RigorMed
← Research resources

Research methods guide

How to check a biomedical paper translation

Review the translation against the source paragraph by paragraph. Check terminology, units, numbers, negation, and qualifiers before polishing the prose.

Reading and reference verification

Suppose you receive an English results paragraph and a draft Chinese translation. Use this synthetic example to check the numbers, comparison direction, and wording separately, then reuse the table for your own review. The study and values are illustrative.

English source, common mistranslation, and a usable revision

Synthetic source: “After adjustment for age and sex, the adjusted OR for 30-day readmission was 1.32 (95% CI 1.08–1.61) for exposed versus unexposed participants. Among participants aged ≥65 years, there was no clear evidence of an association (adjusted OR 1.11, 95% CI 0.89–1.39).”

English source, common mistranslation, and a usable revision
ItemWorked exampleVerification
Effect measureA draft may say “32% higher risk.” A faithful Chinese rendering is “暴露组相对未暴露组的 30 天再入院调整后比值比为 1.32(95% CI 1.08–1.61)”.Match the original table’s OR, RR, or HR label. Preserve the interval and the unexposed reference group.
Adjustment and time windowRestore “调整年龄和性别后” and “30 天再入院” when a draft has only “调整其他因素后” and “再入院”.Check the model footnote and verify the index date for the 30-day window.
Subgroup and strength of evidenceUse “在年龄 ≥65 岁者中,未见明确关联证据(调整后 OR 1.11,95% CI 0.89–1.39)”.Keep ≥ and the interval that crosses 1. Check an interaction analysis before comparing subgroup effects.
Denominator and roundingIn a separate synthetic example, retain “36/428 (8.4%)” and identify 428 as the size of the analysis set.Recalculate 36 ÷ 428 × 100 = 8.41%. For weighted percentages, check the weighting and footnote before attempting this calculation.

Apply the method to your materials

  1. Fix the table and analysis set first

    Use one row for the source page, table, and model, outcome and time point, reference group, estimate and interval, denominator, and adjustment. Enter the source first and record discrepancies between the abstract and table separately.

  2. Check direction and qualifiers sentence by sentence

    Mark versus, after adjustment, no clear evidence, may, and ≥ in the source and locate each in the translation. Preserve the associational wording of an observational finding and check for added causality, certainty, or population changes.

  3. Deliver an auditable edit log

    Reuse the headers “Location, English source, translation, high-risk item, suggested revision, review status.” Mark unresolved model, unit, or terminology questions with the exact location and what needs confirmation.

Questions that arise in practice

Can OR 1.32 be translated as a 32% increase in risk?

State the odds ratio as 1.32. A risk difference or risk ratio requires the appropriate baseline risk and analytical method. An adjusted OR also depends on model covariates and the target population.

What if two Chinese translations are in common use?

Check the journal or institution’s glossary, retain the original term and abbreviation at first mention, and use a consistent translation. Flag ambiguity that changes interpretation.

Verified sources

The official sources below were checked on 2026-11-12. Access conditions and update dates are rechecked before publication.

  1. ICMJE Recommendations for Manuscript Preparation
  2. Bland and Altman · The odds ratio (BMJ)

RigorMed workspace

Start with your research materials

Bring your research question, available materials, and the steps you want to check.