Letter to the Editor,“Early Childhood Exposures to Fluorides and Cognitive Neurodevelopment: A Population-Based Longitudinal Study”

Three types of limitation may explain why Do et al. (2025) was unable to find an association between fluoride and lowered IQ, in contrast to most other studies (Taylor et al. 2025).

First, small subgroup sample sizes and limited exposure contrasts reduce power to detect an effect.

Second, effect sizes were likely attenuated through several sources of nondifferential (random) measurement imprecision, all producing bias toward the null (“no effect”) (Armstrong 1998). These included unmeasured water intake variation, multiyear averaged exposure diluting the more sensitive infancy period, ecological rather than individual biometrics such as urine fluoride, no prenatal exposure information (including from tea), IQ measured at age 16+ y when early life effects become obscured, no control for COVID-19 IQ effects, and adjustment for neurodevelopmental disorders may have “adjusted away” effects because attention-deficit/hyperactivity disorder may be on the causal pathway to IQ.

Third, uncontrolled confounding produced bias away from an adverse effect because Do et al. (2025) did not adjust for Indigenous ethnicity (Aboriginal), and there was likely a higher proportion of Indigenous individuals among the nonfluoridated (Do and Spencer 2016). Likewise, the study did not control for the low child development scores in nonfluoridated Queensland, which is explainable by Queensland’s low preschool attendance and no universal child/parent health services, unlike the fluoridated states (Brinkman et al. 2012).

Also, the authors had undeclared conflicts from editorials and IADR (JDR’s co-owners) position statements supporting fluoridation (Spencer 2013; Do et al. 2022).

Given these serious limitations all biasing results toward “no effect,” Do’s study provides no assurance that artificially fluoridated water is safe.