This talk will describe two statistical analyses intended to estimate the effects of prostate screening. First, latent class models are used to evaluate the accuracy of adjudicating possible prostate cancer deaths. This is motivated by well-known differences in the apparent mortality benefits between large randomised trials of prostate screening in Europe and in the United States, given that subjective variability in the adjudication process might have affected the results. Second, the so-called “catch-up” method is applied to follow-up data, to estimate the extent of over-diagnosis associated with prostate screening.
Stephen Walter received his Ph.D. from the University of Edinburgh. After faculty appointments at the University of Ottawa and Yale University, he joined the faculty of Health Sciences at McMaster University, where he is now a Professor Emeritus. Dr Walter has collaborated in research on internal medicine, evidence-based medicine, developmental pediatrics, environmental health, cancer etiology and medical screening. He is interested in several associated areas of biostatistical methodology, including: the design and analysis of medical research studies; risk assessment; evaluation of diagnostic and screening data; and regional and temporal variation in health. He has published widely on these topics (with almost 500 refereed papers) in the biomedical and statistical literature.
Dr Walter is a past Editor of the American Journal of Epidemiology, and Section Editor for the Wiley Encyclopedia of Biostatistics. He served as Chair of Biostatistics in the International Clinical Epidemiology Network (INCLEN), and has been involved with research capacity development in Asia, Latin America and Africa. Dr Walter is a past coordinator of the Health Research Methods program at McMaster, and has worked with about 100 Masters and Ph.D. students.