1

Introduction

The past century has witnessed remarkable advances in life expectancy, and the United States has shared in that progress. In 2010, however, progress in life expectancy in the United States began to stall despite continuing to increase in other industrialized countries. Alarmingly, U.S. life expectancy fell between 2014 and 2015 (Arias and Xu, 2018) and continued to decrease in the two subsequent years (Arias and Xu, 2018, 2019; Arias, Xu, and Kochanek, 2019).¹ This 3-year period of declining life expectancy represented the longest sustained decline in the United States in a century (since the influenza pandemic of 1918–1919).

The stalling and subsequent decline in life expectancy during the 2010s appears to have been the product of an increase in mortality among middle-age and younger adults, defined as those ages 25–64 (“working age”), which began in the early 2010s. Between 2010 and 2017, the age-adjusted all-cause mortality in this age group increased by 6.0 percent (from 328.5 deaths to 348.2 deaths per 100,000 population) (Centers for Disease Control and Prevention [CDC], 2020b). By contrast, infant mortality decreased by 9.0 percent, and child mortality remained unchanged. Mortality among older adults (ages 65 and over) also continued to decrease over this period (CDC, 2020b).

Explaining why mortality has increased among working-age adults is not straightforward. Mortality is the end result of both acute events and

___________________

¹ U.S. life expectancy increased slightly in 2018 (Xu et al., 2020) and 2019 (Andrasfay and Goldman, 2021) ahead of the COVID-19 pandemic of 2020, which is expected to reduce life expectancy once again.

cumulative, long-term processes involving the interaction of biological, behavioral, social, economic, and environmental factors that develop and unfold over the life course. Establishing the complex relationships among these explanatory factors poses methodological challenges that are complicated by issues of data quality and measurement. Research initially indicated that the phenomenon of rising working-age mortality was occurring primarily among middle-age non-Hispanic Whites (Whites) (Case and Deaton, 2015), particularly women (Gelman and Auerbach, 2016) with a high school degree or less (Case and Deaton, 2015, 2017) and in rural America (Erwin, 2017). In fact, mortality appears to have begun rising in the 1990s among both White and non-Hispanic Black (Black) women with a high school degree or less (Montez et al., 2011); however, research using more recent data suggests that the trend now involves men and most racial/ethnic groups (Curtin and Arias, 2019; Woolf et al., 2018) and has expanded beyond rural America to more populated areas (Elo et al., 2019).

In some ways, these trends have exacerbated long-standing disparities in mortality, such as those between individuals of high and low socioeconomic status (Chetty, Hendren, and Katz, 2016; Cutler, Meara, and Richards-Shubik, 2011; Meara, Richards, and Cutler, 2008; Sasson, 2016) and between urban and rural areas (Monnat, 2020b; Singh and Siahpush, 2014; Vierboom, Preston, and Hendi, 2019). These trends also reflect stagnation in and even reversal of the steady progress made over the previous decades in reducing racial disparities in mortality faced by Black adults (Curtin and Arias, 2019). These effects on disparities in mortality are important in their own right, signaling the effects of deepening inequality across many facets of American life, but also because they help elucidate the underlying processes that may have generated the recent increases in working-age mortality.

Initial studies examining the specific underlying cause or causes of death responsible for increases in mortality focused predominantly on older working-age adults, those ages 45–54. These studies highlighted increases in fatal drug overdoses as a primary factor in rising mortality and found that these increases were driven largely by the less educated, particularly Whites with a high school education or less, for whom mortality had risen sharply since 1999 (Case and Deaton, 2015, 2017). However, this research also pointed to increases in mortality due to alcoholic liver disease and suicide among middle-age adults.

The striking rise in deaths from these three causes—drug poisoning, alcohol-related disease, and suicides—led some researchers to label these causes of death collectively as “deaths of despair” (Case and Deaton, 2017). However, more recent research (Dwyer-Lindgren et al., 2017; Geronimus et al., 2019) shows that mortality rates among working-age adults also increased within a broader age range than 45–54 and for a broad spectrum

of diseases involving multiple body systems (e.g., circulatory, digestive, pulmonary, neurologic, endocrine, and cardiovascular). Neither drugs nor “despair” could fully explain so diverse a phenomenon.

The decline in U.S. life expectancy received national news coverage (e.g., Bernstein, 2016, 2018; Rogers, 2016), as did research identifying drugs, alcohol, and suicide as the causes of death driving the decline (e.g., Achenbach and Keating, 2017; Bernstein and Achenbach, 2015). Despite research implicating a broad range of causes of death in rising mortality rates, media reports focused on the opioid epidemic and its impact on specific demographic groups and communities (e.g., Egan, 2018; Robertson and Trent, 2018). These reports were often set in rural or Rust Belt communities and featured stories of largely White families. The accounts often depicted how residents were affected by the job losses and economic instability resulting from the collapse of manufacturing, mining, and other industries that once provided secure and living-wage jobs to those without a college degree. While this narrative was reflective of early trends in drug overdoses due to prescription opioids, which began to increase in the 1990s primarily among working-age Whites with less than a college degree (Alexander, Kiang, and Barbieri, 2018; Ho, 2017), the rise in drug poisoning was not limited to this population. Although communities in Appalachia, New England, New Mexico, and Utah began experiencing rapid increases in opioid-related mortality in the 1990s (Case and Deaton, 2017; Rigg, Monnat, and Chavez, 2018), these increases were geographically heterogeneous, affecting both metropolitan and nonmetropolitan areas in regionally specific ways (Peters et al., 2020; Rigg, Monnat, and Chavez, 2018). Moreover, although opioid overdoses began noticeably increasing earlier among Whites, these increases were also experienced by Blacks, as well as by American Indian and Alaska Native (AI/AN) populations (Alexander, Kiang, and Barbieri, 2018; Tipps, Buzzard, and McDougall, 2018). Because of the popular focus on predominantly working-class Whites, however, the experiences of racial/ethnic minorities during the overdose epidemic have gone largely ignored.

Moreover, although the media focus on communities marked by growing opioid-related mortality has highlighted the important role of the opioid epidemic in increasing mortality among working-age adults, it has also obscured the broader range of causes of death that have contributed to these mortality increases (Geronimus et al., 2019) and the range of populations affected (Curtin and Arias, 2019; Elo et al., 2019; Woolf and Schoomaker, 2019). The affected age groups also are broader than first reported. Studies initially focused on middle-age adults (ages 45–54), but the data show that mortality rates also increased significantly among younger working-age adults (ages 25–44) between 2010 and 2017 (Curtin and Arias, 2019). Among younger adults, cause-specific mortality, particularly for drug

poisoning, has often increased at a faster pace compared with middle-age adults, and suicide rates in young people have increased as well. Indeed, the phenomenon may be extending into adolescence: the data suggest a recent increase in mortality among those ages 15–24, driven by drug overdoses and suicide (Ali et al., 2019). At the same time that mortality rates among White adults began increasing, the rates among middle-age and younger AI/AN adults were also increasing—at even higher rates (Tipps, Buzzard, and McDougall, 2018; Woolf et al., 2018). And between 2010 and 2017, all-cause mortality rates among working-age Blacks and Hispanics increased as well (Curtin and Arias, 2019).

INDIVIDUAL AND SOCIETAL IMPLICATIONS OF RISING MORTALITY AMONG WORKING-AGE ADULTS

The significance of these ominous trends for the country cannot be overstated. Rising mortality among working-age adults is a population health crisis. The premature death of tens of thousands of Americans in the prime of their lives has profound ripple effects on the well-being of families and the social fabric of communities for generations to come. It affects an age span that encompasses the American workforce, impacting the productivity and competitiveness of U.S. businesses, the economy, and national defense. The health conditions driving these mortality increases are adding to escalating health care costs, posing an unsustainable burden not only on government payers (e.g., Medicaid and Medicare) but also on employers, threatening the U.S. position in the global marketplace (Chernew, Hirth, and Cutler, 2009; Kaiser Family Foundation, 2019; Nunn, Parsons, and Shambaugh, 2020; Office of the Assistant Secretary for Planning and Evaluation [ASPE], 2005; Sood, Ghosh, and Escarce, 2009). And military and national security experts also have raised concerns about these trends (Congressional Budget Office, 2017; Keith, 2011; Riley, 2010).

Rising health care costs are an ongoing concern in America that has been exacerbated by mortality increases and their underlying causes. Per capita national health care expenditures increased more than 220 percent between 2000 and 2017.² While the direct effect of mortality on health care spending likely stems from the high cost of end-of-life care, the increasing burden of health care spending is associated mainly with chronic conditions that develop earlier in life among working-age younger adults and lead to earlier mortality (Einav et al., 2018; French et al., 2017). The Substance Abuse and Mental Health Services Administration (SAMHSA) projects that by 2020, spending on substance abuse and mental health treatment

___________________

² According to the Centers for Medicare & Medicaid Services’ National Health Expenditures (NHE) data, per capita NHE totaled $4,855 in 2000 and $10,739 in 2017.

will total $280.5 billion, a 63 percent increase relative to 2009 (Substance Abuse and Mental Health Services Administration [SAMHSA], 2014). And an analysis by the Urban Institute found that Medicaid spending on three medications used to treat opioid use disorder and overdose increased 136 percent between 2011 and 2016 (Clemans-Cope, Epstein, and Kenney, 2017). Health care spending for these health issues falls on state budgets, largely through Medicaid, and crowds out other important priorities, such as education (Chernew, Hirth, and Cutler, 2009; Medicaid and CHIP Payment and Access Commission [MACPAC], 2016; Rosewicz, Theal, and Ascanio, 2020).

Higher health care spending associated with the causes of working-age mortality also affects the commercial insurance market. Because health insurance pools risk across the population, the costs associated with early death and associated illness for individuals with employer-sponsored coverage are borne by their coworkers. For example, higher insurance premiums reduce wage growth for all workers (Burtless and Milusheva, 2013; Clemens and Cutler, 2014; Kolstad and Kowalski, 2016). Similarly, as premiums rise, employers reduce the generosity of coverage, and the added financial risk is imposed on all workers (Anand, 2017; Kaiser Family Foundation, 2019). In some cases, higher premiums induce employers to drop insurance coverage altogether.

Deteriorating health among working-age Americans will extend the disability rolls to segments of the adult population that heretofore have been disability free (Chen and Sloan, 2015). Transitions to disability status result in declines in earnings, income, and food and housing consumption, especially impacting those with less education (Cutler, Meara, and Richards-Shubik, 2011; Meyer and Mok, 2019; Prinz et al., 2018). Increases in longer-term opioid prescribing may only exacerbate this trend; some estimates using workers’ compensation claims data suggest that such prescriptions roughly triple the duration of temporary disability benefits (Savych, Neumark, and Lea, 2019). Similarly, individual-level survey data on drug use and increases in its intensity are significantly correlated with criminal justice system involvement (Winkelman, Chang, and Binswanger, 2018). Even before the height of the opioid epidemic, cost estimates based on these data suggested that nonmedical use of prescription opioids was responsible for $8.2 billion in criminal justice costs (Hansen et al., 2011). Incarceration also hinders treatment for and recovery from drug use disorders. One study found that only 4.6 percent of justice-referred people received treatment for opioid use, compared with 40.9 percent of other clients (Krawczyk et al., 2017), with significant impacts on individuals and families (see, e.g., National Research Council [NRC], 2014).

Families are especially affected by rising mortality among working-age adults and the conditions leading to those deaths. Children who experience

the suicide of a family member, for example, are four times more likely to commit suicide in their lifetime (Burke et al., 2010). Given that suicide attempts are often related to underlying mental health problems, other studies suggest that children of parents dealing with mental health issues may develop their own issues with depression and suicide ideation later in life (King et al., 2010; Lunde et al., 2018). Similar effects are likely among individuals connected to someone who commits suicide but is outside of the immediate family (Cerel et al., 2016). Beyond suicide, more than one-third of children in foster care were found to have been placed because of parental substance use (Adoption and Foster Care Analysis and Reporting System [AFCARS], 2017), and every 15 minutes a baby is born with exposure to opioids during pregnancy (National Institute on Drug Abuse [NIDA], 2019).

Finally, the economic effects of rising mortality among working-age adults are both direct and indirect. The nation loses valuable workers, their output, and taxes to premature mortality. Moreover, impaired health leads to lower productivity (Currie and Madrian, 1999; Prinz et al., 2018) and increased hospitalizations, which in turn increase unemployment and reduce earnings (Dobkin et al., 2018), and these consequences are amplified for those who lack insurance or lose insurance from their employer. The economic costs due to opioid use are especially illustrative. In 2006, 79 percent of the total cost of nonmedical use of prescription opioids ($53.4 billion) was attributable to lost productivity, primarily through unemployment and subemployment due to opioid abuse (Hansen et al., 2011). Recent data show similar impacts of opioid use and pain management on labor force participation (Harris et al., 2019a), with 40 percent of men ages 25–54 reporting that pain prevented them from working a full-time job (Krueger, 2017). Economic costs will be seen for other causes of premature mortality as well. Indirect costs of lost productivity due to cardiovascular disease are estimated to increase by 55 percent over the coming decades, from $237 billion in 2015 to $368 billion³ in 2035 (Khavjou, Phelps, and Leib, 2016), as metabolic disorders begin to affect the previously healthy years of working-age adults.

Most troubling is the pervasive nature of the rise in mortality, now affecting a broad range of working-age adults in all racial/ethnic groups and in multiple geographic areas of the United States. The rise in mortality could threaten the well-being of individuals, families, health care, criminal justice, and economic systems and the social fabric of communities. Perhaps the greatest threat is posed by not knowing the underlying causes of these mortality trends. As a result, today’s children, whose parents are dying in middle age or as younger adults, may themselves face the insidious causes

___________________

³ Measured in 2015 dollars.

of this phenomenon, carrying the crisis of rising mortality forward into the next generation. Explaining these trends is therefore of paramount importance.

SEARCHING FOR EXPLANATIONS

The increase in mortality rates across multiple conditions among those of working age—but not the very old—precludes easy explanations. A full understanding requires focusing not just on those factors that are most proximate to the specific causes of death (e.g., behavior, psychological factors, health care utilization), but also upstream on the macrostructural causes (e.g., public policies, macroeconomic trends, social and economic inequality) that may affect the health of Americans in multiple ways and through multiple pathways that flow through local community contexts and intersect with the lives of individuals.

Failures in the health care system are conspicuous in the United States and may affect outcomes for multiple conditions. Unlike other industrialized countries, the United States lacks universal access to health care (owing to a lack of health insurance and shortages of providers and facilities in many communities),⁴ relying on a fragmented care delivery system characterized by large disparities in the quality of care and the incidence of medical errors. A growing number of studies have demonstrated that acquisition of health insurance coverage can lead to significant reductions in mortality (Borgschulte and Vogler, 2019; Goldin, Lurie, and McCubbin, 2021; Miller et al., 2019). However, studies suggest that only 10–20 percent of premature deaths are attributable to health care (Kaplan and Milstein, 2019) and that only 13 percent of the improvement in life expectancy since 1990 can be attributed to improvements in medical care, excluding new pharmaceuticals (Buxbaum et al., 2020). Deficiencies in health care are therefore unlikely to fully explain the trend.

Health behaviors, such as tobacco use and sedentary lifestyles, are major influences on health and mortality (Choi et al., 2020; Ford et al., 2012; Franzon et al., 2015; Tencza, Stokes, and Preston, 2014). Moreover, caloric intake in the United States exceeds that in other industrialized countries (Bleich et al., 2008). These unhealthy behaviors may explain an increase in tobacco- or obesity-related deaths, and substance use (problem

___________________

⁴ For example, the United States has fewer physicians, hospitals, and acute care beds per capita than comparable countries (Kamal et al., 2020). The Health Resources and Services Administration maintains a list of U.S. areas that have been designated as Health Provider Shortage Areas and Medically Underserved Areas. See more information here: https://bhw.hrsa.gov/workforce-shortage-areas/shortage-designation#hpsas.

drinking, misuse of analgesics) is a proximate cause of the rise in deaths from alcohol and drugs.

However, extensive research documents that health behaviors (and access to health care) are shaped by people’s environments. Healthy eating, for example, requires access to affordable and nutritious foods, while exercise requires a built environment that is conducive to regular physical activity (MacDonald et al., 2010). Whereas other countries have regulations against advertising junk food to children, food companies in the United States are allowed to market highly processed and high-sugar foods to children aggressively, add sugar to everyday products, and lobby policy makers to subsidize sugar production and promote consumption of high-sugar foods (Freudenberg, 2014). Socioeconomic status, such as educational attainment, income, and wealth, in turn determines the ability to live in a healthy neighborhood, maintain healthy behaviors, and afford health care (Pampel, Krueger, and Denney, 2010). In addition, the social environment, from families to communities, affects health: family and social networks can enhance health and promote resilience (Yang et al., 2016), or they can harm health by exposing people to social isolation, poor health behaviors, dysfunction, trauma, violence, racism, or other forms of discrimination.

Individuals’ status and behaviors and the families and communities in which they live are shaped by macrostructural factors, including public policies (from federal legislation to local zoning laws); economic trends (e.g., decline of manufacturing, rising income inequality); shifting demographics (e.g., immigration, rural outmigration); and business decisions, from how much to pay workers to the marketing of inexpensive calorie-dense foods or highly addictive prescription opioids. These macrostructural factors, in turn, reflect social and cultural values, such as the proper role and size of government, attitudes about social inequality, beliefs in individualism, structural racism, and other forces that shape public policies and spending priorities.

A full understanding of why mortality rates at working ages have increased since the 1990s therefore requires careful study that gives adequate attention to these complex multilevel influences and how they have changed over time, and to the evidence that such changes may have produced the epidemiologic trends occurring today. It is necessary as well to take a life-course perspective—understanding that health evolves over one’s lifespan in a cumulative and interactive fashion, such that events occurring in one life stage shape developmental and health trajectories in subsequent stages. A growing body of research documents the importance of the first 1,000 days of life in shaping growth and development (Crump and Howell, 2019). Exposure to adverse childhood events predicts not only the diseases of young people but also the probability of developing chronic diseases in old age by setting trajectories of risk or resilience in the subsequent life

stages of adolescence, young adulthood, and midlife (Brown et al., 2009; Felitti et al., 1998). This means that the study of rising mortality rates among working-age adults requires consideration of what occurred among the cohorts of Americans born 25–64 years before the trend began and comparing their experiences with those of prior cohorts over time.

CHARGE TO THE STUDY COMMITTEE

In this context, in 2018, the National Institute on Aging and the Robert Wood Johnson Foundation asked the National Academies of Sciences, Engineering, and Medicine to carry out a consensus study on rising rates of midlife mortality and associated socioeconomic disparities. The specific charge to the National Academies is as follows:

To conduct this study, the National Academies appointed the Committee on Rising Midlife Mortality Rates and Socioeconomic Disparities. The committee’s membership included 12 prominent scholars representing a broad range of disciplines—demography, economics, epidemiology, medicine, public health, sociology, and statistics. The committee met six times in person over a 2-year period to complete the study and produce this report.

IMPACT OF THE COVID-19 PANDEMIC ON WORKING-AGE MORTALITY

In 2020, as the committee worked to finalize this report, the COVID-19 pandemic began its spread across the world. Between March 2020 and the start of 2021—when this report went into production—COVID-19 grew from a rare disease in the United States to a leading cause of death (CDC, 2021; Woolf, Chapman, and Lee, 2021). Although all countries were affected by the pandemic, no country suffered as many deaths as did the United States (Bilinski and Emanuel, 2020). As of January 19, 2021, more than 400,000 COVID-19–related deaths, about one-fifth of the global total, had occurred in the United States (Johns Hopkins University, 2021), and the daily death toll from COVID-19 had surpassed the toll for heart disease

and cancer, the nation’s two leading causes of death (Woolf, Chapman, and Lee, 2021).

Moreover, studies of preliminary vital statistics data indicated that excess deaths—the number of deaths beyond what would have been expected without the COVID-19 pandemic—increased by approximately 20 percent in the United States during this period. However, COVID-19 accounted for only about two-thirds of these excess deaths (Rossen et al., 2020; Weinberger et al., 2020; Woolf et al., 2020). Especially during surges, the nation and individual states experienced sharp increases in deaths from other causes, such as heart disease, Alzheimer’s disease, and diabetes (Woolf et al., 2020). Researchers were unable to determine in real time the extent to which excess deaths overall as well as the observed increases in non-COVID-19 deaths occurred among infected patients whose death certificates omitted mention of the virus or uninfected patients who experienced death caused indirectly by disruptions resulting from the pandemic (e.g., inability to access acute emergency services). These deaths also could include deaths due to causes, such as drug overdoses, exacerbated by the pandemic. There is increasing evidence that the pandemic led to increased consumption of alcohol and drugs, including benzodiazepines (antianxiety medications), and in December 2020, the CDC issued a health advisory about the increasing risk of deaths due to these agents during the pandemic (CDC, Emergency Preparedness and Response, Health Advisory, December 17, 2020). While the reasons for this association are not entirely clear, it has been suggested that traumatic stress related to the pandemic in many populations may be partly to blame (Taylor et al., 2021). The pandemic may also have disrupted access to and delivery of substance use and mental health treatment services (Herrera, 2021). And just as the pandemic disrupted the supply chain for many household and food products, it also disrupted the drug supply chain. The result was a decline in the international production and trafficking of heroin in the United States and subsequent increases in fentanyl-adulterated drugs, which pose a greater risk of overdose. Drug shortages have also resulted in increases in the use of injecting drugs and sharing needles, increasing the risk of spreading bloodborne diseases (United Nations Office on Drugs and Crime [UNODC], 2020).

Thus, COVID-19 has reinforced and exacerbated existing mortality disparities within the United States, as well as between the United States and its peer countries. The CDC reported that adults ages 25–44 experienced the largest percentage increases in excess deaths during the pandemic (as of October 2020) (Rossen et al., 2020). COVID-19 mortality was also higher among males, who have long experienced higher mortality than females (Faust et al., 2020). Provisional mortality data from the CDC (2020c) indicated that 54 percent of all COVID-19 deaths had occurred among males. The disparity was even greater for working-age adults, among whom males

represented 65 percent of all COVID-19 deaths reported in the United States.

COVID-19 has disproportionately targeted Hispanic and non-White Americans, particularly Blacks (Ford, Reber, and Reeves, 2020; National Center for Health Statistics [NCHS], 2021). Provisional mortality data (NCHS, 2020) show wide racial disparities in the impact of the disease. Although present and significant at older ages, these disparities are larger among younger Americans, who show the widest racial/ethnic disparities (Ford, Reber, and Reeves, 2020). These disparities are far greater than overall disparities in all-cause mortality among working-age adults. As of early January 9, 2021, 35 percent of COVID-19 deaths among working-age adults had occurred among Hispanic adults and 24 percent among Black adults, even though these groups had experienced only 13 percent and 19 percent of all non-COVID-19 deaths, respectively. In contrast, only 34 percent of COVID-19 deaths among working-age adults had occurred among Whites, who had experienced 63 percent of non-COVID-19–related deaths. Working-age non-Hispanic Asian, AI/AN, and Native Hawaiian and Pacific Islander adults had also experienced disproportionately high mortality from COVID-19 (NCHS, 2021).

These racial disparities are undoubtedly due at least in part to the geographic concentration of the initial waves of the pandemic in large, racially diverse central metropolitan areas, such as New York City, San Francisco, Seattle, and Los Angeles. However, by January 2021, both the COVID-19 case and death rates were higher overall in nonmetropolitan than in metropolitan counties (Ullrich and Mueller, 2021). As the virus spread into less-populated and less racially diverse areas of the country, however, large racial disparities persisted in these areas as well (Cheng, Sun, and Monnat, 2020; Ford, Reber, and Reeves, 2020). Between March 2 and July 25, 2020, for example, average daily increases in COVID-19 death rates were 70 percent higher in rural counties with the largest-percentage Black populations (i.e., those in the top 25th percentile) (Cheng, Sun, and Monnat, 2020). In many ways, the COVID-19 pandemic was ominously poised to exploit and exacerbate existing social and economic inequalities. Because of race-based occupational and residential segregation, Hispanic and Black versus White adults were more likely to be employed in such “essential” occupations as health care, farm work, and food service (Bureau of Labor Statistics, 2019); to live in multigenerational households (Cohen and Casper, 2002); to experience socioeconomic disadvantage; and to lack full access to health care. Moreover, although individual-level data on socioeconomic disparities in COVID-19 mortality are not readily available, individuals of lower socioeconomic status are in general more likely to have comorbidities (Cutler, Meara, and Richards-Shubik, 2011; Pampel, Krueger, and Denney, 2010) that are associated with more severe COVID-19 illness (CDC, 2020a), and

the ability to adhere to social distancing guidelines in daily life depends on financial resources (Weiss and Paasche-Orlow, 2020).

COVID-19, then, will likely magnify the effects of already increasing mortality rates among many subgroups of working-age Americans. These mortality increases are likely to undo years of progress in reducing racial disparities in mortality and to magnify socioeconomic disparities in life expectancy. The important role played by COVID-19 in increasing mortality in the United States in 2020 can already be seen in the percentage of all working-age deaths between January 1, 2020 and January 9, 2021 that were due to COVID-19. Among working-age Whites, 4.3 percent of all deaths involved COVID-19. In contrast, the race-specific contribution of COVID-19 was 10.0 percent among Black adults, 21.4 percent among Hispanic adults, 14.2 percent among AI/AN adults, 13.0 percent among Asian adults, and 16.1 percent among Native Hawaiian and other Pacific Islander adults (NCHS, 2021).

These disparities may continue to shape mortality patterns for decades to come. The burden of COVID-19 is much broader than its direct impact on mortality. As of mid-September 2020, more than 6 million people in the United States had tested positive for the virus (Johns Hopkins University, 2020). Although many people who tested positive experienced no symptoms during their period of infection, growing evidence suggests that many of those who survive COVID-19 infection develop long-term symptoms and health complications that may last far into the future, even when they initially experienced only mild symptoms or were asymptomatic (Pérez-Bermejo et al., 2020; Tenforde et al., 2020). Experience with the COVID-19 pandemic underscores the importance of understanding the underlying causes of current mortality trends, including how racial, socioeconomic, and geographic disparities are consistently produced and reinforced as such novel diseases are introduced into the population.

STUDY METHODS AND LIMITATIONS OF THE EVIDENCE BASE

This report describes and explains the trends in high and rising mortality among working-age adults in the United States and documents the demographic, socioeconomic, and geographic disparities in those trends. To carry out these analyses, the committee developed a multilevel conceptual framework to guide identification of the main drivers of the trends, from upstream macrostructural factors, to local environments in which people live and work, to downstream proximate individual-level factors. Using this framework, the committee reviewed the available evidence on how changes in these factors may have led to increased mortality among working-age adults and how such factors operate across different life stages from

childhood to midlife. This framework also guided the committee’s identification of factors that can be modified through policy, education, or other initiatives to reduce mortality risks and disparities among working-age adults. Through a comprehensive review of the research evidence, the committee was able to identify knowledge gaps and offer recommendations for future research and data collection efforts.

As the committee began its initial task of reviewing previous research documenting the demographic, socioeconomic, and geographic disparities in mortality trends in the United States, it found extensive variation across studies that would have limited its ability to draw clear conclusions about the relative contributions of specific causes of death to changes in all-cause mortality within subpopulations. These studies varied in terms of which populations were included in the analyses, the time periods covered, and the tabulation and presentation of causes of death and demographic characteristics.

Moreover, although approval of prescription opioids occurred in the mid-1990s, the committee found that a number of studies evaluating the relative contribution of drug poisoning to overall changes in mortality began with data from 1999⁵ (Case and Deaton, 2015; Woolf et al., 2018; Woolf and Schoomaker, 2019), thus missing the initial period of the drug overdose epidemic. In addition, some studies combined deaths from cardiometabolic diseases into a single broad category (Masters, Tilstra, and Simon, 2017), while others highlighted more specific causes within this category, such as heart disease (Case and Deaton, 2017); diabetes (Geronimus et al., 2019; Woolf and Schoomaker, 2019); endocrine, nutritional, and metabolic diseases (Woolf et al., 2018); or hypertensive heart disease (Woolf et al., 2018; Woolf and Schoomaker, 2019). Although these differences across studies indicate that the trends in cause-specific mortality were not uniform within this broad category, they also limit the capacity to compare results across studies.

Still another limitation of the evidence base is the significant differences across racial/ethnic groups in what is known about variations in mortality trends by sex, age, and geography: these variations are documented extensively for working-age Whites and working-age adults more generally, and less information is available for Hispanic and non-White populations. Thus, to clearly assess the relative contributions of different causes of death to trends in all-cause mortality and how these contributions differed by sex, age, and geography for different racial/ethnic groups, the committee decided it was necessary to perform independent analyses. Multiple committee members have extensive experience performing such analyses and,

___________________

⁵ In 1999, cause of death on U.S. death certificates began to be classified using the International Classification of Diseases, 10th Revision (ICD-10).

through their existing research projects, had access to restricted mortality data files from the National Vital Statistics System that included detailed geographic information on decedent residence. Multiple committee members were involved in conducting the analyses, thus providing internal checks on the accuracy of the analyses. An exception occurred with respect to the examination of mortality rates by educational attainment. After considering the limitations of the information on educational attainment reported on death certificates, particularly during the 1990s, the committee decided to rely on a review of previous literature for estimates of mortality trends by education rather than produce its own estimates. For this reason, the explication in this report of differences in mortality trends by socioeconomic status is more limited than other aspects of the analysis.

ORGANIZATION OF THE REPORT

The remainder of this report is divided into three parts. Part I evaluates trends and disparities in mortality to identify the origins of the recent (since 2010) increases in mortality among working-age adults. This part of the report also evaluates the strengths and limitations of U.S. mortality data. Chapter 2 compares life expectancy and mortality rates in the United States with those in 16 high-income peer countries, beginning in the 1950s, to establish when the United States’ relative mortality disadvantage first emerged and the important role played by working-age mortality in contributing to this disadvantage. Chapter 3 examines trends in all-cause mortality within the United States between 1990 and 2017 by sex, age, race and ethnicity, socioeconomic status, and geography to provide greater insight into where and among which populations these increases occurred. Chapter 4 focuses on identifying the specific causes of death that contributed to these mortality trends to illuminate which causes made the greatest contribution and how this differs by sex, age, race and ethnicity, and geography. The final chapter of Part I, Chapter 5, evaluates the quality and limitations of mortality data available in the United States and makes recommendations for improving data capacity for future research on mortality.

Having identified in Part I the causes of death that represent the key drivers of recent changes in all-cause working-age mortality, the report turns in Part II to the committee’s evaluation of the quality of current research evidence supporting potential explanations for those changes. Chapter 6 provides an overview of the multilevel conceptual framework the committee applied in evaluating the key contributors to the recent mortality increases. Chapters 7–10 then examine the evidence supporting explanations for the trends in mortality for the key causes of death that drove the increases and provide recommendations for data improvements

and research. Chapters 7–9, respectively, focus on drug poisoning and alcohol-related deaths, suicide, and cardiometabolic diseases. Finally, Chapter 10 evaluates what is known about the broader economic factors that may have contributed to the recent mortality trends.

Part III of the report consists of Chapter 11, which recaps the policy and research implications presented in Part II and the rationale for each and offers new policy and research implications for themes that cut across all of the preceding chapters.

This page intentionally left blank.