Pace of Aging
Pace of Aging estimates how fast multi-system biological decline is unfolding, making it a rate measure rather than a biological-age snapshot.
Also known as: DunedinPACE, DunedinPoAm, pace of biological aging, aging rate, biological aging velocity
Biological-age testing turns aging into a dashboard. Pace of Aging is the slope on that dashboard: not how old a model says the system looks today, but whether multi-system change appears to be moving faster or slower than expected. That makes it useful for research and cautious clinical interpretation. It also makes it easy to oversell when a commercial report treats one number as a verdict.
What It Is
Pace of Aging is a rate concept. It estimates how quickly biological change is accumulating across systems, rather than estimating how old the body appears at one moment. The question is not “what age does the system look like?” The question is “how fast is the system changing?”
The concept came out of the Dunedin Study, a New Zealand birth cohort followed for decades. Researchers measured change across organ-system and blood markers, then asked whether people of the same chronological age were aging at different rates. They were. Later work compressed that longitudinal signal into blood DNA-methylation measures such as DunedinPoAm and DunedinPACE.
The distinction matters because several aging measures sound similar but answer different questions:
| Measurement frame | Question it answers | Better use | Main limit |
|---|---|---|---|
| Biological age | How old does the system look now? | Risk framing at one time point | Can hide whether the slope is improving or worsening |
| Age acceleration | Is the score older or younger than chronological age predicts? | Peer-relative comparison | Depends heavily on the clock and reference population |
| Pace of Aging | How fast is biological change occurring? | Longitudinal risk and trial-endpoint framing | Still indirect, and usually trained against selected markers |
| DunedinPACE | What does blood DNA methylation estimate about aging rate? | Research and repeated-measure tracking | Not a diagnosis and not a complete mechanism map |
This makes Pace of Aging one of the more useful measurement frames in geroscience. It does not claim that a lower number proves longer life. It gives researchers, clinicians, and careful readers a way to ask whether biological change is moving faster or slower than expected, and whether that rate predicts healthspan, disability, disease, cognition, or survival.
Why It Matters
Static biological age is easy to overread. A report says “43.2” or “seven years younger,” and the reader wants to treat the number as a verdict. But aging is not only a position on a scale. It is also a slope.
The slope is the part many interventions claim to change. A fasting-mimicking cycle, exercise block, sleep intervention, rapamycin protocol, or clinic program rarely claims only to make a person look younger on one test day. The stronger claim is that it slows harmful biological change. That claim needs a rate measure, not only a state measure.
Pace language also disciplines evidence claims. A pace measure can be a strong predictor and still be a weak intervention endpoint. If faster Pace of Aging predicts disability and mortality, the measure is useful. If a protocol lowers a pace score over 12 weeks, that is not yet proof that the protocol delayed disability, disease, or death. It is a biomarker movement that needs replication, adverse-event tracking, and harder outcomes.
The reader’s advantage is not that Pace of Aging gives a final answer. It gives a better question. Is the claim about state, rate, risk prediction, or intervention response?
How It Is Measured
The original Pace of Aging measure was built from repeated clinical and biological measures in the Dunedin cohort. Belsky and colleagues tracked change in organ-system integrity markers across young adulthood, then combined individual rates of change into a composite aging-rate measure. That version was powerful because it was longitudinal. It was also hard to use outside a long-running cohort.
DunedinPoAm translated the rate idea into a blood DNA-methylation algorithm. DunedinPACE refined it by training against 20 years of change across 19 indicators, using more measurement occasions and a more reliable set of CpG sites. The point was to make a single blood sample approximate a longitudinal multi-system signal.
DunedinPACE is commonly scaled around 1.0, meaning roughly one year of biological change per chronological year in the source frame. A value above 1.0 is faster; a value below 1.0 is slower. That scaling gives the reader a rate metaphor. It should not be mistaken for an exact personal stopwatch.
The 2025 older-adult analysis broadened the frame beyond methylation. Balachandran and colleagues adapted Pace of Aging for the US Health and Retirement Study and the English Longitudinal Study of Aging, combining longitudinal blood biomarkers, physical measurements, and functional tests. That work supports Pace of Aging as a population-aging method, not only as a midlife Dunedin methylation story.
Recognition starts with the report’s target. A Pace of Aging result should tell the reader which algorithm produced it, what sample was used, which population trained it, whether the lab reports technical repeatability, and which outcomes the measure has predicted in independent cohorts.
A lower Pace of Aging score is not proof that a person has slowed aging in the way that matters most. It is evidence that a model output changed. The stronger claim requires durability, repeatability, and connection to healthspan or survival outcomes.
How It Plays Out
A reader receives a commercial report with a DunedinPACE-style value of 0.92. The restrained interpretation is that the model estimates slower-than-reference biological change. The result is not eight percent more healthspan, and it is not proof that the current stack works. The next questions are whether repeat tests use the same assay, whether the lab reports technical variation, whether illness or weight change affected the sample, and whether the result fits harder markers such as blood pressure, ApoB, VO₂max, strength, sleep, glucose control, and function.
A clinic runs a 12-week protocol and repeats the score. If the score improves, the result is interesting. It still does not isolate cause if the person changed diet, exercise, sleep, medication, weight, supplements, and stress exposure at the same time. The measure can frame a hypothesis. It cannot allocate credit across a stacked intervention.
A researcher uses the measure more cleanly. In a trial, Pace of Aging may sit beside prespecified endpoints: cardiometabolic markers, inflammatory markers, body composition, physical performance, adverse events, and quality of life. In that context, the pace measure is one readout in a bundle. It is useful because it adds a rate estimate, not because it replaces outcomes.
Evidence
Evidence tier: Observational (human, large). Pace-of-aging measures have substantial human cohort support as risk markers. They do not yet prove that any specific intervention slows human aging in a clinically meaningful way.
The first major Dunedin analysis measured biological aging directly from repeated biomarkers. Belsky and colleagues studied 954 participants from the Dunedin birth cohort and tracked marker change from young adulthood into midlife. People with faster measured aging showed worse physical function, cognitive decline, subjective health, and older facial appearance by age 38 (Belsky et al., 2015).
DunedinPoAm translated that longitudinal idea into a blood DNA-methylation algorithm. The 2020 eLife paper reported proof of principle that a single blood test could estimate a person’s pace of biological aging, with validation across cohort studies and the CALERIE caloric-restriction trial setting (Belsky et al., 2020). DunedinPACE then improved the method by training against 20 years of change across 19 indicators and using a more reliable subset of CpG sites. Belsky and colleagues reported strong test-retest reliability and associations with function, morbidity, mortality, and perceived health (Belsky et al., 2022).
The 2021 Nature Aging midlife paper is important because it connected pace differences among same-age adults to future frailty risk and policy questions. It showed that faster biological aging was already visible by midlife, before many chronic diseases would be diagnosed. That finding supports the rate frame: the signal is not only late-life disease burden after the fact (Elliott et al., 2021).
The 2025 Nature Aging study moved the frame into older adults. Balachandran and colleagues implemented an adapted Pace of Aging method in parallel in the US Health and Retirement Study and the English Longitudinal Study of Aging, with 19,045 participants combined. The measure integrated longitudinal blood biomarkers, physical measurements, and functional tests. Faster Pace of Aging predicted mortality, morbidity, disability, and cognitive impairment, and the paper compared the measure with blood-chemistry biological-age metrics and epigenetic clocks (Balachandran et al., 2025).
That 2025 result is the recent shift. Pace of Aging is no longer only a midlife Dunedin-derived methylation story. It is becoming a broader population-aging frame for older adults, healthspan, and lifespan. The boundary remains the same: prediction is not intervention proof.
Caveats and Open Questions
The first caveat is model dependence. Pace of Aging is not one measurement in the way blood pressure is one measurement. The original Dunedin composite, DunedinPoAm, DunedinPACE, and the 2025 older-adult population method are related but not interchangeable. Each has its own training data, input markers, scaling, validation cohorts, and intended use.
The second caveat is mechanism. A pace score can predict risk without naming the pathway to treat. It may summarize inflammation, cell composition, smoking exposure, disease burden, medication effects, weight change, recovery from illness, or technical variation. The result can say “look closer.” It cannot say “add this intervention.”
The third caveat is intervention proof. Pace measures may become useful trial endpoints, but the field has not yet established that lowering one pace marker in a short study reliably translates into longer healthy life. That link is the claim to watch.
Consequences
Benefits. Pace of Aging gives the field a better question. Instead of asking only whether a person looks biologically older or younger at one moment, it asks whether the person’s measured aging process appears faster or slower than expected. That is closer to what geroscience interventions claim to change.
The concept also disciplines Biological Age. Static clocks, PhenoAge, GrimAge, and DunedinPACE-like measures do not answer the same question. A careful reader can ask which claim is being made: state, acceleration, risk prediction, rate, or intervention response.
Liabilities. Pace measures can become another form of Single-Biomarker Tunnel Vision. A person can chase a lower pace number while ignoring blood pressure, ApoB, sleep debt, lean mass, falls risk, alcohol intake, social isolation, or medication side effects. The rate frame is better than a static dashboard, but it is still a dashboard.
The measure can also tempt overclaiming. A protocol that lowers DunedinPACE in a small short study has not shown extended healthy life. It has shown movement in a rate-estimation marker. That may be worth studying. It is not enough to prescribe, market, or personally escalate a protocol without harder outcomes and safety data.
The useful posture is specific: Pace of Aging is a research-grade and increasingly clinic-facing risk vocabulary. It helps interpret whether biological change is moving faster or slower. It does not diagnose a person, choose an intervention, or prove that a protocol added healthy years.
Related Articles
Sources
- Balachandran, Arun, Heming Pei, Yifan Shi, John R. Beard, Avshalom Caspi, Alan A. Cohen, Benjamin W. Domingue, et al. “Pace of Aging Analysis of Healthspan and Lifespan in Older Adults in the US and UK.” Nature Aging 5 (2025): 1132-1142. https://doi.org/10.1038/s43587-025-00866-6
- Belsky, Daniel W., Avshalom Caspi, Renate Houts, Harvey J. Cohen, David L. Corcoran, Andrea Danese, HonaLee Harrington, et al. “Quantification of Biological Aging in Young Adults.” Proceedings of the National Academy of Sciences 112, no. 30 (2015): E4104-E4110. https://doi.org/10.1073/pnas.1506264112
- Belsky, Daniel W., Avshalom Caspi, Louise Arseneault, Andrea Baccarelli, David L. Corcoran, Xian Gao, Eilis Hannon, et al. “Quantification of the Pace of Biological Aging in Humans Through a Blood Test, the DunedinPoAm DNA Methylation Algorithm.” eLife 9 (2020): e54870. https://doi.org/10.7554/eLife.54870
- Belsky, Daniel W., Avshalom Caspi, David L. Corcoran, Karen Sugden, Richie Poulton, Louise Arseneault, Andrea Baccarelli, et al. “DunedinPACE, a DNA Methylation Biomarker of the Pace of Aging.” eLife 11 (2022): e73420. https://doi.org/10.7554/eLife.73420
- Elliott, Maxwell L., Avshalom Caspi, Renate M. Houts, Antony Ambler, Jonathan M. Broadbent, Robert J. Hancox, HonaLee Harrington, et al. “Disparities in the Pace of Biological Aging Among Midlife Adults of the Same Chronological Age Have Implications for Future Frailty Risk and Policy.” Nature Aging 1 (2021): 295-308. https://doi.org/10.1038/s43587-021-00044-4
Medical and Legal Boundary
This entry is a reference, not medical advice. It describes published evidence, regulatory status, and common clinical practice patterns. It does not diagnose, prescribe, or replace a clinician’s judgment for a specific person.