Causal inference with multiple concurrent medications: A comparison of methods and an application in multidrug-resistant tuberculosis

• Type of Material: Article
• Authors: Arman Alam Siddique, McMaster University; Mireille E. Schnitzer, University of Montreal; Asma Bahamyirou, University of Montreal; Guanbo Wang, McGill University; Timothy H. Holtz, Centers for Disease Control and Prevention; Giovanni B. Migliori, Fondazione S. Maugeri; Giovanni Sotgiu, University of Sassari; Neel Gandhi, Emory University; Mario H. Vargas, Instituto Nacional de Enfermedades Respiratorias; Dick Menzies, McGill University; Andrea Benedetti, McGill University
• Language: English
• Date: 2019-12-01
• Publisher: SAGE Publications (UK and US)
• Publication Version: Author Accepted Manuscript (After Peer Review)
• Copyright Statement: © The Author(s) 2018.
• Final Published Version (URL): http://dx.doi.org/10.1177/0962280218808817
• Title of Journal or Parent Work: Statistical Methods in Medical Research
• ISSN: 0962-2802
• Volume: 28
• Issue: 12
• Start Page: 3534
• End Page: 3549
• Grant/Funding Information: This work was supported by the Canadian Institutes of Health Research (CIHR) (project grant 378067 to MES and AB).; NRG is funded in part by the National Institutes of Health (NIH) (K24 award, K24AI114444).; MES is also funded by CIHR (New Investigators Salary Award) and the National Sciences and Engineering Council of Canada (Discovery Grant with Accelerator Supplement).
• Keywords: multidrug-resistant tuberculosis; Causal inference; Mathematical & Computational Biology; concurrent medications; targeted maximum likelihood estimation; PROPENSITY SCORE; Physical Sciences; generalized propensity score; machine learning; Mathematics; Science & Technology; MODELS; Medical Informatics; Life Sciences & Biomedicine; Statistics & Probability; Health Care Sciences & Services
• Research Categories: Health Sciences, Public Health; Statistics

Abstract

This paper investigates different approaches for causal estimation under multiple concurrent medications. Our parameter of interest is the marginal mean counterfactual outcome under different combinations of medications. We explore parametric and non-parametric methods to estimate the generalized propensity score. We then apply three causal estimation approaches (inverse probability of treatment weighting, propensity score adjustment, and targeted maximum likelihood estimation) to estimate the causal parameter of interest. Focusing on the estimation of the expected outcome under the most prevalent regimens, we compare the results obtained using these methods in a simulation study with four potentially concurrent medications. We perform a second simulation study in which some combinations of medications may occur rarely or not occur at all in the dataset. Finally, we apply the methods explored to contrast the probability of patient treatment success for the most prevalent regimens of antimicrobial agents for patients with multidrug-resistant pulmonary tuberculosis.