Educational note: causal decomposition of population health differences using Monte Carlo integration and the g-formula

Abstract

One key objective of the population health sciences is to understand why one social group has different levels of health and well-being compared with another. Whereas several methods have been developed in economics, sociology, demography, and epidemiology to answer these types of questions, a recent method introduced by Jackson and VanderWeele (2018) provided an update to decompositions by anchoring them within causal inference theory. In this paper, we demonstrate how to implement the causal decomposition using Monte Carlo integration and the parametric g-formula. Causal decomposition can help to identify the sources of differences across populations and provide researchers with a way to move beyond estimating inequalities to explaining them and determining what can be done to reduce health disparities. Our implementation approach can easily and flexibly be applied for different types of outcome and explanatory variables without having to derive decomposition equations. We describe the concepts of the approach and the practical steps and considerations needed to implement it. We then walk through a worked example in which we investigate the contribution of smoking to sex differences in mortality in South Korea. For this example, we provide both pseudocode and R code using our package, cfdecomp. Ultimately, we outline how to implement a very general decomposition algorithm that is grounded in counterfactual theory but still easy to apply to a wide range of situations.

The Contribution of Health Behaviors to Depression Risk Across Birth Cohorts

“…We performed the g-computation based causal decomposition for every health behavior (mediator), using an approach that is described in detail by Sudharsanan & Bijlsma. 32 First, we specified the probability of elevated depressive symptoms as our measure of interest, and the relative difference between each birth cohort and the birth cohort with the lowest probability of depressive symptoms in our sample, that is, the 1945 birth cohort, as our contrast.…”

“…Our causal decomposition relies on three core assumptions: exchangeability (no unmeasured confounding), positivity, and consistency. 32 An advantage of causal decomposition is that only the mediator–outcome pathway (e.g., BMI-depression) can be confounded, but not the pathways from exposure to mediator or outcome because the exposure (birth cohort) is a group identifier. 32 Regarding our mediator–outcome pathway, unmeasured confounding might be present due to social or genetic factors.…”

“… 32 An advantage of causal decomposition is that only the mediator–outcome pathway (e.g., BMI-depression) can be confounded, but not the pathways from exposure to mediator or outcome because the exposure (birth cohort) is a group identifier. 32 Regarding our mediator–outcome pathway, unmeasured confounding might be present due to social or genetic factors. We controlled for education, which may not adjust for all possible confounders, resulting potentially in an overestimation of the contribution of health behavior to depression.…”

“…We performed the g-computation based causal decomposition for every health behavior (mediator), using an approach that is described in detail by Sudharsanan & Bijlsma. 32 …”

Rapidly Rising Diabetes and Increasing Body Weight: A Counterfactual Analysis in Repeated Cross-sectional Nationally Representative Data from Bangladesh

“…Our approach for this analysis was inspired by recent methodological developments for causal decomposition analyses. 20,21 This approach required first using the two linear and logistic regression models below (estimated on the pooled 2011 and 2017/2018 data) to predict the observed change in diabetes between 2011 and 2017 (known as the "natural course").…”