CONSORT 2025 Scorer
EQUATOR · CONSORT 2025 · Randomised trials

CONSORT 2025 Scorer

Upload a randomised trial manuscript; anything else is rejected at the door. Each of the 42 checklist items (the 30 numbered CONSORT 2025 items and their parts, e.g. 1a, 1b) is broken into the bullets of the CONSORT 2025 expanded checklist, quoted word for word (212 in all), each judged with a quote and page.

Hopewell et al. BMJ 2025;388:e081123

Manuscript

No manuscript loaded

Checklist ledger

Click any Y / N / NA / ? badge to override the AI's judgement; scores recalculate and overridden badges get an outline.

Export & history

Recent reports on this browser
None yet
How the score is calculated
  • Each item is split into the bullets of the CONSORT 2025 expanded checklist, quoted word for word (212 elements). Every rule is taken from the CONSORT 2025 checklist, expanded checklist, Explanation & Elaboration (E&E) and flow diagram; nothing is added from outside them. Each element gets one of the three verdicts the CONSORT files support — the checklist's "Reported on page no." column is either filled or empty: Y reported, N not reported, NA not applicable (only for bullets the files make conditional). ? for you to decide: cases the CONSORT files do not settle (a bullet written only in part, a bullet about something that did not happen in the trial, a condition the files do not define, information only in a supplement…). The tool does not decide them; it shows the files’ own words and leaves the verdict to you. They are not counted as reported.
  • The three rings count the checklist items that apply to the trial: complete, incomplete, not reported. Items that do not apply are left out. Every item counts once.
  • An item is complete when every applicable element is reported, not reported when none is, incomplete otherwise, and N/A if every element is NA.
  • CONSORT 2025 calls its 30 items "essential items" and does not rank them, so no item is weighted and no grade is given. CONSORT measures how completely a trial is reported; it "is not meant to be used as a quality assessment instrument" (CONSORT 2025 E&E).
  • Eligibility gate: before scoring, the paper is screened. Only primary results reports of randomised trials (parallel, cluster, crossover, factorial, pilot, non-inferiority) are scored. Protocols, secondary analyses, quasi-randomised or single-arm trials, observational studies, reviews and diagnostic studies are rejected without scoring.
  • Item 1b is judged on the abstract alone. "None" statements are counted the way the expanded checklist asks for them (e.g. "If no patient or public involvement, this should be stated").
  • Quote check: every quote the AI gives is searched for in the paper's own text. Quotes that can't be found are flagged in red, so an invented quote can't pass silently.
  • Figures: pages that look like the flow diagram, baseline table or harms table are sent to Claude as images (up to 5) for the results items. Anything still unreadable is marked ? (figure not readable).
  • Supplements: add them after the main article. The expanded checklist names a "supplement to the trial report" as a location only for item 3 (protocol and analysis plan); for every other item the files do not say whether a supplement counts, so information found only in a supplement is marked ? for you to decide.