Targeted Learning on Variable Importance Measure for Heterogeneous Treatment Effect

Quantifying the heterogeneity of treatment effect is important for understanding how a commercial product or medical treatment affects different subgroups in a population. Beyond the overall impact reflected parameters like the average treatment effect, the analysis of treatment effect heterogeneity further reveals details on the importance of different covariates and how they lead to different treatment impacts. One relevant parameter that addresses such heterogeneity is the variance of treatment effect across different covariate groups, however the treatment effect is defined. One can also derive variable importance parameters that measure (and rank) how much of treatment effect heterogeneity is explained by a targeted subset of covariates. In this article, we propose a new targeted maximum likelihood estimator for a treatment effect variable importance measure. This estimator is a pure plug-in estimator that consists of two steps: 1) the initial estimation of relevant components to plug in and 2) an iterative updating step to optimize the bias-variance tradeoff. The simulation results show that this TMLE estimator has competitive performance in terms of lower bias and better confidence interval coverage compared to the simple substitution estimator and the estimating equation estimator. The application of this method also demonstrates the advantage of a substitution estimator, which always respects the global constraints on the data distribution and that the estimand is a particular function of the distribution.