Dietary intake can modify the impact of metals on human health, and is also closely related to
glucose metabolism in human bodies. However, research on their interaction is limited. We used data based on 1738 adults aged ≥20 years from the National Health and Nutrition Examination Survey 2011-2016. We combined linear regression and restricted cubic splines with Bayesian kernel machine regression (BKMR) to identify metals associated with each
glucose metabolism index (P < 0.05 and the posterior inclusion probabilities of BKMR >0.5) in eight non-essential
heavy metals (
barium,
cadmium,
antimony,
tungsten,
uranium,
arsenic, lead, and
thallium) and
glucose metabolism indexes [fasting plasma
glucose (FPG), blood
hemoglobin A1c (HbA1c) and homeostatic model assessment of
insulin resistance (HOMA-IR)]. We identified two pairs of metals associated with
glucose metabolism indexes:
cadmium and
tungsten to HbA1c and
barium and
thallium to HOMA-IR. Then, the cross-validated kernel ensemble (CVEK) approach was applied to identify the specific nutrient group (nutrients) that interacted with the association. By using the CVEK model, we identified significant interactions between the energy-adjusted diet inflammatory index (E-DII) and
cadmium,
tungsten and
barium (all P < 0.05); macro-nutrients and
cadmium,
tungsten and
barium (all P < 0.05); minerals and
cadmium,
tungsten,
barium and
thallium (all P < 0.05); and A
vitamins and
thallium (P = 0.043). Furthermore, a lower E-DII, a lower intake of
carbohydrates and
phosphorus, and a higher consumption of
magnesium seem to attenuate the positive association between metals and
glucose metabolism indexes. Our finding identifying the nutrients that interact with non-essential
heavy metals could provide a feasible nutritional guideline for the general population to protect against the adverse effects of non-essential
heavy metals on
glucose metabolism.