RESTORE trial: clarification of stopping-rule chronology and analysis population

Independent technical commentary on the RESTORE melatonin trial (BMJ Open 2026;16:e107775; DOI 10.1136/bmjopen-2025-107775). The note examines the chronology of futility-rule prespecification, the reported intention-to-treat population and unavailable outcomes, and the interpretation of statistical power. The primary two-sided Fisher exact test is independently reproduced as P = 1.000 from the published aggregate counts. The documentary discrepancy requires dated protocol/SAP amendments and monitoring records; it does not establish outcome-driven stopping or invalid treatment results. Source-access limitations are stated explicitly. No participant-level reanalysis was performed. OpenAI ChatGPT assisted source checking, numerical verification, statistical interpretation, language editing, and preparation of the commentary. Responsibility rests with the named author.