A risk assessor for the U.S. Environmental Protection Agency admitted fluoride is neurotoxic at relatively low levels and the agency’s key expert on fluoride’s neurotoxicity conceded flaws in his own study as the landmark fluoride trial drew to a close.
Editor’s note: The Defender is providing daily updates on the landmark trial pitting Fluoride Action Network against the U.S. Environmental Protection Agency. The trial started Feb. 1. To read previous coverage, click here.
A U.S. Environmental Protection Agency (EPA) risk assessor admitted fluoride is neurotoxic at relatively low levels and the agency’s key expert witness on fluoride’s neurotoxicity conceded flaws in his own study as the landmark fluoride trial drew to a close on Tuesday.
Stanley Barone, Ph.D., a risk assessment scientist, was the EPA’s second and final in-person witness in nine days of testimony before federal Judge Edward Chen in San Francisco.
The trial began in 2020 but was put on hold pending the publication of the National Toxicology Program’s (NTP) report informing the case. It resumed on Jan. 31.
Attorneys deposed Jesús Ibarluzea, Ph.D., lead author of the Spanish INMA cohort study, a key piece of EPA’s evidence, in November 2023. However, he withdrew from further participation in the trial, including not making himself available to testify via Zoom.
Instead, attorneys from both sides edited key parts of his deposition for the court and scheduled the video for public screening during the trial. EPA attorneys moved for the judge to watch it privately, in the interest of speeding along the trial process, they said.
The testimony was shared with The Defender and is reported below.
Food & Water Watch, Fluoride Action Network (FAN), Moms Against Fluoridation and other advocacy groups, mothers and children sued the EPA in 2017 after the agency denied a petition to end water fluoridation under the Toxic Substances Control Act (TSCA).
It is now up to Judge Chen to decide if the EPA should be required to create a rule banning water fluoridation in the U.S.
“Because the regulatory agencies have failed to do their job for decades,” plaintiffs’ attorney Michael Connett told The Defender, “the court is now in the position of having to do it for them.”
“It’s not a job the court takes lightly,” he said. “It’s not a job the court wanted to do, but I think it’s a job the court is prepared to do.”
Chen’s decision could affect more than 200 million Americans who drink fluoridated water. The dental lobby and public health agencies for decades have promoted the practice as one of the 10 great achievements of the 20th century, claiming fluoridation improves dental health.
Fluoride causes ‘neurotoxic harm’
Fluoride causes “neurotoxic harm,” and does so at relatively low levels, Barone admitted under cross-examination.
Although Barone is at the trial to testify for the EPA, plaintiffs called him to the stand last week to explain the EPA’s methods for risk assessment. In several hours of highly technical testimony, he detailed methods the EPA can use to determine hazard, risk, benchmarks and margins of uncertainty.
In his testimony yesterday for the EPA, Barone said there simply isn’t enough data available for EPA to implement its risk assessment process for fluoride. Necessary modeling, namely physiologically-based pharmacokinetic modeling that predicts how a chemical will be absorbed and metabolized by the body, hasn’t yet been done, he said.
But on cross-examination, Connett forced Barone to concede several of the plaintiffs’ key points.
“You do not dispute that fluoride is capable of causing neurodevelopment harm, correct?” Connett asked.
“I do not,” Barone said, adding that he said that in his deposition.
“You agree that the current evidence is suggestive that low-dose fluoride causes neurodevelopmental effects? Correct?” Connett asked.
Barone said the “hazard ID” — the level at which a toxin causes effects — “is probably in the suggestive range but is highly uncertain.”
“You agree that fluoride is associated with neurotoxic effects at water fluoride levels exceeding two parts per million?” Connett asked.
After first evading the question, Barone conceded.
Connett asked if Barone agreed there should be a “benchmark margin of uncertainty” of 10 for fluoride neurotoxicity. That means the lowest allowable human exposure level should be at least 10 times the hazard level, which Barone conceded may be approximately 2 parts per million.
Barone said that is generally true for toxic chemicals under TSCA.
Water fluoridation levels in the U.S. are currently 0.7 parts per million, also referred to as milligrams per liter (mg/L), which would place them well above the allowable level if they were regulated through TSCA’s norms.
Barone also conceded that the NTP’s report linking fluoride to neurotoxicity at 1.5 mg/L is a rigorous, high-quality review and that the NTP is one of the world leaders in doing such reviews.
“Do you feel comfortable as a risk assessor,” Connett asked, “exposing pregnant women to a level of fluoride that is so high that the kidney is oversaturated?”
Barone avoided answering, commenting instead on other foods containing fluoride.
Connett asked a second time, “Are you comfortable then with a pregnant woman having so much fluoride in her circulating system that their kidney has lost the ability to efficiently process it?”
EPA lawyers objected to the question as “vague and argumentative” but Chen overruled.
Barone sat in silence for several seconds before responding, “Again, putting this in context, my comfort level I don’t think is germane.”
Connett then turned to the question of the “data gap” or “uncertainty” that Barone and other EPA experts have argued is the basis for not requiring the agency to regulate fluoride.
Connett asked Barone if he agreed that uncertainty about the threshold level at which a chemical causes harm is not a basis for deciding not to do a risk assessment — the process that would likely lead to chemical regulation.
Barone agreed but said the weight of the evidence was key.
Connett also asked him if he personally agreed that the EPA should “use health protective assumptions” when data is lacking.
He said he did.
Chen intervened to ask Barone why the EPA couldn’t do its risk assessment with the given information, using a “lowest observed effect level,” or LOEL.
“I mean here we have a phenomenon where I think everybody agrees, as you put it, something’s going on,” Chen said, adding:
“And knowing that the EPA is to use health-protective assumptions when the information is lacking, why can’t one approach it from the low-level approach? We seem to know that there’s some level in which something’s going on. There’s adverse effects. We may debate where it is, but wouldn’t it be proper to use even a conservative estimate of LOEL?”
Barone insisted, as he did in earlier testimony, that the data are unclear. But he also conceded the EPA does often use the LOEL in risk assessment.
Throughout Barone’s testimony, Connett drew concessions from Barone through “impeachment” — meaning Barone gave responses under cross-examination that contradicted statements he made in his earlier deposition. Connett read from Barone’s deposition testimony to demonstrate he was misrepresenting his responses.
EPA case rests on data that witness admits is ‘not biologically possible’
No studies exist showing a chemical could increase IQ by 15 points or more, Ibarluzea, principal investigator of the Spanish fluoride study, told Connett during his deposition.
Yet, his study of prenatal fluoride exposure and IQ among a birth cohort in the Basque region of Spain — which found no negative association between fluoride exposure and children’s IQ — found that fluoride exposure can increase IQ in boys from 15 to 28 points.
Ibarluzea told the court the improvement identified is improbably large and “not biologically possible.” But, he said what mattered was that they found an increase in IQ associated with fluoride exposure and he stood by that finding.
Other researchers, he said, like Xu et al. (2020), also found a positive association between fluoride and IQ, although he then conceded that the paper had been retracted.
In his testimony on Monday, EPA expert witness David Savitz, Ph.D., told the court that he thought Ibarluzea’s data was important because if its findings are aggregated with the other cohort studies, it markedly shifts the cumulative evidence such that there would be no association found between maternal fluoride exposure and child IQ, which would support the EPA’s case.
Chen asked Savitz if in making such calculations, he corrected for the “implausible” increase in IQ points. Savitz said he did not.
Ibarluzea also told Connett he had “no interest whatsoever” in pursuing an explanation for certain unlikely numbers in his study that he deemed irrelevant to the overall findings.
That includes the question of how a standard practice of adjusting for creatinine to account for issues with urinary dilution dramatically changed the fluoride association outcomes, transforming what had been a negative relationship between fluoride exposure and IQ into a highly positive one.
However, he admitted there is never any study “that’s ever been done in human history where adjustment for creatinine has had such a dramatic effect on the association between a chemical and health.”
He also did not investigate whether the laboratory may have made an error, he said.
Connett asked Ibarluzea to explain several possible problems with the study that may have influenced the results.
Connett said that according to the Basque government website, consulted at the time of the deposition, the cities in the region stopped fluoridating their water sometime over the last several years. He asked if Ibarluzea was aware of that fact. He said he was.
Connett noted it was true for every town in the Basque region and asked Ibarluzea when that change was made.
“At this point, I am not a representative of the Basque government here in this deposition” Ibarluzea responded. “So I’m not going to give more explanation about this, OK, because it could be very difficult for me personally,” he said.
Connett also informed Ibarluzea that citizens in the Basque region had submitted freedom of information requests to obtain communications Ibarluzea had about the INMA study.
Ibarluzea said he was aware of this request and had appealed so it would not be released, saying he would have sued the Basque government if they had made the requested private information public.
But “not because I have anything to hide,” he added.
Connett also asked Ibarluzea why he did not adjust for fish consumption in his study. Ibarluzea testified that people in the Basque region eat more small, oily fish with high levels of fatty acids known to be beneficial to the brain than almost anyone in the world.
The plaintiffs’ witnesses all testified that this was a concerning confounding factor in the study.
In other research Ibarluzea had done on per- and polyfluoroalkyl substances (PFAS) chemicals, he wrote that when a chemical is found in seafood it can produce a spurious beneficial association between the chemical and IQ.
Ibarluzea said they didn’t consider fish consumption as a variable because they were already looking at a lot of variables.
Because the study found no association between fluoride exposure and loss of IQ in children and because it is the most recently published “high quality” cohort study, it has been a centerpiece of the EPA’s evidence.
Throughout the trial, plaintiffs’ witnesses questioned the study’s validity because of the massive increase in IQ in boys linked to fluoride exposure that all researchers, including Ibarluzea, agreed is not a plausible finding. They also questioned why the study did not control for major factors like fish consumption.
EPA witnesses pointed to the high quality of the study design and discounted the effects for fish, which they argued were accounted for in the adjustments made for mercury.
The INMA fluoride study examined the relationship between fluoride exposure and IQ in children in the only region of Spain, the Basque region, that fluoridated its water, but Ibarluzea’s testimony raises the question of whether the water was fluoridated throughout the duration of the study.