Generalizing Evidence from Randomized Clinical Trials to Target Populations: The ACTG-320 Trial

• Activity Number: 216
• Type: Invited
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Elizabeth Stuart*+ and Stephen R. Cole
• Companies: Johns Hopkins Bloomberg School of Public Health and The University of North Carolina at Chapel Hill
• Keywords: Bias ; Inference ; Weight ; Generalizability ; External validity ; Trial

Abstract

Properly planned and conducted randomized trials are susceptible to a lack of external validity. We propose a model-based method to standardize observed trial results to a specified target population, illustrate the method in a seminal HIV treatment trial, provide Monte Carlo simulation evidence supporting the method, and provide a proof of asymptotic consistency. Results from the trial apply, albeit muted by 12 percent, to the target population, under the assumption that we have measured and correctly modeled the determinants of selection that reflect heterogeneity in the treatment effect. In simulations with a heterogeneous treatment effect, a conventional intent-to-treat estimate was biased with poor confidence limit coverage, but the proposed estimate was largely unbiased with appropriate confidence limit coverage.