PhD defense: Zehao Su - Combined evidence of treatment effects
Date: Thursday, September 11, 2025, at 14:00
Venue: CSS 35.3.13, City Campus, University of Copenhagen Øster Farimagsgade 5, 1353 Copenhagen K
After the defense, a reception will be held in CSS 5.2.46.
Academic Advisors
Frank Eriksson, University of Copenhagen
Henrik Ravn, Novo Nordisk A/S
Helene Rytgaard, University of Copenhagen
Assessment Committee
Brice Ozenne, University of Copenhagen
Christian Dehlendorff, Danish Medicines Council
Shu Yang, North Carolina State University
Summary: Although randomized controlled trials are the gold standard for assessing treatment effects, they can lack external validity and are increasingly costly and time-consuming. Combining existing data provides an opportunity to generate evidence of treatment effects in external populations without the need to conduct new trials. Incorporating external data into the trials can also improve the precision of treatment effect estimates.
This thesis contributes three works of statistical methods for causally interpretable parameters through data combination. The first work establishes the identifiability of the target population average treatment effect in the presence of unmeasured effect modifiers using proxy variables. The second work is concerned with efficient estimation of the marginal treatment effect with data from multiple sources under transportability of the stratified treatment effects. The third work proposes estimators of cumulative incidences in trials with competing risks, incorporating external controls.
The developed methods are illustrated with data from five placebo-controlled trials investigating the effect of semaglutide and liraglutide (glucagon-like peptide-1 receptor agonists) on body weight and cardio-vascular events.