Presentation: Evidence-based Cancer Policy: Taking Statistical Aim at a Perpetually Moving Target
2017-18 Ross L. Prentice Endowed Professor: Ruth Etzioni, Ph.D., Full Member, Biostatistics Program, Public Health Sciences Division, Fred Hutchinson Cancer Research Center, and Affiliate Professor of Biostatistics, University of Washington
Abstract: Cancer policy panels are mandated to produce evidence-based decisions. But what are they to do when published studies disagree? Differences in design, implementation and study populations can all lead to apparently conflicting results. When evidence is a moving target, statistical modeling of the implications of these differences becomes a critical tool for integrating information across studies. Guided by the influential work of Ross Prentice on the discrepancies between observational and clinical trial data on postmenopausal hormone treatment, I will share three stories of our work dealing with—and reconciling—apparently conflicting cancer studies. In prostate cancer, I’ll discuss two trials of primary surgery for early-stage disease and two trials of prostate cancer screening whose primary results famously disagreed. In breast cancer, I’ll examine the difference in published estimates of the risk of overdiagnosis produced by excess-incidence and traditional interval-censored survival methods. In each case I will show how mechanistic understanding and modeling of the process producing the observed data allows us to reconcile differences in findings and move towards an evidence base for policy development.
About Dr. Etzioni: Dr. Ruth Etzioni’s work focuses on statistical and computer modeling for policy development, with a focus on prostate cancer research. Dr. Etzioni’s models of disease have been used to estimate the lifetime probabilities of prostate cancer and its outcomes, study the extent of overdiagnosis associated with prostate cancer screening, and quantify the roles of screening and changes in receipt of initial therapies in explaining population mortality declines. Ruth is a member of three national panels on prostate cancer early detection and serves on the American Cancer Society’s newly formed Guideline Development Panel. Her work on prostate cancer modeling is done as part of the National Cancer Institute’s Cancer Intervention and Surveillance Modeling Network (CISNET) consortium, for which she serves on the steering committee and is the Principal Investigator on the Prostate Cancer Coordinating Center. Dr. Etzioni is also Principal Investigator for the Biostatistics Core of the Pacific Northwest Prostate Cancer SPORE program and an ASA Fellow.