DunedinPACE Predicts Health Risk, Not Your Next Move

A biological-age test can forecast risk. It cannot yet write your health plan.

DunedinPACE, a DNA-methylation measure of the pace of biological aging, has cleared an important test: it predicted later disease burden, modeled healthcare-cost risk, incident chronic disease, and mortality in a real U.S. health-system population. That makes it more than a consumer-facing number with an impressive font. It still does not show that testing people improves treatment decisions, lowers costs, or extends life.

The distinction matters. A biomarker can identify people at higher risk without telling clinicians which intervention will change that risk. Blood pressure works as a clinical tool because trials show that lowering it prevents events. DunedinPACE has promising evidence for forecasting; evidence for acting on its result remains early.

Current evidence supports DunedinPACE as a risk marker, not a prescription engine for supplements, training plans, or medical care.

What DunedinPACE measures

Unlike an epigenetic clock that reports whether a person appears older or younger than their chronological age, DunedinPACE estimates the rate of biological aging from blood DNA methylation. In the original Belsky et al. 2022 study, a score of 1.0 corresponded approximately to one biological year of aging per chronological year. Scores above 1.0 indicate a faster pace.

That calibration is intuitive, but it should not be mistaken for a literal odometer. The test reads methylation patterns associated with the pace of aging. It does not directly measure every organ, prove damage in a particular tissue, or identify one mechanism causing future illness.

Belsky and colleagues developed and validated DunedinPACE as a pace-of-aging biomarker and linked faster scores with worse functional aging, morbidity, disability, and mortality across human cohorts. Those findings established a reason to take the signal seriously. They did not establish that lowering a score will lower disease risk.

The veteran study moves the question closer to clinical reality

The most practical evidence comes from Bourassa et al. 2025, a prospective electronic-health-record study of 2,216 U.S. veterans. After adjustment for demographic factors, biomarkers, and smoking covariates, faster DunedinPACE predicted greater subsequent chronic-disease burden, higher modeled healthcare-cost risk, incident major chronic diseases, all-cause mortality, and chronic-disease mortality.

These are clinically legible endpoints. The study did not merely associate a methylation score with another laboratory marker collected on the same day; it linked faster aging pace to later recorded disease and death. That prospective design reduces, but does not eliminate, the possibility that existing illness fully explains the association.

The healthcare-cost result needs careful reading. The study modeled risk of healthcare costs; it did not test a program in which clinicians used DunedinPACE to prevent spending. Forecasting a more expensive future and saving money by acting on a forecast are different claims.

Other cohorts support a broad morbidity-and-mortality signal

Several studies point in the same general direction while asking different questions in different populations. In the Women's Health Initiative, Jain et al. 2023 found that epigenetic age acceleration was associated with multimorbidity at age 90. The endpoint was accumulated chronic disease in very old women, not incident disease in veterans, so it supports a related rather than identical conclusion: methylation-aging measures track systemic disease burden.

In the Strong Heart Study, Jiang et al. 2023 analyzed arsenic exposure, epigenetic aging, cardiovascular disease, and all-cause mortality. The study places environmental exposure, aging biomarkers, and hard outcomes in the same analytic frame. It supports biological relevance, but it also illustrates the central interpretive problem: toxic exposure can influence both the biomarker and the diseases that follow.

A 2026 study of older U.S. adults by Zhao et al. linked DNA-methylation aging algorithms with type 2 diabetes and mortality risk. Together, these cohorts make a stronger case for prognostic association across settings than any one study could. They do not provide one interchangeable effect estimate or one universal screening use case.

Self-rated health belongs in the conversation

A molecular test is not the only way to detect risk. Li et al. 2024 examined self-rated health, epigenetic aging, and long-term mortality in older Australians. The study puts a simple patient-reported measure and epigenetic aging in the same mortality discussion.

Self-rated health costs almost nothing, takes seconds, and is already familiar in clinical care. It does not replace methylation testing, and the study does not show that combining the two improves care. Still, a person's answer to "How would you rate your health?" may carry useful information that a premium biological-age report should not automatically outrank.

Why association does not settle causation

Epigenetic aging measures respond to a crowded life history. Higher diet quality was associated with slower epigenetic aging in a cross-sectional analysis of 1,995 Framingham Heart Study Offspring participants in Kim et al. 2022. Metal mixtures were associated with biological-aging measures among 2,301 Strong Heart Study participants in Boyer et al. 2023. Studies have also linked epigenetic aging with prenatal growth and maternal energy status, chronic-stress physiology, mental health, neighborhood deprivation, and depressive symptoms.

These findings strengthen the idea that methylation measures capture meaningful cumulative burden. They also make simplistic interpretation harder. A faster score may reflect cardiometabolic risk, toxic exposure, stress, social adversity, current disease, shared causes, or some combination. It is an integrated signal, not a clean readout of one isolated aging process.

This concern becomes concrete in Ammous et al. 2022, where epigenetic age acceleration was associated with cardiometabolic risk factors and clinical cardiovascular-disease risk scores in African Americans. If a biomarker overlaps with blood pressure, glucose, adiposity, and standard risk calculations, it may add useful information—but it may also partly repackage information clinicians already have. Lin et al. 2024 similarly linked epigenetic age acceleration with blood lipid levels in older, multi-ancestry U.S. adults. Cross-sectional metabolic associations cannot determine whether lipids drive the methylation signal, the signal reflects downstream effects, or both arise from common causes.

Limits on general use

Population differences matter. The veteran study involved U.S. veterans. Jain et al. focused on women at age 90. Beynon and colleagues found that epigenetic aging biomarkers predicted mortality in a longitudinal cohort of people with oropharyngeal cancer, but that oncology result does not prove equal value for preventive screening in healthy adults. Disease context, baseline risk, clinical care, and competing causes of death all differ.

Most of this evidence is observational. Statistical adjustment helps, but it cannot fully remove residual confounding or reverse causation. Social conditions deserve particular attention: Joshi et al. 2024 associated neighborhood deprivation and depressive symptoms with epigenetic age acceleration in the Canadian Longitudinal Study on Aging. A high score may therefore identify real risk while also reflecting conditions that a blood test alone cannot solve.

The useful verdict

DunedinPACE has credible human evidence as a risk-forecasting biomarker. Faster scores have tracked worse functional aging, more disease, greater modeled healthcare-use risk, and earlier mortality across observational cohorts, with the veteran study providing particularly relevant prospective follow-up.

For now, a result should prompt better questions rather than a shopping spree: Are blood pressure, lipids, glucose, sleep, activity, smoking, mental health, and preventive care in good shape? How does the person rate their own health? Those measures already guide decisions. No cited study shows that changing care because of a DunedinPACE result improves morbidity, mortality, or costs. Until that evidence arrives, the score is informative context—not a clinical command.