The department of Epidemiology and Biostatistics at Dornsife presents Nandita Mitra, PhD, professor of Biostatistics in the department of Biostatistics at University of Pennsylvania.

 

Bias due to unmeasured confounding is a common concern when researchers estimate a treatment effect using observational data. To address this concern, instrumental variable methods, such as two-stage predictor substitution (2SPS) and two-stage residual inclusion (2SRI), have been widely adopted. In many clinical studies with survival outcomes, 2SRI has been accepted as the method of choice over 2SPS, but a compelling theoretical rationale has not been postulated. We evaluate the bias and consistency in estimating the conditional treatment effect for both 2SPS and 2SRI when the outcome is binary, count, or time to event. We demonstrate analytically that the bias in 2SPS and 2SRI estimators can be reframed to mirror the problem of omitted variables in nonlinear models and that there is a direct relationship with the collapsibility of effect measures. In contrast to conclusions made by previous studies, we demonstrate that the consistency of 2SRI estimators only holds under the following conditions: (1) when the null hypothesis is true; (2) when the outcome model is collapsible; or (3) under the strong and unrealistic assumption that the effect of the unmeasured covariates on the treatment is proportional to their effect on the outcome. We propose a novel dissimilarity metric to provide an intuitive explanation of the bias of 2SRI estimators in noncollapsible models and demonstrate that with increasing dissimilarity, the bias of 2SRI increases in magnitude. Motivating examples from the clinical literature on treatments of kidney cancer will be presented.

Mitra is a professor of Biostatistics and Vice Chair of the Department of Biostatistics, Epidemiology and Informatics at the University of Pennsylvania. She is also the Chair of the Graduate Group in Epidemiology and Biostatistics and Co-Director of the Center for Causal Inference. She received her BA in Mathematics from Brown University, MA in Biostatistics from the University of California, Berkeley and her Ph.D. in Biostatistics from Columbia University. She joined the faculty at the University of Pennsylvania in 2005, after completing a postdoctoral fellowship at Harvard. Her primary research interests include propensity score and instrumental variable methods for observational data, causal inference, and health economics with applications in cancer outcomes and health policy. She is the author of over 190 peer-reviewed journal articles and serves on the editorial boards of the International Journal of Biostatistics and Plos One. She holds several offices in national and international statistical organizations including ASA, IBS and ENAR.