Application of the principal stratification approach to the Faenza randomized experiment on breast self-examination

Many scientific problems require that treatment comparisons be adjusted for posttreatment variables, such as treatment noncompliance, missing outcomes following treatment noncompliance, and "truncation by death". In the last years, there has been substantial progress in the analysis of randomized experiment suffering from noncompliance and missing outcome data, and recent work has addressed the problem of truncation by death; but, we are not aware of any previous work that addresses all these complications jointly. We present an extended framework for the analysis of data from randomized experiments which suffer from treatment noncompliance, missing outcomes following treatment noncompliance, and “truncation by death”. There are two key feature of this framework: we use the principal stratification (Frangakis and Rubin, 2002) approach for comparing treatments adjusting for posttreatment variables, and we adopt a Bayesian approach for inference and sensitivity analysis. This framework is illustrated in the context of a randomized trial of Breast Self-examination. [Working Paper 2005/09, Department of Statistics, University of Florence, Italy]