American Statistical Association New York City Metropolitan Area Chapter Memorial Sloan-Kettering Cancer Center Biostatistics Seminar

Randomization has been long used in therapeutic and intervention trials to form the basis of causal inference regarding treatment effects. The Methods for Improving Reproductive Health in Africa (NIRA) trial, designed to examine the effects of the diaphragm and lubricant gel in reducing sexually transmitted infections, raised issues regarding the standard intention-to-treat analysis due to the role of a secondary intervention on condom use. A further example occurs in the differential use of statins in the RECORD trial examining the safety of Avandia. A similar causal diagram arises in a very different situation involving blinded clinical trials with treatment-related side effects due to the potential for unmasking of treatment assignment. In this case, we motivate and discuss the causal issues based on a randomized, double-blind, placebo-controlled trial of pain medication for patients suffering from diabetic neuropathy.