Fluoride Action Network


Background: Previous systematic reviews of epidemiology studies have found support for a geographical association between high levels of naturally occurring fluoride in water (>1.5 ppm) and lower IQ in children. Most of the evidence from humans is from fluoride-endemic regions having higher background levels of fluoride compared to the fluoride concentrations historically used in community water fluoridation programs (0.7–1.2 ppm). Confidence in this body of evidence is limited, primarily due to poor reporting quality, lack of consideration of confounding (e.g., nutritional status, socioeconomic status, iodine deficiency), and concern for co-exposures to relatively high levels of other known neurotoxicants such as lead or arsenic. A systematic review of experimental animal studies could help in interpreting the human evidence.

Objective: To investigate whether fluoride exposure has detrimental impacts on neurobehavior in laboratory animal studies, prioritizing assessment of learning and memory outcomes. Confidence in the body of evidence was assessed according to one of four statements: (1) High, (2) Moderate, (3) Low, or (4) Very Low/No Evidence Available.

Methods: We included experimental animal studies that used mammalian species (whole organism) exposed during development or adulthood, which compared the effects of oral exposure to various fluoride concentrations to vehicle controls on neurobehavioral responses. The principal outcomes were learning and memory, but other neurobehavioral studies were included (e.g., anxiety, motor activity, aggression, sexual behavior). Studies assessing brain-related cellular, morphometric or histological endpoints were considered beyond the scope of this analysis. A literature search was performed up to January 14, 2016, using PubMed, BIOSIS, EMBASE, Scopus, Web of Science, PsycINFO, and several specialized databases. There were no date or language restrictions, and unpublished data and abstracts were excluded. Risk of bias was assessed regarding randomization, allocation concealment, blinding, exposure characterization, health outcome assessment, incomplete outcome data, selective outcome reporting, and other biases.

Results: The database searches yielded 4,643 unique records and 13 records were identified from other sources. Of the 4,656 studies, we identified 68 studies using mice or rats and testing drinking water or dietary concentrations of 0.45 to 272 ppm fluoride (0.12 to 40 mg/kg-d). Most included studies were published after 2000. Forty-eight studies addressed learning and memory, 16 of which assessed exposure during development.

Synthesis of results: Meta-analysis was not conducted due to the small number of studies that measured endpoints similarly based on study design, that is, dose levels, duration of treatment, lifestage at exposure, species, or differences in measurement of behavioral responses. Relatively few studies provided information on other sources of fluoride (e.g., diet, water source). Most studies were statistically underpowered to detect a <20% change from control groups for behavioral tests. Approximately 30% of the learning and memory studies were considered to have a very serious risk of bias and were excluded from the narrative analysis. Conclusions were reached based on an analysis of 32 studies. Results show low-to-moderate confidence for a pattern of findings suggestive of an effect on learning and memory based on developmental and adult exposure studies. The evidence is strongest (moderate level-of-evidence) in animals exposed as adults and weaker (low level-of-evidence) in animals exposed during development. Level-of-evidence conclusions were rated down due to concern for indirectness and risk of bias. The evidence was strongest and most abundant for adult exposure studies using the Morris water maze. In many cases, across the entire dose range tested, whether the effects were specifically related to learning and memory—versus a possible impact on motor or sensory function that could have impaired the ability of the animal to perform the learning and memory tests as measured—was not possible to discern. This was considered a form of indirectness. Additional studies are required to have higher confidence in the specificity of the responses as learning or memory impairments and in quantitative measures such as the effect sizes, point of departure, identification of no observed effect level or lowest observed effect level doses, or parameters for benchmark dose analysis. Based on control values (means and standard deviations/standard errors) and the number of animals per group, the studies appear statistically underpowered to detect a <10% or <20% change from controls for most behavioral endpoints.

Conclusion: Very few studies assessed learning and memory effects in experimental animals (rats and mice) at exposure levels near 0.7 parts per million, the recommended level for community water fluoridation in the United States. At concentrations higher than 0.7 parts per million, this systematic review found a low to moderate level-of-evidence that suggests adverse effects on learning and memory in animal exposed to fluoride. The evidence is strongest (moderate level-of-evidence) in animals exposed as adults and weaker (low level-of-evidence) in animals exposed during development. Confidence in these findings was reduced primarily based on potential confounding of the learning and memory assessments by deficits in motor function or fear and risk of bias limitations. Additional research is needed, in particular to address potential effects on learning and memory following exposure during development to fluoride at levels nearer to 0.7 parts per million. NTP is conducting laboratory studies in rodents to fill data gaps identified by this systematic review of the animal studies. The findings from those studies will be included in a future systematic review to evaluate potential neurobehavioral effects from exposure to fluoride during development with consideration of human, experimental animal and mechanistic data.