Adjusting for Patients' Unmasking in Clinical Trials: Estimation of Direct Effects in Longitudinal Studies

• Activity Number: 37
• Type: Contributed
• Date/Time: Sunday, July 31, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Farid Jamshidian*+ and Alan Hubbard and Nicholas P. Jewell
• Companies: University of California at Berkeley and University of California at Berkeley and University of California at Berkeley
• Keywords: Causal Effect ; Unmasking ; Perception Effect ; Placebo Effect ; Semi-parametric Estimation ; Gabapentin

Abstract

In blinded clinical trials, treatment assignment is purposefully concealed to prevent patients' outcome from being influenced by any biases due to knowledge of the treatment. However, during the course of a trial some patients may be potentially (if not correctly) unmasked either due to experiencing treatment related side effects, or a lack of efficacy. Previously, we have introduced the related concepts of treatment perception, unmasking, and placebo effects for point treatment studies. Here, we generalize these concepts to longitudinal studies, and use recent developments in causal inference literature to define such effects. Our discussion is motivated by a recent longitudinal study of the effect of gabapentin on pain among diabetic patients with painful neuropathy. We consider two semi-parametric estimation methods of the necessary direct effects, Maximum Likelihood Estimation and Targeted Maximum Likelihood Estimation.