Reducing Bias in Treatment Effect Estimation in Clinical Trials Using Propensity Scores

Propensity score methods are an increasingly popular technique for causal inference. To estimate propensity scores, one must model the distribution of the treatment indicator given a vector of covariates. Much of work has been done in the case of covariates that are fully observed. Many studies, such as longitudinal surveys, suffer from missing covariate. In this paper, different approaches namely randomized comparison, and Propensity Score matching are compared. These methods are applied to assess the relative clinical utility of buprenorphine in comparison to clonidine for short term opiate detoxification.