A Commentary on

Taher M K, Momoli F, Go J et al.

Systematic review of epidemiological and toxicological evidence on health effects of fluoride in drinking water. Crit Rev Toxicol 2024 54 2–34.

GRADE Rating:


The safety of water fluoridation has been frequently questioned and concerns have been raised about many different health effects. Recently, studies investigating neurodevelopment and IQ have generated increasing attention. In the USA, there is an ongoing court case ‘Food & Water Watch, Inc. et al. v. Environmental Protection Agency (EPA) et al.’ in which the plaintiffs, a coalition of anti-fluoridation groups, claim that fluoride poses an ‘unreasonable risk’ to health on the grounds that it is neurotoxic, and should be regulated as such under the Toxic Substances Control Act of 1976 (TSCA)1.

A key piece of evidence in the trial is a US National Toxicology Programme (NTP) systematic review of fluoride exposure and neurodevelopmental and cognitive health effects. The NTP systematic review was started in 2016 and the first draft (2019) concluded that “fluoride is presumed to be a cognitive neurodevelopmental hazard to humans”. The US National Academies of Science, Engineering and Medicine (NASEM) have twice peer reviewed the draft NTP review and requested revisions on the basis that the conclusions were not adequately supported by the evidence2.

An important issue is that many of the studies demonstrating adverse effects of fluoride on IQ are related to very high naturally occurring water fluoride concentrations, far higher than the recommended (and legal, in the UK and Europe) maximum of 1.5 mg F/L, and the 1 mg F/L target for water fluoridation programmes. Additionally, many studies have been conducted in developing countries in populations who may also be exposed to environmental pollutants such as lead, mercury or arsenic through contaminated drinking water or coal smoke pollution2,3. At the time of writing, the final NTP review has not been published, however, the most recently available third draft (Sept 2022) is more cautious, stating with “moderate confidence” that “higher fluoride exposure… [>1.5 mg F/L] is consistently associated with lower IQ in children”, and acknowledges that “more studies are needed to fully understand the potential for lower fluoride exposure [<1.5 mg F/L] to affect children’s IQ”4.

This brings us back to the present systematic review, by Taher et al.5. In terms of the methods, the scope of the review is extensive. The main manuscript is 29 pages, and there are over 1000 pages of supplementary materials. It is therefore surprising that there is no indication that the review protocol was registered in advance, important to maintain rigour and reduce bias. Relatedly, there is no justification given for the ‘top-up search’ between 2021 and Feb 2023, for only IQ-effects studies. It is not clear that data extraction was performed in duplicate, or that the data extraction forms were piloted. Indeed, a large volume of information has been extracted but not all of it is used in the analysis or manuscript, and there are some reporting inconsistencies.

Our main area of concern reflects the risk of bias assessment and how it has been applied. The authors used the OHAT Tool for Human and Animal Studies. A requirement of the tool is that important confounding factors should be agreed in advance with subject matter experts, for each health outcome. There is no indication that this was done, therefore we have limited confidence in the authors’ assessment as we are not sure what confounders they would expect to be considered as a minimum. In addition, the study summary table in the main manuscript presents study type, country, and direction of association for each health outcome, but does not include any indication of the strength of the association or the water fluoride concentration, which makes it difficult to evaluate the relevance of the findings to water fluoridation programmes.

The human evidence included in the Taher review overlaps with the draft NTP review4, and other systematic reviews on the neurological and health effects of fluoride in water, including a recent Irish government review (Lambe et al.6), and a Canadian government review (CADTH)7. There appear to be differing perspectives on the quality of this evidence from reviewers with a toxicology perspective (Taher et al.5 and the draft NTP review), versus those with an evidence-based medicine perspective6,7. The latter have stated that the current evidence base related to IQ is insufficient to draw conclusions, and that further high quality research is needed (Lambe et al. and CADTH)6,7.

To illustrate, a Canadian prospective cohort study by Till. et al.8, assessed as ‘high quality / low risk of bias’ by both Taher et al. and the draft NTP review4, was assessed as ‘low quality / high risk of bias’ by the Lambe et al. and CADTH reviews6,7. Bias concerns stated were: selection bias (low participation rate, high loss to follow-up); validity of the fluoride and IQ measurements; and insufficient adjustment for confounding, including by maternal IQ and marital status6,7. Similar differences in risk of bias assessments are evident for other papers included in the multiple reviews. The paper by Till et al. (2020) uses data from the Maternal-Infant Research on Environmental Chemicals (MIREC) birth cohort database in Canada, as do several other recent fluoride-IQ papers (including Farmus et al.9, also assessed as high quality by Taher et al., and low quality by Lambe et al.6). The MIREC cohort was not designed to evaluate fluoride and the studies using these data have been extensively critiqued2,10, with recent authors stating that these studies should be “considered unacceptable for legal and policy purposes”11.

In conclusion, there is ongoing and high-profile debate regarding the impact of fluoride in drinking water on IQ, with very different perspectives on bias according to discipline. High-quality prospective longitudinal studies based on individual-level exposures, in populations exposed to fluoride concentrations of relevance to fluoridation programmes, and taking account of all important confounding factors are necessary to provide higher-quality information.