4U.S. Trends in Cause-Specific Mortality Among Working-Age Adults

TRENDS IN U.S. WORKING-AGE MORTALITY BY CAUSE OF DEATH

To identify the key underlying causes of death responsible for the changes in all-cause mortality over the 1990–2017 period, this section presents mortality rates (in deaths per 100,000 population) at the beginning (1990–1993) and end (2015–2017) of the period and the changes in cause-specific mortality rates over the period for working-age White, Black, and Hispanic adults in three age groups—24–44, 45–54, and 55–64. The changes over time are presented in terms of both absolute change in mortality rates (in deaths per 100,000 population) and the percentage contribution of each cause-specific change in mortality to the total increase or decrease in all-cause mortality. The latter percentages were calculated by dividing the increase (or decrease) in the cause-specific mortality rate by the total increase (or decrease) across all causes of death that increased (or decreased) between 1990–1993 and 2015–2017.

The results presented in this section reveal dramatic reductions in mortality from the most common (leading) causes of death in 1990–1993, including ischemic heart disease and other circulatory diseases,¹ cancers (excluding liver cancer), and HIV/AIDS. These improvements occurred among both sexes and each of the three racial/ethnic groups the committee examined. However, mortality also increased across a wide range of causes of death, offsetting some, and in some cases all, of these gains. Although there were racial/ethnic differences in which causes of death increased over the period, mortality due to drug poisoning and diseases of the nervous system² increased across all working-age adults, regardless of sex, age group, or race and ethnicity.

Although working-age Black adults maintained the highest mortality throughout the period, they also experienced the largest decreases in mortality across the widest range of causes of death, narrowing the racial gap in mortality between the 1990–1993 and 2009–2011 periods. Working-age Hispanic adults also experienced (comparatively) large reductions in mortality across many causes of death. In contrast, working-age White adults experienced increases in mortality across the widest range of causes of death, and for this reason had higher mortality than working-age Hispanic adults at the end of the period.

Although similarly detailed cause-specific trends in mortality for APIs and AI/ANs could not be included here because of concerns about the quality of racial data on death certificates, Box 4-1 (Asians and Pacific Islanders) and Box 4-2 (American Indians and Alaska Natives) briefly review cause-specific mortality for these populations.

DISPARITIES IN CAUSE-SPECIFIC MORTALITY BY SOCIOECONOMIC STATUS

The literature on trends in educational attainment and working-age cause-specific mortality is fairly extensive; however, only a limited number of these studies focused on socioeconomic disparities in cause-specific mortality rates. Despite their limited number, these studies consistently found that increases in mortality due to drug poisoning, alcohol-induced causes, and suicide were the largest contributors to the growing gap in mortality by education among working-age Whites, and that increases in drug poisoning mortality were increasingly concentrated among working-age Whites with a high school degree or less. This group also experienced larger increases in mortality across a wide range of other causes of death, and this was especially true for White women with less education. In contrast, among working-age Black adults, education-based disparities in cause-specific mortality remained steady over time.

Sasson (2016) used vital statistics data to demonstrate steep increases in educational disparities in life expectancy for White men and women but little change for Black men and women between 1990 and 2010. He showed that the increases in the education-based mortality gap between working-age White men and women involved largely causes of death associated with smoking, external causes (including drug poisoning),⁸ and cardiovascular diseases.

Case and Deaton (2015) delved into the specific causes of death associated with widening educational disparities in working-age mortality, focusing on adults ages 45–54 between 1999 and 2013. They showed that the death rate from poisoning among U.S. adults ages 45–54—which includes deaths from drug poisoning and alcohol-induced causes, both unintentional and of undetermined intentionality—increased during that period for White adults of all educational levels, as well as for Black and Hispanic adults. However, the increase was especially pronounced for White adults with a high school degree or less; in this group the death rate from poisoning increased more than four-fold over the period, from 14 per 100,000 population in 1999 to 58 per 100,000 population in 2013. Death rates due to poisonings among White adults with some college education and those with a college degree or more in this age group also increased rapidly but started at a lower level in 1999, and the increases across the time period were less pronounced. For example, the poisoning death rate for White adults ages 45–54 with some college increased from 6 per 100,000 to 21 per 100,000, while the increase for White adults with a college degree or more was from 3 per 100,000 to 8 per 100,000.

These findings provided the first clear evidence that mortality from drug poisoning among working-age Whites was increasing more rapidly among those with less versus those with more education. Unfortunately, Case and Deaton (2015) did not break down results for working-age Black or Hispanic adults by educational attainment. Notably, though, mortality from poisoning among Black and Hispanic adults ages 45–54 increased between 1999 and 2013, from 18 to 22 and from 10 to 14 per 100,000 population, respectively. Case and Deaton (2017) later updated these descriptive trends, examining them separately by sex, but the main finding of growing education-based disparities in mortality among working-age White adults did not change.

Most recently, Geronimus and colleagues (2019) documented changes in educational disparities in mortality between 1990 and 2015 for working-age (and older) Black and White women and men. They measured educational attainment in quartiles to help account for compositional changes within education categories due to increasing educational attainment across time. Thus, they compared changes in mortality disparities between 1990 and 2015 for those in the bottom 25 percent versus those in the top 25 percent of the educational attainment distribution.

Geronimus and colleagues (2019) found that among White adults, increasing drug-related mortality was especially concentrated among those with less education, accounting for 73 percent and 44 percent of the increased educational disparity in working-age mortality for White men and White women, respectively. White men and women also exhibited modest increases in educational disparities in working-age mortality due to suicide and liver disease. Educational disparities in working-age mortality for White women also widened over the period for a range of causes of death, including cardiovascular disease, nonlung cancers, non-HIV infectious diseases, lower respiratory diseases, and other internal causes, and only (very modestly) narrowed for homicide. Among White men, educational disparities in working-age mortality widened for some of the same causes as those seen among White women (e.g., other cancers, other infectious diseases, other internal causes) but narrowed for others (e.g., lung cancer, accidents).

Thus in all, one-half (White women) to 80 percent (White men) of the increasing educational disparity in working-age mortality over the 1990–2015 period was due to what some researchers have referred to as “despair-related” causes, with educational differences in mortality from drug poisoning being particularly important for understanding the widening of educational disparities in working-age mortality that occurred among White men and women over the period. Furthermore, and particularly for White women, increasing educational disparities in working-age mortality were also seen for a range of other causes. By contrast, there was virtually no change in educational disparities in working-age mortality for Black men and women between 1990 and 2015. Increasing drug-related mortality among Black women and men differed only modestly by educational attainment and thus had very little influence on changing educational disparities in working-age mortality. Geronimus and colleagues (2019) also found only minor changes in educational disparities in mortality from lung cancer, cardiovascular diseases, diabetes, and a range of other causes for Black men and women across the 25-year time period. In other words, declines in working-age mortality over the period for Black men and Black women unfolded in parallel fashion across educational attainment groupings.

CAUSE-SPECIFIC MORTALITY TRENDS BY METROPOLITAN STATUS

All-cause mortality trends were most favorable in large central metropolitan areas (hereafter referred to as “large central metros”) and less favorable in less-populated areas over the period, often leading to a widening mortality gap across these areas (see Chapter 3). The findings reported in this section show that most cause-specific mortality rates followed a similar trend, suggesting that the growing geographic mortality gap was the cumulative result of underlying processes that produced metro status differences for multiple causes of death. In general, when cause-specific mortality rates decreased over the period, they declined the most in large central metros; when they increased, either mortality continued to decline, or the increases were smaller in large central metros. Detailed tables showing the change in cause-specific mortality between 1990–1993 and 2015–2017 by sex, age group, and metropolitan status can be found in the annex at the end of this chapter (Annex Tables 4-1 to 4-3). Cause-specific mortality rates by sex, age group, and metropolitan status are in Appendix A.

This pattern was most consistent for working-age Whites, among whom most causes of death contributed to the growing disparities between large central metros and less-populated areas. In contrast, for working-age Black and Hispanic adults, cause-specific mortality rates did not consistently decline the most in large central metros. Although Black and Hispanic adults in large central metros did experience greater improvements in mortality from many causes of death, those in nonmetropolitan areas (hereafter referred to as “nonmetros”) often experienced larger decreases in mortality from the causes of death that were key drivers of reductions in mortality over the period (e.g., cancers other than liver cancer or ischemic heart disease and other circulatory diseases). Nonetheless, Black and Hispanic adults in large central metros, particularly males, saw much larger declines in mortality from HIV/AIDS and homicides that offset these differences and drove the greater overall improvements in mortality in large central metros.

Areas outside of large central metros were more likely to experience larger increases in mortality across more causes of death. Often, the largest increases in cause-specific mortality occurred within nonmetros. Mortality due to drug poisoning was a notable exception to this pattern. Among White males and older (ages 45–64) Black males and females, large metros saw the biggest increases in mortality from drug poisoning, while nonmetros experienced smaller increases. Among White males, the largest increases occurred in large fringe metropolitan areas (hereafter referred to as “large fringe metros”), while among older Black adults, the largest increases occurred in large central metros. This meant that drug poisoning did not contribute to, and in fact offset, the growing mortality disparity between large central metros and less-populated areas among White males and older (ages 45–64) Black males and females. In contrast, White females, Hispanics, and younger working-age (ages 25–44) Black adults experienced smaller increases in mortality due to drug poisoning overall, with the smallest increases occurring in large central metros. Younger (ages 25–44) Black adults living in nonmetros also experienced a smaller overall increase in drug poisoning mortality.

TEMPORAL PATTERNS IN CAUSE-SPECIFIC MORTALITY TRENDS

The changes described above reflect changes in cause-specific mortality rates over the full period (1990–1993 to 2015–2017). In this section, the total change in each cause-specific mortality rate over the period is decomposed into changes within three periods that correspond roughly to decades. This decomposition makes it possible to determine when the changes occurred for different causes of death and identify the causes of death responsible for those changes. These periods of change are referred to as the 1990s (1990–1993 to 2000–2002), the 2000s (2000–2002 to 2009–2011), and the 2010s (2009–2011 to 2015–2017).

Figures 4-1 and 4-2 (showing changes in cause-specific mortality rates by time period for males and females, respectively) reveal two main findings beyond those already discussed earlier in this chapter. First, among the causes of death for which mortality rates declined in meaningful ways (i.e., HIV/AIDS, nonliver cancers, ischemic heart disease and other diseases of the circulatory system, and homicides [among young Black and Hispanic males]), most improvements occurred in the 1990s or 2000s and stagnated or even reversed in the 2010s. This pattern appeared most consistently for ischemic heart disease and other circulatory diseases, which saw a slowing rate of improvement in mortality over the period among most working-age adults. Among White females, younger Black males, and younger Hispanic males and females, progress in lowering mortality from ischemic heart disease and other circulatory diseases ceased in the 2010s as all-cause mortality rates began to increase.⁹

Most of the progress in reducing mortality from HIV/AIDS occurred in the 1990s among White and Hispanic males and younger Black males. However, mortality from HIV/AIDS continued to increase among Black females and older Black males in the 1990s and did not begin to decrease until the 2000s; among Black females ages 55–64, mortality from HIV/AIDS did not begin to decline until the 2010s. By 2015–2017, mortality from HIV/AIDS among White and Hispanic adults was sufficiently low that continued progress in reducing it would not have substantially affected future mortality trends. In contrast, mortality from HIV/AIDS remained high among working-age Black adults in 2015–2017; therefore, efforts to address this cause of death in this population could affect future mortality trends, as well as mortality disparities between Blacks and Whites. Although homicide rates decreased substantially in the 1990s, especially among Black and Hispanic males, progress slowed in the 2000s. By the 2010s, most groups had experienced at least a small increase in the homicide rate; the increases were particularly large among younger working-age Black males. In contrast to other leading causes of death, progress on lung and other nonliver cancers continued into the 2010s, although possibly at a slower rate.

The second main finding is that mortality rates increased among multiple demographic groups across a wide range of causes. These causes included ENM diseases, hypertensive heart disease, drug poisoning, alcohol-induced causes, mental and behavioral disorders, suicide, and diseases of the nervous system. Some of these causes increased throughout the entire period (1990–1993 to 2015–2017), whereas other causes increased only recently (in the 2010s). For example, mortality from diseases of the nervous system increased in all three periods among most groups, although these increases were generally small among younger working-age adults (ages 25–44). The exception was younger Black males, among whom mortality from diseases of the nervous system decreased slightly in the 1990s and 2000s before increasing in the 2010s.

The timing of the increases in mortality from the two cardiometabolic diseases—hypertensive heart disease and ENM diseases—differed between older and younger working-age adults. Mortality for both increased in all three periods among younger working-age adults. Although older working-age adults saw increases in mortality from ENM diseases in the 1990s and 2010s, they also experienced much larger reductions in between (in the 2000s) that were often large enough to offset the much smaller mortality increases in both the earlier and later periods. Mortality from hypertensive heart disease increased in all three periods among older working-age (ages 45–64) White males and females and Hispanic males, but did not start to rise until the 2000s among older Black males, Black females ages 45–54, and Hispanic females ages 45–54. Black and Hispanic females ages 55–64 continued to experience reductions in mortality from hypertensive heart disease until the 2010s, during which it increased.

As noted earlier, previous studies have considered mortality from drug poisoning, alcohol-induced causes, and suicide together as a group. However, the timing of the increases in these causes of death differed. Increases occurred in drug poisoning mortality in all three periods among White males and females and Hispanic females, but only White males ages 45–54 also saw concurrent increases in mortality from both alcohol-induced causes and suicide. White females ages 25–44 experienced increasing mortality from alcohol-induced causes in all three periods, but their suicide rates did not increase until the 2000s. Neither mortality due to alcohol-induced causes nor suicide rates increased until the 2000s among other working-age White adults, and suicide rates remained flat among Hispanic females in all three periods.

Mortality from drug poisoning also increased in each period among older working-age (ages 45–64) Black males and females and older working-age Hispanic males, although none of these groups experienced an increase in mortality from alcohol-induced causes or suicide. The exception was Hispanic males ages 55–64, who saw a small increase in alcohol-induced mortality in the 2010s. In fact, alcohol-induced mortality decreased among Hispanic males in the 1990s and 2000s. Among Black males ages 45–54, mortality from drug poisoning increased in the 1990s but fell in the 2000s before sharply increasing again in the 2010s, while mortality from alcohol-induced causes decreased in each period, and suicide rates decreased in the 1990s and 2000s. Mortality due to drug poisoning and suicide did not increase among younger Black adults and younger Hispanic males (ages 25–44) until the 2010s, except among younger Black females, who saw no change in suicide rates. Mortality from alcohol-induced causes remained flat throughout the period among these younger working-age adults.

Taken together, the differing trends in mortality for each of these three causes of death (drug poisoning, alcohol-induced causes, and suicide) suggest that there are limitations to considering these causes of death together, particularly when one is examining younger, female, and non-White working-age adults. Drug poisoning was by far the largest contributor to the overall increase in mortality for most working-age adults during the period, but the comparisons in Figures 4-1 and 4-2 demonstrate, importantly, that the timing of the increase in mortality from drug poisoning (the largest contributor to the overall increase in all-cause mortality for several groups) varied by sex, race and ethnicity, and age group.

SUMMARY

The recent trends in all-cause mortality among working-age adults are the result of the confluence of two important trends: (1) rising mortality from drug poisoning and other causes of death, such as nervous system diseases, hypertensive heart disease, and ENM diseases; and (2) slower progress in lowering mortality from heart diseases and other leading causes of death that drove improvements in all-cause mortality rates in prior decades. Table 4-4 summarizes the findings for each of the 20 causes of death considered in this chapter, showing how each contributed to changes in mortality over the period by age group, sex, and race and ethnicity. Subsequent chapters examine in greater detail the trends in the key causes of death that have driven the recent increases in mortality among working-age adults, either through increasing mortality or through a reduction or reversal of progress in reducing mortality in the most recent period. These trends are used to assess how consistent the prevailing explanations for recent increases in mortality in the research literature are with these cause-specific mortality trends.

Based on the findings presented in this chapter, the key drivers of the increases in working-age mortality since 2010 are grouped into three categories, each of which is addressed in detail in Part II of this report. The first category is drug poisoning and alcohol-induced causes (Chapter 7). In addition to mortality from drug poisoning and alcohol-induced causes, this one substance. Thus, drug use is often involved in deaths for which the underlying cause is coded as alcohol-induced, and vice versa.

As noted earlier and discussed in Chapter 7, most deaths for which the underlying cause of death is classified as a mental or behavioral disorder involve either drug poisoning or alcohol (Figure 4-3). In 1990, more than 70 percent of deaths due to a mental or behavioral disorder were due to alcohol, while nearly 15 percent were due to drug use. Over the 1990s and early 2000s, the percentage of these deaths due to alcohol decreased steadily, reaching a low of 55 percent in 2007. At the same time, the percentage due to drug poisoning increased steadily, reaching 23 percent in 2006. Throughout the period, more than 70 percent of all deaths due to a mental or behavioral disorder were due to either alcohol or drug use. For this reason, the explanations for mortality due to drug poisoning, alcohol-induced causes, and mental and behavioral disorders are discussed together in Chapter 7.

The second key driver of mortality evaluated by the committee in detail is suicide. Arguments can be made for examining suicide alongside mortality due to drugs and alcohol. For one, it can be difficult for medical examiners to distinguish between accidental and intentional drug poisoning, leading some suicides to be misclassified as accidental poisoning and vice versa. However, the committee sidestepped this classification problem by categorizing all mortality due to drug poisoning, both intentional and category also includes mortality due to mental and behavioral disorders, which often involve drugs or alcohol (see Figure 4-3). The second category includes suicides that do not involve drug poisoning (Chapter 8). The third is mortality due to cardiometabolic diseases, which include ENM diseases, hypertensive heart disease, and ischemic heart disease and other circulatory diseases. The results presented in Table 4-4 indicate that in addition to these key causes of death, several other causes—including infectious and parasitic diseases other than HIV/AIDS, liver cancer, diseases of the nervous system, transport accidents, and homicide—also contributed to rising mortality over the period.

The results presented in this chapter demonstrate that drug poisoning mortality rose throughout the study period and was the single largest contributor to the overall increases in mortality among working-age adults, except older (ages 45–64) Hispanics. The largest increases during the period occurred among White adults, particularly White males and older Black males. Among working-age Whites, increases in mortality due to drug poisoning were largest among younger males (ages 25–44), those with a high school degree or less, and those living in large metropolitan areas. In contrast, among Black adults, the largest increases in mortality occurred among older males (ages 55–64) in large central metros, but there was no difference in drug poisoning mortality by educational attainment. Alcohol-induced mortality also increased among working-age Whites throughout the period, while increases among Black and Hispanic adults did not begin until the 2010s. Mortality from alcohol-induced causes declined among working-age Black and Hispanic males throughout the 1990s and early 2000s, but these declines leveled off during the 2000s and began to increase in the 2010s. Moreover, the increases in alcohol-induced mortality among working-age Whites followed different patterns than the increases in mortality from drug poisoning, which could reflect temporal differences in the etiology of these causes of death. Mortality due to alcohol-induced causes increased more among older working-age Whites than among other groups and outside of large central metros.

Despite these different trends in drug- and alcohol-induced mortality, there are important reasons to consider the explanations for these trends in concert with each other. For example, Case and Deaton (2015, 2017, 2020) posit that these deaths are the result of an underlying root cause: the erosion of economic and social stability within the White working class has increased physical, emotional, and psychological pain, leading to increases in substance use and mortality, particularly among less-educated White men. Considering these causes of death together allowed the committee to better evaluate the evidence underlying this “deaths of despair” hypothesis. However, a second, more practical, reason to consider these causes of death in parallel is that most substance-induced deaths involve more than accidental, as drug poisoning. Moreover, the underlying trends for suicide differ in important ways from those for drug poisoning and alcohol-induced mortality. Unlike increased mortality due to drug poisoning, increases in suicide deaths occurred primarily among working-age Whites, particularly White men. Among White adults, suicide rates were highest in nonmetros and lowest in large central metros—whereas the largest increases in mortality from drug poisoning were within large central metros—and were higher among older (ages 45–64) White adults compared with their younger counterparts. These differences between the trends in suicide and in drug poisoning deaths suggest that the explanations for the two causes may differ. Potential explanations for suicide trends are therefore evaluated separately in Chapter 8.

The final category of causes of death evaluated by the committee in detail is mortality due to cardiometabolic diseases. This category encompasses two causes of death that increased among most working-age adults over the study period (hypertensive heart disease and ENM diseases), as well as a cause of death that had previously seen dramatic improvements but on which progress stalled or reversed (the combined category of ischemic heart disease and other circulatory diseases). Although the overall trends in mortality for the three cardiometabolic causes of death differ in direction, all experienced a common slowdown or reversal of progress in reducing mortality, and there is reason to suspect that they share an underlying proximate cause, such as obesity, that justifies examining them together.

Mortality due to hypertensive heart disease and ENM diseases increased among most working-age males and White females. These increases were larger outside of large central metros, particularly among White males and females, contributing to the growing mortality gap between large central metros and less-populated areas. As discussed at length in Chapter 9, although mortality due to the combined category of ischemic heart disease and other circulatory diseases declined dramatically overall during the study period among working-age adults, these gains slowed and, in some cases, reversed in the 2010s. Even among those working-age adults for whom mortality from this cause continued to decrease, such as Black females, it decreased at a slower rate in that period, and it began to increase among many younger (ages 25–44) adults and older (ages 45–64) White females. Trends in metropolitan areas differed by race and ethnicity but generally contributed to mortality gaps by metropolitan status. Working-age Whites in large central metros continued to see reductions in mortality due to ischemic heart disease and other circulatory diseases, while those in nonmetros were most likely to experience increasing mortality from this set of causes. The combination of these trends contributed to an expanding mortality gap between large central metros and nonmetros among White adults. In contrast, older Black and Hispanic adults in nonmetros experienced larger reductions in mortality relative to those in more populous areas.

As noted above, other causes of death—including infectious and parasitic diseases (excluding HIV/AIDS), liver cancer, diseases of the nervous system, transport accidents, and homicide—contributed to increasing working-age mortality during the 1990–2017 period. Although these other causes of death are not addressed in detail in this report, they do merit attention. The committee therefore included detailed tables with cause-specific trends for these causes of death in Appendix A.

Although some of these other causes of death did not, on their own, contribute meaningfully to the recent alarming increases in mortality, the results presented in Tables 4-1 to 4-3 demonstrate that their combined influence on working-age mortality trends was not trivial. Moreover, important details about deaths from certain causes shed light on their potential role, or lack thereof, in explaining increases in working-age mortality over the 1990–2017 period or ongoing racial/ethnic disparities in mortality. For example, mortality rates for HIV/AIDS, as well as other infectious and parasitic diseases, increased predominantly in the 1990s and subsequently fell; therefore, they do not help explain the current rise in working-age mortality, although the delayed progress in reducing mortality from HIV/AIDS among older working-age Black males and females has contributed to continuing racial/ethnic disparities in mortality.

In a similar vein, mortality due to transport accidents and homicides decreased overall between 1990 and 2017, primarily as a result of large reductions during the 1990s. In the 2010s, however, these gains began to reverse, particularly among younger Black and Hispanic working-age males, who otherwise experienced decreases in overall mortality during this period. Overall, neither transport accidents nor homicides were a significant contributor to the recent increase in mortality among working-age White males and females, but like the slower progress in reducing mortality from HIV/AIDS among older Black males and females compared with Whites, these causes of death contributed to mortality disparities between younger Black and White males. The reasons for these recent changes in transport accidents and homicides are not well understood. Research suggests that a recent increase in fatal police shootings is a leading cause of homicide among young Black males (Edwards, Lee, and Esposito, 2019), but inconsistency across states in data collection on these shootings makes it difficult to assess whether this increase can explain the rise in homicides among younger Black and Hispanic males.

Mortality due to nervous system diseases increased among working-age adults regardless of age, sex, and race and ethnicity, although these increases were often very small. Recent studies have noted that similar increases in mortality due to nervous system diseases occurred internationally (Pritchard et al., 2017), but the committee is not aware of research offering an explanation for this trend. The increase in deaths from neurologic diseases among the elderly is an expected outcome of an aging population, but reasons for the increase among working-age adults, before age 65, are less clear.

Many of the cause-of-death categories included in this report are broadly defined, often by the body system affected, and encompass a wide range of diseases and disorders. For example, mortality from ischemic heart disease and other circulatory system diseases was examined as a combined group, but “other circulatory system diseases” encompassed all circulatory system diseases besides ischemic heart disease and hypertensive heart disease, including arrhythmias, cardiomyopathy, heart failure, cardiac arrest, stroke, intracerebral hemorrhage, and pulmonary embolism. Because the committee’s purpose was to identify the key drivers of recent changes in working-age mortality rather than to fully explore recent changes in all causes of death, this report focuses on mortality trends for entire body systems, such as the nervous, genitourinary, respiratory, and digestive systems, without detailing more pronounced increases in mortality from specific diseases within these body systems. However, readers should note that within each of the broad cause-of-death categories, there may be heterogeneity in the magnitude—and sometimes the direction—of the changes in cause-specific mortality that occurred between 1990 and 2017, particularly for rare causes of death that result in only a small number of deaths per year among working-age adults.

It will be important for future research to seek explanations for the increase in working-age mortality across these conditions. To some extent, increased death rates in working age may be coincidental and reflect independent causal pathways. For example, increases in working-age mortality from cerebral palsy may reflect medical advances that have enabled children with these conditions to survive into adulthood (Woolf et al., 2018). Increased cellphone use could contribute to the increase in transport injuries but would not explain deaths from chronic diseases.

Another possibility is that some increases in working-age mortality may be secondary to the primary causes this report examines—drugs, alcohol, suicide, and cardiometabolic diseases. Deaths from other causes that stem from these primary causes might be considered “collateral” deaths. Mortality due to liver cancer, for example, increased among older working-age adults (ages 45–64) regardless of race and ethnicity. This increase could potentially have resulted from several underlying causes, including increasing alcohol and drug use, as well as viral hepatitis, diabetes, and non-alcohol-induced fatty liver disease. This complexity in potential etiology complicates any attempt to use the trends in liver cancer to evaluate potential explanations for the increase in alcohol- and drug-related deaths. However, to the extent that the recent increases in mortality due to liver cancer are linked to drug and alcohol use or diabetes, the explanations for these trends in Chapters 7 and 9, respectively, may be relevant.

This “collateral” mortality effect could also have contributed to changes in more detailed causes of death contained within the 20 broad cause-of-death categories examined in this chapter. For example, increased use of injection drugs could explain not only overdose deaths but also increases in deaths from viral hepatitis, infectious valvular heart diseases, and other drug-related complications. Likewise, alcohol use can increase an individual’s risk of death from atrial fibrillation and other arrhythmias, transport accidents, and other causes for which working-age mortality has increased. People who initially survive a suicide attempt may die in the hospital from secondary complications. And obesity and other contributors to deaths from cardiometabolic diseases could help explain increased mortality from renal failure.

It is difficult, however, to identify all of the detailed causes of death that could potentially be considered collateral consequences of the larger trends in substance use, suicide, and cardiometabolic diseases. In addition, because death is the result of complex processes that unfold over the life course, trends in mortality are rarely so simply explained. Despite these complications, the evidence gaps noted above and the loss of life involved provide a strong argument for a research agenda to seek the underlying explanations for the large number of causes of death for which mortality has been increasing. Of necessity, the next chapters focus on the main drivers of increasing working-age mortality—drugs, alcohol, suicide, and cardiometabolic diseases—but the committee encourages the research community to continue the work of exploring the explanations underlying increases in working-age mortality due to the range of other causes of death identified in this report.

ANNEX 4-1. Trends in Cause-Specific Mortality Among American Indians and Alaska Natives

The American Indian/Alaska Native (AI/AN) population has the highest mortality rates for more causes of death of all racial/ethnic groups. Epsey and colleagues (2014) (reproduced in Sancar, Abbasi, and Bucher, 2017) present leading causes of death for 1999–2009 among AI/ANs living in 637 Contract Health Service Delivery Areas (CHSDAs). These data were taken from death certificates, corrected for misclassification of AI/AN identity and from National Vital Statistics Reports for 2017 for the entire United States, without such correction. Annex Figure 4-1 presents the leading causes of death for the CHSDA counties in 1999–2009. AI/AN individuals are more likely to die from diabetes, chronic liver disease, and suicide than are Whites, and also more likely to die from these causes than are non-Hispanic Blacks and Hispanics.

CAUSE-SPECIFIC MORTALITY TRENDS BY METROPOLITAN AREA STATUS