A Systematic Review and Dose-response Meta-analysis on Fluoride Exposure and Neurodevelopmental Toxicity in Children

Background The possible harmful effect of fluoride exposure on neurodevelopment in children is widely discussed, however, the magnitude of this relation is still controversial. The aim of this systematic review and meta-analysis was to characterize this relation through a dose-response approach. Methods/Approach We performed an online literature search in PubMed, Web of Science and Embase up to September 10, 2022. From each eligible study we extracted the effect size as a mean difference (MD) of intelligence quotient (IQ) scores and its 95% confidence intervals (CI). We performed a meta-analysis comparing the highest versus lowest fluoride exposure using a random-effects model, stratifying for type of exposure (fluoride in drinking water or urinary fluoride). Finally, we quantitively assessed such relation by performing a one-stage dose-response meta-analysis based on a cubic spline random-effects model. Results Out of 1955 potentially relevant records retrieved through a systematic literature search, we identified 34 eligible studies for review; 32 were also included for meta-analysis. The analysis comparing the highest-versus-lowest exposure showed decreased IQ scores for both water and urinary fluoride with MD of -6.62 (95% CI -8.61, -4.63) and -3.58 (95% CI -7.28, 0.12), respectively. The dose-response analysis showed a decrease in IQ score from 1 mg/L, considerably steeper above 2.5 mg/L for water fluoride. The dose-response curve for urinary fluoride showed a weaker but constant decrease in IQ scores with increasing fluoride levels. Conclusions Overall, a substantial consistency in the effect direction emerged, with adverse effects of different magnitude on children’s intelligence, arising from rather low fluoride exposure previously considered safe from a public health perspective. However, potential methodological limitations, such as heterogeneity in outcome measures and lack of consideration of possible confounders in some of the studies, may have affected these findings.