The Authors Spent Four Pages Begging You Not To Use Their Result
Fifty-nine studies, pooled, showing a clear effect at high exposure. At the concentrations North American water systems actually use, the confidence interval crosses zero. The dose–response curve is the finding. The headline number is not.
Background written by the news staff. Contains no argument or recommendation.
ITHACA, N.Y. — The paper is real, the effect it reports is real, and almost every account of it this newspaper has read gets the same thing wrong in the same way.
It reports a dose–response relationship. Coverage reports a finding.
Those are not the same object and the difference is the whole story.
What the paper did
Fifty-nine studies, pooled, examining childhood exposure to fluoride against measured cognitive outcomes.
The pooled effect is negative and statistically significant. That sentence is true and is where most coverage stops.
The curve
The authors then did the thing that makes the paper worth reading: they split by exposure.
| Exposure band | Studies | Effect | Confidence interval |
|---|---|---|---|
| Above 4.0 mg/L | 21 | clear negative | does not cross zero |
| 2.0 – 4.0 mg/L | 17 | negative | does not cross zero |
| 1.5 – 2.0 mg/L | 12 | small negative | narrowly excludes zero |
| Below 1.5 mg/L | 9 | very small negative | crosses zero |
North American community water systems operate between 0.6 and 0.8 mg/L. The two municipal engineers this newspaper telephoned gave 0.7 for both of their systems and both said the target had not changed in a decade.
That is in the bottom band. The bottom band is the one where the interval crosses zero — which means the pooled data cannot distinguish the effect from no effect at all.
The high-exposure studies come overwhelmingly from regions with naturally high groundwater fluoride, at three to eight times the concentration any North American system adds deliberately.
What the authors said about their own paper
Dr. Perpétua Almeida-Nyongo, the paper's second author, was asked what she wanted a general reader to take from it and answered without hesitating.
"That the dose matters and that we said so in four pages of discussion which nobody reads," she said. "We were extremely careful. We wrote that the low-exposure estimate is imprecise, that eleven studies use area-level exposure which is weak, and that the confidence interval at community concentrations includes no effect. All of that is in the paper. None of it was in the coverage."
Asked whether she regretted publishing it, she was sharp: "Absolutely not. High natural fluoride is a genuine public health problem for tens of millions of people and this paper is about them. It got turned into a story about somebody else."
The critic
Dr. Werner Achebe-Lindholm is not an author and thinks the meta-analysis is too permissive.
His objection is specific: eleven of the fifty-nine studies classify exposure by area rather than by individual measurement, which means a child is assigned the fluoride level of their district and not of the water they drank. He thinks those eleven should have been excluded and that including them widens everything.
He also said, unprompted, that he agrees with the authors about the dose–response shape and that he thinks the coverage has been worse than the paper.
"I am arguing with them about inclusion criteria," he said. "That is a normal argument between epidemiologists. It is not the argument people think is happening."
What this piece does not say
It does not say fluoridation is safe. It does not say it is unsafe. It is an explainer and it takes no position, because taking a position would be an opinion and this newspaper does not print opinion on the news pages.
What it says is narrower: that the paper reports an effect that scales with dose; that the dose at which the effect is clear is several times the dose used in North American water systems; that at those systems' concentrations the pooled interval includes zero; and that the paper's own authors have been saying this, in print, for longer than anyone has been quoting them.
The six o'clock edition
Three stories a morning, in your inbox before the coffee is made.
Sources & Method
An explainer, written by the news staff. It takes no position on what anyone should do. We read the paper rather than the press release, telephoned an author, telephoned a critic who is not an author, and printed the dose–response curve because the curve is the finding and the headline number is not.
Who we spoke to
- Dr. Perpétua Almeida-Nyongo, Epidemiologist; second author of the meta-analysis. Interviewed by video call 26 June 2026 Asked us to lead on the dose–response curve rather than the headline effect. We have.
- Dr. Werner Achebe-Lindholm, Environmental health researcher, not an author. Interviewed by telephone 29 June 2026 Is critical of the meta-analysis's inclusion criteria and his criticism is printed.
- Two municipal water engineers, Named in our notes. Telephone July 2026 Supplied operating concentrations for their systems.
What we could not confirm
- Whether the residual effect at low exposure is real. The confidence interval crosses zero and the authors say so.
- The exposure classification in eleven of the fifty-nine studies, which relies on area-level rather than individual measurement. Both the authors and their critic flagged this independently.
How was this story?
We publish the result, whatever it is. Reader verdicts appear on the front page and in our newsroom metrics.
Readers' Letters 0
Printed at once under the name you give and read by the desk afterwards; anything unfit is removed, with a note saying so, and nothing else is ever deleted — only corrected. Letters that changed something in the story carry a mark saying so, and there are 5 of those across the archive.