STROBE · Observational Studies

The STROBE checklist, section by section.

STROBE is the reporting standard for cohort, case-control, and cross-sectional studies — every observational design where investigators do not assign exposures. Twenty-two items cover what reviewers want to see, with a small number of design-specific entries layered on top. This is a complete walkthrough.

Check Your ManuscriptLearn About AI Peer Review
The guideline

What STROBE is.

STROBE — Strengthening the Reporting of Observational Studies in Epidemiology — is the international reporting guideline for observational research. It exists because observational study reports historically omitted the items most needed to assess internal validity: how participants were selected, how exposures and outcomes were measured, what confounders were considered, and how missing data and bias sources were handled. Twenty-two items address these, organized into the same six manuscript sections every reader scans in order.

STROBE is maintained by the STROBE Initiative, listed under the EQUATOR Network. The current widely adopted version is the 22-item core checklist with the original Explanation and Elaboration document. Most major biomedical journals require STROBE compliance for observational studies, and many ask for a completed checklist at submission.

STROBE applies to any study where investigators observe — rather than assign — exposures. If you randomized participants to interventions, you need CONSORT instead. If you synthesized evidence from prior studies, you need PRISMA.

The variants

Three design-specific checklists.

The 22 STROBE items are largely shared across designs, but cohort, case-control, and cross-sectional studies each have a dedicated checklist that emphasizes the items that matter most for that design. PeerReviewAI selects the correct variant from the manuscript automatically.
STROBE Cohort
For cohort studies — prospective or retrospective. Defines exposed and unexposed groups at baseline and follows them forward to outcome. STROBE places extra weight on follow-up time, loss to follow-up, and how exposure was operationalized at baseline.
STROBE Case-Control
For case-control studies. Cases must be defined precisely; controls must represent the source population that generated the cases. STROBE places extra weight on case definition, control selection, matching criteria, and ensuring exposure ascertainment is identical for cases and controls.
STROBE Cross-Sectional
For cross-sectional studies — exposure and outcome assessed at one time point. STROBE places extra weight on the inability to establish temporality, on selection of the source population, and on how exposure and outcome were measured at the same visit.
The checklist

STROBE, section by section.

The 22 items are organized into title and abstract, introduction, methods, results, discussion, and funding. These are the topics each item asks you to cover — what is required, and why reviewers care.
1–2
Title, abstract, and introduction
The title or abstract must identify the study's design (cohort, case-control, cross-sectional) using a term standard readers will recognize. The abstract must give a balanced summary of what was done and what was found. The introduction must establish scientific background, rationale, and pre-specified objectives, including any pre-specified hypotheses. Burying the design in the methods, or stating the rationale only after results are summarized, both fail this item.
3
Methods — study design and setting
Present key elements of the study design early in the methods. Describe the setting, locations, and relevant dates — periods of recruitment, exposure, follow-up, and data collection. Studies that fail to specify when participants were enrolled, when exposure was measured, or how long follow-up lasted cannot be evaluated for temporality or generalizability.
4
Methods — participants
Eligibility criteria and the sources and methods of selection. For cohort studies: how exposed and unexposed groups were defined, and methods of follow-up. For case-control: how cases were identified, how controls were selected, and the rationale for the choice of controls. For cross-sectional: source population, sampling frame, and selection method. Vague selection ("patients seen in our clinic") is a flagged item.
5
Methods — variables
Clearly define all outcomes, exposures, predictors, potential confounders, and effect modifiers. Give diagnostic criteria where applicable. A study that examines "smoking" without defining what counts as smoking — pack-years, current vs. former, intensity, recency — cannot be reproduced and cannot be compared across the literature.
6
Methods — data sources and measurement
For each variable of interest, give sources of data and details of methods of assessment (measurement). Describe comparability of assessment methods if there is more than one group. Self-reported, registry-linked, lab-measured, and clinically-adjudicated data have very different validity profiles — STROBE asks reviewers to know which one is being used.
7
Methods — bias
Describe any efforts to address potential sources of bias. STROBE expects authors to name the bias sources their design is exposed to — selection, information, recall, ascertainment, immortal-time, healthy-worker — and describe what was done about each one. Generic statements ("limitations are discussed in the discussion") do not satisfy this item.
8
Methods — study size
Explain how the study size was arrived at. This is the STROBE equivalent of CONSORT's sample size justification. Reviewers expect a power calculation for a primary analysis where the study size was chosen prospectively, or a precision-based rationale for studies built on secondary data.
9
Methods — quantitative variables
Explain how quantitative variables were handled in the analyses. If applicable, describe which groupings were chosen and why. Categorizing continuous exposures into deciles without justification, or choosing cut-points after seeing the data, both raise the spectre of multiplicity and post-hoc analysis.
10
Methods — statistical methods
All statistical methods, including those used to control for confounding. Methods used to examine subgroups and interactions. How missing data were addressed. (For cohort) any analyses accounting for loss to follow-up. (For case-control) how matching of cases and controls was addressed. (For cross-sectional) analyses accounting for sampling strategy if relevant. Sensitivity analyses. This is the most commonly under-reported section in observational manuscripts.
11
Results — participants
Report numbers of individuals at each stage — for example, numbers potentially eligible, examined for eligibility, confirmed eligible, included in the study, completing follow-up, and analyzed. Give reasons for non-participation at each stage. Consider use of a flow diagram. The STROBE flow diagram is not formally required, but reviewers strongly prefer it for cohort and case-control studies.
12
Results — descriptive data
Give characteristics of study participants (demographic, clinical, social) and information on exposures and potential confounders. Indicate the number of participants with missing data for each variable of interest. For cohort: summarize follow-up time. Missing-data counts per variable are easy to omit and frequently flagged.
13
Results — outcome and main results
For cohort: report numbers of outcome events or summary measures over time. For case-control: report numbers in each exposure category, or summary measures of exposure. For cross-sectional: report numbers of outcome events or summary measures. Give unadjusted estimates and, if applicable, confounder-adjusted estimates and their precision (such as 95% confidence interval). Make clear which confounders were adjusted for and why they were included.
14
Results — other analyses
Report other analyses done — analyses of subgroups and interactions, and sensitivity analyses. Pre-specified vs. exploratory analyses must be distinguished. Subgroup findings without interaction tests, or sensitivity analyses chosen to support the main result, are common overreaches.
15–17
Discussion — key results, limitations, interpretation
Summarize key results with reference to study objectives. Discuss limitations of the study, taking into account sources of potential bias or imprecision, and the direction and magnitude of any potential bias. Give a cautious overall interpretation considering objectives, limitations, multiplicity of analyses, results from similar studies, and other relevant evidence. STROBE asks authors to name the bias direction — not merely acknowledge bias exists.
18
Discussion — generalizability
Discuss the generalizability (external validity) of the study results. A finding in a single-center clinic population does not transfer to a national registry sample without justification, and STROBE asks authors to make the transferability case explicitly.
19
Other information — funding
Give the source of funding and the role of the funders for the present study and, if applicable, for the original study on which the present article is based. Underspecified funder roles, especially industry-sponsored secondary analyses, are flagged.
The contrast

Adequate vs. inadequate reporting.

Three of the STROBE items most consistently missed in submitted observational manuscripts. Each shows the version that fails review next to the version that passes — with the reasoning a reviewer would use.
Exposure definition and ascertainment
Inadequate
"We compared smokers and non-smokers for risk of lung cancer."
Adequate
"Smoking status was defined at baseline as: current smoker (≥1 cigarette/day for ≥6 months and smoking at the time of enrollment), former smoker (any history of regular smoking but no smoking in the 6 months before enrollment), or never smoker (lifetime <100 cigarettes). Status was self-reported via a structured questionnaire and validated by serum cotinine in a 10% random subsample (95% agreement). Cumulative exposure was calculated in pack-years."
Why it mattersSTROBE Item 5 (variables) and Item 6 (data sources / measurement) together require this level of detail. The inadequate version makes misclassification a near-certainty — "smokers" could mean anything from one cigarette ever to two packs a day. The adequate version operationalizes the exposure, names the measurement instrument, and reports a validation check.
Confounding control
Inadequate
"We adjusted for age, sex, and other relevant variables in a multivariable model."
Adequate
"The primary analysis adjusted for pre-specified confounders identified a priori from a directed acyclic graph: age (continuous), sex, baseline BMI, smoking status, alcohol intake, physical activity, baseline blood pressure, baseline LDL cholesterol, and socioeconomic status (deprivation index). Variables were chosen as confounders if they were associated with both exposure and outcome based on prior literature and the DAG; mediators on the causal pathway were excluded. A sensitivity analysis additionally adjusted for [variable], which approached significance in univariate analysis but was not pre-specified."
Why it mattersSTROBE Item 10 (statistical methods) requires confounding control to be specified, not gestured at. "Other relevant variables" is uninterpretable and signals data-driven model selection. The adequate version pre-specifies confounders, names the framework (DAG), and separates pre-specified from post-hoc adjustment — which is exactly what a careful observational analysis looks like.
Participant flow and missing data
Inadequate
"We included 2,400 patients in the analysis."
Adequate
"Of 3,150 individuals potentially eligible from the source registry, 2,610 met all eligibility criteria (excluded: 312 with prior outcome, 158 with missing baseline exposure data, 70 lost to baseline visit). Of these, 2,400 were included in the primary analysis (loss to follow-up: 210, 8.0%; reasons reported in Supplementary Figure S1). Missing data on individual covariates ranged from 0% to 6.8%; multiple imputation under a missing-at-random assumption was used for the primary analysis (20 imputations, MICE), with complete-case analysis reported as a sensitivity check."
Why it mattersSTROBE Item 11 (participants — numbers at each stage) and Item 12 (descriptive data — missing values per variable) together require this. The inadequate version skips from "source" to "analyzed" with no accountability for the 750 who dropped out at each stage. The adequate version makes participation transparent and pre-specifies missing-data handling — which is the difference between a study reviewers trust and one they do not.
The check

How PeerReviewAI evaluates STROBE compliance.

STROBE compliance is checked on every observational manuscript submitted to AI Peer Review — including the lowest tier. No add-on required.
01
Auto-detected observational design
Upload your manuscript and cohort, case-control, or cross-sectional design is identified automatically. The matching STROBE variant is used as the reporting checklist — you do not have to choose it.
02
Item-by-item qualitative assessment
Each STROBE item is evaluated against your manuscript text. The output is a qualitative judgment per item — adequate, incomplete, or missing. Items most often under-reported in observational research (variable definitions, confounder handling, bias sources, missing data) get explicit attention.
03
Specific, locatable feedback
For every item flagged as incomplete or missing, the feedback says what STROBE requires, what your manuscript currently states, and what is missing. You can act on it directly without cross-referencing the official Explanation and Elaboration document.
04
Included in every tier
STROBE compliance checking is part of Essentials ($2.99), Peer Review ($29), and Author Review ($79). The deeper tiers add Major/Minor Issue analysis and (for Author Review) a compliance audit against your target journal's author instructions.
Related guidelines

If your study is not observational.

CONSORT 2025
Randomized controlled trials
If your study assigned participants to interventions rather than observing them, CONSORT 2025 is the standard. Read the deep-dive on CONSORT.
Read more
PRISMA 2020
Systematic reviews & meta-analyses
If you synthesized evidence from prior studies rather than collecting primary data, PRISMA 2020 is the standard. Read the deep-dive on PRISMA.
Read more
Hub
All reporting guidelines
Not sure which guideline applies? The hub maps every study type to its checklist — CONSORT, STROBE, PRISMA, STARD, ARRIVE, and 15+ more.
Read more
STROBE · checked on every observational study

Check your study against STROBE.

FAQ

Questions, answered.

Don't see yours? Email us — we read every one.