Observational Study Report Template (STROBE)

Cohort, case-control or cross-sectional report by STROBE topics: variables, bias, OR/RR formulas, adjusted estimates with CI

科学研究フルSTROBEobservational study math更新 2026/10/20

エディタで開く.mdをダウンロード

使い方: Use when reporting a cohort, case-control or cross-sectional study. Pick the design first and define every variable and the confounder set before writing results. Main pitfall: causal language and adjusted estimates without saying which confounders were included and why.

プレビュー

Observational Study Report: [Short title of the study]

Report template for cohort, case-control and cross-sectional studies. The sections follow the topics of the STROBE Statement (title and abstract, introduction, methods, results, discussion, other information), written in original words for this collection. Use it with the official STROBE checklist for your design from strobe-statement.org. This is a reporting format, not clinical, epidemiological or statistical advice. Replace every [bracketed] value and every row marked Example.

Document control

Field	Value
Document ID	[OBS-STUDY-001]
Study design	[Cohort / Case-control / Cross-sectional]
Protocol or registration	[Registry, OSF link or "not registered"]
Ethics reference	[REC / IRB number or exemption]
Version	[0.1]
Status	[Draft / In review / Approved]
Owner	[@principal-investigator]
Approvers	[@study-statistician], [@co-investigator]
Last updated	YYYY-MM-DD
Classification	[Confidential until publication]

Approvals

Role	Name	Decision	Date
Principal investigator	[@name]	[Approved]	YYYY-MM-DD
Study statistician	[@name]	[Approved]	YYYY-MM-DD
Data custodian	[@name]	[Approved for release of aggregate results]	YYYY-MM-DD

Revision history

Version	Date	Author	Change summary
0.1	YYYY-MM-DD	[@name]	First draft after analysis
0.2	YYYY-MM-DD	[@name]	Example: Added E-value sensitivity analysis for unmeasured confounding
1.0	YYYY-MM-DD	[@name]	Approved for submission

Title and abstract
Introduction
Methods
Results
Discussion
Other information
Appendices

1. Title and abstract

The title states the design; the abstract gives an informative, balanced summary.

Title: [Exposure] and [outcome] in [population]: a [prospective cohort / population-based case-control / cross-sectional] study

Heading	Content
Objective	[Research question in one sentence]
Design and setting	[Design, data source, country, years]
Participants	[Who and how many]
Exposure and outcome	[Definitions in brief]
Results	Example: Among 18,240 adults followed for a median 6.1 years, 1,032 developed [outcome]. The adjusted hazard ratio for [exposure] was 1.34 (95% CI 1.15 to 1.56).
Conclusions	[Interpretation proportionate to the design]

2. Introduction

The introduction explains the background and states what the study set out to find.

2.1 Background and rationale

[Summarise what is known about the exposure-outcome relationship, the gaps or inconsistencies in the evidence and why an observational design is appropriate.]

2.2 Objectives

ID	Objective	Pre-specified hypothesis
OBJ-01	Example: Estimate the association between [exposure] and incident [outcome]	Higher exposure is associated with higher risk
OBJ-02	[Secondary objective]	[Hypothesis or "exploratory"]

3. Methods

This section explains how the study was designed and analysed.

3.1 Study design

[State the design and its key elements, for example: prospective cohort nested within [cohort name], with exposure measured at baseline and outcome ascertained through linkage to [registry].]

The design determines which measures and checklist items apply:

Design	Participants sampled by	Typical measure	Key validity question
Cohort	Exposure status, then followed for outcome	Risk ratio, rate ratio, hazard ratio	Was follow-up complete and similar across exposure groups?
Case-control	Outcome status, then exposure measured	Odds ratio	Do controls represent the population that produced the cases?
Cross-sectional	Neither; one point in time	Prevalence, prevalence ratio, odds ratio	Can the order of exposure and outcome be established?

3.2 Setting

Item	Detail
Location	[Country, region, facilities]
Recruitment period	YYYY-MM-DD to YYYY-MM-DD
Exposure assessment period	[Dates]
Follow-up period	Example: From baseline until outcome, death, emigration or 2024-12-31, whichever came first
Data collection	[Dates]

3.3 Participants

Design	Item	Detail
Cohort	Eligibility, sources and selection	[Criteria and methods]
Cohort	Follow-up methods	[Active follow-up, linkage]
Cohort (matched)	Matching criteria, number exposed and unexposed	[Criteria]
Case-control	Case ascertainment and diagnosis	[Definition and source]
Case-control	Control selection and rationale	Example: Four population controls per case, matched on age (+/- 2 years), sex and region, sampled by incidence density
Cross-sectional	Eligibility, sources and selection	[Criteria and sampling frame]

3.4 Variables

ID	Variable	Role	Definition	Type
VAR-01	Example: Incident type 2 diabetes	Outcome	First diagnosis code [codes] or HbA1c of at least 48 mmol/mol	Time to event
VAR-02	[Exposure]	Exposure	[Definition, categories]	[Categorical]
VAR-03	Age	Confounder	Years at baseline	Continuous
VAR-04	[Smoking status]	Confounder	[Never / former / current]	Categorical
VAR-05	[Sex]	Effect modifier	[Recorded sex]	Binary

A directed acyclic graph helps justify the confounder set:

3.5 Data sources and measurement

Variable	Source	Measurement method	Comparability across groups
[Exposure]	[Questionnaire / records / biomarker]	[Instrument, validation reference]	[Same method for all]
[Outcome]	[Registry / adjudication]	[Coding system, adjudication process]	Example: Adjudicators blind to exposure status
[Confounder]	[Source]	[Method]	[Comment]

3.6 Bias

Bias	How it could arise	Mitigation	Likely direction if residual
Selection bias	[Healthy volunteer effect]	[Population-based sampling]	[Towards null]
Information bias	Example: Recall of exposure differs between cases and controls	Exposure taken from records made before diagnosis	Not applicable
Confounding	[Unmeasured lifestyle factors]	[Adjustment; E-value]	[Away from null]
Loss to follow-up	[Differential attrition]	[Linkage-based follow-up]	[Unclear]

3.7 Study size

[Explain how the size was determined. For example: all eligible members of the cohort were included; with an expected 1,000 events, the study could detect a hazard ratio of 1.20 with 80% power at a two-sided alpha of 0.05.]

3.8 Quantitative variables

Variable	Handling	Rationale
[Age]	Example: Restricted cubic spline with 4 knots	Non-linear relationship expected
[Exposure]	[Quartiles of the distribution in controls]	[Rationale]
[Body mass index]	[WHO categories]	[Clinical convention]

3.9 Statistical methods

The measures of association used depend on design. From a 2 x 2 table with exposed cases $a$ , exposed non-cases $b$ , unexposed cases $c$ and unexposed non-cases $d$ :

\text{RR} = \frac{a/(a+b)}{c/(c+d)}, \qquad \text{OR} = \frac{a\,d}{b\,c}

The 95% confidence interval for the odds ratio (Woolf method) is:

\exp\!\left(\ln \text{OR} \pm 1.96\sqrt{\tfrac{1}{a} + \tfrac{1}{b} + \tfrac{1}{c} + \tfrac{1}{d}}\right)

Item	Method
Main model	Example: Cox proportional hazards with age as the time scale; proportional hazards checked with Schoenfeld residuals
Confounder adjustment	[Covariates chosen from the DAG in 3.4]
Subgroups and interactions	[Pre-specified; likelihood ratio test]
Missing data	[Multiple imputation by chained equations, 20 data sets]
Loss to follow-up / matching	[Censoring rules; conditional logistic regression for matched sets]
Survey sampling	[Weights, strata, clusters for cross-sectional surveys]
Sensitivity analyses	[Complete case; excluding first 2 years of follow-up; E-value]
Software	[R / Stata / SAS version]

4. Results

This section reports who was studied and what was found.

4.1 Participants

The flow diagram shows numbers at each stage of the study.

Stage	n	Reasons for exclusion or non-participation
Potentially eligible	[24,105]
Excluded	[5,865]	Example: 3,210 prevalent outcome at baseline; 1,904 missing exposure; 751 declined
Included	[18,240]
Analysed	[17,828]	[Reasons]

4.2 Descriptive data

Characteristic	Exposed (n = [x])	Unexposed (n = [y])	Missing, n
Age, years, mean (SD)	Example: 57.3 (8.9)	55.1 (9.4)	0
Female, n (%)	[Value]	[Value]	[0]
Current smoker, n (%)	[Value]	[Value]	[214]
[Confounder]	[Value]	[Value]	[n]

Follow-up: [median (IQR) years and total person-years].

4.3 Outcome data

Group	Events / participants	Person-years	Rate per 1,000 person-years
Exposed	Example: 612 / 8,904	53,420	11.5
Unexposed	420 / 8,924	55,310	7.6

4.4 Main results

Model	Estimate (95% CI)	Adjusted for
Unadjusted	Example: HR 1.52 (1.34 to 1.73)	None
Model 1	HR 1.41 (1.24 to 1.61)	Age, sex
Model 2 (main)	HR 1.34 (1.15 to 1.56)	Model 1 plus smoking, socioeconomic status, [others]

Category boundaries for categorised continuous variables: [list]. Absolute risk difference over [10 years]: [value (95% CI)].

4.5 Other analyses

Analysis	Pre-specified?	Result	Comment
By sex	Yes	[HR women 1.29; men 1.40; interaction P = 0.52]	No effect modification
Excluding first 2 years	Yes	[HR 1.31 (1.11 to 1.54)]	Reverse causation unlikely
E-value for main estimate	No	[2.02]	[Interpretation]

5. Discussion

The discussion interprets the findings with appropriate caution.

5.1 Key results

[Summarise the main findings with reference to the study objectives.]

Example: In this population-based cohort, adults with [exposure] had a 34% higher adjusted rate of [outcome] than those without, corresponding to about 3.9 additional cases per 1,000 person-years. The association was consistent across sex and age groups and persisted after excluding the first two years of follow-up.

5.2 Limitations

[Source of potential bias or imprecision, its likely direction and magnitude]
[Residual and unmeasured confounding]
[Measurement error in exposure, for example a single baseline measurement that may not reflect long-term exposure, which usually biases estimates towards the null]
[Missing data and the plausibility of the missing-at-random assumption]
[Multiple comparisons across secondary outcomes and subgroups]

For each limitation, state whether it is likely to have inflated or attenuated the main estimate and, where possible, by how much. A quantitative bias analysis or an E-value is more informative than a general statement that confounding cannot be excluded.

5.3 Interpretation

[Give a cautious overall interpretation considering objectives, limitations, multiplicity of analyses, results from similar studies and other relevant evidence. Avoid causal language unless justified.]

5.4 Generalisability

[Discuss the external validity of the results: to which populations and settings they apply.]

Consider how the study population differs from the target population in age, ethnicity, healthcare access and baseline risk; whether the exposure definition matches how the exposure occurs in practice; and whether the calendar period of the study still reflects current care. Distinguish between transporting a relative effect, which may hold across settings, and an absolute effect, which depends on local baseline risk.

6. Other information

Transparency information for readers:

Item	Detail
Funding	[Funder, grant number]
Role of funder	[Role in design, analysis, interpretation, publication, or none]
Original study funding	[If the present article is based on an earlier study]
Ethics	[Approval reference, consent or waiver]
Data availability	[Access route, conditions, governance]
Code availability	[Repository link]
Conflicts of interest	[Statement]

7. Appendices

Supporting material for reviewers:

Appendix	Content
A	Completed STROBE checklist for the design, with page numbers
B	Code lists for exposure, outcome and covariates
C	Full model output and sensitivity analyses
D	Glossary: [CI confidence interval, DAG directed acyclic graph, HR hazard ratio, OR odds ratio, RR relative risk]

Standard and licence note

Based on the STROBE Statement: von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP, for the STROBE Initiative, The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies (PLoS Med 2007;4(10):e296 and parallel publications), https://www.strobe-statement.org. The checklists are published open access; this template uses their item topics only, paraphrased in original text, and reproduces no checklist wording. Use the official checklist for your design when submitting. This is a reporting format, not clinical, epidemiological or statistical advice. Template licence: MIT.