RESUMO
The topic of cardiovascular hazards from multiple metal (loid)s exposures has attracted widespread attention. Here, we measured concentrations of twenty-three urinary metal (loid)s and mean platelet volume (MPV), an early cardiovascular damage biomarker, for 3396 Chinese adults. We aimed to comprehensively assess the associations of single metal (loid) and multiple metal (loid)s (as a mixture) with MPV by combined use of five statistical methods, including general linear models, Bayesian kernel machine regression (BKMR), weight quartile sum (WQS) regression, quantile g-computation (QGC), and adaptive elastic network regression (AENR). And based on that, we hope to provide insight into assessing the health effect of multipollutant exposure. After adjustment for potential covariates, at least three methods jointly suggested that of twenty-three metal (loid)s, iron, arsenic, and antimony were positively while aluminum, tungsten, and thallium were inversely associated with MPV. The environmental risk score of metal (loid)s construed by AENR was significantly positively associated with MPV, while the association between overall twenty-three metal (loid)s mixture and MPV was neutralized to be insignificant in QGC and BKMR. Conclusively, single metal (loid) may be inversely (iron, arsenic, and antimony) and positively (aluminum, tungsten, and thallium) associated with early cardiovascular damage, while the association of overall twenty-three metal (loid)s mixture with MPV was insignificant when concurrent exposures exist. It is crucial to select appropriate statistical methods based on study purpose and principles/characteristics of statistical methods, and combined employment of multimethod is insightfully suggested when assessing health effects of multipollutant exposure.