Estimators of Effect Modification of Cumulative Incidence: Aalen-Johansen and Targeted Minimum Loss-based Estimator

Abstract

In time to event settings there are many occasions in which multiple events can occur. This competing risks phenomenon is often encountered in vaccine trials when interest is given to efficacy against a specific subtype of the disease, particularly those included (or not) in the formulation of the vaccine. In these settings, the parameter of interest is cumulative incidence - the probability of being infected by a particular subtype by a certain time. In this thesis we present and compare two estimators of cumulative incidence in various simulated scenarios and compare additive and multiplicative treatment effects as well as effect modification across groups. We apply these estimators to two phase-III double-blind placebo controlled trials of Sanofi Pasteur's dengue vaccine, which has recently been approved in Mexico and is likely to gain approval from the FDA and many other countries in 2016.