Discussion of: Statistical and Regulatory Issues with the Application of Propensity Score Analysis to Nonrandomized Medical Device Clinical Studies

摘要

As indicated by Dr. Yue, nonrandomized studies are more common in medical device studies than in drug and biological product studies. Informative analyses of such registries compared to prior historical controls are challenged by the rapid pace of change in medical device technologies. There is relatively strong consensus that propensity scores overall provide a useful adjustment approach across nonrandomized cohorts. There may be some controversy, however, in assessments of the relative strengths and weaknesses of various propensity score methods for any specific application. In this discussion, we would like to expand upon some of these issues and suggest some strategies for minimizing certain problems.