A randomized trial has produced an unusually intriguing result in sleep and aging research. Older adults who received cognitive behavioural therapy for insomnia showed a more favourable change in one epigenetic measure of aging than people who received sleep education.
That is stronger evidence than the familiar studies that simply find poor sleepers tend to have worse health. Random assignment gives researchers a better chance of isolating the effect of the treatment itself.
But the result still requires restraint. The trial did not show that anyone lived longer, avoided disease, or gained additional years of healthy life. It found a difference in DunedinPACE, a DNA methylation measure designed to estimate the pace of biological change.
What the trial tested
The analysis came from a randomized clinical trial involving adults aged 60 or older who met diagnostic criteria for insomnia. Participants received either cognitive behavioural therapy for insomnia, usually called CBT-I, or sleep education therapy.
CBT-I is a structured, non-drug treatment. It helps people change behaviours and thought patterns that can keep insomnia going. Common elements include keeping a consistent sleep schedule, limiting time awake in bed, reducing anxiety about sleep, and rebuilding the connection between bed and sleep.
Sleep education provides useful information, but it is generally less intensive. In this study it served as an active comparison rather than leaving the control group untreated.
Researchers analysed paired blood samples from 92 participants, including 47 in the CBT-I group and 45 in the sleep education group. The samples were collected before treatment and again roughly 20 to 39 months later.
At follow-up, 34 percent of the CBT-I group was in insomnia remission, compared with 13 percent of the sleep education group.
One aging measure changed
The researchers examined three DNA methylation measures: DunedinPACE, GrimAge, and a principal-component version of PhenoAge.
DNA methylation is a chemical marking system that helps regulate how genes are used. Patterns in these marks can be combined into statistical models often called epigenetic clocks. Different clocks are trained on different outcomes, so they do not all measure the same thing.
DunedinPACE was built to estimate how quickly a set of age-related physiological changes is occurring. A value near 1 is often interpreted as roughly one year of biological change per chronological year. It is better understood as a research estimate of pace than as a literal reading of a person’s biological age.
Compared with sleep education, CBT-I was associated with a 0.02 lower DunedinPACE value. The 95 percent confidence interval ranged from 0.01 to 0.04 lower, and the result remained statistically significant after correction for multiple comparisons.
The other clocks did not show statistically significant differences. GrimAge moved in a favourable direction, but its confidence interval included no effect. PCPhenoAge also did not differ significantly between groups.
Why randomization matters
People with chronic insomnia can differ from good sleepers in many ways. They may have more illness, stress, pain, depression, medication use, or socioeconomic disadvantage. In an observational study, those differences can make it difficult to tell whether insomnia contributes to faster aging or simply accompanies other causes of poor health.
Randomization helps distribute measured and unmeasured differences between groups. When one assigned treatment produces a different result from another, the case for causation becomes stronger.
Even so, this was a secondary analysis of an older trial rather than a study designed from the beginning around aging outcomes. Only participants with suitable paired samples were included. That makes the finding more informative than an association, but less definitive than a large trial built specifically to test aging biology.
A changed clock is not a changed lifespan
The central uncertainty is whether a small movement in DunedinPACE reflects a durable change in the biology that determines disease, disability, and survival.
Epigenetic clocks can predict health risks across populations. That does not automatically make every short-term change in a clock a validated surrogate for a meaningful clinical benefit. A biomarker can respond to shifts in immune cells, inflammation, medication, or other temporary conditions without capturing a lasting change in the aging process.
The disagreement among the three clocks matters for the same reason. If CBT-I had broadly altered aging biology, researchers might hope to see a coherent pattern across multiple measures. The isolated DunedinPACE result could mean that this measure is especially responsive to sleep improvement. It could also represent a narrow biological effect or a chance finding in a small sample.
Why sleep still matters
None of this makes insomnia unimportant. Persistent insomnia can damage quality of life and is associated with depression, cardiovascular disease, cognitive problems, and other health risks. CBT-I is already recommended because it can improve sleep without the long-term drawbacks that accompany some sedative medicines.
Sleep also affects immune activity, metabolism, stress hormones, and the brain’s maintenance systems. It is biologically plausible that treating serious insomnia could influence processes connected to aging.
The current result adds a useful clue. It is not the reason to seek effective insomnia care, and it should not be sold as an anti-aging treatment.
What needs to happen next
A stronger test would enrol a larger and more diverse group, collect repeated blood samples at standardized intervals, and track several validated biomarkers alongside clinical outcomes. Researchers would also want to know whether the molecular difference persists and whether it predicts fewer cardiovascular events, less cognitive decline, better physical function, or longer disability-free survival.
Replication is especially important because the analysis included only 92 people and only one of the three clocks produced a statistically significant result.
For now, the fairest conclusion is modest. CBT-I improved insomnia remission and changed one promising aging-related biomarker. That is a credible signal from a randomized human study. It is not proof that the participants aged more slowly in the way that ultimately matters.
Primary study: https://doi.org/10.1016/j.lanhl.2026.100861
Related reading: https://thelifespanbrief.com/2026/09/07/rentosertib-proteomic-aging-clocks-phase-2a-trial/