Multiple plasma metal exposures and incident anemia in middle-aged and elderly individuals: The Dongfeng-Tongji cohort

Anemia is a global health concern. However, prospective studies on the relationships between multiple metal exposures and incident anemia remain limited. This study included 3928 participants from the Dongfeng-Tongji cohort. Using the generalized linear (GLM) model to estimate the associations between individual metals and anemia. The least absolute shrinkage and selection operator (LASSO) regression model identified metals associated with anemia. Additionally, Bayesian kernel machine regression (BKMR) and Quantile g-computation (Q-g) regression models were employed to identify the significant contributing metals. Compared to the lowest quartiles, the odds ratios (ORs) and 95 % confidence intervals (CIs) for the fourth quartiles were: aluminum 1.64 (1.17, 2.32), arsenic 1.57 (1.12, 2.20), barium 1.56 (1.12, 2.19), molybdenum 1.92 (1.38, 2.69), rubidium 0.71 (0.51, 0.98), strontium 1.55 (1.13, 2.14), vanadium 1.67 (1.18, 2.37). The ORs and 95 % CIs for the risk of incident anemia per one interquartile range (IQR) increase were: selenium 0.88 (0.78, 1.00), and zinc 1.20 (1.04, 1.38). LASSO regression models identified eight plasma metals potentially associated with incident anemia. Six metals, significant in both the GLM and LASSO models, were selected for further analysis. The BKMR model demonstrated a positive association between mixed metals and anemia. Q-gcomp regression indicated rubidium mainly contributed to negative weights, while molybdenum and vanadium contributed to positive weights. This study found that elevated plasma levels of arsenic, molybdenum, strontium, and vanadium were associated with higher risks of incident anemia, while rubidium and selenium had a protective role. These findings indicated that controlling metal exposures would be important in preventing anemia.