John Petkau, SSC Award for Impact of Applied and Collaborative Work 2014
The 2014 recipient of the Statistical Society of Canada Award for Impact of Applied and Collaborative Work is A. John Petkau, Professor in the Department of Statistics at the University of British Columbia, Vancouver. The award recognizes outstanding contributions by members of the SSC in collaborative research and applied work, the importance of which derives primarily from its relatively recent impact on a subject area outside of the statistical sciences, on an area of application, or on an organization.
John was born in 1950 in Carman, Manitoba and grew up on three different farms in the vicinity of the neighbouring town of Elm Creek, a community about 70 kilometres west of Winnipeg. As a lad, he benefited from the tutelage of his older sister, who taught him to speak English before he started school at Wingham Elementary. His later education was at Elm Creek Elementary and High Schools, where he was one of three males in his 1967 graduating class of twelve students.
John began his university education at the University of Manitoba. When he was searching for additional science courses to take in first year, a future brother-in-law suggested he enroll in a second year introductory statistics course, giving him the opportunity to become aware of a discipline he had no idea existed. During his subsequent undergraduate studies concentrating in Mathematics and Statistics and completed in 1971, he was challenged and advised by many excellent instructors, among others B. Johnston, B.K. Kale, P. McClure, N.S. Mendelsohn, K. Mount, J.N.K. Rao, and R. Wong.
John continued to a PhD program in Statistics at Stanford, graduating in 1975. Here he was inspired by the many outstanding faculty and benefitted from valuable mentoring not only by his PhD supervisor, Herman Chernoff, but also by others including Lou Gordon, Paul Switzer and Ingram Olkin. In his thesis research, he formulated the problem of designing a clinical trial as a Bayes sequential decision problem, utilized a connection to optimal stopping problems for Wiener processes and developed numerical methods for evaluating approximate solutions. These types of problems were the focus of much of his early research and his interest in sequential approaches continues to the present.
John moved to MIT to take up an instructorship in 1974, accompanying Chernoff who was to establish a statistical presence within the Department of Mathematics. Beyond the opportunity to continue to collaborate closely with Chernoff, the two years spent at MIT provided valuable experience in developing courses and initiating programs.
John’s first study leave after moving to UBC in 1976 was spent in the Harvard Department of Biostatistics in 1981-82, where he had the opportunity to advise and collaborate with clinical researchers at the Massachusetts General Hospital. The stimulation provided by this experience prompted efforts to become more involved with collaborative research at UBC. This led to contact with Professor Don Paty who had recently moved to Vancouver to head the Multiple Sclerosis (MS) clinic at the UBC Hospital. This somewhat random initial event has led to many collaborations over subsequent years, both at UBC and elsewhere, and has led to the longest thread in John’s research path and the one with the largest impact.
John’s MS collaborations have addressed issues of specific concern to the discipline, particularly issues related to improving how MS clinical trials are designed, executed and analyzed. These subject-area issues have provided many opportunities for methodological research of immediate applicability. A theme running through this work is the development of approaches that make better use of the rich longitudinal data collected on MS patients. John has made contributions via modeling longitudinal data with mixed and hidden Markov models, via improving experimental designs, and in general bringing best statistical practices to MS research. His work has helped researchers to identify factors that counteract the effects of very expensive medication, providing guidance to allow people with MS to switch from futile medications to something more effective. In recent work with UBC MS researcher Dr Helen Tremlett, John contributed to the surprising observational study that showed there was little evidence of the long-term effectiveness of the commonly used beta-interferon drugs. His work on the use of magnetic resonance imaging (MRI) outcomes as potential surrogates has been influential in phase II clinical trials. Currently he is developing a new safety monitoring approach for detecting unusual increases in the MRI lesion counts of individual patients that might indicate an increased risk of imminent clinical worsening. The respect John has in the MS community is evidenced by his many collaborations, his invitations to subject-area workshops and conferences, and his extensive service on advisory boards and monitoring committees.
Another subject-area in which John has had major collaborations is environmental epidemiology. John was among the earliest statisticians to study connections between air pollution and human health, in particular the health effects of inhalable particulate matter. The analysis of the longitudinal data was complicated by the presence of missing data and the computational challenges of many long sequences. To handle these challenges, John enlisted the department’s Statistical and Consulting Research Laboratory (SCARL) to develop suitable software to implement the then new methodology of generalized estimating equations. John’s involvement in environmental epidemiology diminished as his involvement in MS continued to expand.
John has contributed in many ways to the UBC Statistics Department, including as Head (1989-1994) and as frequent Faculty Advisor to SCARL. Perhaps most importantly, his vision of the statistical sciences is felt throughout the Department – in its faculty, in its curriculum and in the importance of SCARL.
John has also contributed to the discipline of statistics at the national level. He has served the SSC as a Regional Representative on the Board of Directors (1982-84, 2003-05), as President of the Biostatistics Section (1996-97), and as Chair of the Pierre Robillard Committee (1983-84), the Program Committee for the 1988 Annual Meeting, the Awards Committee (1990-91) and the Research Committee (1999-2002). He also served as Chair of the NSERC Statistical Sciences 1989-90 GSC and 2000-02 Reallocations Exercise Steering Committee. Currently, he is the SSC President-Elect.
No assessment of John’s impact is complete without considering his influence on young statisticians – his mentoring and the example he sets with his high standards. His impact on applied statistical methodology is far-reaching through his instilling in students the importance of analyzing data in a way that honours the data and addresses the scientific question of interest.
John is greatly indebted to his UBC colleagues for the stimulating and supportive work environment in the Department of Statistics, to his many wonderful collaborators at UBC and elsewhere, and to the many outstanding students who have been part of his academic life. He credits much of his success to his wife Barbara, a native Vancouverite, whom he met on a volleyball court in 1977. Barbara has been a constant source of exceptional support and has also succeeded in keeping John involved in a variety of non-academic activities. John enjoys travelling, attending the theatre, hiking, snowshoeing, reading and escaping to their cabin on Lummi Island, the “tranquil and forgotten” San Juan Island.
The citation for the award reads:
“To John Petkau, for contributions to the development, implementation and dissemination of statistical methodology related to the health sciences; for helping to bring a deeper understanding of the health effects of pollution and, through his work on design and analysis, an insight into the disease course of Multiple Sclerosis and the effectiveness of various treatments for this debilitating disease; for inspiring decades of statistics students to collaborate effectively with researchers to answer important subject-area questions.”