The directed acyclic graph helped identify confounders in the association between coronary heart disease and pesticide exposure among greenhouse vegetable farmers

Abstract

To explore the causal pathways associated with coronary heart disease (CHD) and pesticide exposure using a directed acyclic graph (DAG) analysis and to investigate the potential benefits of DAG by comparing it with logistic regression. This cross-sectional study enrolled 1368 participants from April 2015 to May 2017. Trained research investigators interviewed farmers using a self-administered questionnaire. Logistic regression and DAG models were used to identify the associations between CHD and chronic pesticide exposure. A total of 150 (11.0%) of the 1368 participants are characterized as having CHD. High pesticide exposure (odds ratio = 2.852, 95% confidence intervals: 1.951–4.171) is associated with CHD when compare with low pesticide exposure by both DAG and logistic analyses. After adjusting for the additional potential influence of factors identified by the DAG analysis, there is no significant association, such as the results in logistic regression: ethnicity, education level, settlement time, and mixed pesticide status. Specifically, age, meal frequency, and consumption of fresh fruit, according to the DAG analysis, are independent factors for CHD. High pesticide exposure is a risk factor for CHD as indicated by both DAG and logistic regression analyses. DAG can be a preferable improvement over traditional regression methods to identify sources of bias and causal inference in observational studies, especially for complex research questions.