Overview
Cross-sectional study using NHANES 2017-2018 data (n=1,648 adults with directly measured PFAS and blood metals) to assess the joint association of PFOA, PFOS, lead, cadmium, and mercury with prevalent diabetes. The analysis used survey-weighted logistic regression, WQS regression, quantile g-computation, and both design-aware and naive BKMR. Survey-weighted diabetes prevalence in the sample was 11.4%.
Key numbers
n=1,648 US adults, NHANES 2017-2018.
Mixture findings:
- Lead and PFOS emerged as the dominant contributors in BKMR, with posterior inclusion probabilities of 1.00 and 0.997, respectively.
- In WQS, PFOS and PFOA were the primary drivers of the positive mixture index.
- BKMR exposure-response functions for lead and cadmium were non-monotonic: associations with diabetes are nonlinear rather than uniformly directional across the exposure range.
- Survey-weighted diabetes prevalence: 11.4%.
The non-monotonic finding for lead and cadmium is important: it indicates that conventional logistic regression with a single interaction term cannot capture the structure of the dose-response relationship, and that mixture methods reveal associations that single-pollutant models miss.
Evidence Fitness
Context only. Cross-sectional design; temporal precedence not established. The study specifically tests a metal-PFAS joint mixture, not metals alone. The BKMR non-monotonic finding for lead and cadmium illustrates that mixture effects can diverge from what single-metal models predict, which is relevant to the underprotection synthesis without itself being a food concentration measurement.
Methods (brief)
NHANES 2017-2018. Blood metals by ICP-MS; serum PFAS by HPLC-MS/MS. Diabetes: physician diagnosis or HbA1c at least 6.5% or fasting glucose at least 126 mg/dL or antidiabetic medication use. Lab-confirmed sensitivity analysis used HbA1c at least 6.5% or antidiabetic medication. Survey-weighted analyses to account for NHANES complex design. BKMR, WQS, and quantile g-computation triangulated.
Implications
Certification: Reinforces the case for multi-analyte monitoring; lead’s dominant role in a metal-PFAS joint mixture for diabetes risk is consistent with HMT&C Tier-1 prioritization of lead.
Courses: Useful for illustrating non-monotonic dose-response relationships and why standard logistic regression can underestimate mixture effects.
App: No food concentration data.
Microbiome: Not applicable.
Related evidence
- Lead
- Cadmium
- Mercury, Total
- Single-metal regulatory limits systematically underprotect against dietary heavy-metal exposure
Update history
No substantive edit history is available in this build. The full commit record is available in git.