OBJECTIVES : Marginal structural models (MSM) are a class of causal models for the estimation of the causal effect of a time-dependent exposure in the presence of time-dependent covariates that may be simultaneously confounders and intermediate variables. We developed best practices for the application of MSM methodology to studies using real-world data.

METHODS : A systematic literature review was conducted of the Current Index to Statistics database and OVID’s MEDLINE, Embase, and Cochrane database of systematic reviews using a combination of relevant search terms. The search was restricted to publications no earlier than 1997. Additionally, abstracts from the most recent 3 years of meetings from the International Society for Pharmacoeconomics and Outcomes Research and the International Society for Pharmacoepidemiology were reviewed.

RESULTS : A total of 31 methodological articles met the inclusion/exclusion criteria and were included in the review. Critical appraisal of the following aspects of MSM was conducted: causal effect analysis in time-dependent confounding scenarios; methodology assumptions; implementation in missing data; dynamic treatment regimens; targeted maximum likelihood estimator; and other applications of MSM in longitudinal data. Recommendations were developed for each aspect.

CONCLUSIONS : MSM is used to account for time-dependent confounding and to estimate causal effects in real-world data. We developed recommendations for applying this methodology as best practice guidance for pharmacoepidemiologists and outcomes researchers.