A Hierarchical Potential Outcomes Model for Estimating Both Causal and Intention-to-Treat Effects from Repeated Measures Under 'Double Sampling'

• Activity Number: 341
• Type: Invited
• Date/Time: Wednesday, August 14, 2002 : 2:00 PM to 3:50 PM
• Sponsor: ENAR
• Author(s): Joseph Hogan*+
• Affiliation(s): Brown University
• Keywords: neurocognitive functioning ; clinical trials ; Bayesian hierarchical model

Abstract

Longitudinal clinical trials with long follow-up periods almost invariably suffer from both loss to follow-up and noncompliance with assigned therapy. An example is Protocol 128 of the AIDS Clinical Trials Group (ACTG 128), a five-year equivalency trial comparing reduced dose Zidovudine (ZDV) to the standard dose for treatment of pediatric AIDS patients. We examine the effects of dose on neurocognitive functioning, measured by longitudinal IQ scores. A Bayesian hierarchical model is used to estimate both the intent-to-treat effect (ITE) and average causal effect (ACE) of reducing the prescribed dose of ZDV by 50 percent. The ITE quantifies the causal effect of assigning the lower dose, whereas the ACE represents the causal effect of actually taking the lower dose. We adopt a potential outcomes framework where, for each individual, the existence of a different potential outcomes process at each level of time spent on treatment is assumed. The joint distribution of the potential outcomes and the time spent on assigned treatment is formulated using a hierarchical model: The potential outcomes distribution is given at the first level, and dependence between the outcomes and time on treat