Pre-Submission Review for Epidemiology Papers

Reviewers often ask:

• Is the study design appropriate for the question?
• Are exposure and outcome definitions credible?
• How were confounders selected and measured?
• Is selection bias or missing data handled openly?
• Does the analysis distinguish association from causation?
• Is the reporting guideline correct for the design?
• Are sensitivity analyses meaningful rather than decorative?
• Are conclusions proportionate to the uncertainty?

The manuscript can be well written and still fail if the inference is overconfident.

In our pre-submission review work, epidemiology manuscripts usually become vulnerable when the result is framed more strongly than the design allows.

The common failure patterns are:

• Causal drift: the paper starts with association but concludes like causation has been shown.
• Confounder opacity: authors adjust for variables without explaining why those variables belong in the model.
• Exposure fuzziness: the exposure definition is too broad, unstable, or inconsistently measured.
• Missing-data silence: missingness is described but not interpreted as a possible bias source.
• Checklist mismatch: authors use generic reporting language when STROBE, RECORD, STROBE-MR, CONSORT, or PRISMA is the real standard.

A good epidemiology review should tell authors which of those issues would make a reviewer lose trust first.

The STROBE Statement was created to strengthen reporting of observational studies in epidemiology and provides a checklist for what should appear in reports of such research. EQUATOR lists STROBE for observational studies, CONSORT for randomized trials, PRISMA for systematic reviews, STARD for diagnostic/prognostic studies, and other study-design-specific guidelines.

The International Journal of Epidemiology requires STROBE for observational studies and RECORD when routinely collected health data are used. American Journal of Epidemiology guidance emphasizes rigorous statistical methods and discourages using "effect" as a proxy for "association" in single observational studies.

Those signals point to the same rule: methods language matters because inference language matters.

Send the manuscript, target journal, study protocol if available, statistical analysis plan, STROBE or other reporting checklist, variable definitions, exposure and outcome definitions, missing-data approach, sensitivity analyses, and any data-access limitations.

For registry or routinely collected data studies, include data provenance and coding rules. For causal claims, include the authors' causal diagram or rationale for the adjustment set if one exists.

Before submission, epidemiology manuscripts often need:

• a clearer exposure and outcome definition
• a stronger explanation of confounder selection
• sensitivity analyses tied to the main bias concern
• more transparent missing-data handling
• effect estimates and uncertainty interpreted together
• causal language narrowed to match the design
• STROBE, RECORD, or another checklist completed with page references

These changes can do more for acceptance risk than another line edit.

Use epidemiology-specific review when:

• the paper makes a risk, association, policy, or causal claim
• observational design choices are likely to be debated
• the target journal has strong methods expectations
• the paper uses routinely collected health data
• missing data, selection bias, or confounding could dominate review
• the result will be read by clinicians, policymakers, or public-health decision makers

General editing will not solve those issues.

The deliverable should say whether the inference is ready for the target journal.

For epidemiology, the main commercial value is avoiding a slow review where the first methods reviewer says the analysis cannot support the conclusion.

Many epidemiology papers become vulnerable because the introduction promises causal insight, the methods deliver association, and the discussion slides back into causal language. Reviewers notice this quickly. A pre-submission review should mark every place where "effect," "impact," "leads to," or "prevents" appears and ask whether the design can support that wording.

When the design is observational, the safer language often focuses on association, risk pattern, prediction, surveillance, or hypothesis generation. If the authors are making a causal claim, the paper needs stronger design logic, transparent assumptions, and sensitivity analyses that make the claim defensible.

Before choosing the target journal, ask:

• Is the paper primarily epidemiologic methods, clinical epidemiology, public health, or policy?
• Does the target journal expect causal inference, prediction, surveillance, or descriptive evidence?
• Will readers care about the population, outcome, and exposure outside the study setting?
• Is the data source strong enough for the journal tier?
• Are the reporting and transparency requirements stricter than the current package?

Those questions keep the review focused on publication readiness, not only statistical correctness.