ESTIMATING THE AVERAGE TREATMENT EFFECT IN RANDOMIZED CLINICAL TRIALS WITH ALL-OR-NONE COMPLIANCE

• 成果类型: Article
• 署名作者: Zhang, Zhiwei; Hu, Zonghui; Follmann, Dean; Nie, Lei
• 署名单位: National Institutes of Health (NIH) - USA; NIH National Cancer Institute (NCI); NIH Division of Cancer Treatment & Diagnosis; National Institutes of Health (NIH) - USA; NIH National Institute of Allergy & Infectious Diseases (NIAID); US Food & Drug Administration (FDA)
• 刊物名称: ANNALS OF APPLIED STATISTICS
• ISSN/ISSBN: 1932-6157
• DOI: 10.1214/22-AOAS1627
• 发表日期: 2023
• 页码: 294-312
• 关键词: doubly robust estimation; Causal Inference; mortality; models

摘要

Noncompliance is a common intercurrent event in randomized clini-cal trials that raises important questions about analytical objectives and ap-proaches. Motivated by the Multiple Risk Factor Intervention Trial (MRFIT), we consider how to estimate the average treatment effect (ATE) in random-ized trials with all-or-none compliance. Confounding is a major challenge in estimating the ATE, and conventional methods for confounding adjustment typically require the assumption of no unmeasured confounders which may be difficult to justify. Using randomized treatment assignment as an instru-mental variable, the ATE can be identified in the presence of unmeasured confounders under suitable assumptions, including an assumption that limits the effect-modifying activities of unmeasured confounders. We describe and compare several estimation methods based on different modeling assump-tions. Some of these methods are able to incorporate information from auxil-iary covariates for improved efficiency without introducing bias. The different methods are compared in a simulation study and applied to the MRFIT.