Adjusting for Partial Compliance in SMARTs: a Bayesian Semiparametric Approach

TL;DR

This paper introduces a Bayesian semiparametric method to analyze longitudinal treatment data with partial compliance, improving the estimation of optimal strategies in substance use disorder trials.

Contribution

It develops a novel Bayesian semiparametric approach that accounts for partial compliance in longitudinal studies, addressing a key gap in adaptive treatment analysis.

Findings

Method effectively estimates treatment outcomes under different compliance strata.

Application to ENGAGE study shows treatment strategy varies with compliance levels.

Simulation studies confirm robustness and accuracy of the proposed approach.

Abstract

The cyclical and heterogeneous nature of many substance use disorders highlights the need to adapt the type or the dose of treatment to accommodate the specific and changing needs of individuals. The Adaptive Treatment for Alcohol and Cocaine Dependence study (ENGAGE) is a multi-stage randomized trial that aimed to provide longitudinal data for constructing treatment strategies to improve patients' engagement in therapy. However, the high rate of noncompliance and lack of analytic tools to account for noncompliance have impeded researchers from using the data to achieve the main goal of the trial. We overcome this issue by defining our target parameter as the mean outcome under different treatment strategies for given potential compliance strata and propose a Bayesian semiparametric model to estimate this quantity. While it adds substantial complexities to the analysis, one important feature of our work is that we consider partial rather than binary compliance classes which is more relevant in longitudinal studies. We assess the performance of our method through comprehensive simulation studies. We illustrate its application on the ENGAGE study and demonstrate that the optimal treatment strategy depends on compliance strata.