Fluoride and children's IQ: evidence of causation lacking

Human, animal, and in vitro studies. Extensive literature search of multiple bibliographic databases, trial registries, major grey literature sources and bibliographies of identified studies.

The authors aimed to identify studies which could be used to determine the maximum safe level for fluoride in drinking water. To identify new studies published since a 2016 Australian review, the search period was 2016 to July 2021. Studies which evaluated the association between either naturally or artificially fluoridated water (any concentration) and any health outcomes were included. No restrictions on study design or publication status. Articles published in a 'non-Latin language' were excluded. Screening of abstracts and full texts was in duplicate. For IQ and dental fluorosis, a top-up search was conducted between 2021 and Feb 2023.

Extensive data extraction. Risk of bias assessment using the OHAT tool. A narrative synthesis of the results was carried out.

The review included 89 studies in humans, 199 in animals and 10 reviews of in vitro studies. Where there was consistent evidence of a positive association, in relation to a water fluoride concentration of <20 ppm (mg F/L), and where studies were judged to be acceptable or high quality, health effects were taken forwards for further examination of causality using Bradford Hill's 9 criteria. Of the 39 health outcomes reviewed, 4 were further assessed for causality. The authors reported 'strong' evidence of causality for dental fluorosis and reductions in children's IQ scores, 'moderate' strength evidence for thyroid dysfunction, 'weak' for kidney dysfunction, and 'limited' evidence for sex hormone disruption.

The authors conclude that moderate dental fluorosis and reductions in children's IQ scores are the most appropriate health outcomes to use when setting an upper safe level of fluoride in drinking water. For reductions in children's IQ, the authors acknowledge a biological mechanism of action has not been elucidated, and the dose response curve is not clear at lower concentrations, limiting the ability to set an upper safe threshold.