Large US study finds no link between prenatal water fluoride and child behaviour

A large US cohort study published on 3 September in JAMA Network Open found no overall association between prenatal exposure to fluoride in regulated public drinking water and children's emotional and behavioural problems [s1]. The result lands in the middle of an unusually heated policy argument, and its details matter more than its headline.

What the study did

Researchers used data from the Environmental Influences on Child Health Outcomes (ECHO) Cohort, pooling observational pregnancy-cohort records from 20 sites across 34 US states [s1]. The analysis covered 5,520 children aged 1.5 to 17 years, with a mean age of 5.8 years; 2,849 (51.6%) were boys [s1]. The underlying pregnancy data ran from 1 January 2005 to 31 December 2019 [s1].

Exposure was not measured in each mother's body but estimated from where she lived: the team calculated area-level, time-weighted mean fluoride concentrations in public water based on residential addresses during pregnancy [s1]. The mean prenatal water fluoride level was 407.9 µg/L, with a wide range from below 1.4 to 3,272.5 µg/L [s1]. For reference, most participants sat below the US Public Health Service recommended level of 700.0 µg/L [s1]. Children's outcomes were caregiver reports on the Child Behaviour Checklist, split into internalising problems (such as anxiety) and externalising problems (such as aggression) [s1].

What it found

Across the main models, the study found nothing. Per 500.0-µg/L higher prenatal water fluoride, the adjusted mean difference in problem scores was −0.15 (95% CI, −0.85 to 0.56) for internalising problems and 0.06 (95% CI, −0.44 to 0.55) for externalising problems [s1]. Both intervals straddle zero comfortably, which is the statistical signature of no detectable effect [s1].

There was one exception, and it is the part that will be quoted in both directions. In a secondary "change point" analysis — which tests whether an effect appears only above some threshold rather than across the whole range — the authors found a positive association for externalising problems above an inflection point: an adjusted mean difference of 1.59 (95% CI, 0.72 to 2.47) per 500.0-µg/L increase [s1]. Some subgroups also showed positive associations [s1].

How to read the split result

The honest summary is that the primary, prespecified comparison was null, and a positive signal surfaced only in a model designed to look for thresholds. That does not make the change-point finding meaningless — thresholds are biologically plausible for many exposures — but a positive result that appears in a secondary analysis and not in the main one carries less weight than a result that shows up everywhere. The authors themselves conclude that prenatal exposure to fluoride in regulated public water "was not associated with" internalising and externalising problems overall, while flagging the change-point signal as a reason for further study [s1].

Several limits are built into the design. Exposure was inferred from residential water systems, not individual intake, so it cannot capture fluoride from toothpaste, food, tea or supplements — a point the authors note by calling for study of other sources [s1]. The data are observational, so they describe associations, not cause and effect. And the outcome is a caregiver questionnaire, a useful but imperfect instrument. An accompanying commentary in the same journal framed the ecological fluoride and neurodevelopment literature as "an important starting point," language that itself concedes how far the evidence still has to travel [s2].

Why it matters

This is squarely a "what the study actually says" case. Earlier work associating higher fluoride exposure with neurodevelopmental effects has been used to argue against community water fluoridation, and this study will be enlisted by both camps — as reassurance that regulated levels show no overall behavioural signal, or as fresh evidence of harm above a threshold. Both readings overreach the data. What the ECHO analysis adds is a large, multi-state, US-specific null at the population level, with a narrower question left open about higher exposures and specific outcomes [s1].

The results, the authors write, "could inform future policy decisions regarding fluoride in the US public water systems" [s1]. They do not settle them, and this article offers no advice on individual fluoride intake. The value here is precisely the discipline of separating a null primary result from a secondary signal — the distinction that tends to vanish once a fluoride study reaches the headlines.

Sources

• [s1] Hernandez A, Nigra AE, Simon KR, et al. "Prenatal Water Fluoride Exposure and Children's Behavioral Problems." JAMA Network Open. 2026;9(9):e2629389.
• [s2] "Ecological Fluoride and Neurodevelopment Studies—An Important Starting Point" (Invited Commentary). JAMA Network Open. 2026;9(9):e2629226.