Prenatal PFAS and two child outcomes: sleep scores and bone density

PFAS cross the placenta, and they are persistent enough that a single maternal serum measurement represents years of accumulated exposure rather than a recent event [s1]. That combination has made prenatal exposure one of the most heavily studied questions in the field — and one of the hardest to get a clean answer out of, because the outcomes of interest in children are numerous, the effect sizes are small, and the number of compounds measured keeps growing.

Two papers published a week apart illustrate both the appeal and the difficulty.

Sleep, in 2,836 pairs

The larger study drew on the Environmental influences on Child Health Outcomes cohort, using 2,836 mother-child pairs from 14 study sites [s1]. Maternal serum concentrations of PFOA, PFOS, PFNA and PFHxS were measured during pregnancy, and outcomes were Child Behavior Checklist Sleep Problems T-scores assessed between ages one and five [s1]. Linear mixed-effects models with random intercepts estimated associations with log2-transformed PFAS concentrations, with covariate adjustment [s1].

The sample is unusually diverse for this literature: mean maternal age 29.6 years, 43% non-Hispanic White, 24% Black and 23% Hispanic, with 27% holding advanced degrees and 36% reporting annual incomes under $30,000 [s1].

One of the four compounds produced a signal. Higher prenatal PFOA concentrations were associated with higher sleep problem scores in early childhood, at a coefficient of 0.13 (95% CI 0.05 to 0.21) [s1]. Associations for PFOS, PFNA and PFHxS were small and not statistically significant [s1].

The authors describe the magnitude of the PFOA association as modest, and set out precisely what would be needed to firm it up: harmonised trimester-specific exposure assessment, objective sleep measures, and mixture-based analytic approaches [s1].

That last list is the most useful part of the paper. A CBCL Sleep Problems T-score is a parent-reported measure, not a measurement of sleep. It captures what a parent notices and reports about a child's sleep-related behaviour, which is a legitimate outcome but a different one from sleep duration, latency or fragmentation. The coefficient of 0.13 is on a T-score scale where the population standard deviation is large; this is a shift in a distribution, not a description of any individual child.

Bone density, in 755 pairs

The second study measured ten PFAS in first-trimester maternal plasma by high-performance liquid chromatography with tandem mass spectrometry among 755 mother-child pairs in the Shanghai Birth Cohort, and assessed childhood lumbar spine bone mineral density by dual-energy X-ray absorptiometry [s2]. Analysis used multivariable linear regression, Bayesian kernel machine regression and quantile g-computation [s2].

In the overall analysis, perfluoroheptanoic acid was suggestively associated with higher total bone mineral density, at 0.13 (95% CI 0.00 to 0.27) — an interval whose lower bound sits exactly on the null [s2].

The sex-stratified results run in opposite directions. Among girls, PFDA and PFDoA showed inverse associations with total bone mineral density, at -0.29 (95% CI -0.57 to 0.00) and -0.27 (-0.50 to -0.04) [s2]. Among boys, PFHpA showed a positive association, at 0.20 (0.04 to 0.37) [s2]. The Bayesian kernel machine regression analyses suggested mixture effects tending inverse among girls and positive among boys at higher quantiles, and in the quantile g-computation models the positive association was most evident for L3 bone mineral density among boys, at 0.23 (0.00 to 0.45) [s2].

The authors call this preliminary evidence of possible sex-specific patterns [s2], which is the right register. Three of the intervals quoted above touch or include zero.

The common problem

Both studies test many compounds against outcomes that can be sliced several ways, in samples where the true effects, if they exist, are small. The ECHO analysis tested four compounds; the Shanghai analysis ten, plus mixtures, plus sex strata, plus multiple bone sites [s1] [s2]. Under those conditions, some associations will reach conventional significance whether or not anything is happening.

That is not an accusation against either paper — both report their full result sets and both hedge appropriately. It is a statement about what a reader can take from a single such study, which is: less than the abstract's first sentence suggests.

The opposite-direction finding by sex in the bone study is a specific case worth flagging. A plausible biological story exists for it — PFAS are endocrine-active and bone accrual is hormone-dependent — but a plausible story can be constructed for almost any sex difference after the fact. The finding earns attention as a hypothesis. It does not yet describe a phenomenon.

Where this literature actually stands

The defensible summary across both papers is narrow. Prenatal PFOA showed a small association with parent-reported sleep problems in a large, diverse US cohort, and three other PFAS did not [s1]. Prenatal PFAS showed inconsistent, sex-divergent associations with childhood bone mineral density in a smaller Chinese cohort, with most estimates close to the null [s2].

Nothing here supports a claim about what any individual pregnancy or child faces, and neither study was designed to answer that. What both point toward — and what the ECHO authors ask for explicitly — is a next generation of studies with repeated exposure measurement across trimesters and objective rather than reported outcomes [s1]. Until those arrive, the honest position on prenatal PFAS and these particular endpoints is that the associations are small, unstable across compounds, and not yet replicated.

Sources

• [s1] Prenatal PFAS exposure and early childhood sleep quality in the ECHO Cohort. Environmental Research, 13 August 2026. https://doi.org/10.1016/j.envres.2026.125429
• [s2] Prenatal exposure to per- and polyfluoroalkyl substances and child bone mineral density from the Shanghai Birth Cohort. Journal of Hazardous Materials, 8 August 2026. https://doi.org/10.1016/j.jhazmat.2026.143230