Association between exposure to a mixture of phenols, pesticides, and phthalates and obesity: Comparison of three statistical models

Abstract

Background: The evaluation of the chemical impact on human health is usually constrained to the analysis of the health effects of exposure to a single chemical or a group of similar chemicals at one time. The effects of chemical mixtures are seldom analyzed. In this study, we applied three statistical models to assess the association between the exposure to a mixture of seven xenobiotics (three phthalate metabolites, two phenols, and two pesticides) and obesity. Methods: Urinary levels of environmental phenols, pesticides, and phthalate metabolites were measured in adults who participated in the U.S.-based National Health and Nutrition Examination Survey (NHANES) from 2013 to 2014. Body examination was conducted to determine obesity. We fitted multivariable models, using generalized linear (here both logistic and linear) regression, weighted quantile sum (WQS) regression, and Bayesian kernel machine regression (BKMR) models to estimate the association between chemical exposures and obesity. Results: Of 1269 individuals included in our final analysis, 38.5% had general obesity and 58.0% had abdominal obesity. In the logistic regression model established for each single chemical, bisphenol S (BPS), mono (carboxyoctyl) phthalate (MCOP), and mono (2-ethyl-5-carboxypentyl) phthalate (MECPP) were associated with both general and abdominal obesity (fourth vs. first quartile). In linear regression, MCOP was associated with BMI and waist circumference. In WQS regression analysis, the WQS index was significantly associated with both general obesity (OR = 1.63, 95% CI: 1.21–2.20) and abdominal obesity (OR = 1.66, 95% CI: 1.18–2.34). MCOP, bisphenol A (BPA), bisphenol S (BPS), and mono ethyl phthalate (MEP) were the most heavily weighing chemicals. In BKMR analysis, the overall effect of mixture was significantly associated with general obesity when all the chemicals were at their 60th percentile or above it, compared to all of them at their 50th percentile. MCOP, BPA, and BPS showed positive trends. By contrast, MECPP showed a flat and modest inverse trend. Conclusion: When comparing results from these three models, MCOP, BPA, and BPS were identified as the most important factors associated with obesity. We recommend estimating the joint effects of chemical mixtures by applying diverse statistical methods and interpreting their results together, considering their advantages and disadvantages. Keywords: Chemical mixture, Weighted quantile sum (WQS) regression, Bayesian kernel machine regression (BKMR), Obesity, Obesogen