Friday, November 13, 2020 - 10:00am
The Department of Biostatistics and Epidemiology welcomes Alisa J. Stephens-Shields, PhD, Assistant Professor of Biostatistics at the University of Pennsylvania Perelman School of Medicine. Stephens-Shields will deliver a talk titled "Standardized Direct Effects to Address Intercurrent Unblinding in HIV Prevention Research" on Friday, November 13, beginning at 10:00 AM via Zoom.
Although the primary target of inference in randomized trials is often some version of an average causal effect, recent guidance in the practice of clinical trials has emphasized alternative estimands to enhance the handling of intercurrent events such as treatment switches or discontinuation. We use standardized direct effects within the context of mediation analysis to explicate observed treatment differences in the Phambili study, a double-blind, randomized trial of an experimental vaccine to reduce HIV infection. Initial treatment comparisons revealed an elevated rate of infection among participants randomized to vaccine compared with placebo. Interpretation of this effect, however, was challenged by the intercurrent effect of post-randomization unblinding due to futility in a parallel trial. Considering post-randomization sexual behavior, such as total number of partners and condom use, we use additive hazard models to estimate direct effects of vaccine when sexual behavior is standardized to the blinded period. Our analyses demonstrate the complementary value of this estimand in understanding the vaccine effect amid unblinding.
Alisa J. Stephens-Shields, Ph.D., is an Assistant Professor of Biostatistics at the University of Pennsylvania Perelman School of Medicine, where she was a recipient of the inaugural Department of Biostatistics, Epidemiology, and Informatics Distinguished Faculty Award. Her research focuses on clinical trials, clustered and longitudinal data analysis, and causal inference, including methods for flexible and efficient inference in randomized trials, methods for causal inference for the effect of time-varying exposures, and the development of patient reported outcomes. Dr. Stephens-Shields collaborates in several clinical areas, including pediatrics, chronic pain, pharmacoepidemiology, and behavioral economics. She is a member of the editorial boards of Biostatistics and Pharmacoepidemiology and Drug Safety and an elected officer of the American Statistical Association Section on Statistics in Epidemiology and the ENAR Regional Committee. Dr. Stephens-Shields holds Ph.D. and A.M. degrees in biostatistics from Harvard University and a B.S. in mathematics with minor in Spanish from the University of Maryland, College Park.
Email Tonya Menard to pre-register and receive the Zoom link.