Reporting Case-Control and Cohort Studies

Overview

Transparent reporting allows readers to determine what investigators intended, who entered the study, how exposure and outcome were measured, what follow-up occurred and whether the analysis corresponds to the design. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement provides a 22-item framework for cohort, case-control and cross-sectional reports, supported by an explanation and elaboration document.[1,2]

STROBE is a reporting guideline, not a quality score or a recipe that converts weak methods into valid evidence. Authors should use the design-specific checklist to disclose decisions and limitations, while reviewers should appraise the underlying study rather than count checked boxes.

This lecture focuses on reporting details that are especially important for case-control and cohort studies.

Why this matters

Terms such as “retrospective”, “matched” or “cohort” do not tell readers enough to reproduce the design. Reports may omit:

• the study base that produced cases and controls;
• whether cases were incident or prevalent;
• the control source and sampling scheme;
• matching criteria and matched analysis;
• cohort time zero and baseline outcome-free status;
• person-time, censoring and loss to follow-up;
• exposure windows and data provenance;
• missing-data denominators;
• crude and adjusted estimates with precision; or
• reasons participants were excluded after selection.

Omission prevents assessment of selection bias, temporal ordering, measurement and applicability.

Key concepts

Identify the design in the title or abstract

Readers should know whether the report is a cohort or case-control study. “Retrospective study” is insufficient because it describes timing, not the sampling structure. The abstract should state setting, participants, exposure, outcome, main estimate and uncertainty.

Describe the design early

The Methods section should explain design features before technical detail. For case-control studies, state the study base, case definition, control source and sampling. For cohorts, state cohort entry, exposure definition, follow-up and outcome ascertainment.

Report what happened, not only what was planned

A protocol describes intentions; the manuscript reports actual implementation. Deviations, changed sites, unavailable specimens, unmatched cases, loss to follow-up and analytical exclusions should be disclosed.

Separate participants, observations and events

A person may contribute several observations, repeated control selections or person-time intervals. Tables and flow diagrams should make the analytical unit and denominators clear.

Using STROBE appropriately

DECISION GUIDE

A practical reporting workflow

• Select the correct STROBE checklist for the observational design.
• Retrieve the protocol, analysis plan, data dictionary and flow logs.
• Draft the design and population before the statistical methods.
• Map each checklist item to a manuscript page or supplement.
• Use the explanation and elaboration document for item intent and examples.
• Report design-specific departures and sensitivity analyses.
• Submit the completed checklist when requested, but do not use it as a validity score.

STROBE includes title and abstract, background, objectives, methods, results, discussion and funding. Some items have separate instructions for cohort and case-control studies, especially participant selection, matching, follow-up and reporting of numbers.[1,2]

Reporting case-control studies

DESIGN PROFILE

Minimum case-control reporting logic

Core structure to reveal: Study base → case ascertainment → control sampling → exposure assessment → design-aware analysis.

Best suited for reporting: Incident or prevalent case status, control source, matching, index dates and odds-ratio interpretation.

Important limitation to disclose: Sampled case and control counts do not directly estimate population risk.

Cases

Report diagnostic criteria, incident or prevalent status, source, dates, sites, confirmation, exclusions and numbers. State whether assessors knew exposure history.

Controls

Explain why the control source represents the population that produced cases. Report the frame, eligibility at the index date, sampling method, number per case, replacement, participation and repeated selection where relevant.[3]

Matching

State individual or frequency matching, factors, categories or calipers, target and actual ratio, unmatched cases and the corresponding analysis. “Matched for age and sex” is incomplete.

Exposure assessment

Describe the etiologically relevant window, source hierarchy, interviewer blinding, validation and missingness separately by outcome group.

Reporting cohort studies

DESIGN PROFILE

Minimum cohort reporting logic

Core structure to reveal: Cohort entry → exposure or prognostic classification → at-risk follow-up → event or censoring → risk, rate or survival analysis.

Best suited for reporting: Time zero, follow-up duration, person-time, event ascertainment, loss and changing exposure.

Important limitation to disclose: Differential surveillance and informative censoring can threaten temporally ordered data.

Cohort entry and exposure

Report target and source populations, baseline eligibility, sampling, recruitment dates, time origin, at-risk status and exposure measurement. If exposure is determined over a grace period or landmark, explain alignment with follow-up.

Follow-up

State scheduled and passive sources, event definitions, censoring, competing events, administrative end, median or distribution of follow-up, person-time, losses and numbers at risk.[4]

Historical data

Describe databases, coverage, linkage, code lists, look-back, validation, calendar changes and whether data were collected for care or research. A supplementary code list or repository can improve reproducibility without overwhelming the main text.

Participant flow and descriptive tables

PROTOCOL CHECKLIST

Numbers to report

For case-control studies:

• potential and confirmed cases;
• eligible, approached, participating and analysed cases;
• controls sampled, eligible, approached, participating and analysed;
• matching success and set sizes;
• exclusions and missing key variables by group.

For cohort studies:

• frame or source population;
• eligible, invited, enrolled and included at baseline;
• exposure-group counts;
• events, deaths, withdrawals and losses;
• person-time or follow-up distribution;
• numbers included in each analysis.

Flow diagrams are useful even when not formally required. Baseline tables should describe groups and missing denominators but should not be dominated by significance tests. Differences are interpreted through subject-matter relevance and confounding structure, not p-values comparing baseline characteristics.

Design-specific Methods checklist

illustrative

Table 1. Information needed to reconstruct the design

Domain	Case-control report	Cohort report
Population	Study base and case-detection pathway	Source population and cohort entry
Comparison	Control source and sampling scheme	Exposure or prognostic groups
Time	Case index and control reference dates	Time zero, follow-up and censoring
Measurement	Prior exposure window	Baseline or time-varying exposure and outcomes
Analysis	Matching and odds-ratio interpretation	Risk, rate or survival estimand

Note: This table is illustrative and complements the STROBE checklist.

Statistical reporting

Report effect estimates with 95% confidence intervals and clearly named reference categories. Provide crude and adjusted estimates where they serve distinct purposes. State the adjustment set and rationale, not only “potential confounders”. Do not select variables solely because a univariable p-value passed a threshold.

For case-control studies, name logistic or conditional logistic methods and connect the odds ratio to the control-sampling design. For cohorts, identify risk, rate, survival or hazard measures, time scale, censoring and proportional-hazards assessment where applicable.

Missing data should be reported by variable and analysis. If multiple imputation, weighting or sensitivity analyses were used, provide the model variables and assumptions. Report all prespecified primary and secondary outcomes, not only statistically significant findings.

Bias, interpretation and generalisability

General epidemiological appraisal remains essential beyond checklist completion.[5]

The Discussion should address likely direction and magnitude of key biases where possible. Avoid a generic list of “recall bias, selection bias and confounding”. Explain the mechanism in the actual study and what was done to reduce it.

Generalisation should refer to the target and source populations, participation, settings and calendar period. A single-centre study may still be informative, but its applicability should not be claimed from sample size alone.

APPRAISAL CHECKLIST

Reviewing an observational manuscript

• Can the source population and participant pathway be reconstructed?
• Are time origin, index dates and exposure windows clear?
• Are measurements and data sources sufficiently reproducible?
• Are matching, sampling, clustering and repeated observations reflected in analysis?
• Are denominators and missing data visible throughout?
• Are effect estimates reported with uncertainty and correct interpretation?
• Do the limitations describe mechanisms relevant to the design?
• Are conclusions no stronger than the design and assumptions allow?

Ethical and governance considerations

Interface with statistical analysis

The Methods section should name the estimand before software. Report how matching, sampling weights, clustering, repeated selection, time-varying exposure, censoring and competing events were addressed. Provide enough detail for an informed analyst to reproduce the model.

The Results section should lead with the estimate and confidence interval, then the exact p-value where useful. Avoid presenting only a model coefficient without clinical units or only a p-value without direction and magnitude.

Reproducible supplements and codeMMed extension

Complex cohort and sampled case-control studies benefit from supplements containing code lists, causal diagrams, matching specifications, flow algorithms, variable derivations and analysis code. Sharing should comply with privacy and licences. When individual data cannot be shared, metadata and executable synthetic examples can still improve reproducibility.[2]

Authors should complete the checklist during drafting, not after acceptance. Early use often reveals missing counts, unclear dates and undocumented analytical decisions while source files are still accessible. A senior investigator should then compare the final manuscript with the protocol and analysis plan, identify deviations and ensure that supplementary material contains information that is essential for reproducibility rather than using the supplement to conceal core design details. The completed report should also identify where the protocol, analysis plan, code lists and any accessible analytical materials can be found.

Worked example

WORKED EXAMPLE

Repairing two incomplete Methods sections

The case-control manuscript replaces “cases and controls were matched” with a description of incident case surveillance, the district population register, age and district calipers, index dates, random control selection, actual set sizes and conditional logistic regression. A flow diagram shows cases without valid controls and non-participation separately.

The cohort manuscript replaces “patients were followed retrospectively” with cohort entry at six months postpartum, exposure classification before the landmark, linked follow-up sources, outcome definitions, censoring, death, loss of coverage and five-year administrative end. It reports person-years, median follow-up, numbers at risk and exposure-specific loss.

Both papers present crude and prespecified adjusted estimates with confidence intervals, identify missing-data denominators and explain the principal selection and surveillance mechanisms. The completed STROBE checklists point to manuscript pages but are not presented as quality scores.

Suggested reporting wording

Common mistakes

Knowledge check

What is the main purpose of STROBE?

Answer and reasoning

It improves transparent reporting of observational studies so readers can understand and appraise the methods and findings. It is not a tool that certifies validity or assigns a study-quality score.

What matching information should a case-control report include?

Answer and reasoning

State individual or frequency matching, factors, categories or calipers, intended and achieved ratios, unmatched cases, set handling and the design-appropriate analysis.

Why is 'five years of follow-up' insufficient for a cohort report?

Answer and reasoning

Readers also need entry and exit rules, actual follow-up distribution, person-time, events, losses, deaths, competing events and numbers supporting later estimates.

Should baseline tables test every group difference statistically?

Answer and reasoning

Usually no. Baseline characteristics should describe the groups and missingness. Confounding decisions should follow causal and subject-matter reasoning rather than automated significance testing.

What should a limitations paragraph say beyond naming selection bias?

Answer and reasoning

It should explain the study-specific mechanism, such as lower registry coverage among an exposure group, the likely consequence, mitigation undertaken and remaining uncertainty.

Key take-home messages

SUMMARY

Key take-home messages

• Use the design-specific STROBE checklist and explanation document while drafting.
• Reporting should expose the study base, selection, timing, measurement and analytical logic.
• Case-control reports require detailed case, control, matching and index-date information.
• Cohort reports require clear time zero, follow-up, person-time, censoring and loss information.
• Report denominators, missing data, crude and adjusted estimates and uncertainty transparently.
• Reporting completeness supports appraisal but does not establish methodological validity.

The current design-specific checklists are maintained through the official STROBE and EQUATOR resources.[6]

References

1. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology statement: guidelines for reporting observational studies. PLoS Med. 2007;4(10):e296. https://doi.org/10.1371/journal.pmed.0040296
2. Vandenbroucke JP, von Elm E, Altman DG, Gøtzsche PC, Mulrow CD, Pocock SJ, et al. Strengthening the Reporting of Observational Studies in Epidemiology: explanation and elaboration. PLoS Med. 2007;4(10):e297. https://doi.org/10.1371/journal.pmed.0040297
3. Wacholder S, McLaughlin JK, Silverman DT, Mandel JS. Selection of controls in case-control studies. I. Principles. Am J Epidemiol. 1992;135(9):1019–1028. https://doi.org/10.1093/oxfordjournals.aje.a116396
4. Xue X, Agalliu I, Kim MY, Wang T, Lin J, Ghavamian R, et al. New methods for estimating follow-up rates in cohort studies. BMC Med Res Methodol. 2017;17:155. https://doi.org/10.1186/s12874-017-0436-z
5. Rothman KJ, Greenland S, Lash TL. Modern epidemiology. 3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2008.
6. EQUATOR Network. STROBE statement: reporting guidelines for observational studies. Oxford: EQUATOR Network; accessed 6 August 2026.