Exposure to glyphosate, the most used herbicide in the United States, is not well characterized. We assessed glyphosate exposure in a representative sample of the U.S. population ≥ 6 years from the 2013–2014 National Health and Nutrition Examination Survey.
We quantified glyphosate in urine (N = 2,310) by ion chromatography isotope-dilution tandem mass spectrometry. We conducted univariate analysis using log-transformed creatinine-corrected glyphosate concentrations with demographic and lifestyle covariates we hypothesized could affect glyphosate exposure based on published data including race/ethnicity, sex, age group, family income to poverty ratio, fasting time, sample collection season, consumption of food categories (including cereal consumption) and having used weed killer products. We used multiple logistic regression to examine the likelihood of glyphosate concentrations being above the 95th percentile and age-stratified multiple linear regression to evaluate associations between glyphosate concentrations and statistically significant covariates from the univariate analysis: race/ethnicity, sex, age group, fasting time, cereal consumption, soft drink consumption, sample collection season, and urinary creatinine.
Glyphosate weighted detection frequency was 81.2 % (median (interquartile range): 0.392 (0.263–0.656) μg/L; 0.450 (0.266–0.753) μg/g creatinine). Glyphosate concentration decreased from age 6–11 until age 20–59 and increased at 60+ years in univariate analyses. Children/adolescents and adults who fasted > 8 h had significantly lower model-adjusted geometric means (0.43 (0.37–0.51) μg/L and 0.37 (0.33–0.39) μg/L) than those fasting ≤ 8 h (0.51 (0.46–0.56) μg/L and 0.44 (0.41–0.48) μg/L), respectively. The likelihood (odds ratio (95 % CI)) of glyphosate concentrations being > 95th percentile was 1.94 (1.06–3.54) times higher in people who fasted ≤ 8 h than people fasting > 8 h (P = 0.0318).