Methods of Analysis in Randomized Noninferiority Trials: Methodological Survey Review Protocol

Noninferiority (NI) trial designs that investigate whether an experimental intervention is no worse than the standard of care have been used increasingly in recent years. The robustness of the conclusions depends in part on the analysis population set used. In NI settings, the intention-to-treat (ITT) and per-protocol (PP) analysis sets are most common. The ITT analysis has been considered anticonservative compared with the PP analysis.

This study aimed to conduct a methodological review assessing the analysis population sets used in contemporary NI trials.

A comprehensive electronic search strategy will be conducted to identify studies indexed in MEDLINE, Embase, Emcare, and Cochrane CENTRAL. Studies will be included if they are NI trials published in 2024. The primary outcome is the analysis population used for the primary analysis of the trial (ITT, PP, or as-treated). Secondary outcomes include the NI margin, effect estimates, point estimates, and corresponding CIs. Analyses will be performed using descriptive statistics.

The comprehensive search initially identified 1209 studies, of which 403 trials were eligible for data extraction. Data extraction began in January 2025 and is expected to be completed in January 2026.

This methodological survey of NI trials will describe the analysis population used in the primary analysis and assess factors that may be associated with each analysis method.