Technology Category
Machine Learning: Causal LearningReasoning under Uncertainty: CausalitySearch and Optimization: Algorithm Configuration
Causal Inference with Missing Exposures, Missing Outcomes, and Dependence
🤖 AI Summary
This study addresses the challenge of causal inference in public health observational studies featuring multivariate missingness in exposure, outcome, and baseline risk variables, coupled with intra-family clustering dependence. Motivated by the Ugandan SEARCH-TB cohort—designed to assess the effect of alcohol consumption on tuberculosis infection risk—we propose a novel methodological framework. First, we introduce a causal graph model that jointly encodes both the missingness mechanism and the clustering structure, establishing formal identifiability theory for causal effects under multivariate missingness. Second, we extend Targeted Minimum Loss-based Estimation (TMLE) to accommodate simultaneous missingness in exposure, outcome, and baseline risk variables, integrating Super Learner for flexible, explicit modeling of the missingness mechanism and robust estimation. Applied to the SEARCH-TB data, our approach estimates a 49% increased relative risk of TB infection associated with alcohol use (RR = 1.49, 95% CI: 1.39–1.59), demonstrating superior precision and stability compared to inverse probability weighting and complete-case analysis.
Technology Category
Application Category
📝 Abstract
Missing data are ubiquitous in public health research. The missing-completely-at-random (MCAR) assumption is often unrealistic and can lead to meaningful bias when violated. The missing-at-random (MAR) assumption tends to be more reasonable, but guidance on conducting causal analyses under MAR is limited when there is missingness on multiple variables. We present a series of causal graphs and identification results to demonstrate the handling of missing exposures and outcomes in observational studies. For estimation and inference, we highlight the use of targeted minimum loss-based estimation (TMLE) with Super Learner to flexibly and robustly address confounding, missing data, and dependence. Our work is motivated by SEARCH-TB's investigation of the effect of alcohol consumption on the risk of incident tuberculosis (TB) infection in rural Uganda. This study posed notable challenges due to confounding, missingness on the exposure (alcohol use), missingness on the baseline outcome (defining who was at risk of TB), missingness on the outcome at follow-up (capturing who acquired TB), and clustering within households. Application to real data from SEARCH-TB highlighted the real-world consequences of the discussed methods. Estimates from TMLE suggested that alcohol use was associated with a 49% increase in the relative risk (RR) of incident TB infection (RR=1.49, 95%CI: 1.39-1.59). These estimates were notably larger and more precise than estimates from inverse probability weighting (RR=1.13, 95%CI: 1.00-1.27) and unadjusted, complete case analyses (RR=1.18, 95%CI: 0.89-1.57). Our work demonstrates the utility of causal models for describing the missing data mechanism and TMLE for flexible inference.
Problem
Research questions and friction points this paper is trying to address.
Addressing missing exposures and outcomes in causal inference studies
Handling confounding and data dependence in observational research
Estimating causal effects under missing-at-random assumptions
Innovation
Methods, ideas, or system contributions that make the work stand out.
Uses causal graphs for missing data mechanism
Applies TMLE with Super Learner for estimation
Addresses confounding and dependence flexibly
🔎 Similar Papers
No similar papers found.
