Norman Edward Breslow–Key figure in the foundation of modern biostatistics

M. Schumacher

DOI: https://doi.org/10.1002/bimj.201600196

2017-05-01

Abstract:The title of this contribution was adopted from one of the many obituaries that were published after the death of Norman Breslow on December 9, 2015, aged 74 years. It is remarkable that not only his home institution, the University of Washington, the International Biometric Society and other statistical societies, but also The Lancet as one of the most prominent, general medical journals published a one-page obituary especially highlighting his contributions to medical research (Watts, 2016). Most biostatisticians will know at least some of his groundbreaking contributions to survival analysis, the most popular of which is his investigation on the statistical properties of the Kaplan–Meier estimator for censored survival data that he published jointly with John Crowley (Breslow and Crowley, 1974). This publication is most likely the major reason, that the Kaplan–Meier estimator was rediscovered for medical statistics and became the accepted standard method of survival analysis since then. There were, however, further contributions of similar importance to this field and closely connected with his name: Sequential analysis of survival data (Breslow, 1969), a K-sample test (“the Breslow test”) for censored survival data (Breslow, 1970), an estimator for the baseline cumulative hazard (“the Breslow estimator”) in a Cox proportional hazard model (Breslow, 1972) and the first modern treatment of survival data in the presence of competing risks (Prentice et al., 1978). Colleagues working in the field of epidemiology will probably know two other publications even better: the IARC monographs on “Statistical Methods in Cancer Research” that he worked out together with Nick Day during his many visits at the International Agency for Research on Cancer (IARC) in Lyon, France. Volume I on “The Analysis of Case-Control Studies” (Breslow and Day, 1980) and Volume II on “The Design and Analysis of Cohort Studies” (Breslow and Day, 1987) served for decades and still serve as the standard textbooks for researchers in the field of cancer epidemiology. I met Norman Breslow for the first time in the early 1980s as I worked at the Institute of Medical Statistics of the Heidelberg University. My project was focused on statistical modeling for clinical trials in breast cancer and was part of the Collaborative Research Center “Stochastic Mathematical Models.” Norman spent a year in Heidelberg at the German Cancer Research Center as a fellow of the Alexandervon-Humboldt Foundation. During this time, I remember one of his seminars on “Estimation of the odds ratio” promoting the use of the Mantel–Haenszel estimator in sparse data situations (Breslow, 1981), a topic that he further developed jointly with Sander Greenland and Jamie Robins (Robins et al., 1986). I was, however, even more impressed by another lecture of Norman on goals and achievements of the National Wilms Tumor Study where he was the responsible statistician. As a member and an integral part of the national study group for more than 30 years, he published about 94 papers (37 of which as first or last author) devoted to the etiology, diagnosis, treatment, and long-term prognosis of a rare pediatric kidney tumor showing his in-depth subject-matter knowledge and dedication to a longlasting cooperation with his clinical colleagues. During the time he spent in Heidelberg, I had regular