STROBE for observational studies. JAMA, for example, names STROBE in its instructions for authors.

In practice, reviewers open the checklist next to the manuscript and spot-check the items that are most often weak: the flow of participants (13), missing data (12c, 14b), confounder selection (16a) and the limitations (19). A checklist that says "Methods" for every item, or page numbers that point to the wrong place, signals that the checklist was filled in without reading the paper. That costs credibility before the science is even read.

CliniPaper checks a manuscript against STROBE item by item, points to where each item is reported or missing, and recomputes the participant counts across the abstract, flow diagram and tables. Authors stay responsible for the content. If you use AI tools while writing, disclose it as your target journal requires.

Common mistakes reviewers flag

• The design is not named in the title or abstract, or it is called "retrospective study" without saying cohort, case-control or cross-sectional.
• Numbers that do not add up between the abstract, the flow diagram, Table 1 and the results text.
• Missing data are not reported, or a complete-case analysis is used without saying how many participants were dropped.
• Only adjusted estimates are reported, or the adjustment set is not explained.
• Confounders chosen by stepwise selection with no rationale, which reviewers question under item 16a.
• No study-size rationale. "All eligible patients" is acceptable if you say so, along with the period.
• Causal language ("X reduces Y") in a cross-sectional or retrospective design, against item 20's call for a cautious interpretation.
• The wrong guideline. A trial reported with STROBE, or an observational study with CONSORT. For randomised designs, use the CONSORT 2025 checklist.
• A checklist that points to nothing, or page numbers left over from an earlier version.