Comparison of Methods for Estimating the Causal Effect of a Treatment in Randomized Clinical Trials Subject to Noncompliance

• Published in: Biometrics Vol. 65; no. 2; pp. 640 - 649
• Main Authors: Little, Roderick J., Long, Qi, Lin, Xihong
• Format: Journal Article
• Language: English
• Published: Malden, USA Blackwell Publishing Inc 01.06.2009; Wiley-Blackwell Publishing; Blackwell Publishing Ltd
• Subjects: Algorithms; Analytical estimating; As-treated analysis; Biometric Practice; Biometry - methods; Biostatistics; Causal inference; Causality; Clinical trials; Comparative studies; Computer Simulation; Data Interpretation, Statistical; Efficacy; Epidemiologic Research Design; Estimating techniques; Estimation bias; Estimation methods; Estimators; Instrumental variables; Instrumental variables estimation; Models, Statistical; Outcome Assessment, Health Care - methods; Parametric models; Patient Compliance - statistics & numerical data; Per-protocol analysis; Principal stratification; Propensity scores; Proportional Hazards Models; Random allocation; Randomized Controlled Trials as Topic - methods; Statistical analysis; Statistical methods; Treatment compliance
• ISSN: 0006-341X; 1541-0420; 1541-0420
• DOI: 10.1111/j.1541-0420.2008.01066.x

We consider the analysis of clinical trials that involve randomization to an active treatment (T = 1) or a control treatment (T = 0), when the active treatment is subject to all-or-nothing compliance. We compare three approaches to estimating treatment efficacy in this situation: as-treated analysis, per-protocol analysis, and instrumental variable (IV) estimation, where the treatment effect is estimated using the randomization indicator as an IV. Both model- and method-of-moment based IV estimators are considered. The assumptions underlying these estimators are assessed, standard errors and mean squared errors of the estimates are compared, and design implications of the three methods are examined. Extensions of the methods to include observed covariates are then discussed, emphasizing the role of compliance propensity methods and the contrasting role of covariates in these extensions. Methods are illustrated on data from the Women Take Pride study, an assessment of behavioral treatments for women with heart disease.