The Estimation of Causal Effects of Multiple Treatments in Observational Studies Using Bayesian Additive Regression Trees

TL;DR

This paper introduces a Bayesian nonparametric method using BART to estimate causal effects of multiple treatments with binary outcomes, demonstrating superior accuracy over existing methods through simulations and real data analysis.

Contribution

The paper develops and evaluates a novel Bayesian Additive Regression Trees approach for causal inference with multiple treatments and binary outcomes, filling a methodological gap.

Findings

BART shows low bias and mean-squared errors in simulations.

BART outperforms propensity score-based methods in estimating treatment effects.

Application to SEER-Medicare data illustrates practical utility.

Abstract

There is currently a dearth of appropriate methods to estimate the causal effects of multiple treatments when the outcome is binary. For such settings, we propose the use of nonparametric Bayesian modeling, Bayesian Additive Regression Trees (BART). We conduct an extensive simulation study to compare BART to several existing, propensity score-based methods and to identify its operating characteristics when estimating average treatment effects on the treated. BART consistently demonstrates low bias and mean-squared errors. We illustrate the use of BART through a comparative effectiveness analysis of a large dataset, drawn from the latest SEER-Medicare linkage, on patients who were operated via robotic-assisted surgery, video-assisted thoratic surgery or open thoracotomy.