What is STROBE? Start by choosing the version for your design
STROBE is a guideline for reporting observational studies so that readers can understand and assess them. The core checklist has 22 items, from the title and abstract to funding. It does not tell you how to design or conduct a study, and it is not a score of study quality. See the official scope and the 2007 statement.
The official checklist page offers a combined checklist, a fillable combined version, separate PDF and Word versions for cohort, case-control and cross-sectional studies, and a checklist for conference abstracts. If your target journal provides its own form, check that version and its submission requirements first.
- Cohort: a defined group is observed for exposures and later outcomes. A retrospective cohort built from existing medical records is still a cohort.
- Case-control: cases and controls are selected by outcome status and their exposures are compared. Comparing two groups does not make a study case-control on its own.
- Cross-sectional: exposures and outcomes are observed at one time point or period. Describe recruitment and sampling clearly.
Do not decide the design from the word “retrospective” or a single word in the title. Confirm the actual selection, measurement and observation structure with your co-authors. Whether you also need an extension for a particular data source or field, or a different reporting guideline, depends on the study and your journal.
The 22 STROBE items
To turn reporting items into a draft, see Methods subheadings and English examples and connecting the background, research gap and objective.
Item 1 covers the title and abstract, items 2–3 the introduction, items 4–12 the methods, items 13–17 the results, items 18–21 the discussion, and item 22 funding. The explanations below are based on the original guideline by von Elm and colleagues (2007); the common omissions and self-check questions are our own.
Check each sub-item (a, b, c and so on) under a number as well. Asterisked items in the original ask for information to be given separately for cases and controls, or for exposed and unexposed groups, where relevant. A single total cannot show every sub-item.
1. Title and abstract
Make the title and abstract show the design and main findings from the start.
Where it goes: Title / Abstract
- 1a: Name the study design with a commonly used term in the title or abstract.
- 1b: Give a balanced summary of what was done and what was found. Do not pick only favourable results.
Common omission: The main text describes a cohort, but the abstract does not name the design, or the number analysed differs from the main text.
Self-check: Is the design clear, and do the abstract’s population, methods, numbers and conclusion match the main text?
2. Background and rationale
Explain the problem the study addresses and what existing knowledge leaves open.
Where it goes: Introduction
- Connect the scope of previous studies to the questions that remain.
- State specifically which gap this study tries to fill.
Common omission: The introduction describes the general importance of the disease or topic at length but gives no reason for this study.
Self-check: Does the existing evidence lead to the reason this study was needed?
3. Objectives and hypotheses
State specific objectives and any prespecified hypotheses, so readers have a reference point for the results.
Where it goes: End of the Introduction
- Link the objective to the main population, exposure or predictor, and outcome.
- If there were prespecified hypotheses, state them and keep them separate from exploratory analyses.
Common omission: An explanation formed after seeing the results is written as if it were the original hypothesis.
Self-check: Is the main objective clear, and are prespecified hypotheses separated from post hoc exploration?
4. Study design
Present the key elements of the design early in the Methods so readers can interpret the rest.
Where it goes: Early in the Methods
- Describe the actual design, such as cohort, case-control or cross-sectional, and its selection and observation structure.
- Do not stop at the time direction (“prospective” or “retrospective”).
Common omission: The statistical methods are detailed, but it is unclear who was observed and how.
Self-check: Can readers understand the actual selection and observation structure early in the Methods?
5. Setting
Report where and when the study took place, and what each period means.
Where it goes: Methods — Setting
- Give the institutions, locations, care setting and relevant dates.
- If recruitment, exposure, follow-up and data collection periods differ, state each one.
Common omission: The date of record extraction is presented as the observation period, or the end of follow-up is missing.
Self-check: Are the location and the recruitment, exposure, follow-up and data collection periods distinguished?
6. Participants
Explain, in the way that fits the design, from which population and by which criteria the participants were selected.
Where it goes: Methods — Participants
- 6a: For a cohort, describe selection and follow-up methods; for a case-control study, how cases and controls were defined and selected and why; for a cross-sectional study, the sources and methods of selection.
- 6b: If matching was used, check the requirement for your design, such as the matching criteria and number exposed and unexposed (cohort) or the number of controls per case (case-control).
Common omission: Only the analysed sample is presented, with no source of controls or exclusion criteria.
Self-check: Did you describe eligibility, sources of selection, follow-up or control selection, and matching?
7. Variables
Define the main variables so that each term means the same in the text and tables.
Where it goes: Methods — Variables
- Clearly define outcomes, exposures, predictors, potential confounders and effect modifiers.
- Where relevant, give the diagnostic criteria and how each was determined.
Common omission: The definition of the outcome differs between the Methods and a table footnote.
Self-check: Can readers check the key variables and their diagnostic or classification criteria?
8. Data sources and measurement
Explain where each variable came from and how it was measured.
Where it goes: Methods — Data sources / Measurement
- Give the data source and the method of measurement or assessment for each variable.
- If measurement differed between groups, describe comparability. Check the original checklist’s asterisked items, which ask for information by group.
Common omission: The manuscript says data came from medical records but omits how they were measured, extracted or classified.
Self-check: Can readers find the source and measurement method for each variable and any differences between groups?
9. Bias
Report what was actually done to address potential sources of bias.
Where it goes: Methods — Bias
- Consider how selection, measurement and other biases could arise in this study.
- Describe the steps you took, and do not add steps that were not taken.
Common omission: Bias is mentioned in the limitations, but the Methods do not show how it was addressed.
Self-check: Did you describe the actual steps taken to address potential bias?
10. Study size
Explain how the study size was actually arrived at.
Where it goes: Methods — Study size
- If you calculated a sample size, give the assumptions and the basis for the decision.
- If you used all available data for a fixed period, say so and give its scope.
Common omission: The origin of the sample size is not explained, or a power calculation that was never done is added afterwards.
Self-check: Did you report the real basis for the study size?
11. Quantitative variables
Explain how continuous measurements were handled in the analysis.
Where it goes: Methods — Quantitative variables
- Describe the actual handling: original scale, transformation or grouping.
- If you created groups, give the cut-off values and why they were chosen.
Common omission: Tables show “high” and “low” groups, but there is no cut-off or reason for it.
Self-check: Are the transformation or grouping of numerical variables, the cut-offs and their rationale clear?
12. Statistical methods
Report the analysis methods needed to understand the results, from the main analysis to missing data and design-specific handling.
Where it goes: Methods — Statistical analysis
- 12a–12b: Describe the statistical methods, including control of confounding, and the methods for subgroups and interactions.
- 12c: Explain how missing data were addressed.
- 12d: Explain, as fits the design, how the analysis handled loss to follow-up (cohort), matching (case-control) or the sampling strategy (cross-sectional).
- 12e: If you ran sensitivity analyses, describe them.
Common omission: Only the software and significance level are given, with no model, confounder adjustment or handling of missing data.
Self-check: Did you describe adjustment, subgroups, missing data, design-specific handling and sensitivity analyses as needed?
13. Participants (flow)
Connect the numbers and reasons for exclusion from the first people screened to the final analysed sample.
Where it goes: Results — Participants / flow diagram
- 13a–13b: Report the number at each stage and the reasons for non-participation or exclusion.
- 13c: Consider using a flow diagram. Not every design needs the same diagram.
Common omission: There is no reason for the drop from the starting number to the final number, or the numbers differ between the flow diagram and the tables.
Self-check: Do the numbers and reasons connect through selection, exclusion, follow-up and analysis?
14. Descriptive data
Show the characteristics of the study participants and reveal gaps in information for each variable.
Where it goes: Results — baseline characteristics table
- 14a–14b: Report demographic, clinical and social characteristics, information on exposures and potential confounders, and the number with missing data for each variable.
- 14c: For a cohort, also summarise follow-up time.
Common omission: All tables appear to share one denominator, but missing numbers per variable are not shown.
Self-check: Can readers see the group characteristics, missing data per variable and, for a cohort, follow-up time?
15. Outcome data
Present what was observed, in which group and how often, in the form that fits the design.
Where it goes: Results — outcome / exposure data
- For a cohort, report numbers of outcome events or summary measures over time; for a case-control study, numbers in each exposure category or summary measures of exposure.
- For a cross-sectional study, report numbers of outcome events or summary measures.
Common omission: Only percentages are given, without the denominator, the number of events or the period they cover.
Self-check: Are the event or exposure data right for the design, with clear denominators and periods?
16. Main results
Report which analysis each main estimate comes from and how uncertain it is.
Where it goes: Results — main results
- 16a: Give unadjusted estimates and, where relevant, adjusted estimates with their precision, which confounders were adjusted for, and why they were included.
- 16b: If continuous variables were categorised, report the category boundaries.
- 16c: If relevant, consider translating relative risk into absolute risk for a meaningful time period, based on the actual data.
Common omission: Only p values are given, without effect estimates, confidence intervals or the reason for the adjustment variables.
Self-check: Are the estimates, uncertainty, adjustment factors and category boundaries linked to the results?
17. Other analyses
Report the results of analyses beyond the main analysis, together with their purpose.
Where it goes: Results — other analyses
- Describe the additional analyses actually done, such as subgroup, interaction and sensitivity analyses.
- Separate prespecified analyses from post hoc exploration.
Common omission: A sensitivity analysis described in the Methods has no result, or an exploratory finding becomes the main conclusion.
Self-check: Do the methods and results of additional analyses correspond, and is their exploratory nature clear?
18. Key results
Summarise the key results with reference to the study objectives.
Where it goes: Start of the Discussion
- Answer the research question instead of repeating the results tables.
- Separate the main results from secondary observations.
Common omission: The Discussion opens with a result that does not relate to the stated objective.
Self-check: Does the summary of key results answer the original objective directly?
19. Limitations
Explain specifically how bias and imprecision could affect the results.
Where it goes: Discussion — Limitations
- Where possible, discuss the direction and size of any potential bias.
- Connect constraints such as small samples, missing data or measurement limits to how they restrict interpretation.
Common omission: The limitations only list “retrospective” or “single-centre” without saying how this affects the findings.
Self-check: Did you explain how the main limitations arise and how they could affect the results?
20. Interpretation
Interpret the results cautiously, considering the objectives, limitations, multiple analyses and evidence from other studies.
Where it goes: Discussion / Conclusion
- Do not present an observed association as a causal effect that the design cannot support.
- Consider agreement and disagreement with previous studies, multiplicity of analyses and remaining uncertainty.
Common omission: Only significant results are emphasised, or the conclusion suddenly states a clinical effect as established.
Self-check: Is the strength of the claims balanced against the design, results and limitations?
21. Generalisability
Discuss to which settings and populations the results may apply.
Where it goes: Discussion
- Base the scope on participant selection, institutions, period and setting.
- State the limits of interpretation for groups not included in the data.
Common omission: Results from a limited sample are applied to all patients and regions.
Self-check: Did you explain the populations and settings the results apply to, and the limits?
22. Funding
Give the source of funding and the role of the funders in the study.
Where it goes: Funding / Declarations
- Report the funding and the funders’ role for the present study.
- If applicable, also report the funding of the original study on which the present article is based.
Common omission: Only the grant name is given, with no role of the funder in design, analysis or publication.
Self-check: Did you state the funding source and the funders’ role accurately?
Cohort, case-control and cross-sectional: what is different?
The numbers are the same, but items 6, 12, 14 and 15 ask for different information by design. Compare the differences below with the original. This table summarises key differences and does not replace every sub-item. Design-specific checklists
| Item | Cohort | Case-control | Cross-sectional |
|---|---|---|---|
| 6 Participants | Selection and follow-up methods; matching criteria and numbers in each group if matched | Definition and selection of cases and controls and the rationale; matching criteria and number of controls per case if matched | Sources and methods of selection |
| 12d Statistical methods | How loss to follow-up was addressed | How matching was addressed | How the sampling strategy was taken into account |
| 14c Follow-up time | Summary of follow-up time, such as average and total amount | Do not apply the cohort follow-up requirement automatically | Do not apply the cohort follow-up requirement automatically |
| 15 Outcome data | Numbers of outcome events or summary measures over time | Numbers in each exposure category, or summary measures of exposure | Numbers of outcome events or summary measures |
For example, applying the cohort follow-up requirement to case-control data produces a false “missing” judgement. Conversely, if you matched, you need to describe both the matching criteria and how matching was handled in the analysis. “Not applicable” is not a convenient choice; you should be able to explain it from the study structure.
One-page self-check for your design: cohort · case-control · cross-sectional
Self-check before submission and how to record page and line numbers
Copy this table into your own document and record the page and line numbers in your manuscript and what you still need to add. Check the sub-items in the explanations above and in the original checklist. This table is a learning summary and does not replace a journal’s submission form.
On small screens, scroll the table sideways. Record each status as checked, needs work, undecided, or not applicable, and note why an item does not apply. These are the table’s own labels, not Paper Buddy’s AI results.
Prints only this self-check table on a single page. Choose “Save as PDF” in the print dialog to get a PDF file. Choose a design first to print the version for that design.
STROBE self-check · CombinedCohortCase-controlCross-sectional
Manuscript: ____________________Checked by: ____________________Date: ____________________
Status: checked · needs work · undecided · not applicable (note why)
| Item | Question to ask of your manuscript | Location · status · notes |
|---|---|---|
| 1. Title and abstract | Is the design clear, and do the abstract’s population, methods, numbers and conclusion match the main text? | Location: ______ Status: ______ Notes: ______ |
| 2. Background and rationale | Does the existing evidence lead to the reason this study was needed? | Location: ______ Status: ______ Notes: ______ |
| 3. Objectives and hypotheses | Is the main objective clear, and are prespecified hypotheses separated from post hoc exploration? | Location: ______ Status: ______ Notes: ______ |
| 4. Study design | Can readers understand the actual selection and observation structure early in the Methods? | Location: ______ Status: ______ Notes: ______ |
| 5. Setting | Are the location and the recruitment, exposure, follow-up and data collection periods distinguished? | Location: ______ Status: ______ Notes: ______ |
| 6. Participants | Did you describe eligibility, sources of selection, follow-up or control selection, and matching?
| Location: ______ Status: ______ Notes: ______ |
| 7. Variables | Can readers check the key variables and their diagnostic or classification criteria? | Location: ______ Status: ______ Notes: ______ |
| 8. Data sources and measurement | Can readers find the source and measurement method for each variable and any differences between groups? | Location: ______ Status: ______ Notes: ______ |
| 9. Bias | Did you describe the actual steps taken to address potential bias? | Location: ______ Status: ______ Notes: ______ |
| 10. Study size | Did you report the real basis for the study size? | Location: ______ Status: ______ Notes: ______ |
| 11. Quantitative variables | Are the transformation or grouping of numerical variables, the cut-offs and their rationale clear? | Location: ______ Status: ______ Notes: ______ |
| 12. Statistical methods | Did you describe adjustment, subgroups, missing data, design-specific handling and sensitivity analyses as needed?
| Location: ______ Status: ______ Notes: ______ |
| 13. Participants (flow) | Do the numbers and reasons connect through selection, exclusion, follow-up and analysis? | Location: ______ Status: ______ Notes: ______ |
| 14. Descriptive data | Can readers see the group characteristics, missing data per variable and, for a cohort, follow-up time?
| Location: ______ Status: ______ Notes: ______ |
| 15. Outcome data | Are the event or exposure data right for the design, with clear denominators and periods?
| Location: ______ Status: ______ Notes: ______ |
| 16. Main results | Are the estimates, uncertainty, adjustment factors and category boundaries linked to the results? | Location: ______ Status: ______ Notes: ______ |
| 17. Other analyses | Do the methods and results of additional analyses correspond, and is their exploratory nature clear? | Location: ______ Status: ______ Notes: ______ |
| 18. Key results | Does the summary of key results answer the original objective directly? | Location: ______ Status: ______ Notes: ______ |
| 19. Limitations | Did you explain how the main limitations arise and how they could affect the results? | Location: ______ Status: ______ Notes: ______ |
| 20. Interpretation | Is the strength of the claims balanced against the design, results and limitations? | Location: ______ Status: ______ Notes: ______ |
| 21. Generalisability | Did you explain the populations and settings the results apply to, and the limits? | Location: ______ Status: ______ Notes: ______ |
| 22. Funding | Did you state the funding source and the funders’ role accurately? | Location: ______ Status: ______ Notes: ______ |
- Choose the design and the form first. Even with the combined checklist, mark the requirements for your design.
- Find the sentence that reports it. Record the actual page and line, table, footnote or supplement, not a broad location such as “Methods”. Items 12c and 12d may be in different places.
- Find out why a cell is empty. Was it done but not described, is the information unavailable, or does the item not apply to the design? Never describe a procedure that was not done just to fill a cell.
- Check again in the final file. If editing or Word export moves text, update the locations from the submission version. Refer to supplements in the way the journal allows.
Finally, compare the sample sizes, event counts, denominators and percentages across the abstract, text, tables and flow diagram. If the units of analysis differ, such as patients versus eyes, tests or specimens, the difference should be visible. When numbers do not match, check the original data and analysis; do not just edit the sentences until they agree.
Download the official PDF or Word checklist for submission →
Two ways to check STROBE with AI
Give a general chatbot the manuscript and the original checklist
In a service your institution allows, provide the parts of the manuscript you need and the checklist for your design, and ask for the supporting sentence, location and open questions for each item. Below is an example request we wrote. Check whether the tool read the whole document and what was attached, and do not enter identifiable patient information or data you are not allowed to share.
The study design of this manuscript, as confirmed by the authors, is [design]. For each applicable item of the attached STROBE checklist, find the supporting sentence and its location. Distinguish content that cannot be confirmed from the text provided from content that is actually missing, and give the reason when you cannot judge. Do not fill gaps by inventing study procedures or numbers.
Review key reporting items with Paper Buddy Health Check
Paper Buddy Health Check reviews key STROBE-based reporting items in the manuscript, together with the consistency of statistical reporting, claims and evidence, and what to revise first. It takes the observational design into account, but authors make the final call on the applicable guideline and whether the study is cohort, case-control or cross-sectional. See the feature description and a real result excerpt for its scope.
Results are shown as met, missing, unclear, not evaluable or not applicable. “Not evaluable” means the item could not be confirmed from the manuscript text or evidence provided; do not read it as “not done”. The counts in the published example come from that one manuscript and are not a general accuracy or performance figure.
This guide explains all 22 items, but the product checks selected key reporting items. It does not check every item on the official checklist or guarantee compliance. Authors still need to check remaining sub-items, any extensions and the journal’s submission requirements.
Whatever tool you use, compare the results yourself
Read whether the sentence the AI quoted actually supports the whole item, and check facts outside the manuscript against the real data. Complete reporting and a valid study or analysis are different judgements. Do not use AI output as a statistical reanalysis of raw data or a prediction of acceptance.
Record the tools you used and what they did, and check ICMJE guidance on AI use by authors and your journal’s current disclosure policy. See our AI disclosure guide for where the statement goes, and the comparison of AI checking approaches.
Frequently asked questions
What is the STROBE checklist and how many items does it have?
STROBE is a guideline of 22 items to check when reporting cohort, case-control and cross-sectional studies. Several numbers have sub-items, so checking the main numbers alone does not cover every requirement. It is also not a scoring tool for study design or quality.
Which STROBE checklist should I use for a retrospective chart review?
“Retrospective” does not decide the design on its own. Look at the actual selection and observation structure: do you follow exposure to outcome, select cases and controls by outcome, or observe status at one point in time? Settle the design with your co-authors and a methods reviewer, then choose the matching checklist.
How do I fill in the page number column?
Record the page and line, or the table or supplement, where each sub-item is actually described in the final submission file. Do not submit the locations from an earlier draft. For items that do not apply, note why; for items you cannot locate, check what was actually done and then add the description.
Is a STROBE flow diagram required?
Item 13c asks authors to consider using a flow diagram. It does not require the same figure in every manuscript. When selection, exclusion, follow-up and analysis steps are complex, a flow diagram can show the numbers and reasons clearly. Check your journal’s own requirements as well.
Can AI check the STROBE checklist?
You can give a general chatbot the manuscript and the checklist for your design, or use a dedicated checking feature. Paper Buddy Health Check also reviews key STROBE-based reporting items. None of this is a complete automated check of all 22 items or a guarantee of compliance; authors compare the results with the original checklist and the manuscript.
Has STROBE been revised recently?
This guide follows the core STROBE checklist on the official website and the 2007 statement. Some fields, data sources and designs have extension guidelines. An extension is not a revision of core STROBE; check which guideline and version your journal names.
Sources and how we checked them
The Paper Buddy team wrote these explanations and questions from the official sources below, checked on October 18, 2026. Before you submit, recheck your target journal’s current instructions and the original checklist.
- STROBE official website — purpose, scope and use
- STROBE official checklists — combined and design-specific PDF and Word versions
- von Elm et al. The STROBE Statement (2007), PLOS Medicine — original guideline and checklist
- ICMJE — AI use by authors and disclosure
- Paper Buddy — what Health Check covers and its limits