Optimal Balancing of Time-Dependent Confounders for Marginal Structural Models
This addresses a methodological bottleneck in causal inference for researchers, offering a more robust alternative to existing weighting methods, though it is incremental as it builds on prior work like IPTW and CBPS.
The paper tackles the problem of high-variance and model misspecification in marginal structural models for causal effects of time-varying treatments by proposing Kernel Optimal Weighting (KOW), which optimally balances time-dependent confounders and controls precision, showing improved performance in simulations and applications to HIV treatment and election advertising.
Marginal structural models (MSMs) estimate the causal effect of a time-varying treatment in the presence of time-dependent confounding via weighted regression. The standard approach of using inverse probability of treatment weighting (IPTW) can lead to high-variance estimates due to extreme weights and be sensitive to model misspecification. Various methods have been proposed to partially address this, including truncation and stabilized-IPTW to temper extreme weights and covariate balancing propensity score (CBPS) to address treatment model misspecification. In this paper, we present Kernel Optimal Weighting (KOW), a convex-optimization-based approach that finds weights for fitting the MSM that optimally balance time-dependent confounders while simultaneously controlling for precision, directly addressing the above limitations. KOW directly minimizes the error in estimation due to time-dependent confounding via a new decomposition as a functional. We further extend KOW to control for informative censoring. We evaluate the performance of KOW in a simulation study, comparing it with IPTW, stabilized-IPTW, and CBPS. We demonstrate the use of KOW in studying the effect of treatment initiation on time-to-death among people living with HIV and the effect of negative advertising on elections in the United States.