Original Investigation
Fluoride Exposure and Children’s IQ Scores.
Kyla W. Taylor, PhD; Sorina E. Eftim, PhD; Christopher A. Sibrizzi, MPH; Robyn B. Blain, PhD; Kristen Magnuson, MESM; Pamela A. Hartman, MEM; Andrew A. Rooney, PhD; John R. Bucher, PhD. JAMA Pediatrics. January 6, 2025.
Related Articles
• Comment. Assessing Fluoride Exposure and Children’s IQ Scores.
James W. Antoon, MD, PhD, MPH; Jayanth V. Kumar, DDS, MPH. JAMA Pediatrics.
• Comment. Assessing Fluoride Exposure and Children’s IQ Scores.
Robert C. Speth, PhD. JAMA Pediatrics.
In Reply
We appreciate the opportunity to address questions raised by the letter writers concerning our meta-analysis.1 Speth invokes the concept of hormesis and asks whether alternative dose-response models were adequately considered and whether a threshold exists. The dose-response meta-analysis considered 3 separate models—linear, quadratic, cubic splines—all are presented in supplemental materials; both latter models can be used to detect thresholds. For both drinking water and urine, the linear model provided the best fit in all dose-response analyses with statistically significant associations. When estimating fluoride exposure solely from drinking water, none of these models were statistically significant at less than 1.5 mg/L, and there was no evidence to support a threshold.
Antoon and Kumar (henceforth “Antoon”) express concern about the conduct of the meta-analyses; however, their comments contain multiple inaccuracies, uncited results, unsupported claims, and statements that indicate a lack of understanding of standard methodologies and statistical approaches used in meta-analysis (eg, calculation of standardized mean differences [SMDs] does not involve “doses,” nor are there “unadjusted” SMDs, and no studies involved population surveys).
Antoon suggests there is misplaced confidence in cross-sectional studies and inaccurately states that cross-sectional studies are generally excluded from meta-analyses. It is a misconception that cross-sectional studies cannot establish temporality, a key element in causal inference. We describe how temporality was determined in such studies. Judging or excluding studies solely by their study-design label (eg, cross-sectional, cohort), as Antoon suggests, is inappropriate. Doing so can introduce publication or selection bias and risks discarding important evidence.2 Instead, to follow best practices (eg, Cochrane), each study should be individually and systematically evaluated, as we did.1
Antoon cites values that do not appear in the published literature (eg, not in Yu et al3) while claiming the meta-analysis does not report some results, even though all of those results are available in the article or supplemental materials. Antoon’s comments include terminology that is not recognized in the field of environmental epidemiology (eg, “secondary cohort studies,” “etiologic meta-analysis,” “nonlinear regression coefficients”).
Additionally, some comments echo concerns from the Levy4 Editorial, with which we disagree, whereas others are unsupported (concerns with the meta-analysis by Kumar et al5 are outlined in a letter to the editor6). For example, both Antoon and Levy imply that any form of “heterogeneity” is problematic. But unlike meta-analyses of randomized trials, heterogeneity in methods and study populations is expected in observational studies and can be a strength: In our work,1 associations remained inverse despite heterogeneity in methods and across different study populations, supporting the robustness of associations. Their claim that spot urine samples are not valid does not reflect the scientific consensus (eg, regulatory organizations like the Environmental Protection Agency routinely rely on urinary measurements as exposure estimates in risk assessments) and ignores the evidence that results were also consistent across other exposure matrices (eg, water, fluoride intake). For exposure misclassification to systematically bias results, it would have to occur differentially by outcome across the entire body of evidence—a highly unlikely prospect given the diverse array of studies in our work.1
Confidence in the inverse association between fluoride exposure and children’s IQ in our meta-analysis is based on the clear demonstration of this association in the studies with low risk of bias, as well as the consistency of the association across different countries; across wide variations in exposure source, exposure levels, and exposure metrics; across different instruments that measure IQ; and across different designs in studies that have both high and low risk of bias.
While research continues, it is worth emphasizing the importance of limiting total fluoride intake during pregnancy, infancy, and early childhood, known critical periods of brain development.
___________________________________________________________
Corresponding Author: Kyla W. Taylor, PhD, PO Box 12233, Mail Drop K2-04, Research Triangle Park, NC 27709 (kyla.taylor@nih.gov).
Published Online: May 12, 2025. doi:10.1001/jamapediatrics.2025.0935
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported by the Intramural Research Program (ES103316, ES103317) at the National Institute of Environmental Health Sciences (NIEHS), National Institutes of Health and was performed for NIEHS under contract GS00Q14OADU417 (order HHSN273201600015U).
Role of the Funder/Sponsor: The funders had no role in the preparation, review, or approval of the manuscript or decision to submit the manuscript for publication.
Additional Contributions: We appreciate the helpful input from Christopher A. Sibrizzi, MPH (ICF), Robin B. Blain, PhD (ICF), Kristen Magnuson, MESM (ICF), Pamela A. Hartman, MEM (ICF), and John R. Bucher, PhD (retired).
1. Taylor KW, Eftim SE, Sibrizzi CA, et al. Fluoride exposure and children’s IQ scores: a systematic review and meta-analysis. JAMA Pediatr. 2025;179(3):282-292. doi:10.1001/jamapediatrics.2024.5542
ArticlePubMedGoogle ScholarCrossref
2. Savitz DA, Wellenius GA. Can cross-sectional studies contribute to causal inference? it depends. Am J Epidemiol. 2023;192(4):514-516. doi:10.1093/aje/kwac037PubMedGoogle ScholarCrossref
3. Yu X, Chen J, Li Y, et al. Threshold effects of moderately excessive fluoride exposure on children’s health: a potential association between dental fluorosis and loss of excellent intelligence. Environ Int. 2018;118:116-124. doi:10.1016/j.envint.2018.05.042PubMedGoogle ScholarCrossref
4. Levy SM. Caution needed in interpreting the evidence base on fluoride and IQ. JAMA Pediatr. 2025;179(3):231-234. doi:10.1001/jamapediatrics.2024.5539
ArticlePubMedGoogle ScholarCrossref
5. Kumar JV, Moss ME, Liu H, Fisher-Owens S. Association between low fluoride exposure and children’s intelligence: a meta-analysis relevant to community water fluoridation. Public Health. 2023;219:73-84. doi:10.1016/j.puhe.2023.03.011PubMedGoogle ScholarCrossref
