Epidemiology and Biostatistics Seminar Series Presents: Elizabeth Handorf, PhD Associate Professor Biostatistics Fox Chase Cancer Center

 

Cox proportional hazards models are often used to compare survival outcomes in observational studies, but when the hazards are non-proportional, using such a model is problematic. Here, we discuss several methods which do not rely on the assumption of proportionality, and use propensity score weighting to adjust for confounding. Using simulations, we compare the performance of each method under a range of scenarios for various survival outcomes of interest, and demonstrate how these results can guide choice of model in relevant clinical analyses of survival in kidney cancer and sarcoma. Dr. Handorf is an Assistant Research Professor of biostatistics at Fox Chase Cancer Center. Dr. Handorf joined Fox Chase after she received her PhD in biostatistics from the University of Pennsylvania in 2012. Her research interests include analysis of large observational datasets, cost-effectiveness analysis, and the application of statistical methods to cancer research.