Abstract / Summary
Protocol for a systematic review and meta-analysis of whether the operational definition of antibiotic de-escalation explains the inconsistency of the pooled estimate of all-cause mortality in hospitalised adults. The protocol was finalised on 24 September 2026 and specifies eligibility criteria; the coding of de-escalation definitions into four reproducible archetypes; a four-database search strategy (PubMed/MEDLINE, Embase, Cochrane CENTRAL, Web of Science Core Collection) supplemented by trial-register searching, citation tracking and author contact; duplicate-independent screening and extraction; RoB 2 and ROBINS-I assessment with an additional definition-reporting-quality item; a random-effects analysis using the Paule-Mandel estimator with Hartung-Knapp-Sidik-Jonkman confidence intervals; and a pre-specified meta-regression entering the definition archetype as a categorical moderator. This review was not registered in PROSPERO, because PROSPERO does not accept registration of reviews in which data extraction has already started. The quantitative analysis under the finalised protocol was completed on 28 September 2026, before this deposit, which is therefore a retrospective, version-controlled deposit made to keep the pre-specified plan publicly inspectable and citable. A preliminary extraction round preceding protocol finalisation is disclosed in Section 3.12 of the protocol; none of its results are reported in the review.