Fluoride Action Network

Fluoride Action Network applauds NTP’s systematic review on Fluoride’s neurotoxicity but corrects its biased risk assessment

Fluoride Action Network | Bulletin | October 28, 2020

NTP National Toxicology Program Fluoride Neurotoxicity Review


The Fluoride Action Network (FAN) is pleased that the updated NTP Review has reconfirmed that fluoride is “presumed to be a cognitive neurodevelopmental hazard to humans.” We would also underline that:

The NTP rating of “presumed” is the highest confidence rating possible without doing a controlled human experiment, which would be unethical. Human experiments are not allowed for testing toxicity of a chemical. 

However, FAN is concerned with NTP’s highly flawed “risk assessment” which claims that:

The evidence that fluoride lowers IQ in children is less robust and less consistent at levels below 1.5 ppm.

This claim is not supported by the science that their review used. The highest quality studies, including several funded by the National Institute for Environmental Health Sciences, have shown that IQ (or cognitive development) is lowered at exposure levels experienced in communities that have fluoridation programs which add 0.7 ppm fluoride to the water (Bashash 1 2017; Green 2 2019 and Till 3 2020).


The National Toxicology Program (NTP) performed a Systematic Review of Fluoride Exposure and Neurodevelopmental and Cognitive Health Effects (the “Review). This Review was initiated in response to a nomination from FAN. The NTP began the Review in 2016 and produced its first draft in 2019. The first draft Review was sent to the National Academies of Sciences, Engineering and Medicine (NASEM) for peer review. The NASEM Committee (see members below) that performed the peer-review recommended that NTP make several changes. The NTP incorporated the changes requested and released a second draft Review in September 2020 which was sent to NASEM for a second peer-review.

The NASEM Committee held a zoom meeting on October 19 that was open to the public and gave FAN’s Science Director, Chris Neurath, the opportunity to communicate our agreement with their overall finding but challenge their flawed risk assessment.

Making The Best Studies Disappear

At this meeting FAN was shocked when Kyla Taylor, of the NTP, showed a slide of a table of results from a NTP meta-analysis of IQ studies that indicated that the Bashash 2017 study did not find a lowering of IQ!  See figure A5-6 on page 268 of the Review:

This figure above shows a Forrest plot for the range of IQ loss for a meta-analysis of all the high-quality studies. The bottom axis shows the IQ points lost. The NTP authors show the IQ values in the Bashash, 2017 study range from a loss of 0.18 to an increase of 0.42 IQ points with a mean IQ above the No effect level (0). Such a selected data set is miss-leading and a miss-representation of the full Bashash study. Even so the overall SMD for all the studies combined is -0.50 which is equivalent to a loss of 7-8 IQ points. So while this meta-analysis does not affect the overall – and vitally important – conclusion that fluoride is a presumed developmental neurotoxicant, this manipulation of the Bashash study helped the NTP to incorrectly question the robustness of the studies performed at less than 1.5 ppm and their relevance to the US fluoridation program.

How Could This Have Happened?

For those who have followed the studies on fluoride’s neurotoxicity, Bashash 2017 had been the big breakthrough study showing a strong relationship between pregnant women’s exposure to fluoride and 1) a lowered cognitive function (CGI) in their offspring at 4 years of age and 2) a lowered IQ measured between 6 and 12 years of age. It was the breakthrough moment when they confirmed that the critical period of vulnerability to fluoride’s neurotoxicity was during fetal development.  So how could the NTP now suggest that the Bashash study does not show a lowering of IQ?

Neurath has delved into how the NTP reached this “extraordinary” conclusion about the Bashash 2017 findings in particular and the studies at lower fluoride exposures generally?  The good news from a science perspective is that the NTP is wrong.  However, FAN is deeply concerned about why such disturbing manipulations occurred. It leaves us questioning whether the CDC or other pro-fluoridation forces have corrupted the NTP’s analysis.

While this downplaying of the Bashash 2017 study did not alter the overall conclusion that fluoride is a presumed developmental neurotoxin, it is a key factor in the NTP’s tacked-on claim that at exposure levels relevant to the US and artificial fluoridation, the evidence is “inconsistent” and “unclear.”

This claim arises in a section the NTP added that they call “Generalizability to the US”. It is essentially a risk assessment.  This section was never within the planned scope nor was it in the protocol, so it violates a fundamental principle of systematic review. The NAS peer review panel itself said in 2019 it was not appropriate.  But worse than doing an unplanned risk assessment, the NTP used improper methods that underestimate the confidence that fluoride causes developmental neurotoxicity at levels relevant to the US and to artificial fluoridation – which is what the strongest studies have found.

FAN’s Submission to the NTP/NAS

Chris Neurath, on behalf of FAN, submitted formal written comments to NASEM and NTP, as well as the short oral comments during the public meeting:

  1. Transcript of Chris’ 5-minute oral presentation at the Oct 19 meeting (see above).
  2. Outline summary of findings.
  3. Full written submission to the NTP and NAS.

Chris identified and explained how the NTP downgraded evidence in the 2020 review. The following are the downgrades (fully discussed in Chris’ submission – see 3. above)

NTP Downgraded the Following Evidence in their 2020 Review.

The NTP:

• Excluded the largest effect in the strongest study: Bashash 2017.
• Excluded strongest low-dose studies from dose-response meta-analysis.
• Unnecessary division of studies lowers power in dose-response meta-analyses.
• Simplistic exposure assessment underestimates hazard at doses relevant to US.
• NTP’s simplistic dose-response meta-analysis methods underestimated effects at low doses because they used the mean exposure while most studies had individual-level exposures that ranged well below the mean.
• Improperly downgraded the animal evidence to “inadequate” despite the NTP 2016 review of the animal evidence concluding it was “low to moderate”.
• The NTP monograph deviated from the OHAT guidelines in its section “Generalizability to the U.S. Population”.
• Meta-analyses have inadequate documentation.

Downgrading the Bashash Study

On the specific case of the “downgrading” of the Bashash study discussed above, Neurath gives several examples, here are two:

  1. The NTP discriminated against the findings in the Bashash study by ignoring neurocognitive development findings for 4-year-olds in favor of the IQ studies for 6-12-year-olds. They used the latter finding to conclude a threshold for lowering of IQ at 0.8 ppm in maternal urinary levels. Whereas, the excluded analysis for cognitive index (GCI) in 4-year-olds found a linear dose-response relationship with no threshold down to the lowest measured level of 0.2 ppm. This bizarre exclusion was based on the NTP’s characterization of GCI (General Cognitive Index) scores as an “other outcome” rather than a measure of neurocognitive development. These GCI measures are used for young children where IQ measures are less appropriate. GCI is generally considered as a valid measure of neurocognitive development and the NTP was wrong to exclude this key analysis.
  2. Also for the Bashash study, the NTP focused on a secondary analysis which used childhood urine fluoride levels dichotomized into just two levels (below 0.8 ppm versus 0.8 ppm and above) as the measure of fluoride exposure, instead of the primary analyses, which used maternal urine fluoride levels. The former only found a small non-significant effect, the latter a large and significant effect (11 IQ points lost per 1 ppm maternal urine fluoride).

Bashash 2017 was a landmark study in part because

• it was the first to focus directly on fetal fluoride exposure
• it was a longitudinal cohort study (of the subtype prospective cohort)
• it had individual-level exposure data
• it controlled for a wide range of other factors
• it was very rigorously conducted

Many of the previous ecological studies in China had been restricted to children who were mostly life-long residents of a single location with a constant water F concentration.  For those studies, their prenatal exposure would have been similar to their exposure measured when they were older children.  So, the prenatal period was captured in those studies and probably explains why the IQ loss was so great in many of them, with an average of -7 IQ points

Paul Connett’s Submission

A separate submission by FAN executive director, Paul Connett, PhD, emphasizes that NTP’s conclusion of harm to the developing brain at levels at or above 1.5 ppm offers no adequate margin of safety to protect all children in the USA. It is standard practice in Risk Assessment to divide a dose that finds harm in a human study by 10 to determine a safe dose sufficient to protect the whole population. This is especially needed In the case of fluoride, not only to allow for the anticipated range of genetic susceptibility in a large population of children but also to allow for wide range of doses they are likely to experience. If we ignore the difference between dose and concentration, 1.5 ppm divided by 10 would indicate that in order to protect the whole population the level of fluoride allowed in water should not be higher than 0.15 ppm.

A simple way of demonstrating why a safety factor is necessary is to look at the EPA’s exposure assessment handbook where it shows that the 5% of Americans who drink the most water drink twice as much as the average.  Therefore, the NTP’s conclusion that water concentrations of 1.5 ppm are presumed to cause neurotoxic harm to those who consume AVERAGE amounts of water, would also apply to anyone drinking twice as much water with a concentration of around 0.7 ppm.  Their internal dose would be roughly the same as those drinking water with 1.5 ppm.

From even such a simple but valid Risk Assessment, it is clear that adding fluoride to water at 0.7 ppm (the current recommended level) is likely to be harming some children. Paul Connett’s commentary on Margin of Safety can be viewed on video below:

FAN was pleased to see that Dr. Linda Birnbaum, former head of both the NIEHS and the NTP, submitted her Oct 19 Op-ed to the NASEM panel for their consideration. In this Op-ed she advised that there was enough solid evidence in the available neurocognitive studies for regulatory bodies to recommend to pregnant women that they avoid fluoridated water and parents not use fluoridated tap water to make up baby formula.

We have provided links to all the documents pertaining to the NTP’s systematic review and our criticisms. See the Overview of the NTP study that contains the documents, news articles, and videos.

See all FAN bulletins online