Pace of Aging Analysis of Healthspan and Lifespan in Older Adults in the US and UK

Geriatrics Neurology Hematology Environmental and Public Health Genetics Life Sciences Respiratory System Medicine
pace of aging healthspan biomarkers epidemiology United States United Kingdom aging population

— Longitudinal Population Analysis Based on the “Pace of Aging” Method

1. Research Background and Scientific Significance

With the acceleration of global population aging, objectively measuring and improving the health status of the elderly has become an important issue in social policy and public health worldwide. Traditionally, measurements of population aging have primarily relied on indicators such as “lifespan” and “healthspan,” but these metrics have certain limitations—particularly their inability to effectively distinguish between “health disparities caused by early life factors (such as prenatal care, early nutrition)” and “modifiable health changes during adulthood and old age resulting from ongoing physiological aging processes.” Traditional metrics also struggle to reflect the impact of late-life interventions in a timely or sensitive manner, and they are limited in dissecting the intrinsic mechanisms behind group health inequalities.

To address this, researchers have actively explored how to quantify individual and population “pace of aging,” that is, the rate at which biological functions decline over time. This indicator focuses more on capturing dynamic changes in organs, tissues, and functional capacity with aging, rather than static health status or endpoints like death. By measuring the pace of aging, policymakers can more precisely grasp population health trends, support the design and evaluation of public health interventions, and advance the science of “healthy longevity.” Recently, various research groups have developed metrics such as Biological Age and Epigenetic Clocks, but these are mostly static measurements, emphasizing the accumulation of health disparities at stages, rather than the dynamic rate of aging.

In response to the above issues, this study proposes and validates an “adapted pace of aging method” suitable for large-scale, national cohort longitudinal studies, with the aim of providing a more sensitive and practically useful tool for modern population health science and healthy aging policy.

2. Paper Provenance and Research Team

This is an original research article titled “Pace of aging analysis of healthspan and lifespan in older adults in the US and UK,” published in the internationally renowned journal Nature Aging in June 2025.
The author team hails from several world-leading research institutions, including the Columbia University Mailman School of Public Health, Duke University, Stanford University, the National Institute on Aging, and the Norwegian Institute for Public Health; the corresponding author is Daniel W. Belsky (Columbia Aging Center).

3. Research Process and Technical Roadmap

1. Overall Structure and Design

This study implemented a standardized, generalizable quantitative scheme for the pace of aging using long-term follow-up data from two nationally representative cohorts: the U.S. Health and Retirement Study (HRS) and the English Longitudinal Study of Aging (ELSA). The core process is as follows:

(1) Sample Selection and Collection

  • HRS Cohort: Included data from 2006-2016, selecting individuals above age 40 who had at least two measurements of blood, physical, and functional indicators over the period, with coverage of at least six biological markers (biomarkers), totaling 13,358 participants.
  • ELSA Cohort: Included data from 2004-2012, valid sample size of 5,687, with a process essentially consistent with HRS.
  • Both cohorts are nationally representative panels with high biomarker measurement coverage and follow-up data on survival, function, disease, and other health outcomes.

(2) Integration and Standardization of the Biomarker System

According to consensus in aging and health science, nine blood, physiological, and functional test indicators showing significant age-related changes and longitudinal tracking in both cohorts were chosen, including: - C-reactive protein (CRP, inflammation marker) - Cystatin-C (kidney function marker, replaced with Hemoglobin in ELSA where unavailable) - Glycated hemoglobin (HbA1c, glucose metabolism) - Diastolic blood pressure - Waist circumference - Peak flow (lung capacity) - Tandem balance (balance capacity) - Grip strength - Gait speed

All indicators underwent sex- and age-specific standardization and directional adjustment to enable unified comparison (reverse coding for indicators of functional decline).

(3) Longitudinal Analysis and Calculation of the “Pace of Aging” Metric

  • Using linear mixed-effects models, for each biomarker a multi-level model was established (“individual-sex-follow-up years-baseline age”) to obtain the time-change slope (i.e., annual rate of decline in biological function) for each participant.
  • The nine slopes were then aggregated to yield each individual’s “pace of aging” score (using the annual change rate among those under 65 and same sex as reference; score 1 = average for that group, >1 = faster aging, = slower aging).
  • The same measurement system was developed and validated for the ELSA cohort, enabling direct cross-country comparison.

(4) Collection and Tracking of Relevant Health Outcome Data

  • Mortality: HRS follow-up to 2021.
  • Incident chronic diseases: Physician-diagnosed/self-reported new-onset diseases.
  • Incident disability (ADL/IADL): Limitations in activities/instrumental activities of daily living.
  • Cognitive impairment and dementia: Based on cognitive assessment and interview.
  • ELSA, due to data only to 2018, did not include part of mortality and dementia classification, focusing mainly on cognitive performance scores.

(5) Validation of Metric Sensitivity, Robustness, and Socioeconomic Associations

  • Examined the predictive relationship between “pace of aging” and indicators such as mortality, disease, disability, cognitive impairment (including ROC curve, risk ratio, correlation).
  • Explored indicator performance and trends in different gender, age, ethnicity, education level subpopulations.
  • Compared multiple mainstream “biological age” measures (blood chemistry, physiological age models, epigenetic clocks) in their ability to predict health, judging the unique value of this approach.

2. Data Analysis and Algorithm Highlights

(a) Unique “Adapted Pace of Aging” Calculation Method

  • Innovatively analyzes slopes of biomarker change (rate of biological decline), rather than absolute values, to measure the dynamic nature of individual aging.
  • Flexibly adapts to varying data densities and indicator systems, allowing global multi-center cohort comparison.
  • In response to non-linear physiological changes due to interventions, unsuitable indicators (like blood lipids, systolic blood pressure) are pre-excluded.

(b) Scientific and Rigorous Statistical Models

  • Integrates mixed-effects models, Poisson regression, Cox proportional hazards models, ROC curves, and more, ensuring that results withstand multi-angle scrutiny and have quantitatively strong effect sizes.
  • Systematic adjustment for uncontrollable confounders (e.g., age, sex, smoking history, BMI, education level), effectively excluding the impact of sample selection bias.

4. Main Findings and Data Demonstration

1. Distribution Characteristics of Pace of Aging in Populations

  • HRS data showed that among 13,358 American middle-aged and elderly, pace of aging values were approximately normally distributed. Men had faster aging than women, and older participants had faster aging than those in younger old groups (mean 1.49, SD 0.89).
  • Using “white” as reference, Black and Hispanic American populations had faster pace of aging (Cohen’s d 0.20 and -0.07, respectively).
  • ELSA analysis showed similar patterns (high inter-indicator correlations), though gender and ethnicity differences were slightly smaller.

2. Association Between Pace of Aging and Health Outcomes

  • The faster the pace of aging, the significantly higher the 10-year risk of mortality (HR = 1.83).
  • Incidences of chronic disease (IRR = 1.08), daily activity disabilities (ADL IRR = 1.58, IADL IRR = 1.49), cognitive impairment (IRR = 1.57), etc., all increased significantly with higher pace of aging.
  • Correlations were consistent across subgroups (by sex, age, education, ethnicity) and were more predictive than mainstream measurements such as epigenetic clocks and conventional biological age.

3. Tests of Metric Sensitivity and Stability (Robustness)

  • In “leave-one-out” analyses, where any single indicator was removed, overall predictive power remained stable, indicating the method has strong robustness to data sources.
  • Baseline characteristics such as BMI only affected certain outcomes (e.g., cognitive ability in ELSA); in most cases, pace of aging showed strong independence.

4. Comparison with Biological Age and Epigenetic Clock Methods

  • Correlation between blood chemistry-based biological age and pace of aging is moderate (r = 0.3-0.4), though biological age metrics themselves are more correlated (r = 0.6-0.8).
  • Pace of aging was superior (or comparable) to other methods for predicting mortality, disability, and cognitive impairment.
  • Epigenetic clocks (DunedinPACE, pc Grimage) correlated with pace of aging at r = 0.34 and r = 0.20, but pace of aging had greater advantages in health outcome prediction and relative independence.
  • ROC curve results showed that for all health outcomes except chronic disease, pace of aging yielded the highest AUC.

5. Socioeconomic and Population Differences Found

  • Lower education, Black/Hispanic ethnicity, and male groups had significantly faster pace of aging, directly reflecting that social health inequality is manifested not just in end-of-life lifespan, but in the speed of functional decline.
  • These findings reveal that interventions aimed at the middle-aged and elderly hold practical value for narrowing group health differences.

5. Interpretation of Conclusions and Scientific/Application Value

1. Main Conclusions

  • The “adapted pace of aging” can serve as a novel, sensitive, dynamic biomarker-based aging measurement, suitable for large sample, multi-ethnic, and multi-socioeconomic background research on elderly health.
  • The indicator is closely associated with elderly health outcomes like mortality, chronic disease, disability, and cognitive impairment, and significantly outperforms traditional biological age and epigenetic clock metrics.
  • Pace of aging itself embodies pronounced social stratification and inequality, able not only to capture shared trends of aging societies but also to quantify the actual effect of intervention measures.
  • This method provides powerful, effective tool support for government policymakers, public health practitioners, and life science researchers in the evaluation and monitoring of healthy aging.

2. Major Innovations and Research Highlights

  • Adaptive measurement system: Does not rely on singular or costly “omics” data; only requires long-term blood/physical/functional testing for cross-country and cross-cohort comparability.
  • Dynamic predictive ability: Pioneers the introduction of “time-varying rates of biological decline” at the group/individual level into population health monitoring, gaining methodological advantages over static biological age.
  • Progress in quantifying social inequality: Based on biological metrics, it more accurately uncovers the deep links among social structure, behavior, and healthy aging trajectories.

3. Other Valuable Information

  • The study’s database and analysis code are fully open (GitHub link provided in the text), facilitating reproducibility and dissemination by the international community.
  • The innovative method is also applicable to dozens of major global population cohorts and policy needs in the future, and is likely to become a new standard for “healthy aging” research and evaluation.
  • Team members are inventors of DunedinPACE and other epigenetic clocks, with substantial cross-cohort research background and methodological innovation experience.

6. Outlook and Limitations

  • The frequency of sampling is relatively low (mostly three measurements); with more data, the model will support more refined nonlinear change exploration.
  • The proportion of the very old (over 80 years) is low; expansion of the sample is needed to test broader applicability.
  • Some biomarkers were adjusted for drug intervention and other factors, but the system still supports a comprehensive assessment of multisystem aging rate.

7. Summary

This study provides a solid theoretical and empirical foundation for the methodological system of population-level pace of aging measurement, addressing a key metric challenge in healthy aging research. The method not only suggests that health disparities in the elderly can be dynamically quantified and specifically intervened upon, but also provides an innovative tool for scientific decision-making in global healthy aging strategies and precision public health management. In the future, as cohort observation periods are extended and data further enriched, the pace of aging measurement method is expected to become an important cornerstone for measuring and promoting human healthy longevity.