SCIENTIFIC OPINION: Updated consumer risk assessment of fluoride in food and drinking water including the contribution from other sources of oral exposure.

Abstract

Updated Risk Assessment at
https://efsa.onlinelibrary.wiley.com/doi/full/10.2903/j.efsa.2025.9478

This updated risk assessment evaluated evidence on potential adverse health effects of fluoride related to all sources of oral exposure as mandated by the European Commission. Fluoride benefit assessment was not included. Effects on the central nervous system, thyroid and bone were prioritised. Evidence from human studies indicates that total fluoride intake is associated with adverse effects on the developing brain at drinking water concentrations >1.5 mg/L. The evidence of such associations below 1.5 mg/L was not sufficiently consistent to draw conclusions for risk assessment. Using drinking water concentration of 1.5 mg/L as a reference point, a safe level of intake including all sources of oral exposure of 3.3 mg/day was established for pregnant women to protect the fetus. This safe level of intake was extended to apply to other adults and children >8 years. It is considered protective also against possible adverse effects on thyroid function and bone mineralisation, for which associations have been observed at water concentrations >1.5 mg/L. Dental fluorosis was considered the most sensitive endpoint for children <8years. Tolerable upper intake levels (UL) of 1.0, 1.6 and 2.0 mg/day were established for infants, toddlers and children 4–8years, respectively. These ULs are considered protective against other possible adverse effects of fluoride, including neurodevelopmental outcomes. Aggregate exposure included intake of fluoride from food, drinking water, discretionary salt and (ingested) dental care products. Aggregate exposure based on the mean concentration of fluoride in EU drinking water (submitted data) was below the above health-based guidance values (HBGVs) for all age groups. Aggregate exposure exceeds the HBGVs at the 95th percentile of intake in the scenario of the P95 concentration of fluoride in EU drinking water, for all age groups except adolescents. The risk assessment suggests that the current legal limit for drinking water (1.5 mg/L) in the EU is not sufficiently protective.

SUMMARY

In the context of recent studies suggesting a relationship between intake of fluoride from drinking water with less than 1.5 mg fluoride/L and neurotoxicity in children, EFSA was mandated by the European Commission to carry out a consumer risk assessment for fluoride in food and drinking water, taking into account all relevant hazard information and all sources of oral exposure. The Scientific Committee considers that dental care products and fortified foods are major relevant sources and they were included in the assessment.

Chemistry

The scope of this update includes elemental or ionic fluoride as well as compounds that can release either of these fluoride forms. It does not cover fluorinated compounds where fluoride is covalently bound or those with a toxicological profile not attributed to fluoride. To ensure accurate fluoride quantitation, correct sample preparation and measurement interpretation are critical to avoid matrix effects or interferences and ensure correct recovery.

Previous assessments by EFSA and other agencies

Fluoride benefits and risks have been previously assessed by EFSA and other agencies. Although the beneficial properties of fluoride are out of scope of this mandate, the Scientific Committee noted that benefits for prevention of dental caries have previously been acknowledged by establishment of an adequate intake by the EFSA Panel on Nutrition, Novel Foods and Food Allergens (NDA Panel) in 2013. In 2005, the NDA Panel also established tolerable upper intake levels (ULs)¹ of 1.5 and 2.5 mg/day for children of 1–3 and 4–8 years, respectively, based on less than 5% prevalence of moderate to severe dental fluorosis. For children and adults aged 9–14 and > 15 years, ULs of 5 and 7 mg/day were established, respectively, based on increased risk for bone fractures observed in randomised controlled trials among postmenopausal women. Among more recent assessments, the U.S. National Toxicology Program concluded, in their assessment finalised in 2024, that fluoride exposure corresponding to water concentrations of > 1.5 mg/L were associated with adverse neurodevelopmental outcomes in children. Health Canada concluded in their 2023 assessment that ‘there is not a sufficient basis at this time to recommend a specific point of departure and health-based value for neurocognitive effects’. In their assessment Health Canada identified water concentration of 1.56 mg F/L ‘as the point of departure for deriving the health-based value’, based on dental fluorosis. However, Health Canada has not yet established a ‘health-based value’.

Prioritisation of health effects for systematic literature review

Among all endpoints reported in the literature to be associated with fluoride exposure, those related to the central nervous system (CNS), thyroid and bone were prioritised by the Scientific Committee for a systematic literature review and comprehensive risk assessment. Studies in humans and experimental animals were assessed as separate lines of evidence through the weight of evidence approach. Other lines of evidence included information on fluoride kinetics and relevant in vitro evidence. New evidence on well-established adverse effects of fluoride, such as skeletal and dental fluorosis, was not subject to systematic literature review for several reasons. First, the direct link (or causality) between fluoride and these outcomes is well established. Second, in line with the EFSA NDA Panel assessment of 2005, it was concluded that skeletal fluorosis likely occurs at higher exposures compared to exposures resulting in bone fractures for which the UL for adults was established in 2005. Third, the UL for children was already based on dental fluorosis and screening of the new literature did not suggest that new evidence would challenge studies published prior to 2003.

Association of fluoride intake with central nervous system (CNS) development

The available prospective studies examining associations with CNS development were judged to be relevant and reliable. These studies primarily examined associations between exposure to fluoride during pregnancy in relation to later assessment of IQ in the offspring. Those studies were conducted in populations exposed to relatively low concentration of fluoride in drinking water (< 1.5 mg/L). Although associations with lower IQ scores were observed in some studies, the overall evidence from these studies was judged to be inconclusive. To provide more clarity around this conclusion, further studies addressing neurodevelopmental outcomes in children exposed pre-and/ or postnatally to low levels of fluoride (< 1.5 mg/L in drinking water) are recommended.

The available cross-sectional studies were judged to be of moderate relevance and reliability, given that exposure to fluoride from drinking water in these studies is expected to have been ongoing, thereby preceding the outcome. The cross-sectional studies were most often conducted in populations exposed to relatively high fluoride concentrations in drinking water (> 1.5 mg/L). A large majority of those studies reported associations between fluoride concentrations in drinking water and lower IQ in children. The relative consistency of these findings observed across different populations increases the confidence in their findings. It was therefore concluded that exposure to fluoride in drinking water higher than 1.5 mg/L during fetal development and/or postnatally is associated with lower IQ scores in children. The cross-sectional studies did not provide further clarity on possible associations with lower IQ at concentrations of fluoride in drinking water < 1.5 mg/L.

In the human studies, the mean water fluoride concentrations correlated with mean urinary and/or serum fluoride concentrations across studies. Although concentrations of fluoride in urine and serum can be considered an indirect marker for total fluoride intake, the total intake of fluoride received by the study participants is uncertain. For the purpose of this assessment, total intake was estimated using default factors and (conservative) assumptions (see below and Section 3.8.2 in the opinion).

Studies in experimental animals provide evidence that adverse effects of fluoride on neurobehavioural indicators begin to occur at dose levels above 2.1 mg/kg bw per day, with adverse effects at organ level (brain) reported at doses of 3.5 mg/kg bw per day and higher. However, the mechanistic data and the non-specific effects reported at molecular or cellular level do not provide reliable additional support for effects of fluoride on the brain or a mode of action relevant to effects in the CNS. As concentration of fluoride in control feed was rarely reported, the total fluoride intake of the experimental animals in the in vivo studies is uncertain and may be overestimated with the use of default feed concentrations. Fluoride accesses the brain during development and possibly in adulthood, but the fraction of the dose reaching the brain could not be estimated. Therefore, it was not possible to estimate the target tissue concentrations that are associated with the effects reported in vivo.

Overall, for neurodevelopment and neurotoxicity, the Scientific Committee concluded that there is reasonable confidence in the evidence from both human and animal studies suggesting that an association with neurodevelopmental outcomes may occur at relatively high fluoride exposures, i.e. above 1.5 mg/L in drinking water. The conclusion is primarily based on the evidence from human studies with supporting evidence from animal studies. The evidence for associations at lower concentrations is inconclusive.

Association of fluoride intake with thyroid function

Human cross-sectional studies suggest that living in areas with relatively high fluoride concentrations in drinking water (> 1.5 mg/L) is associated with increased serum TSH. The biological relevance of the TSH increases reported in these studies is subject to some uncertainty due to the modest effect sizes and inconsistent associations with concentrations of thyroid hormones (T3 and T4). Results from few, mostly Tier 3 (high risk of bias), animal studies and the available in vitro studies provide insufficient evidence to support adverse effects of fluoride on thyroid function or a mode of action on the thyroid.

Association of fluoride intake with bone health

New evidence on the relationship between fluoride exposure and bone health from human studies suggests that changes in bone mineral density and increased risk of fractures may occur below the UL of 7 mg/day (at ~ 3 mg/day) established by the EFSA NDA Panel (2005). However, the new evidence was not sufficiently robust to identify a more precise estimate of oral intake at which changes in bone mineral density become adverse, or to provide a more precise estimate of intake at which fracture risk might start to increase. The new evidence does not suggest that adverse associations between fluoride and measures of bone health occur at fluoride concentrations < 1.5 mg/L in drinking water. Relevant and reliable evidence from animal studies consistently shows effects of fluoride on bone starting from about 6 mg/kg bw per day. Consistent with the previous EFSA conclusion, an association of fluoride exposure with bone cancer risk is not supported by new studies published since 2005.

Synthesis of the evidence and selection of critical endpoints

The weight of evidence in this Opinion indicates that adverse effects on neurodevelopment are observed at intakes below the UL of 7 mg/day established by the EFSA NDA Panel in 2005, and there are some indications that this also applies to thyroid and bone health. The fetus of exposed pregnant women is most sensitive to the potential adverse effects of fluoride on neurodevelopment due to the vulnerability of the developing human brain. Since development continues postnatally, fluoride effects on the CNS may also be relevant for the developing offspring from exposure occurring after birth. However, there was insufficient evidence to directly assess this relationship. Hence, effects on the developing CNS were selected as the critical endpoint applicable to pregnant women and the developing offspring. Based on the weight of evidence, adverse effects on thyroid function or bone health are unlikely to occur at fluoride intakes below those identified as critical for neurodevelopment.

For toddlers 1–3 years and children 4–8 years, the available evidence on dental fluorosis is still considered robust and relevant. This critical endpoint is also relevant for infants < 1 year, as mineralisation of permanent teeth begins after birth. Based on the weight of evidence, adverse effects on the CNS, thyroid function or bone health are unlikely to occur in children at exposures to fluoride below those associated with dental fluorosis. Therefore, the Scientific Committee considered dental fluorosis also to be a relevant and reliable critical endpoint infants, toddlers and children up to 8 years.

Derivation of reference point

The Scientific Committee concluded, based on a number of studies in children, that lower IQ scores in children are consistently reported at drinking water fluoride levels above 1.5 mg/L. There were not sufficient data for dose–response modelling of this endpoint. Hence, it is important to note that the concentration in drinking water where adverse effects begin to occur is subject to uncertainty. Adverse associations have been reported at lower concentrations in some prospective studies, but the evidence is not conclusive. As a result, it was judged appropriate to use the value of 1.5 mg/L drinking water as a reference point (RP) for neurodevelopmental effects, supported by the overall evidence on the prioritised endpoints from human and animal studies.

Benchmark dose modelling was possible for drinking water concentration data (benchmark concentration, BMC) and data on dental fluorosis from the Dean, 1942 study and was based on two criteria: the severity grade (mild, moderate, severe) and the prevalence reflected in the BMR. Moderate to severe dental fluorosis representing onset of adversity and a benchmark response (BMR) of 1% resulted in a lower bound of the BMC (BMCL) of 1.7 mg/L in drinking water. This was not considered sufficiently protective since more recent evidence indicates that moderate fluorosis may occur at lower concentrations of fluoride in drinking water (considering total fluoride intake has changed since the 1942 study). The Scientific Committee selected a more conservative approach based on the combined prevalence from mild to severe dental fluorosis, with a BMR of 5%, that resulted in a BMCL of 1.4 mg/L in drinking water, where the probability for cases of moderate severity in the population is minimal. A conservative approach is justified due to the uncertainty about the total intake in the population of the pivotal study, as well as the cumulative and non-reversible nature of the adverse effect. The Scientific Committee considers that the BMCL of 1.4 mg/L in drinking water is an appropriate RP from which to establish ULs that are protective against dental fluorosis in infants (< 1 year), toddlers (1–3 years) and children (4–8 years). This reference point is slightly lower, and therefore more protective, than the value of 1.5 mg/L that is used as reference point for possible effects on neurodevelopment.

Establishing health-based guidance values²

The Scientific Committee notes that the effects reported in the literature are not the result only of fluoride present in drinking water, but of the total fluoride intake of the participants in the human studies and the experimental animals. Therefore, the health-based guidance value (HBGV) must represent the total intake to which drinking water is but one contributor. The estimated total intake of study participants included contributions from food and dental care products. The RPs for both dental fluorosis and for effects in the CNS are obtained from human studies. Hence, no interspecies uncertainty factor is needed. The Scientific Committee concluded that an uncertainty factor to account for interindividual variability is not needed because the evidence for both endpoints is obtained from diverse, relevant and sensitive populations.

Because the RP for effects on the CNS is derived based on evaluation of the overall evidence and is not derived numerically though dose–response characterisation, a UL could not be established. Instead, the Scientific Committee established a ‘safe level of intake’³ of 3.3 mg/day for pregnant women based on CNS effects in the offspring following exposure to fluoride from maternal intake. Considering that other evidence indicates that associations with changes in thyroid function and bone mineralisation in adults also occur at drinking water fluoride concentrations > 1.5 mg/L, the safe level of intake established for pregnant women is considered protective against other potential adverse effects and is applicable to other adults and children > 8 years.

The Scientific Committee established ULs of 1.0, 1.6 and 2.0 mg/day for infants < 1 year, toddlers 1–3 years and children 4–8 years, respectively, based on dental fluorosis. Although effects on the CNS are also in principle applicable to infants, toddlers and children, the Scientific Committee concluded that the reference point for dental fluorosis is more robust (obtained from dose–response data) and an UL based on dental fluorosis should also be protective against CNS and other potential adverse effects of fluoride.

An uncertainty analysis of the HBGVs was performed based on expert knowledge elicitation. Assessment of overall uncertainties was based on the contribution of all lines of evidence. The results of the uncertainty analysis are presented in detail in Section 5.1 of the opinion. The established HBGVs represent the best outcome of the assessment (protective values of highest confidence) based on the weight of the available evidence. The credible intervals of the HBGVs represent the uncertainty around the HBGV contributed by different lines of evidence of lower confidence and the information (or lack information) they provided.

Dietary and non-dietary exposure assessment

Dietary and non-dietary exposure assessment was conducted using EFSA databases and methods for age groups older than 12 weeks. Data relevant to exposure through dental care products and fluoridated salt were obtained from the literature and previous EFSA assessments. Aggregated exposure was assessed by age group based on fluoride from major sources, including food, drinking water, fluoridated discretionary salt (not accounted for in food) and dental care products (assuming 100% of product is ingested by children < 8 years and 10% is ingested by adults and children > 8 years). Intake from drinking water was based on data submitted to EFSA of current drinking water fluoride concentrations in European countries representing naturally occurring fluoride in drinking water or water fluoridation. Based on drinking water concentration data submitted to EFSA, > 86% of samples contain < 0.3 mg/L fluoride and > 97% of samples contain <?0.7 mg/L fluoride.  Intakes based on these data are presented as the ‘basic scenario’, when using the mean concentration of fluoride in water estimated from the submitted data and ‘water P95 scenario’, when using the P95 concentration of fluoride in water estimated from the submitted data. Intake from drinking water was also assessed according to scenarios assuming that all drinking water consumed contained fluoride up to the current legal limits of 1.5 mg/L (tap or bottled water, ‘legal limit 1’ scenario) or 5 mg/L (bottled water, ‘legal limit 2’ scenario) in the European Union. The drinking water scenarios referring to legal limits of water fluoridation indicate the highest potential intakes and overestimate actual intakes.

The contribution of the mean dietary exposure (food and drinking water) to the mean total aggregated oral exposure to fluoride ranged from 19% in children to 66% in adults. Main contributing food categories (> 10% to the total exposure in the basic scenario) included ‘Grains and grain-based products’, ‘Milk and dairy products’, ‘Tea beverages’ and ‘Drinking water’ for all age groups. ‘Food products for young population’ was also a main contributor for infants. Under the ‘basic scenario’ of water fluoride, contributions to the mean total aggregated oral exposure of fluoride from all sources ranged as follows: mean exposure from discretionary salt (not accounted for within the food categories) from 15% in ‘other children’ (4–8 years) to 33% in adults (this source is not applicable to infants), and mean exposure from ingested dental care products from 15% in adults to 75% in infants and toddlers. In the ‘water P95 scenario’ and two water fluoridation scenarios up to current legal limits (‘legal limit 1’ and ‘legal limit 2’ scenarios), the contribution of dietary sources (food and drinking water) increased, reaching up to 74%, 84% and 89%, respectively, and the contribution of the other sources decreased respectively.

An uncertainty analysis of exposure by age group was performed based on expert knowledge elicitation. Uncertainties in exposure assessment were related to analytical methods used for food composition data, amount of fluoridated salt consumed, amount of toothpaste used and fraction of ingested product. The impact of each source of uncertainty was estimated separately for each age group, resulting in 90% certainty ranges for the respective aggregated exposures and for each source of exposure (drinking water, food (excluding drinking water), fluoridated salt and dental care products). The results of the uncertainty analysis are presented in detail in Section 5.2 of the opinion. Considering that sources of fluoride exposure vary among EU Members States and additional sources of fluoride intake may be applicable in some Member States, the related sources of uncertainty and their respective ranges can be used to support risk management decisions under different exposure scenarios.

Risk characterisation

The estimated mean and P95 intakes based on the ‘basic scenario’ are below the HBGVs for all age groups, except for the top of the P95 intake exceeding the ULs of toddlers (slightly) and children 4–8 years. Exceedances of the HBGVs for infants (slightly), toddlers, children and adults are noted for aggregate oral exposures at the P95 intake level of the ‘water P95 scenario’. Intakes at the mean and P95 ranges of ‘legal limit 1’ scenario exceed the HBGVs for all populations groups. Intakes at the mean and P95 of ‘legal limit 2’ scenario exceed the HBGV for adults and children > 8 years (legal limit 2 scenario is not applicable to younger age groups).

An uncertainty analysis of the risk characterisation was performed based on expert knowledge elicitation. The results of the uncertainty analysis are presented in detail in Section 5.3 of the opinion. The analysis provides the likelihood of no concern or of likely exceedance of the established HBGVs to support risk management decisions under different exposure scenarios and level of certainty.

Conclusions

The Scientific Committee concluded that the mean and P95 intakes of fluoride based on typical drinking water concentrations in Europe (basic water scenario) do not exceed the established HBGVs, except for the top of the P95 intake that exceeds the UL for toddlers (slightly) and children 4–8 years. The ULs are exceeded for infants (slightly), toddlers and children 4–8 years at the highest drinking water concentrations in Europe (P95 water scenario). The exceedance of UL in these age groups suggests that mild forms of dental fluorosis may occur. For children 4–8 years, dental fluorosis may occur in the molar teeth which develop during this period. A conservative assumption of 100% ingestion of fluoridated dental care products is included.

For adults the safe level of intake is exceeded at levels of exposure associated with high contributions from all the following sources combined in descending order: drinking water, food, fluoridated discretionary salt and dental care products. Such exceedances in the subgroup of pregnant women would indicate increased risk of adverse effects on fetal neurodevelopment.

Because the RP derived for CNS effects (1.5 mg fluoride/L) and the RP for dental fluorosis (1.4 mg/L) are respectively equal or nearly equal to fluoride concentration corresponding to the current legal limit for drinking water, aggregate intake under the ‘legal limit 1’ scenario will lead to exceedances of the HBGVs. As a result, the current legal limit for fluoride in drinking water is not considered sufficiently protective.

The Scientific Committee concluded that the risk for adverse effects in the CNS from fluoride exposure is related to ingested fluoride. It does not apply to exposure through topical applications of fluoridated dental care products (recommended use), as long as the products are not ingested.

Notes

¹ The UL is the maximum level of total chronic daily intake of a nutrient (from all dietary sources) which is not expected to pose a risk of adverse health effects to humans (EFSA NDA Panel 2024). The UL is also defined as a level of intake above which the risk of adverse effects begins to increase.

² The term HBGV is used in this opinion when referring to both safe level of intake and UL established for fluoride.

³ A safe level of intake is the only type of HBGV proposed when the RP is not derived from dose–response data and, therefore, a UL cannot be established. It is generally defined as ‘the maximum amount that can be confidently concluded poses no risk of adverse effects in the population’ (EFSA NDA Panel 2024). In the case of fluoride, the evidence below the safe level of intake does not allow conclusion on safety, but it is nonetheless considered inconsistent and insufficient to result in a lower HBGV.

EFSA Scientific Committee, Susanne Hougaard Bennekou, Ana Allende, Angela Bearth, Josep Casacuberta, Laurence Castle, Tamara Coja, Amélie Crépet, Ron Hoogenboom, Helle Knutsen, Claude Lambré, Søren Saxmose Nielsen, Dominique Turck, Antonio Vicent Civera, Roberto Villa, Holger Zorn, Jacqueline Castenmiller, Karlien Cheyns, Keyvin Darney, Mary Gilbert, Jean-Charles Leblanc, Haakon Meyer, Evangelia Ntzani, Martin Paparella, Marco Vinceti, Heather Wallace, Maria Anastassiadou, Maria Bastaki, Irene Cattaneo, Luna Greco, Anna Lanzoni, Francesca Riolo, Olaf Mosbach-Schulz, Andrea Terron, Thorhallur Halldorsson