Serum cotinine and bone mineral density among U.S. women aged 50 years or older: a complex-survey cross-sectional analysis of NHANES 2013–2014 and 2017–2018

Serum cotinine objectively reflects recent nicotine uptake, but a single measurement does not capture cumulative smoking exposure. We evaluated the cross-sectional association with bone mineral density (BMD) among U.S. women aged 50 years and older while accounting for the NHANES complex survey design. This cross-sectional analysis pooled NHANES 2013–2014 and 2017–2018 and accounted for the multistage complex survey design. Serum cotinine was log2-transformed; coefficients represent the adjusted BMD difference per twofold higher concentration. Primary outcomes were total femur, femoral neck, and the official total L1–L4 lumbar-spine BMD. Survey analyses used WTMEC2YR/2, the NHANES masked variance strata and primary sampling units as documented in the analysis code, Taylor series linearization, and domain estimation. The prespecified primary family comprised women-specific model-2 estimates adjusted for age and body-mass-index splines, race/ethnicity, cycle, poverty-income ratio, alcohol use, physical activity, diabetes, estimated glomerular filtration rate, reproductive status, and hormone therapy; Benjamini–Hochberg false-discovery-rate (FDR) correction covered the three sites. Outcome-specific women samples were 1,699 for the two hip sites and 1,065 for official spine BMD. Per cotinine doubling, adjusted differences were − 0.0010 g/cm² (95% CI − 0.0023 to 0.0003; FDR q = 0.415) for total femur, − 0.0005 (− 0.0019 to 0.0009; q = 0.576) for femoral neck, and 0.0006 (− 0.0017 to 0.0029; q = 0.576) for the spine. Restricted cubic splines did not support overall or nonlinear associations after FDR correction. Cotinine interactions with sex, menopausal status, and cycle were also not supported. A 50-imputation sensitivity analysis yielded no FDR-significant primary result. In the NHANES 2013-2014 and 2017-2018 cycles, we found no statistically robust evidence, after adjustment for the prespecified model-2 covariates and multiplicity correction, that serum cotinine was associated with BMD among women aged 50 years or older. These findings do not establish clinical equivalence and should not be extrapolated to lifetime smoking burden, fracture risk, or the skeletal safety of specific nicotine products. Longitudinal studies with repeated, product-specific biomarkers and lifetime exposure histories are needed.