Association Between Maternal Fluoride Exposure and Child IQ—Reply

In Reply We appreciate the widespread interest in our study and the opportunity to respond to the critiques and clarify some facts.

1. Fundamentals of epidemiology

We agree that our study reported an association, not causation. Still, observational studies are optimal for testing many important questions in public health, and they are often used to infer causality. Indeed, if we dismissed observational studies, we would have no evidence that water fluoridation prevents tooth decay in children.

Several commentators raised questions about the effect of residual confounding. Our results, as with all observational studies, are subject to confounding by unmeasured factors. Still, we considered many potential confounders. Unless unmeasured factors are identified and shown to be associated with both fluoride exposure and IQ, these criticisms are speculative.

Our conclusions did not “simply rely on P values.” We established our hypotheses a priori, carefully interpreted effect estimates and confidence intervals, conducted a comprehensive set of sensitivity analyses, acknowledged the limitations of our study, and considered the results of other studies before concluding that pregnant women should reduce their fluoride intake.

Others were concerned about imprecise results. Keep in mind that (1) the fitted regression line is covariate-adjusted, whereas individual data points in the figure are not, and (2) individual data points close to the regression line may represent numerous overlapping cases. Also, the outliers in our data set have low influence but good leverage because their inclusion improves the estimate’s precision. Nonetheless, we ran secondary analyses without the 2 very low IQ scores and the fluoride effect remained.

2. This is not a subtle finding

Is a 4.5 IQ decrement per 1 mg/L increase in urinary fluoride weak? A decrement of 4.5 IQ points among many boys can have a substantial effect on the population distribution of IQ scores.¹ With 14% of Canadian women living in fluoridated regions having urinary fluoride of at least 1 mg/L, we would expect even larger IQ decrements for some children.

Our study is consistent with other high-quality research; it is not a standalone study. More than 50 studies have shown an association between higher fluoride exposure and lower IQ. All 3 prospective birth cohort studies^2,3 reported that higher prenatal fluoride exposure led to lower IQ scores in children.

3. Clarifying misconception

We did not set out to compare IQ scores in children who lived in fluoridated vs nonfluoridated communities. We used a biomarker of fluoride exposure that accounts for all sources. However, we did find consistent IQ decrements associated with fluoride intake and water fluoride concentrations.

Some commentators were troubled by the reported sex differences. Male individuals are often more susceptible to toxicants, including prenatal fluoride exposure in some animal studies.⁴ Moreover, the National Toxicology Program (2016) called for sex-based analyses for fluoride.⁵ Thus, we planned to test sex differences. We thank Waugh for his suggestion that male mitochondria may be more sensitive to fluoride.

Blendoe et al suggest that the IQ differences are explained by sex differences. Sex differences in IQ scores do not explain the exposure-response association that we observed in boys using urinary fluoride concentrations.

We collected 3 urine samples (1 for each trimester) and adjusted them for urinary dilution, an improvement over other population-based studies on fluoride neurotoxicity. Urinary fluoride is not a perfect measure of fluoride intake, but it is reliable, and it is the optimal biomarker of concurrent fluoride exposure.⁶

We cited a dental fluorosis prevalence rate of 68% in adolescents based on a prior study using National Health and Nutrition Examination Survey data. As noted by Gehani et al, the quality of these data was assessed in an April 2019 report. The prevalence of fluorosis, which is now estimated to be 41%,⁷ is still considerably higher than when water fluoridation was introduced.

We acknowledge that IQ tests have been criticized as having cultural biases. However, our analyses did not compare groups that differed by culture. Moreover, we controlled for ethnicity and city.

Finally, several commentators requested results that we did not report in our manuscript. We endeavor to make our analyses as transparent as possible and have posted supplemental material on the Open Science Framework (https://osf.io/7vfu5). Access to the data to validate our findings is available to independent researchers who adhere to the MIREC Biobank policies outlined here: http://www.mirec-canada.ca/en/research/.

We agree with Howard et al. Ten years ago, many of the same criticisms, such as bias, imprecision, unmeasured confounding and subtle effects, were raised about studies of low-level lead toxicity.

To conclude, some commentators argued that studies that showed fluoride diminishes children’s intellectual abilities should be interpreted cautiously. We agree. We should also take precautions to protect children’s brains from a potential toxicant, especially when there is no benefit of fluoride for preerupted teeth.