YSPH Biostatistics Seminar: “Statistical Methods for Design and Analysis of Pragmatic Trials with Treatment Effect Heterogeneity"

NOTE: BIS 525 students are required to attend in person. Others are invited to attend in person, but may also attend via Zoom.

SPEAKER: Guangyu Tong, PhD, Assistant Professor of Cardiovascular Medicine and Biostatistics, Yale University

TITLE: “Statistical Methods for Design and Analysis of Pragmatic Trials with Treatment Effect Heterogeneity"

ABSTRACT: Pragmatic clinical trials are designed to inform practice in diverse, real-world settings. A central challenge in these studies is that the effects of interventions rarely operate uniformly across populations—understanding treatment effect heterogeneity (HTE) is therefore essential for tailoring care and guiding policy. In this seminar, I will present a body of work that advances both the design and analysis of trials to better capture and interpret heterogeneity. On the design side, I have developed new methods for powering HTE analyses in cluster randomized and group treatment trials, accounting for unequal cluster sizes, attrition, and complex correlation structures. These methods offer practical tools for planning adequately powered studies while reflecting the realities of implementation research. On the analytic side, I will highlight recent innovations using Bayesian models and machine learning to estimate heterogeneous causal effects in settings complicated by clustering, non-ignorable missingness, and outcomes truncated by death. Applications from large-scale pragmatic trials, including studies in heart failure and acute lung injury, illustrate how these methods reveal important subgroup differences that remain hidden under average effect estimates. Together, this work demonstrates how advancing methodology for HTE can strengthen the scientific and clinical value of pragmatic trials.