Concerns About Data and Analyses Used in Assessing Fluoride Exposure and Children’s IQ Scores.

Abstract

Original Investigation

Fluoride Exposure and Children’s IQ Scores. A Systematic Review and Meta-Analysis.
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.

• Response. Concerns About Data and Analyses Used in Assessing Fluoride Exposure and Children’s IQ Scores.
Kyla W. Taylor, Sorina E. Eftim, Andrew A. Rooney. JAMA Pediatrics.

To the Editor

We have several concerns about the confidence Taylor et al¹ placed in cross-sectional and secondary data cohort studies. Cross-sectional studies are generally excluded from an etiologic meta-analysis because they cannot account for reverse causality or time-varying confounding. The OHAT Handbook also assigns low confidence ratings to population surveys. Unlike adjusted regression coefficients, unadjusted standardized mean difference (SMD) estimates derived from unrepresentative samples are not an accurate effect measure. The study findings may be mistakenly interpreted that water fluoridation causes IQ deficits, contradicting other meta-analyses.²

The SMD dose-response analysis relied on water fluoride concentration in countries where the water intake differs substantially.³ Therefore, it misrepresents the dose of fluoride used to calculate the effect estimates. Furthermore, the meta-analysis of urinary fluoride studies is based on 1 to 3 spot urine samples, often not adjusted for dilution, and is not a valid exposure measure.⁴

Concerns about data included in the meta-analysis that are different from the original publications also exist. Taylor et al¹ should provide more details to ensure the study’s replicability.

• Seven of 13 urinary coefficients are not found in the original studies. The nonlinear regression coefficients, when combined using the random-effects model vs fixed-effects model, show vastly different coefficients (Yu et al, reference 33 in the original article, -1.65 vs -0.17). It is unclear why the coefficient that better describes an approximately 1- vs 9–IQ point decline over the 5.6-mg/L change in urinary fluoride is not used.
• The log-transformed B for Lin et al, reference 10 in the original article, differs from that provided to us by the study authors (B, -1.67; 95% CI, -6.36 to 3.02). How did the authors calculate this coefficient for other studies?
• Unlike Yu et al, the other publications from Tianjin City (Zhang et al, Cui et al, and Zhao et al, references 35, 99, 100 in the original article, respectively) did not directly assess associations between urinary fluoride and IQ. The inclusion of these studies undermines the validity of the synthesized estimates.

Substantial unexplained heterogeneity across studies directly contradicts the authors’ statements about the consistency of findings. Publication bias, as in the case of the unpublished ELEMENT study⁵ that reported a positive association (B, 1.32, P = .33), could show consistent inverse associations. The inconsistency becomes more apparent if the positive coefficients in the nonlinear relationship are included in the Figure 2 forest plot. Subgroup analysis to assess heterogeneity by endemic-vs-nonendemic areas reveals inverse (pooled B_, -2.18; 95% CI, -3.32 to -1.03) and positive associations (pooled B, 0.27; 95% CI, -0.76 to 1.29), raising questions about the utility of the pooled overall estimates.

Corresponding Author: James W. Antoon, MD, PhD, MPH, Department of Pediatrics, Vanderbilt University Medical Center, 2200 Children’s Way, Nashville, TN 37232 (james.antoon@vumc.org).

Published Online: May 12, 2025. doi:10.1001/jamapediatrics.2025.0932

Conflict of Interest Disclosures: Dr Antoon reported previously serving as a member of an AstraZeneca scientific advisory board and being partially supported by the National Institute for Allergy and Infectious Diseases of the National Institutes of Health (NIH; K23 AI168496). Dr Kumar reported being a former employee of the California Department of Public Health, being a member of the National Fluoridation Advisory Committee for the American Dental Association, and being a senior volunteer with the National Institute of Dental and Craniofacial Research (NIDCR) Fellowship Program .

Disclaimer: The findings and conclusions in this letter are those of the authors and do not necessarily represent the views or opinions of the Association of State and Territorial Dental Directors, NIH, California Department of Public Health, or NIDCR.