Issues in Adjusting for Covariates Arising Postrandomization in Clinical Trials

Abstract

While there is broad consensus on analytic techniques for adjusting for covariates at baseline, the situation for covariates arising postrandomization is considerably more difficult. Examples include the level of patient “compliance” measured through pill counts and other biochemical markers, the occurrence of missing data over patient followup, and early withdrawal from medication. The “intention-to-treat” (ITT) principle requires that all randomized patients be included in all analyses irrespective of their confounder experience. This approach, however, seems at odds with good scientific method and is a considerable source of friction with medical investigators. In this paper, we review the interpretation of this analysis strategy and suggest that the statistical community has been careless in its interpretation of these results. We outline a conservative strategy that is consistent with ITT principles. Nevertheless, any analysis that adjusts for these covariates must be considered speculative in nature and followed by a properly designed confirmatory study. For this reason, we argue that these analyses are of greater relevance early in a drug development program.