A claim-to-source review checklist: population, endpoint, timepoint, and precision

1. Capture the exact claim, not a remembered summary

Copy the current sentence into your review log and record its slide, page, or section. Keep the draft version alongside it. If a sentence makes several assertions, examine them separately: a result, a comparison, and a statement about tolerability may require different passages. Do not let support for one clause stand in for the others.

Record the candidate source and a precise location before judging the wording. Include the table or figure number and any footnotes that change interpretation. A reference-list entry identifies a document; it does not show where an assertion is supported. Retain enough surrounding text for another reviewer to understand the evidence without guessing.

2. Population: who is actually represented?

Compare the people named or implied in the claim with the source population. Look for age range, condition, prior treatment, eligibility restrictions, and the analysis set used for the result. A statement about all enrolled participants is not interchangeable with a result reported only for respondents or a subgroup.

Write the mismatch as an editorial question, not an improvised scientific conclusion. For example: “The draft says participants; the table names questionnaire completers. Should the sentence specify that group?” If the source does not make the population clear at the selected location, inspect the methods or ask the responsible reviewer rather than filling the gap from memory.

3. Endpoint: what was measured and how?

Check the endpoint’s full name, measurement scale, unit, and definition. Distinguish a change in a score from the proportion reaching a defined threshold. Check whether the result concerns an individual component or a composite measure. Plain-language wording can be helpful, but it should not silently replace a measured outcome with a broader promise.

Watch the verbs and adjectives as closely as the nouns. “Improved,” “maintained,” “prevented,” and “was associated with” make different assertions. If the source reports an observation, do not introduce a causal claim through a stronger verb. Escalate questions about study design and interpretation to an appropriate reviewer; a wording checklist cannot resolve them by itself.

4. Timepoint and comparator: when, and against what?

Find the exact assessment timepoint or follow-up interval. A result at a single visit does not automatically describe the full study period. If the draft says “sustained,” locate evidence covering the period that word implies. Separate the timing of treatment, measurement, and follow-up when the source distinguishes them.

Identify the comparison behind the statement: baseline, another study group, a predefined threshold, or something else. A within-group change is not automatically a between-group difference. Check that the comparator named in the sentence matches the relevant table column and analysis, rather than a nearby result that happens to contain a similar number.

5. Precision: preserve the meaning of the numbers

Compare each number with its unit, denominator, direction, and statistical label. Keep percentages distinct from percentage-point differences, and relative changes distinct from absolute changes. Check whether the value is a mean, median, count, estimate, or range. A matching numeral alone is not enough to establish that the draft represents the source correctly.

Follow the agreed editorial policy for rounding, and retain the original source value in your review record. Inspect confidence intervals, uncertainty statements, and analysis qualifications when they bear on the wording. Do not calculate a new comparison from displayed values and present it as a reported finding without making the derivation and its review requirements explicit.

6. Try the checklist on an intentionally imperfect sentence

Fictitious example for editorial practice only; this is not a research finding: the draft reads, “Treatment A improved daily functioning in all participants throughout follow-up.” The invented source instead describes a change from baseline in a named questionnaire score at week twelve among questionnaire completers. It supplies no between-group comparison at that location.

The review log would flag four questions: Does “all participants” match the analysis set? Does “daily functioning” accurately describe the named endpoint? Is “throughout follow-up” justified by a single timepoint? Does “Treatment A improved” imply an effect the cited analysis establishes? The next step is source review and wording revision, not automatic acceptance of a shorter sentence.

7. Close with a traceable next action

Record what you found, what remains uncertain, and who needs to respond under your team’s process. Useful editorial labels might be “wording to revise,” “additional source needed,” and “ready for expert review.” These are working-log labels, not claims about product approval features. When the sentence changes, check the revised wording against the source again.

• Exact claim and draft version captured.
• Population and analysis set aligned.
• Endpoint, unit, timepoint, and comparator checked.
• Numbers and qualifying context inspected together.
• Source locator and unresolved questions recorded.