Court-documented child maltreatment did not register on two epigenetic clocks

That childhood maltreatment shortens lives is not in dispute. It is associated with increased risk of chronic disease and premature mortality, and one hypothesised mechanism is accelerated biological ageing [s1]. Epigenetic clocks — DNA methylation-based estimates of biological age — are the measurement technology the field has reached for to test that idea, and prior studies using adults' retrospective self-reports of maltreatment have reported small to moderate associations with clock measures [s1].

A study published in JAMA Network Open in August 2026 tested the same question with a much stronger exposure measure, and did not find it.

The design

Participants came from a prospective cohort of people whose childhood maltreatment was substantiated in court records between 1967 and 1971, alongside demographically matched controls recruited from elementary school registers and the birth hospitals of the exposed cases [s1]. The documented cases comprised physical abuse (17%), neglect (82%) and sexual abuse (14%) [s1].

Court documentation is the methodological point. It removes the recall and reporting problems that haunt retrospective self-report, in which the people who report worse childhoods may also differ in present-day health and mood.

At follow-up in 2022 and 2023, participants gave saliva for methylation analysis on the Illumina EPIC v1 array [s1]. Biological ageing was quantified with GrimAge, for cumulative lifetime burden of ageing-related damage, and DunedinPACE, for the current rate of deterioration [s1]. The authors chose these blood-based clocks over clocks designed for saliva because the saliva clocks lack a track record of association with health-risk exposures or later-life health outcomes [s1]. Values were adjusted for technical covariates including estimated buccal epithelial cell proportion and standardised to z scores [s1].

A total of 276 people were analysed: 112 controls with a mean age of 60.0 years (SD 3.7), 53 of them male (47%), and 164 in the maltreated group with a mean age of 60.0 (SD 3.6), 69 male (42%) [s1].

The result

Adults with court-documented maltreatment histories showed older GrimAge and faster DunedinPACE than controls, but the differences were small and not statistically distinguishable from zero at the .05 threshold [s1].

For GrimAge, Cohen d was 0.25 with a 95% confidence interval of −0.01 to 0.50 and P = .06; for DunedinPACE, d was 0.18 (−0.06 to 0.43), P = .14 [s1].

Adjusting for smoking flattened both: GrimAge fell to d = 0.05 (−0.194 to 0.287, P = .70) and DunedinPACE to 0.08 (−0.166 to 0.333, P = .51) [s1]. Adjusting for body mass index instead left them closer to the base model — GrimAge 0.23 (−0.033 to 0.499), DunedinPACE 0.13 (−0.126 to 0.386) [s1]. Applying inverse probability weighting to account for differential attrition did not change the picture [s1].

Two ways to read it

The first reading is that the clocks are not capturing what maltreatment does to a life. Effect sizes of 0.25 and 0.18 with intervals crossing zero, in 276 people, are consistent with a small real effect the study was not powered to resolve — but they are also consistent with very little.

The second is more specific and probably more useful: smoking absorbed nearly all of it. GrimAge is built partly to capture smoking-related methylation, so an association that collapses under smoking adjustment may be telling us the pathway rather than nullifying the finding. Childhood adversity raises the likelihood of smoking; smoking accelerates biological ageing; the clock reads the smoking. That is a mechanism, not an absence — but it is not the direct biological embedding of adversity that the accelerated-ageing hypothesis proposes.

Why the tissue matters

There is a further caveat the authors handle explicitly, and it matters for anyone reading a commercial clock result. These clocks were trained on blood; the samples here were saliva [s1]. A 2026 meta-analysis of more than 15,000 human methylation profiles across 17 tissues found that ageing produces both conserved and tissue-specific epigenetic signatures [s2] — which is to say that a clock's behaviour is not automatically portable between the tissue it was built on and the tissue a sample happens to come from.

The study's own COI statement records that one author holds a licence agreement, as inventor of the DunedinPACE clock, with a commercial epigenetic testing company [s1]. That the disclosure sits under a null result is worth noting, since the incentive ran the other way.

What it adds up to

The clearest conclusion is a methodological one. When maltreatment is measured from court records rather than from what adults remember, the association with epigenetic age gets smaller. Studies relying on self-report may be measuring something about the reporter as well as something about the childhood.

For the reader, the transferable point is about what clocks can carry. A measure that does not separate two groups with genuinely different life histories, in a well-matched cohort at 60, is not a measure that can be expected to adjudicate an individual's biography.

This article describes research findings and is not medical advice.

Sources

1. [s1] Epigenetic Clock Analysis of Biological Aging in Adults With Documented Childhood Maltreatment. JAMA Network Open, published August 10, 2026. https://doi.org/10.1001/jamanetworkopen.2026.28033
2. [s2] Meta-analysis of DNA methylation aging signatures in 17 human tissues. Nature Aging, published online June 26, 2026. https://doi.org/10.1038/s43587-026-01164-5