TITLE: Win statistics (win ratio, win odds and net benefit): introduction, properties, implementations, and applications
SPEAKERS: Gaohong Dong, BeiGene and Margaret (Meg) Gamalo, Pfizer
MODERATOR: Jingjing Ye
Over the past decade, the win ratio (ratio of win proportions, Pocock et al. 2012), the net benefit (difference in win proportions, Buyse 2010) and the win odds (odds of win proportions by dividing a tie into two half wins and assigning a half win to each treatment group, Dong et al. 2020) have been developed and comprehensively studied. The concept of “win” from Pocock et al. (2012) is very intuitive and attractive. The pioneering work of Finkelstein and Schoenfeld (1999) is equivalent to the test of the difference in the number of wins between the two treatment groups. The win statistics (win ratio, win odds and net benefit) are based on the generalized pairwise comparisons (GPC) (Buyse, 2010), for the analyses of multiple outcomes using their hierarchical importance order. Under the GPC, each patient in the Treatment group is compared with every patient in the Control group; and each pairwise comparison starts with the most important outcome (e.g., death), then less important endpoints (e.g., a non-fatal outcome such as disease progression in oncology studies) are considered only if the higher priority outcomes do not result in a win. Therefore, the win statistics have advantages compared to the conventional time-to-first-event analyses (e.g., Kaplan-Meier estimates, log-rank test, and Cox models).
Moreover, as nonparametric methods, the win statistics can avoid multiplicity issue of multiple outcomes, they can handle semi-competing risk situations (i.e., fatal outcomes plus non-fatal outcomes) and non-proportional hazards situations (e.g., delayed treatment effect typically seen in Immuno-Oncology). Their flexibility allows a composite of multiple endpoints in any data type (e.g., time-to-event, continuous, ordinal). The win statistics have been applied in practice (e.g., design and analysis of Phase III trials) and in support of regulatory approvals (e.g. tafamidis for treatment of cardiomyopathy per the ATTR-ACT trial).
In this short course, we will focus on introduction, properties, implementations, and applications of win statistics.
Gaohong Dong, PhD has 20 years of experience in the pharmaceutical industry. He is a Director of Biostatistics at BeiGene. Prior to joining BeiGene, he worked at Novartis, then worked as a consultant under his own entity of iStats Inc. Gaohong has been supporting drug development in multiple therapeutic areas including solid organ transplant, stem-cell transplant, infection disease, and oncology. He is a co-author of many highly cited medical papers. Gaohong has a great passion on statistical research. He published peer-reviewed statistical journal papers and book chapters on Bayesian-Frequentist design, adaptive design, missing data imputation, meta-analysis, and composite of prioritized multiple outcomes. During the recent years, he has been focusing on the win statistics (win ratio, win odds, and net benefit). His research of the stratified win ratio and the win odds have been applied to designs and analyses of clinical trials including phase III studies. Gaohong has been serving as an Associate Editor of the Journal of Biopharmaceutical Statistics since 2017.
Margaret (Meg) Gamalo, PhD is Statistics Head for Inflammation and Immunology in Pfizer Global Product Development. She combines expertise in biostatistics, regulatory science and adult and pediatric clinical development. Prior to joining Pfizer, she was Research Advisor, Global Statistical Sciences at Eli Lilly and Company and as Mathematical Statistician at the Food and Drug Administration. Meg leads the Complex Innovative Design Task Force at the Biotechnology Innovation Organization. She also actively contributes to research topics within the European Forum for Good Clinical Practice – Children’s Medicine Working Party. Meg is currently Editor-in-Chief of the Journal of Biopharmaceutical Statistics and is actively involved in many statistical activities in the American Statistical Association. Recently, she was elected Fellow of the American Statistical Association.