Life Expectancy and Health Expectancy in the Twenty-first Century: The Unthinkable, the Inconceivable, and the Unknowable

Abstract

The last century witnessed an unprecedented rise in life expectancy; however, in recent decades the “unthinkable” has occurred—life expectancy stagnation, a dramatic drop in the U.S. international life expectancy ranking, rising midlife death rates, and widening socioeconomic and geographic disparities. The “inconceivable” has occurred with the high level of mortality from the COVID-19 pandemic in the United States, which further exacerbated racial, ethnic, and socioeconomic disparities and highlighted the vulnerabilities of long-term care systems and fragmented health policies. The “unknowable” future of mortality is explored through the lens of emerging work in geroscience based on an integration of biology with studies of aging populations, which offers some promise of potential interventions in the process of aging that underlies chronic disease resulting in mortality at older ages. However, transformative changes in social policy, health equity, behaviors, and legal rights are needed for the United States to improve its current situation. While the integration of biological understanding is likely to point to new avenues for improving population health and life expectancy, without immediate social changes, only a portion of the U.S. population is likely to be able to take advantage of these improvements, and the United States is likely to lag other countries in the level of life expectancy.

Introduction

Mortality has not been the focus of the Population Association of America presidential address since Evelyn Kitagawa’s address in 1977 (Kitagawa 1977). At that time, individual-level data were just beginning to be used by demographers, making the approach to studying mortality very different from that of today. Most research examined age, sex, and cause of death to highlight population differences and trends and to make projections. The substantive issues of interest were different then as there was not yet recognition of the dramatic reductions in mortality at older ages and from heart disease that were already underway; however, there was a focus on the adverse levels of infant mortality as well as social inequality in life expectancy, and these remain issues today (Singh and Yu 1995). Until recent years, the expected future for U.S. mortality generally assumed continued improvement, but this is no longer true. Improving life expectancy and health in the United States, as well as the relative situation of the country compared with the rest of the world, requires a reduction in social differentials in health and mortality, which in turn requires changes in social policy, health care access, behaviors, and laws in the United States. Scientific advances have put us on the verge of future interventions in the aging process, which underlies chronic disease mortality. Yet, without changes in social policy, such interventions will improve the health and mortality of only a portion of the population. The current decade is a time of great uncertainty about the future of health and mortality for the American population.

The Unthinkable

The most important accomplishment of the last century was an almost doubling of life expectancy as we experienced a steady increase from the 40s to almost 80 years (Oeppen and Vaupel 2002) as a result of the decrease in infectious conditions during the first half of the century and then, in an almost seamless transition, subsequent decades when mortality decline was driven by the decrease in mortality from chronic conditions, particularly heart disease. This led most demographers to expect continuous mortality decline and increase in life expectancy during the twenty-first century; the question was whether change would be relatively rapid or slower.

But the unthinkable has happened in recent decades—increases in life expectancy virtually slowed to a halt, even before the COVID-19 pandemic of 2020 (Beltran-Sanchez et al. 2015; Preston and Vierboom 2021). The peak U.S. life expectancy of 78.9 occurred in 2014; for three years, from 2015 to 2017, the U.S. population experienced small decreases in life expectancy. In 2018 and 2019, there were infinitesimal increases of 0.1 year, or about a month per year (Preston and Vierboom 2021). The years 2020 and 2021 saw life expectancy drop owing to the COVID-19 pandemic (Andrasfay and Goldman 2021a, 2021b; Woolf et al. 2022). The decrease from 78.86 years in 2019 to 76.99 years in 2020 represented a loss of 1.87 years in a single year (Woolf et al. 2022). By 2023, life expectancy had risen to 78.4, still lower than the peak. I use the word “unthinkable” because demographers did not expect this poor performance in a developed country, even though Russia had previously seen declines in life expectancy with social disruption; we never thought it would happen in the United States (Leon et al. 1997; Notzon et al. 1998).

Another unthinkable aspect of current trends is the poor ranking of the United States in life expectancy relative to other countries. This was the focus of a National Academy of Sciences report about 15 years ago that focused on the older population (Crimmins et al. 2011); this report was followed by a related report that focused on the population younger than 50 (Woolf and Aron 2013) and then another that focused on the working-age population (Harris et al. 2021). These reports focused on the low ranking and poor trajectory of mortality in the United States relative to OECD peers; but since these analyses were published, the relative situation of the United States has continued to deteriorate. Figure 1 is an update of a figure from Crimmins et al. (2011) showing U.S. life expectancy at birth relative to 22 OECD peers (the United States is the red dot and the other countries appear in gray). The United States is now below the group once considered our peers. In fact, our “peers” in life expectancy have changed: the 2023 life expectancy rankings show the United States at 43rd, below many countries formerly not considered our peers—including Singapore, South Korea, Israel, Slovenia, Qatar, Costa Rica, and Chile (United Nations 2024).

Fig. 1.

Life expectancy from 1980 to 2023 for 23 OECD countries. Source: Updated figure from Crimmins et al. (2011). Data source: United Nations (2024).

Why Is the United States in Such Relatively Bad Position?

Demographers usually begin to understand trends and differences in mortality by looking at causes of death. Differences in life expectancy between the United States and other countries are due to several causes: Fenelon et al. (2016) found that three causes—motor vehicle traffic accidents, drug poisoning, and firearm-related deaths—accounted for 50% of the differences among males and 19% of the differences among females between the United States and 12 other countries. High rates of death from traffic fatalities have been blamed on higher speeds, larger vehicles, more miles driven, less use of public transportation, and more impaired driving after drinking with less enforcement of laws.

Gun-related deaths have long been a negative influence on the relative ranking of the United States in international comparisons. Kalesan et al. (2019) estimated that national life expectancy loss from firearms was 2.48 years (with 0.98 from assault and 1.43 from suicide). These numbers clarify that firearm-related mortality should be considered a public health issue and addressed appropriately; the time has come to consider gun control when talking about maintaining and improving relative length of life in the United States.

Much has been written about the opioid epidemic as a source of reduction in American life expectancy (Barbieri 2018; Case and Deaton 2021; Harris et al. 2021). The opioid epidemic was not experienced in the same way in other countries, for two reasons. First, the availability and use of these drugs were manipulated by pharmaceutical companies in the United States and so their availability was much greater than in other countries (Ho 2019). Second, the “despair” resulting from reduced circumstances from the worldwide recession was less among young adults in other countries. While some European nations were more economically affected than the United States in the recession beginning in 2008, leading to unemployment rates among young adults in the 30% range in several countries, the recession in Europe appears not to have been accompanied by as much psychological distress and rise in mortality among the younger population from “deaths of despair” (Mackenbach et al. 2018). One hypothesis is that those whose livelihoods were threatened in Europe were more protected by multiple social programs. U.S. policies on the distribution of pharmaceuticals, as well as federal policies on social support for those with financial hardship, appear to have led to the much greater impact of difficult economic times on health and mortality.

The United States has long experienced some of the highest infant and maternal mortality rates among the well-off countries of the world, and the U.S. rankings continue to deteriorate (Chen et al. 2016; Collier and Molina 2019). Higher mortality levels are assumed to reflect the effects of poverty and a lack of health care utilization, as well as excessive use of cesarean sections in childbirth.

There are also differences between the United States and other countries in chronic disease mortality, which lowers relative life expectancy and contributes to adverse trends, most notably in the prevalence of diabetes and heart disease (Acosta et al. 2022; Crimmins 2021; Glei 2022). Poor health behaviors in the United States, including smoking and a rise in obesity, have also played a role in diabetes and heart disease trends, as has the country’s relatively poor health behavior overall.

Recent declines in U.S. life expectancy are largely due to increases in mortality among individuals younger than 60; before the pandemic, older individuals continued to improve in life expectancy, although at a slower rate than in the past (Barbieri 2019). The United States scores poorly in mortality at every age below old age—as was shown in the earlier National Academy of Sciences report (Woolf and Aron 2013) and also by Ho and Preston (2010). Figure 2 updates earlier data to show that among 22 countries, the United States ranks highest—22nd or 21st—in mortality at each age up to the old-old (80+); in the oldest age groups, the United States ranks near the top.

Fig. 2.

Among 22 OECD peer countries, in 2022 the United States ranks as worst or second to worst in mortality at ages up to old age. The countries include Australia, Austria, Belgium, Canada, Denmark, Finland, France, Greece, Iceland, Ireland, Italy, Japan, Luxembourg, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, and the United States. Source: Updated figure from Woolf and Aron (2013). Data source: Human Mortality Database (2025); Australia data are for 2021, Greece for 2019, and New Zealand for 2021.

Mortality in Subgroups of the Population

Most of the causes of death that have contributed to the adverse trends and poor relative status among Americans are more likely to occur among those of low socioeconomic status. Rising inequality in life expectancy is the other unthinkable trend in recent decades. The gain in life expectancy for individuals in the top 5% of the income distribution between 2001 and 2014 was 2.6 years, but the gain was less than two months for individuals in the bottom 5% (Chetty et al. 2016, 2017). The poor and uneducated are leading increasingly shorter lives relative to the better off and better educated in the United States (Hayward and Farina 2023). Figure 3 presents the change in life expectancy over almost three decades for the White population. With higher levels of education, increases in life expectancy occurred, but at the lowest level, we see a decrease among women and only a small increase among men.

Fig. 3.

Change in life expectancy at age 25, by education: 1990 to 2017. Sources: Sasson (2016) and Sasson and Hayward (2019).

While trends in mortality were worse for women, initially they were also worse for middle-aged and White individuals during the initial slowdown. After 1980 and before the pandemic, gains in White life expectancy were slower than those for African Americans (Figure 4). The unacceptable long-term race gap in life expectancy had been closing as a result of the earlier, more rapid decline in mortality among Black than White individuals, in part owing to the initial concentration of the opioid epidemic among Whites; however, the adverse trend in life expectancy has spread to other racial and ethnic groups (Woolf et al. 2018), and the pandemic eliminated much of the progress that had been made in reducing the racial gap in life expectancy.

Fig. 4.

Life expectancy at birth among Black and White males and females: 1980 to 2022. Source: https://www.cdc.gov/nchs/products/life_tables.htm.

Related to these growing socioeconomic differentials are growing geographic differentials (Dwyer-Lindgren et al. 2017). County life expectancy differed by as much as 25 years in 2019 (68.2 to 93.2), and only 85% of U.S. counties saw improvement between 2000 and 2019 (Sylte et al. 2025). Several scholars have suggested that increased focus on geographic differences in recent years is useful in pointing out the potential for targeted interventions (Monnat 2018; Montez et al. 2020; Montez et al. 2016).

In terms of trends in life expectancy, the United States is divided into two countries, and the same can be said for provision of health care, economic growth, poverty, and psychological well-being. Maps depicting lack of health insurance, unemployment, family disruption, obesity, smoking, and drug use would overlap significantly with maps of low life expectancy and high rates of mortality.

All countries have socioeconomic differences in health and mortality, but the U.S. differences are larger—about twice as large as in countries we once considered our peers (Crimmins et al. 2011). One should note, however, that there are differences between the United States and other countries in health even at high levels of socioeconomic status (Avendano et al. 2009). While it is literally unhealthy to be poor or uneducated in the United States, the well-off and highly educated do not do well in comparison with their counterparts in other countries.

The somewhat recent focus on deaths of despair has brought to our attention the link between psychological well-being and stress and mortality—something that many social scientists have been noting for decades but that has only recently been integrated into demographic analyses (Richmond-Rakerd et al. 2021; Thoits 2010). Demographers need to more fully incorporate stress and psychological determinants of health to better understand the future impacts of social differences for life expectancy. An analysis of job loss in U.S. men older than 50 showed that loss of a job was associated with a potential 10% increase in annual mortality rate (Michaud et al. 2016). Increasingly, the adverse psychological states associated with low socioeconomic status and the mediating role of psychological states in subsequent health and mortality are being documented (Kiecolt-Glaser et al. 2002; Mitchell et al. 2018).

Proximate causes of the recent declining trend in mortality among those younger than 60 are the increase in despair, the opioid epidemic, and poor health behaviors, along with lack of health care access or treatment for some conditions (Barbieri 2019). Root causes are worsening economic conditions for a segment of the population, characterized by a loss of jobs and a lack of income growth, related family stresses, and lack of health care and supportive social policies. All of these are reasons the United States has fallen behind other countries. Improving life expectancy by reducing deaths from the causes of mortality mentioned so far does not require scientific and medical advances, but rather changes in social, public health, legal, and health care policies, as well as changes in individual behavior (Avendano and Kawachi 2014; Mokdad et al. 2024).

The Inconceivable

One of the fundamental lessons from the field of demography was that the age of infectious conditions was over, that we had largely eliminated infectious deaths with public health procedures and economic development and now our focus was to be on chronic diseases. The potential for infectious diseases to play a role in life expectancy trends was always mentioned as a possibility, but usually as an after-thought. In 2020, the SARS COVID-19 pandemic showed how short-sighted this view was. The demographic perspective on the decline in infectious diseases never made a clear distinction between the ability to prevent and treat bacterial as compared with viral infections, which posed the potential for the resulting worldwide pandemic.

The American performance during the pandemic was worse than we could have imagined—more cases and deaths than in most other countries, especially among resource-rich countries (Campbell et al. 2022; Wang et al. 2022; Woolf et al. 2022). The country experienced one of the largest drops in life expectancy among high-income nations—1.5 years overall in 2020, compared with an average of 0.22 years across 21 peer countries (Woolf et al. 2022). The decline in life expectancy was much greater among racial and ethnic minorities, about 2.5 years for African Americans and 3.5 years for Hispanics (Andrasfay and Goldman 2021b). Racial and ethnic differences in the relative risk of death by age were striking (Figure 5). Blacks and Hispanics were overrepresented in deaths, particularly at ages younger than 85, making clear why African Americans and Hispanics had much greater decreases in life expectancy and also why years of life lost might be a better indicator than life expectancy at showing the differential devastation wrought by the pandemic (Arolas et al. 2021; Goldstein and Lee 2020; Luck et al. 2022; Preston and Vierboom 2021). People who were of lower SES, essential workers, and those who lived in more crowded circumstances were more likely to contract the disease, and although vaccination reduced overall risk, persistent racial and ethnic disparities—shaped by geography and structural factors such as workplace exposure and housing—revealed that vaccination alone could not eliminate the inequities of the disease (Lee et al. 2022). Once again, tremendous social differentials and overall death rates were highly affected by social policies (Jones et al. 2023); COVID was an acute demonstration of U.S. health policy failures even as the country demonstrated scientific prowess by rapidly developing vaccines.

Fig. 5.

Differences between the percentage of COVID-19 deaths and the percentage in the population distribution, by race and ethnicity, according to age group. Source: Deaths as of December 10, 2020. https://www.cdc.gov/coronavirus/2019-ncov/community/health-equity/racial-ethnic-disparities/disparities-deaths.html (data are no longer available).

Overall, age was the greatest predictor of death from COVID-19, with 81% of the deaths in the first year of the pandemic occurring among those aged 65+ (Crimmins 2020b). There are biological reasons why the old were more vulnerable: they are more likely to have senescence of the immune system, making it more difficult to fight off novel infections; and older persons are more likely to have dysregulated cytokines, which may have resulted in inflammatory storms that were the cause of many COVID deaths (Crimmins 2020b). Older persons and African Americans are also more likely to have one of the chronic diseases or conditions that made it harder to recover from COVID-19.

There were also social reasons why older persons died more frequently. What is most striking is the COVID death rate among the population living in long-term-care facilities. Less than 1% of the U.S. population lived in such facilities, but in the first year of the pandemic, they accounted for 34% of COVID-19-related deaths (COVID Tracking Project 2021). Nursing home residents could not socially distance from workers and were often in the presence of infected persons for prolonged periods. Infections in nursing home patients and workers were highly influenced by facility policies and the socioeconomic characteristics of workers and their jobs. Policy failures relative to COVID-19 hit the most vulnerable and least healthy segments of American society the hardest. The pandemic set back relative life expectancy even further compared with most other countries, and it exacerbated racial and socioeconomic differentials in life expectancy.

One must conclude that the United States’ low relative ranking internationally, a significant portion of the country’s adverse trends, and increasing socioeconomic inequality could be improved with substantive changes in health and social policies. Such changes would reduce the inequities in American society. However, the pandemic has made it clear that attitudes and behaviors influencing public health are now also political and not merely scientific. Current health and social policies do not appear to be changing in any way that would promote improved health and longer life expectancy in the near future.

The Unknowable: The Future of Health and Mortality

This now leads me to the unknowable—the future of life expectancy—where science as well as policy is playing a role. Until the last decade, demographers thought future trends in life expectancy would be determined by mortality at older ages from chronic diseases. I still believe this is the case, although the likelihood that the pandemic will reoccur or the possibility that we will continue to have worsening mortality from the causes of death that have become more important in middle age must be recognized.

Before the pandemic and the adverse trends in middle-aged mortality, demographers were generally modestly optimistic about the future of life expectancy, although some demographers and many biologists were wildly optimistic about the future (Christensen et al. 2009; de Grey and Rae 2008; Oeppen and Vaupel 2002; Sinclair and La Plante 2019). These optimistic projections included assertions that more than half of children born today will live to be 100—or that the first person to live to 150 has been born. These were highly unlikely outcomes even before recent trends.

I think the best evidence undermining such projections is the slow historical improvement in life expectancy at the highest ages. In the last 55 years, life expectancy at age 90 in the United States has increased by only about 1.13 years, or about one week per year. During this time, the likelihood of getting old has increased and the likelihood of then dying at an older age among the old has increased, but the likelihood of making it to a very old age—above 100—remains quite low. I believe this will remain the case unless we have major scientific breakthroughs—on the order of treatment of bacterial infections—that affect the onset and progression of aging itself.

The most important recent trend for evaluating the future of life expectancy may be the slowdown in the decline in heart disease mortality. Heart disease remains the most important cause of all mortality and the most important cause of the decline that occurred over the past 50 years. The annual reduction in heart disease mortality in the United States went from very fast—almost 4% decline a year from 2000 to 2010—to increasing at a rate of 0.4% a year in 2013–2016 (Ma et al. 2015). Comparison with other OECD countries shows that reductions in the cardiovascular mortality decline have occurred in the entire group in recent years, but the United States sank to the bottom of the pack with the worst performance in the most recent period (Figure 6). So, the United States is experiencing a similar trend as other countries but is exceptional in how the rate of improvement has faltered.

Fig. 6.

Average annual percentage decrease in age-standardized death rate from cardiovascular diseases: 1990–2000, 2000–2010, and 2010–2019, in 23 OECD countries. The countries include Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Japan, Luxembourg, the Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, the United Kingdom, and the United States. Source: https://ourworldindata.org/grapher/age-standardized-death-rate-cardiovascular-disease?time=earliest..latest.

This slowdown in heart disease mortality decline has resulted in a reduction in the rate of improvement in life expectancy in all of these countries. In 20 of 23 OECD countries, the six years from 2011 to 2017 have seen less increase in life expectancy than the earlier six-year period; the United States has performed the worst but others are shifting in the same direction (Crimmins 2021).

There are a number of potential explanations of this slowdown in the decrease of cardiovascular disease around the world. We may have little potential remaining for decline; since 1970, heart disease deaths are down by about two thirds. Some of the medical advances that fueled the decline may be losing their effect on the overall trend because they have played out. For instance, interventions such as the use of antihypertensives and statins have become increasingly widespread. Many individuals with high levels of risk have been successfully medicated, leaving little room for improvement.

Changes in smoking may no longer have as much influence on the U.S. trend, and the different timing of similar changes in other countries may be affecting their mortality declines. The rising prevalence of obesity may be related to deteriorating trends in many countries and may be a source of adverse effects in the United States relative to other countries. Although most nations are experiencing increases in the level of obesity, the United States leads in this trend.

In addition, the role of early-life factors and past experiences with infectious disease may have been a powerful force in reducing old age mortality, particularly from heart disease. In countries like the United States, these forces leading to mortality decline among successive cohorts have largely played out (Finch and Crimmins 2004). The cohorts now reaching old age in developed countries never had much infectious disease exposure, which may have resulted in better physical and mental development for physiological systems.

As mortality became increasingly concentrated at older ages and in deaths from chronic conditions, a question arose as to whether life expectancy was the best metric for assessing population health or whether we should focus on health expectancy or the length of healthy life. In contrast to an earlier epidemiological time when life expectancy was rising because of the prevention or cure of infectious conditions, in a chronic disease regime, life expectancy is often increasing not because diseases have been cured or prevented but because death has been prevented. This shift has led demographic analysis to focus on additional dimensions of health and their links to mortality or health expectancy, as well as life expectancy. This newfound focus also reflects the increasing availability of population data on various aspects of health available for demographic analysis.

The Morbidity Process

Health, or its absence—morbidity—has multiple dimensions that I have called the morbidity process (Crimmins 2015). In aging populations, changes with age occur first in biological aging, which includes molecular and cellular change and then physiological dysregulation, proceeding to disease onset, then disability, and then mortality. Behavioral and medical interventions can affect the process of change at the onset of each dimension, in death rates for any dimension and in the links between the dimensions.

Trends in the length of life with disability and diseases provide some insight into what has been happening to these dimensions of population health. In general, there has not been much change in the overall age-specific prevalence of disability from 1970 to 2010 (Crimmins et al. 2019; Crimmins et al. 2016). There were small changes in some types of disability and in some age groups, but overall not large change. The lengths of life with and without disability have increased in sync, maintaining the relative length of life with disability over this time.

On the other hand, the prevalence of many chronic diseases has increased, largely because mortality decreased as we saved people with diseases from dying but did not prevent the onset of the diseases (Crimmins et al. 2016). Figure 7 shows change over the 20 years from 1998 to 2018 in the number of years after age 65 lived with and without each of six diseases for older American men. On average, more years were lived with disease in the later year for hypertension, diabetes, cancer, heart disease, and arthritis; years lived after stroke did not increase significantly. It is also true that, on average, we are living with more diseases: in 1998, U.S. men over age 50 had an average of 1.1 of these six conditions, compared with about 1.5 conditions in 2018. We appear to have slowed the progression from disease to disability and perhaps reduced the severity of disease; however, in not delaying the onset of disease, we have not reduced the time spent with disease or the number of diseases.

Fig. 7.

Years lived with and without disease after age 65, among U.S. males. Key colors apply to both years, but the 1998 bars have lighter shading. Source: Health and Retirement Study (https://hrs.isr.umich.edu/about).

Realization of this situation is one of the sources of the current geroscience focus on understanding the aging processes at biological levels that precede diagnosed disease and disability (Kennedy et al. 2014). Gaining an understanding of how aging processes begin early in life and are influenced by social, economic, and psychological exposures from the earliest ages is an additional reason to redirect our focus to changes that occur before the onset of disease and disability (Raffington et al. 2021). The desire to increase the health span and reduce social differences by intervening early in life has been an incentive for social scientists to join with biologists to employ indicators of the molecular and cellular mechanisms by which life circumstances “get under the skin” to cause health change with age.

So for both social scientists and biologists, there is a new focus on “aging” per se. The argument is that the aging process itself sets in motion the changes that result in disease, disability, and mortality, and the only way to make significant improvements in both health span and life span is to directly affect aging rather than to try to treat or prevent one disease at a time. Over the last couple of decades, I and many of my colleagues have been focused on introducing the ability to study biological aging in large populations. In her presidential address 15 years ago, Kathleen Mullan Harris argued for an integrated biological, psychological, and social approach to health (Harris 2010). Since that time, extensive progress has been made in integrating biological data with social data in large population studies (Crimmins et al. 2025; Freilich et al. 2024; Kivimäki et al. 2025; Ravi et al. 2024) to clarify the multisystem components of physiological change, as well as some basic cellular and molecular mechanisms that encourage the progression of morbidity, disability, and mortality, and so play a role in population disparities (Crimmins 2020a; Kivimäki et al. 2025). Population studies now routinely include complete genome-wide association studies (or GWAS), as well as indicators of DNA methylation and the transcriptome. Indicators of the microbiome, the proteome, and the metabolome are starting to be included. This shift has resulted from scientific progress in understanding the underlying factors affecting disease and aging and technical progress in measurement and computing ability that has allowed the integration of these biological factors into population studies. Many of the now recognized molecular and cellular changes associated with aging, as well as many indicators of physiological dysregulation, are likely to be treated in the coming decades just as high cholesterol, hypertension, and adverse glucose levels are now. A number of mechanisms are currently being tested in clinical trials (Sehgal et al. 2024), which leaves us poised to actually intervene to delay the aging process, including the onset of disease and disability, resulting in longer health and life expectancy.

I will provide just three examples of biological measures with large social differences that are derived from data on individuals aged 56 or older in the Health and Retirement Study: measures based on indicators of physiological dysregulation, epigenetics, and transcriptomics. All three of these measures have been developed to characterize biological age relative to chronological age to identify individuals who are biologically older than expected in the three dimensions—call them fast agers—and those who are aging slower than expected. The indicator of physiological dysregulation is based on 22 indicators, including markers of cardiovascular, metabolic, immune, inflammation, renal, kidney, and lung functioning, as well as hematological measures (Crimmins et al. 2021). The epigenetic measures are based on DNA methylation, which is sometimes characterized as indicating which genes are turned on or off. Epigenetic age acceleration is an example of a measure at the molecular level that indicates methylation changes in the genome occurring in relation to life circumstances and aging itself (Belsky et al. 2022; Crimmins et al. 2024; Levine et al. 2018; Lu et al. 2019). The recently developed transcriptomic measure is called TraMA and was developed to predict mortality (Klopack et al. 2025). Transcriptomic measures reflect gene expression more directly.

Differences in accelerated physiological dysregulation by education, race, and ethnicity are shown in Figure 8 (panel a). People with the lowest education are about 1.7 years older physiologically than chronologically, and people with higher education are about 2.2 years younger, resulting in a four-year difference between high and low education (p < .0001). Non-Hispanic Black individuals are about 1.9 years older at a given age than non-Hispanic Whites (p < .0001).

Fig. 8.

Accelerated biological age among individuals aged 56 or older, by education and by race and ethnicity. (a) Years of accelerated expanded biological age are based on 22 markers of physiological dysregulation. (b) Accelerated epigenetic age uses an epigenetic factor score based on GrimAge, PhenoAge, and DunedinPACE, measured in standard deviation units. (c) Years of accelerated transcriptomic age (TraMA).

Panel b of Figure 8 shows accelerated epigenetic aging based on a factor score combining three widely used epigenetic acceleration measures: GrimAge, PhenoAge, and DunedinPACE. The patterns are consistent with those based on physiological dysregulation: individuals with the least education are 0.27 standard deviations (SD) above the age-specific mean, indicating accelerated epigenetic age, while those with 16+ years of education are 0.39 SD below the mean. This represents a 0.66-SD difference or just over 2 years of difference between the lowest and highest education groups (p < .0001). Racial and ethnic disparities are again apparent, with non-Hispanic Black individuals showing the most epigenetic age acceleration—a 0.43-SD shift relative to non-Hispanic Whites (p < .0001).

As shown in panel c of Figure 8, the educational gradient in TraMA acceleration is also steep: individuals with 11 or fewer years of education are nearly 1.9 years older biologically than their chronological age, while those with 16+ years are about 2 years younger (p < .0001). Racial and ethnic disparities are also pronounced in transcriptomic age. Non-Hispanic Black individuals are more than 3 years older biologically relative to non-Hispanic Whites (p < .0001), while Hispanics show modest age acceleration. These three measures are all part of the aging process, so it is not surprising that they are related to each other (with correlations among them ranging from .5 to .6).

To illustrate the potential for increasing our understanding of the process of health change leading to mortality and health differentials, we indicate the relative importance of these three biological measures in explaining the variance in subsequent mortality over six years. We also include measures of multimorbidty and disability, so indicators of the entire process are in the equations. Mortality was regressed on all five health measures, age, sex, race and ethnicity, and education. The relative importance of each variable, based on a Shapley decomposition, is shown in Figure 9 (see details in the online supplementary material). Among the indicators of health or morbidity, transcriptomic age and physiological dysregulation explain the largest amount of variance in mortality (R² = .04 each), followed by functioning loss (R² = .03) and the epigenetic factor and multimorbidity (R² = .02). With the inclusion of the measures of health and the biological measures, race and ethnicity, education, and sex are no longer significant, leading to the conclusion that the process of health change leading to differentials in mortality by these characteristics is accounted for by these differences in biological and health measures.

Fig. 9.

Variability (proportion of R² attributed to each category of measures) in subsequent mortality attributed to age, sex, race/ethnicity, education, accelerated transcriptomic age, accelerated epigenetic age, accelerated physiological dysregulation, multimorbidity, and functioning problems

This is just one example of how the integration of biological measures into population data allows the assessment of the relative importance of risk factors—social, environmental, and biological. Numerous clinical trials are currently underway to clarify the efficacy of the treatment of “aging” by targeting these measures with the intent to improve the number of healthy years, which would result in both increasing life expectancy overall and decreasing social differentials in health (Sehgal et al. 2024).

Conclusion

After a century of rapidly increasing life expectancy, the past few decades have seen mixed trends in health as well as mortality—with a lengthening of the morbidity process and a slowing of progression from one stage to the next. The causes of death responsible for both adverse trends and the poor U.S. performance relative to other countries indicate the importance of the social, psychological, and policy determinants of health and mortality.

Although demographers have increasingly integrated social and psychological determinants of health into individual models of health outcomes and mortality, these factors have largely not been integrated as determinants of population trends. There needs to be greater integration of our understanding of individual life circumstances, behaviors, and policy in forecasting future population trends. Recently, we missed the importance of the growing socioeconomic differentials in health and mortality and the increasing role of psychological factors as a cause of mortality trends. We have tended to think health and mortality can only be bettered with scientific advances, akin to the revolution in antibiotics. Past projections for life expectancy did not assess the social, psychological, and health costs of having a fragmented system of social support, behavior that is health damaging, and a safety net that does not include universal health care on national trends in health and mortality, which are proving lethal at ages up to old age. However, a recent projection based on the Global Burden of Disease Study 2021 took these trends into account and projected a U.S. life expectancy of 80.4 years in 2050, with a continued decline in relative ranking (Mokdad et al. 2024; Rysava and Thompson 2024). If this projection is accurate, it would put the United States about 50 years behind Japan in life expectancy level. This is clearly not acceptable for the country that spends the most on health care in the world.

What is happening at older ages is scientifically remarkable and offers the promise of a better future with longer life for those who reach old age. This potential has come about because of the scientific progress resulting from the support of the U.S. National Institutes of Health for research over many decades. Whether improvement in the current state of U.S. health and life expectancy will be achieved will depend on factors largely outside the purview of science.

Supplementary Material

ELECTRONIC SUPPLEMENTARY MATERIAL The online version of this article (https://doi.org/10.1215/00703370-12185960) contains supplementary material.