Associations Between Prenatal Fluoride Exposure and Performance IQ in Canadian Preschool Aged Children: A Multilevel Modeling Approach

Abstract

BACKGROUND AND AIM: Prenatal fluoride exposure is associated with reduced IQ in children, especially Performance IQ (PIQ; nonverbal intelligence), but studies have not accounted for hierarchical clustering. We examined the association between sex-specific prenatal fluoride exposure and PIQ using multilevel modeling (MLM) to account for nesting of 448 mother-child pairs within 184 postal codes and six cities. METHODS: Our sample was from the Maternal-Infant Research on Environmental Chemicals (MIREC) prospective Canadian birth cohort. Variables included maternal urinary fluoride (MUF; adjusted for dilution), PIQ scores at age 3-4 years from the WPPSI, and participant level covariates (HOME scores, maternal education level, and maternal race). A series of three-level MLMs were fit to estimate associations between within-postal code and within-city centered MUF and PIQ. RESULTS:Median MUF concentration was 0.44 mg/L (IQR= 0.35), with average PIQ of 102 (SD= 14) for boys and 105 (SD= 14) for girls. A three-level intra-class correlation revealed that 3.8% and 9.1% of the variance in PIQ was accounted for by nesting of dyads in postal codes and cities, respectively. The interaction between child sex and MUF and the covariate effects were fixed across postal codes and cities in the optimal model. The interaction between child sex and within-postal-code MUF was significant (p .05).For boys, a 1 mg/L increase in within-postal-code MUF was significantly associated with PIQ, B= -7.9 (95% CI: -13.6, -2.2); for girls, the association was not significant, B= 1.7 (95% CI: -5.0, 8.5). Centering MUF within city resulted in a weaker effect among boys (B= -5.7 95% CI: -10.2, -1.2), and a stronger, but nonsignificant effect among girls (B= 3.3, 95% CI: -2.7, 9.3). CONCLUSIONS:For boys, the stronger effect of MUF in postal codes than cities shows the importance of accounting for hierarchical clustering in the MIREC cohort to examine associations between prenatal fluoride exposure and PIQ. KEYWORDS: biomarkers of exposure, children’s environmental health, epidemiology, fluoride, neurodevelopmental outcomes