Combined exposure to multiple metals on serum uric acid in NHANES under three statistical models

Abstract

Background: There are rare researches on the correlations between metals exposure and serum uric acid (SUA), and existing research has only investigated the single metal effect. This study aimed to investigate the combined effects of metal mixtures on SUA and hyperuricemia using three statistical models.

Methods: In this study, the data were extracted from three cycle years of the National Health and Nutrition Examination Survey (NHANES). Subsequently, generalized linear regression, weighted quantile regression (WQS) and Bayesian kernel machine regression (BKMR) models were fitted to evaluate the correlations between metal mixtures and both SUA and hyperuricemia.

Results: Of 3926 participants included, 19.13% participants had hyperuricemia. It was found using multi-metals generalized linear regression models that there were positive correlations of arsenic and cadmium with both outcomes. The negative correlations were identified in cobalt, iodine, and manganese with SUA concentration, whereas only cobalt was negatively correlated with hyperuricemia. Based on the WQS regression model fitted in positive direction, it was suggested that the WQS indices were significantly correlated with SUA (β = 6.64, 95% CI: 3.14-10.13) and hyperuricemia (OR = 1.25, 95% CI: 1.08-1.44); however, the result achieved by using the model fitted in negative direction indicated that the WQS indices were only significantly correlated with SUA (β = -5.29, 95%CI: 8.02 ∼ -2.56). With the use of the BKMR model, a significant increasing trend between metal mixtures and hyperuricemia was found, while no significant overall effect of metal mixtures on SUA was identified. The predominant roles of arsenic, cadmium, and cobalt in the change of SUA and hyperuricemia risk were found using all three models.

Conclusion: The finding of this study revealed that metal mixtures might have a positive combined effect on hyperuricemia. The mutual verification of two outcomes using the three different models provided strong public health implications for protecting people from heavy metal pollution and preventing hyperuricemia.

Keywords: Bayesian kernel machine regression model; Generalized linear regression model; Heavy metals; Serum uric acid; Weighted quantile regression model.