Background: Exposure to individual metals (and metalloids; hereafter 'metals') is associated with adverse cardiometabolic outcomes. Specifying analytic models to assess relationships among metal mixtures and cardiometabolic outcomes requires evidence-based models of the (assumed) causal structures; however, such models have not been previously published.
Methods: We conducted a systematic literature review to develop an evidence-based directed acyclic graph (DAG) identifying relationships among metals, cardiometabolic health indicators, and potential confounders. To evaluate the consistency of the DAG with data from 1797 participants in the San Luis Valley Diabetes Study (SLVDS; mean age=54 years, 53% women, 48% Hispanic), we tested conditional independence statements suggested by the DAG and by 100 DAGs with the same structure but randomly permuted nodes using linear (continuous outcomes), logistic (dichotomous outcomes), or Bayesian kernel machine regression (BKMR; statements with metal coexposures) models. Based on minimally sufficient adjustment sets identified by the DAG, we specified BKMR models assessing associations between urinary metal mixtures and cardiometabolic outcomes in the SLVDS population.
Results: Twenty-nine articles met the inclusion criteria for the systematic review. From these articles, we developed an evidence-based DAG with 382 testable conditional independence statements (71% supported by SLVDS data). Only 3% of the DAGs with randomly permuted nodes indicated more agreement with the data than our evidence-based DAG. Applying the evidence-based DAG in a pilot analysis, we did not observe evidence for an association among metal mixtures and cardiometabolic outcomes.
Conclusions: We developed, tested, and applied an evidence-based approach to analyze associations between metal mixtures and cardiometabolic health.