Estimating the treatment effect on the treated under time-dependent confounding in an application to the Swiss HIV Cohort Study

TL;DR

This paper introduces a new two-step method using additive hazard regression and linear increments models to estimate the treatment effect on the treated in the presence of time-dependent confounding, demonstrated on Swiss HIV data.

Contribution

It presents a novel, easy-to-implement approach for estimating ATT with time-dependent confounders, applicable to mediation analysis and using standard R packages.

Findings

Effective estimation of treatment effects in HIV cohort data.

Method outperforms existing approaches in handling time-dependent confounding.

Applicable to real-world clinical data for treatment effect analysis.

Abstract

When comparing time-varying treatments in a non-randomised setting, one must often correct for time-dependent confounders that influence treatment choice over time and that are themselves influenced by treatment. We present a new two step procedure, based on additive hazard regression and linear increments models, for handling such confounding when estimating average treatment effects on the treated (ATT). The approach can also be used for mediation analysis. The method is applied to data from the Swiss HIV Cohort Study, estimating the effect of antiretroviral treatment on time to AIDS or death. Compared to other methods for estimating the ATT, the proposed method is easy to implement using available software packages in R.