Missing Americans: Early death in the United States-1933-2021 — TEI-tagged view

Jacob BorpersName, Andrew C StokespersName, Julia RaifmanpersName, Atheendar VenkataramanipersName, Mary T BassettpersName, David HimmelsteinpersName, Steffie WoolhandlerpersName (2023date). “Missing Americans: Early death in the United States-1933-2021.” PNAS Nexustitle·j.

Object of study — excess US deathsterm·dom relative to 21 other wealthy nationsterm·key, 1933-2021

Research question — how many US deaths each year would have been averted with the mortalityterm·key rates of peer nations

Contribution — the missing-Americans metric and its long series, showing an unprecedented and growing US mortality disadvantageterm·dom

Full text — tagged

The reference list is not part of the text we hold for this paper. Our copy was extracted from the publisher's page and the bibliography did not survive that extraction, so the article below ends before its references. Nothing has been removed by us.

Missing Americansterm·dom: Early death in the United States-1933-2021 Abstract We assessed how many US deaths would have been averted each year, 1933date–2021date, if US age-specific mortalityterm·key rates had equaled the average of 21 other wealthy nationsterm·key. We refer to these excess US deathsterm·dom as “missing Americansterm·dom.” The United States had lower mortalityterm·key rates than peer countries in the 1930s–1950s and similar mortalityterm·key in the 1960s and 1970s. Beginning in the 1980s, however, the United States began experiencing a steady increase in the number of missing Americansterm·dom, reaching 622,534 in 2019 alone. Excess US deathsterm·dom surged during the COVID-19term·key pandemic, reaching 1,009,467 in 2020 and 1,090,103 in 2021date. Excess US mortalityterm·key was particularly pronounced for persons under 65 years. In 2020 and 2021date, half of all US deaths under 65 years and 90% of the increase in under-65 mortalityterm·key from 2019 to 2021date would have been avoided if the United States had the mortalityterm·key rates of its peers. In 2021date, there were 26.4 million years of life lostterm·key due to excess US mortalityterm·key relative to peer nations, and 49% of all missing Americansterm·dom died before age 65. Black and Native Americans made up a disproportionate share of excess US deathsterm·dom, although the majority of missing Americansterm·dom were White.

One million US deaths in 2020 and 1.1 million US deaths in 2021date would have been averted if the United States had the mortalityterm·key rates of other wealthy nationsterm·key. About half of these missing Americansterm·dom died before age 65. The number of excess US deathsterm·dom relative to peers is unprecedented in modern times, at least since the 1930s. These excess US deathsterm·dom were a result of a decades-long divergence in mortalityterm·key from other wealthy nationsterm·key, beginning in the 1980s, and were further exacerbated by the COVID-19term·key pandemic. The use of an international benchmarkterm·mth highlights unfavorable mortalityterm·key trends involving all US racial/ethnic groups and disproportionately affecting younger and working-age adults.

Introduction US residents died at higher rates during the COVID-19term·key pandemic than did residents of other wealthy nationsterm·key ( 1 , 2 ). This pattern of early death is not new ( 3–9 ). US life expectancyterm·key has been falling behind peer countries' since the 1980s and has stagnated in the last decade ( 10 ).

Excess mortalityterm·key in the United States relative to other nations has been linked to proximate factors including drug overdoses, firearm injuries, vehicle accidents, cardiometabolic disorders, and smoking ( 3–7 ) and to fundamental causes including structural racism, economic inequality, gaps in access to medical care, and underinvestment in public health and social safety net programs in the United States ( 11 ). These factors have contributed to the large population-level impact of the COVID-19term·key pandemic in the United States as well as its disproportionate toll on people with lower income and education levels and Black, Hispanic, and Native American individuals ( 12 , 13 ).

In this paper, we quantify the number of US deaths that would have been averted each year, from 1933date through 2021date, if US age-specific mortalityterm·key rates (ASMRs) had equaled the average of 21 other wealthy countries. We refer to these excess US deathsterm·dom relative to peer countries as “missing Americansterm·dom.” We note that “missing Americansterm·dom” is a statistical construct; we cannot specify which deaths would have been averted, only the number that would have been averted if the United States had mortalityterm·key rates of its peers.

Our use of an international benchmarkterm·mth—the average of other wealthy nationsterm·key—to estimate “excess US deathsterm·dom” differentiates our analysis from the large number of studies estimating “excess mortalityterm·key” during the COVID-19term·key pandemic relative to prepandemic US death ratesterm·mth ( 14–18 ). By taking pre-COVID-19 death rates as “normal,” these approaches ignore longstanding mortalityterm·key differences across nations and render invisible fundamental causes ( 19 ) of excess US mortalityterm·key compared with other nations before and during the pandemic. Peer comparisons showcase the kind of mortalityterm·key rates that are achievable in high-income countries and provide a better benchmark to judge where the United States stands. “Missing Americansterm·dom” offers an easily interpretable metric to quantify the US mortality disadvantageterm·dom and and augments the existing literature on life expectancyterm·key comparisons ( 20 ).

We extend prior analyses of excess US deathsterm·dom in three directions ( 7 , 8 , 11 , 21 ). First, we estimate long-term historical trends in excess US deathsterm·dom, starting in 1933date (when the US mortalityterm·key series begins). The longer view paints a clearer picture of the timing of the mortalityterm·key divergence between the United States and other wealthy nationsterm·key and variation in timing by age group. It also encompasses World War II (WWII), the largest mortalityterm·key crisis prior to COVID-19term·key among today's wealthy nationsterm·key ( 22 ).

Second, we examine the impact of COVID-19term·key on excess US deathsterm·dom: the number of missing Americansterm·dom in 2019, on the eve of the COVID-19term·key pandemic; the change in mortalityterm·key in 2020 and 2021date associated with the COVID-19term·key pandemic in the United States and peer countries; and the impact of the pandemic on the number of missing Americansterm·dom. We stratify our analyses by age, focusing in particular on adults under 65 years, who were already experiencing rising mortalityterm·key prior to COVID-19term·key ( 5 , 7 , 8 ) and who are largely excluded from Social Security and Medicare and from social programs targeted to children.

Third, we use an international benchmarkterm·mth to assess trends in excess mortalityterm·key borne by US racial and ethnic groups, given persistent racial disparities in life expectancyterm·key ( 23 ), rising mortalityterm·key among White Americans ( 5 ), and differential mortalityterm·key impacts of the pandemic ( 24 , 25 ). Whereas studies of US health disparitiesterm·key typically focus on differences between US racial and ethnic groups, the declining health of White US residents has made it difficult to interpret trends in disparities. External comparisons shed light on the exceptional nature of US mortalityterm·key trends and the experiences of different US racial and ethnic groups.

We conduct each of these analyses using recently released mortalityterm·key data through 2021date for the United States and other wealthy nationsterm·key, compiled by the Human Mortalityterm·key Database (HMD) ( 26 ) and the US Centers for Disease Control and Prevention (CDC) ( 27 ).

Methods We compared mortalityterm·key trends in the United States with mortalityterm·key trends in 21 other wealthy countries. The comparison set of countries included Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Iceland, Ireland, Italy, Japan, Luxembourg, The Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, and the United Kingdom. This comparison set included all countries with age-specific mortalityterm·key data available from HMD beginning in 1960 or earlier and ending in 2020 or later, after excluding former-Soviet/Eastern Bloc nations. All countries included had 2021date gross domestic product (GDP) per capita >US$24,000 ( 28 ). Our comparison set makes comprehensive use of available data on peer countries, following recent work on mortalityterm·key trends ( 1 , 17 , 29 ). However, we also present missing Americansterm·dom estimates using two alternate comparison groups that have been used in international mortalityterm·key comparisons: the other Group of Seven (G7) countries—Canada, France, Germany, Italy, Japan, and the United Kingdom—used in prior work by members of this study team ( 11 ) and the five largest Western European countries—France, Germany, Italy, Spain, and the United Kingdom—used by Preston and Vierboom ( 8 ) and Heuveline ( 21 ).

Data on deaths and population denominators starting in 1933date were obtained from the HMD, which compiles official data from national vital registries and censuses. We augmented the HMD standard long-term mortalityterm·key series (available through 2019 to 2021date for the countries in the sample) with data from the HMD's Short-Term Mortalityterm·key Fluctuations (STMF) database, which includes more complete data for 2020 and 2021date. Both databases were accessed 2023 January 3; STMF was last updated 2022 December 17; HMD long series was last updated 2022 December 27. Details on the panel of countries are provided in Table S1 . All countries were observed starting in 1933date with the exception of Portugal (1940), Austria (1947), Japan (1947), New Zealand (1948), Ireland (1950), Germany (1956), and Luxembourg (1960). The Germany series combines East and West Germany prior to 1989. All countries were observed through 2021date, with the exception of Japan and Ireland, for which we carried forward 2020 ASMRs to 2021date.

For each country, we extracted annual ASMRs, deaths, and denominators (exposure time) for 5-year age groups from the HMD long series. Seven countries had detailed data from the HMD long series through 2021date. All remaining countries had data at least through 2019 (Table S1 ). For countries missing 2020 and/or 2021date data, we extracted data on deaths using the original input data from the STMF database. Countries reported deaths for different age bands in the STMF database, and we allocated deaths within STMF database age groups to 5-year age groups based on the age distribution of deaths in the last year available from the HMD long series. Denominators for 2020 and 2021date were obtained by fitting a linear trend to the last 5 years of exposure data available from the HMD long series (e.g. Italy, 2015–2019) for each 5-year age group in each country to obtain predictions for the most recent years. For country-years with overlapping data between the STMF database and HMD long series, ASMRs were very similar, supporting the validity of our approach (Table S1 ). Finally, we aggregated deaths and exposure time to the 10-year age bands reported by the United States in the STMF database: 0–4, 5–14, 15–24, …, and 85 + years. We also constructed six broader age groups to facilitate presentation of age-stratified analyses: 0–14, 15–44, 45–64, 65–74, 75–84, and 85 + .

As a benchmark against which to compute excess US mortalityterm·key, we calculated the average ASMR across the 21 other wealthy countries for each age group and year (excluding countries with no data available for that year). We weighted countries by population [following ( 8 )], comparing the mortalityterm·key risks faced by US residents with a representative set of residents of other wealthy nationsterm·key. (We present unweighted results in the supplement.) We then computed the number of deaths that would have occurred in the United States if the US population had experienced the age-specific mortalityterm·key profile of peer nations. We subtracted this number from the actual number of US deaths at each age to compute the number of excess US deathsterm·dom each year, i.e. the number of missing Americansterm·dom.

We also calculated excess US years of life lostterm·key (YLLterm·mth). YLLterm·mth captures the number of years that a person would have been expected to live had they survived, placing greater weight on deaths at younger ages. We defined “excess US YLLterm·mth” as the YLLterm·mth that would be averted if the United States had the ASMRs of other wealthy nationsterm·key. We estimated YLLterm·mth associated with observed US deaths, YLLterm·mth associated with US deaths under counterfactual (peer country) ASMRs, and “excess US YLLterm·mth” computed as the difference between US observed and counterfactual YLLterm·mth estimates. For each of these estimates, we weighted deaths by the age-specific life expectancyterm·key of other wealthy nationsterm·key in the year the death occurred (details are provided in the supporting information).

Next, we compared ASMRs for US racial/ethnic groups relative to the average of other wealthy nationsterm·key, combining all racial/ethnic groups in those countries into a single benchmark. We obtained data on US mortalityterm·key and population denominators for 1999–2021date, stratified by age and race/ethnicity, from the Multiple Cause of Death Files hosted on the CDC WONDER database (downloaded: 2022 January 5). We computed ASMRs for non-Hispanic Black (henceforth Black), non-Hispanic White (White), Hispanic, non-Hispanic Asian/Pacific Islander (Asian/Pacific Islander), and non-Hispanic American Indian/Alaskan Native (Native American) persons. Data from 2021date were provisional. Details are provided in the supporting information. For each racial/ethnic group, ASMRs were extracted for 10-year age bands, aligned with the data from the HMD. To avoid unstable estimates in race-by-age strata with low mortalityterm·key rates, we aggregated to wider age bands: 0–14, 15–44, 45–64, 65–74, 75–84, and 85 + years. We computed mortalityterm·key rate ratios and excess deathsterm·key for each US racial/ethnic group relative to the average of other wealthy nationsterm·key. Replication data and code for all tables and figures have been posted in a public data repository ( 30 ).

Results Long-term mortalityterm·key trends in the United States and other wealthy nationsterm·key prior to the COVID-19term·key pandemic Figure 1 displays long-term trends, 1933date–2021date, in age-standardized mortalityterm·key rates for each of the countries in the study. Mortalityterm·key has fallen dramatically across all countries. The early time series bears the imprint of WWII and its aftermath, with high mortalityterm·key rates observed in Europe and Japan. The United States had similar age-standardized mortalityterm·key rates as other wealthy nationsterm·key in the 1960s–1970s. However, since the 1980s, the position of the United States relative to other wealthy nationsterm·key has progressively deteriorated.

Age-standardized mortalityterm·key trends in the United States and other wealthy nationsterm·key. Figure shows deaths per 100,000 person-years: A) 1933date–2021date and B) 1980–2021date. The solid thick red line is the United States, the dashed thick grey line is the population-weighted average of 21 other wealthy nationsterm·key, and the thin grey lines are country-specific trends for each of the other nations. Total mortalityterm·key was age-standardized to the 2000 US population age distribution.

Figure S1 stratifies these trends by age group, revealing substantial age heterogeneity in the US position relative to other countries. Despite similar aggregate mortalityterm·key rates in the 1960s and 1970s, the United States had higher mortalityterm·key at ages 15–64 years and lower mortalityterm·key among people ages 75 years and older during this period. These patterns changed in the 1980s and 1990s as the mortalityterm·key advantage of older US residents disappeared. During the period 2000–2021date, the gap between the United States and other wealthy countries widened particularly among nonelderly adults (15–64 years). Mortalityterm·key among US nonelderly adults slowed its decline after 2000 and increased in absolute terms starting in 2010 (Fig. S1 ).

The changing position of the United States relative to other wealthy nationsterm·key is visualized in Fig. S2 , which displays trends in age-specific mortalityterm·key ratios between 1933date and 2021date. The US mortalityterm·key advantage in the late 1930s and 1940s reflects the disproportionate mortalityterm·key shock of WWII on younger age groups in many of the comparison countries. Post-WWII, however, younger US age groups had elevated mortalityterm·key rates relative to peer nations, while older US age groups had lower mortalityterm·key. This pattern began to shift in the late 1970s as the US advantage at older ages started to diminish. Since 2000, trends in rate ratios for different age groups have diverged, increasing rapidly for persons under 65 years while remaining approximately stable (until the onset of the COVID-19term·key pandemic) for the older age group.

The COVID-19term·key pandemic appears as a spike in mortalityterm·key in 2020 and 2021date (Fig. 1 ). The size of the 2020 increase differed by country and age group. While many countries experienced mortalityterm·key increases among persons 75 years and older, the United States far outpaced other countries in the mortalityterm·key among adults under 65 years (Fig. S1 ). From 2020 to 2021date, mortalityterm·key rates in the United States and most other peer nations declined in older age groups (over 75 years). However, in the United States, mortalityterm·key continued to rise in younger age groups.

Figure 2 A shows, for each year 1933date–2021date, the number of deaths that occurred in the United States and the counterfactual, the number of deaths that would have occurred if the United States had the ASMRs of other wealthy nationsterm·key. Figure 2 B shows the number of missing Americansterm·dom each year. Since the early 1980s, the United States has experienced a steady and approximately linear increase in the number of excess deathsterm·key relative to other wealthy nationsterm·key. In 2019 alone, on the eve of the COVID-19term·key pandemic, there were 622,534 excess US deathsterm·dom (Table 1 and Fig. 2 B). Between 1980 and 2019, 11.0 million excess US deathsterm·dom occurred (Fig. 2 and Table S3 ).

Excess deathsterm·key and YLLterm·mth in the United States relative to other wealthy nationsterm·key, 1933date–2021date. Source: Human Mortalityterm·key Database. Note: The figure shows number of deaths in the United States and the counterfactual, the number of deaths that would have occurred in the United States if the United States had ASMRs equal to the average of 21 other wealthy nationsterm·key. The average of other wealthy nationsterm·key excludes Portugal prior to 1940, Austria and Japan prior to 1947, New Zealand before 1948, Ireland before 1950, Germany prior to 1956, and Luxembourg prior to 1960. From 1960, all countries are represented. The panels show: A) deaths in the United States and the counterfactual; B) the difference between the two, i.e. the number of excess US deathsterm·dom or “missing Americansterm·dom” each year; C) and D) analogous plots for YLLterm·mth, where each death is weighted by the age-specific life expectancyterm·key of other wealthy nationsterm·key in the year it occurred. Plotting points for all figures are provided in Table S3 .

Excess US mortalityterm·key relative to other wealthy nationsterm·key: 2019–2021date.

The rising number of missing Americansterm·dom was driven by a reduction in the US mortalityterm·key advantage over 65 years in the 1980s and 1990s and increases in the US mortality disadvantageterm·dom under 65 years since 2000 (Fig. S3 ). By 2019, half (51%) of these excess US deathsterm·dom were in persons younger than 65 years. In 2019, 43% of all US deaths under 65 years and 14% of all US deaths over 65 years would have been avoided if US mortalityterm·key rates had equaled the average of other wealthy nationsterm·key (Tables 1 and S2 ).

The increase in excess US mortalityterm·key is even more stark when using a YLLterm·mth metric. Figure 2 C shows YLLterm·mth for observed deaths and the counterfactual. YLLterm·mth increased in the United States after 1980; however, this increase would not have occurred if the United States had the ASMRs of other wealthy countries. Figure 2 D shows the difference in YLLterm·mth associated with excess US deathsterm·dom. The gap in YLLterm·mth between the United States and the counterfactual began to grow in the 1960s but has accelerated since the 1990s. Between 1980 and 2019, YLLterm·mth attributable to excess deathsterm·key increased from 3.2 million per year to 15.8 million per year, claiming a cumulative total of 334.8 million YLLterm·mth (Table S3 ). The rise in excess US deathsterm·dom and excess US YLLterm·mth was not explained by population aging, as shown in Fig. S4 , which shows age-standardized trends.

Figure 3 presents detailed age profiles of US excess mortalityterm·key for selected years: 1933date, 1960, 1980, 2000, and 2019–2021date. The figure displays mortalityterm·key rate ratios, excess US deathsterm·dom, and excess YLLterm·mth in the United States relative to other wealthy nationsterm·key. In all plots, higher numbers reflect worse outcomes for the United States relative to its peers. The divergence of the United States from peer countries occurred at different points in time for different age groups. As shown in Fig. 3 A, in both 1933date and 1960, ASMRs were broadly similar in the United States and the average of peer countries. These years bookend a period, during WWII and its aftermath, when US mortalityterm·key rates were substantially lower than those of peer nations. From 1960 to 1980, US-to-peer country mortalityterm·key rate ratios increased for younger age groups (15–34 years). From 1980 to 2000, mortalityterm·key rate ratios increased for older age groups (65 + years), although they remained lower than the mortalityterm·key rate ratios for younger adults. From 2000 to 2019, US mortalityterm·key rate ratios increased dramatically among young and middle-aged adults.

Excess US mortalityterm·key by age group for select years: 1933date–2021date. Source: Human Mortalityterm·key Database. Note: The figure shows A) mortalityterm·key rate ratios, B) excess US deathsterm·dom, and C) YLLterm·mth in the United States relative to what would have occurred if the United States had ASMRs equal to the average of 21 other wealthy nationsterm·key. The group of other wealthy nationsterm·key excludes Portugal, Austria, Japan, New Zealand, Ireland, and Germany in 1933date. From 1960, all countries are represented. Age groups are 0–4, 5–14, 15–24, …, and 85 + years.

Figure 3 B shows the number of excess US deathsterm·dom by age and year. These absolute magnitudes are greater at older ages when most deaths occur. Figure 3 C shows these differences using the YLLterm·mth metric, which places additional weight on deaths occurring earlier in life. Figure 3 B and C affirms that the absolute increase in excess US deathsterm·dom and YLLterm·mth between 1980 and 2000 occurred at older ages, while the increase in excess US deathsterm·dom and YLLterm·mth between 2000 and 2019 occurred primarily among non-elderly adults.

Impact of COVID-19term·key pandemic on the US mortality disadvantageterm·dom The last 2 years of our mortalityterm·key series reflect the toll of the COVID-19term·key pandemic, which in 2020 caused the largest single-year increase in mortalityterm·key for the United States and many other wealthy countries since WWII. From 2019 to 2020 and 2021date, mortalityterm·key increased in all US age groups aged 15 and older (Fig. S1 and Table 1 ).

Other countries experienced smaller increases in AMSRs in 2020 and 2021date than did the United States. If the United States had the ASMRs of other wealthy nationsterm·key, the increase in US deaths between 2019 and 2020 would have been 73% smaller, and the increase between 2019 and 2021date would have been 78% smaller. During this period, the number of missing Americansterm·dom grew dramatically, from 622,534 in 2019 to 1,009,467 in 2020 and 1,090,103 in 2021date. These excess US deathsterm·dom accounted for 30% and 32%, respectively, of all US deaths (from all causes) in 2020 and 2021date (Tables 1 and S2 ).

The COVID-19term·key pandemic took a much larger toll on young and working-age adults in the United States than in other wealthy nationsterm·key (Fig. S1 ). If the United States had the ASMRs of its peers, 88% of the 2019 to 2020 increase in US mortalityterm·key under age 65 and 92% of the 2019 to 2021date increase in US mortalityterm·key under age 65 would have been avoided. Excess US deathsterm·dom accounted for 50% of all US deaths under age 65 (from all causes) in 2020 and 55% of all US deaths under age 65 in 2021date (Tables 1 and S2 ). During 2020 and 2021date, there were 968,261 excess US deathsterm·dom under 65 years, nearly half of the total number of missing Americansterm·dom during this period.

Annual YLLterm·mth associated with the US mortality disadvantageterm·dom increased from 15.8 million in 2019 to 22.5 million in 2020 and 26.4 million in 2021date, highlighting both the greater increase in mortalityterm·key in the United States relative to peer countries and the younger ages at which these deaths occurred (Fig. 2 and Table S3 ). Estimates of excess US deathsterm·dom in 2019–2021date were qualitatively similar when using the alternative international comparison sets of the G7 and five western European countries and when using the unweighted average of peer nations (Table S4 ). Between 1980 and 2021date, there were a total of 13.1 million excess US deathsterm·dom and 383.7 million excess YLLterm·mth.

Who are the missing Americansterm·dom? Excess mortalityterm·key by US race/ethnicity Figure 4 A presents ratios comparing ASMRs for each US racial/ethnic group with the average of other wealthy countries. Data are presented for 4 years—1999, 2019, 2020, and 2021date—enabling assessment of trends in excess US deathsterm·dom by race/ethnicity both before and during the COVID-19term·key pandemic. The figure shows important commonalities across different US racial/ethnic groups. First, mortalityterm·key rate ratios relative to other nations increased over time for all race/ethnicities, indicating worsening position relative to other wealthy nationsterm·key. Second, mortalityterm·key rate ratios were largest among younger adults, indicating that the pattern of elevated early- and mid-life US mortalityterm·key is not limited to one specific race or ethnicity.

Excess mortalityterm·key of US racial/ethnic groups relative to the average of other wealthy nationsterm·key, 1999–2021date. Sources: Multiple Cause Mortalityterm·key Files from CDC Wonder and Human Mortalityterm·key Database. The figure shows A) mortalityterm·key rate ratios comparing US racial/ethnic groups with the average of 21 other wealthy nationsterm·key and B) numbers of excess US deathsterm·dom for each group, stratified by age: 0–14, 15–44, 45–64, 65–74, 75–84, and 85 + years. US racial/ethnic groups are Hispanic, non-Hispanic Black (Black), non-Hispanic White (White), non-Hispanic Asian/Pacific Islander (Asian/Pacific Islander), and non-Hispanic American Indian/Alaskan Native (Native American).

However, there were also key differences in levels and trends (Fig. 4 , Table 2 ). In 2019, Black and Native American US residents aged 15–44 years had mortalityterm·key rates 3.8 and 5.3 times higher, respectively, than the average of other wealthy countries. Mortalityterm·key among White and Hispanic US residents ages 15–44 years was also elevated at 2.5 and 1.8 times the average of other nations, respectively. In contrast, in 2019 Asian/Pacific Islander US residents had mortalityterm·key rates below the average of other nations. White and Native Americans experienced the largest relative increases in mortalityterm·key from 2000 to 2019. Figure 4 B displays the age profile of excess deathsterm·key by year and race/ethnicity, revealing how differences in the population size and age distribution shape racial/ethnic differences in excess deathsterm·key.

Excess US deathsterm·dom by race/ethnicity relative to other wealthy nationsterm·key: 2019–2021date.

During the COVID-19term·key pandemic, mortalityterm·key rates rose for all US racial/ethnic groups (Fig. 5 ). Native American, Black, and Hispanic US residents experienced the largest increases, and nonelderly adults of these racial and ethnic groups experienced the largest relative increases compared with other wealthy nationsterm·key. For example, from 2019 to 2021date, age-standardized, under-65 mortalityterm·key increased by just 6 deaths per 100,000 in other wealthy nationsterm·key. However, in the United States, under-65 mortalityterm·key increased by 254 deaths per 100,000 among Native American residents, 143 deaths per 100,000 among Black residents, 108 deaths per 100,000 among Hispanic residents, 72 deaths per 100,000 among White residents, and 36 deaths per 100,000 among Asian residents (Fig. 5 ). The increase in Hispanic Americans' death rates during the pandemic largely erased their previous survival advantage relative to non-Hispanic White Americans and pushed their mortalityterm·key rates above the average of other wealthy countries at all but the oldest ages. By 2021date, Black and Native Americans ages 15-44 had mortalityterm·key rates 5.3 and 8.3 times higher than the average of other wealthy nationsterm·key; White and Hispanic Americans ages 15–44 had mortalityterm·key rates 3.0 and 2.7 times the peer country average (Fig. 4 A, Table 2 ).

Age-standardized mortalityterm·key rates for US racial and ethnic groups and the average of other wealthy nationsterm·key: 2019, 2020, and 2021date. Source: CDC Wonder and Human Mortalityterm·key Database. Notes: OWN, other wealthy nationsterm·key. Figure shows age-standardized mortalityterm·key rates for A) all ages, B) under 65 years, and C) 65 + years. Mortalityterm·key rates are standardized to the US population age distribution in each year. Mortalityterm·key rates are shown for the following US populations: Hispanic, non-Hispanic White (White), non-Hispanic Black (Black), non-Hispanic Asian/Pacific Islander (Asian), and non-Hispanic American Indian/Alaskan Native (AIAN). Dashed line shows 2019 (pre-COVID) mortalityterm·key rates for the average of other wealthy nationsterm·key.

The data reveal noteworthy changes between 2020 and 2021date. Whereas mortalityterm·key rates declined among Black Americans in the second year of the pandemic, mortalityterm·key continued to increase among White and Native Americans (Fig. 5 A), leading to large increases in excess United States deaths in these populations. From 2020 to 2021date, all racial/ethnic groups experienced a substantial increase in mortalityterm·key rates among persons under 65 years (Fig. 5 B). This occurred amidst falling mortalityterm·key at older ages among US Black and Hispanic residents and little change in over-65 mortalityterm·key for other racial/ethnic groups (Fig. 5 C). White US adults over 65 saw no reduction in mortalityterm·key between 2020 and 2021date.

Table 2 and Fig. 4 B show the absolute number of missing Americansterm·dom by US racial/ethnic group. Despite higher mortalityterm·key rates among Black and Native Americans, in absolute numbers, the majority of missing Americansterm·dom were White, reflecting the much larger size and older age distribution of the US White population. While White Americans accounted for 61% of the population and experienced 69% of excess deathsterm·key in 2021date, Black Americans accounted for 13% of the population and 21% of excess deathsterm·key. Hispanic Americans, who are younger on average, made up 19% of the population and 8% of excess deathsterm·key (Table 2 ).

Discussion The current US mortality disadvantageterm·dom is unprecedented in modern times: there were more “missing Americansterm·dom” in 2021date than at any other point going back to 1933date. The United States had lower death rates than other wealthy nationsterm·key in the 1930s through 1950s and similar death rates in the 1960s and 1970s. US death rates began to diverge in the 1980s, and this divergence has accelerated in the 21st century. When COVID-19term·key emerged, Americans were already dying at much higher rates than residents of other wealthy nationsterm·key. There were 622,534 excess US deathsterm·dom in 2019 alone and 11.0 million total excess US deathsterm·dom between 1980 and 2019. The annual death toll from the US mortality disadvantageterm·dom increased to 1,009,467 in 2020 and 1,090,103 in 2021date, reflecting the catastrophic loss of life the United States experienced during the pandemic, far in excess of other wealthy nationsterm·key. Missing Americansterm·dom represent a large share of under-65 mortalityterm·key in the US. In 2020 and 2021date, more than half of all US deaths under 65 years would have been averted if the United States had experienced the mortalityterm·key rates of its peers. During the first two years of the COVID-19term·key pandemic, there were nearly 1 million excess US deathsterm·dom under 65 years. The 1.1 million missing Americansterm·dom in 2021date would have enjoyed an estimated 26.4 million additional years of life had they survived.

Our findings are consistent with recent reports indicating a growing life expectancyterm·key gap between the United States and other countries ( 17 , 22 , 29 , 31 , 32 ). Using a similar set of comparator countries, Woolf and colleagues estimated that the US life expectancyterm·key deficit increased from 1.9 years to 3.1 years between 2010 and 2018 and widened further to 4.7 years in 2020 ( 17 ) and 5.3 years in 2021date ( 29 ). While period life expectancyterm·key is a widely used summary measure of mortalityterm·key, it may be misinterpreted as reflecting mortalityterm·key differences at end of life rather than at younger ages where the greatest differences lie between the United States and peer nations.

“Missing Americansterm·dom” offers an alternative, easily interpretable metric: the number of US deaths that would have been averted if the United States had the ASMRs of its peers. Our analysis builds on our ( 11 ) and others' ( 7 , 8 ) previous analyses documenting excess US deathsterm·dom prior to COVID-19term·key and during the pandemic ( 21 ), as well as research on the racial distribution of excess deathsterm·key before the pandemic ( 11 , 33 ) and in pandemic-associated mortalityterm·key ( 24 , 25 , 34 , 35 ).

Other studies have computed “excess deathsterm·key” relative to different benchmarks. Our choice of an international comparator offers insights not available from comparison groups within the United States. For example, studies computing excess mortalityterm·key relative to pre-COVID-19 trends within the same country take prepandemic mortalityterm·key as given ( 14 ). However, US mortalityterm·key has diverged from other wealthy nationsterm·key since the 1980s, after tracking close to other nations for most of the post-WWII period. As we show, there were over 600,000 missing Americansterm·dom in 2019 even before COVID-19term·key. High levels of excess US mortalityterm·key are likely to continue even as COVID mortalityterm·key wanes. Other studies have compared mortalityterm·key rates across racial/ethnic or socioeconomic groups within the United States ( 24 ). However, life expectancyterm·key of White Americans—particularly those with less than a college degree and those residing in nonmetro areas—has stagnated or fallen since the 1990s ( 5 , 36 ). Using White Americans as the reference group or benchmark for comparison renders invisible the mortalityterm·key crisis among White US residents and underestimates excess mortalityterm·key for other US racial and ethnic groups.

About half of the missing Americansterm·dom died before reaching age 65. The United States has had higher working-age mortalityterm·key than other wealthy nationsterm·key since the 1960s, but the gap has widened dramatically in recent years. Excess US deathsterm·dom under 65 doubled between 2000 and 2019 and then increased by an additional two-thirds between 2019 and 2021date. During the pandemic, half of all US deaths under 65 years—and 90% of the increase in under-65 mortalityterm·key since 2019—would have been averted if the United States had the ASMRs of its peers.

Reports by the National Academies of Science, Engineering, and Medicine have identified multiple proximate causes for excess US mortalityterm·key at working ages ( 3 , 7 ). Higher rates of homicides, motor vehicle fatalities, sexually transmitted infections including HIV/AIDS, drug-related mortalityterm·key, obesity and diabetes, heart disease, and lung disease explained much of the US mortality disadvantageterm·dom during the 1990s and early 2000s ( 3 ). From 2000 to 2019, the increase in excess US mortalityterm·key under 65 years was due largely to increases in drug overdoses, alcohol-related mortalityterm·key, suicides, and cardiometabolic diseases ( 7 ). In addition to having shorter lives, US working-age adults experience greater morbidity than do residents of peer countries. The World Health Organization estimates than in 2019, US “health-adjusted life expectancyterm·key” (HALE) lagged unadjusted life expectancyterm·key by 12.4 years—more life-years spent in poor health than any other wealthy nationterm·key ( 37 ). Higher prevalence of diabetes and obesity in the United States likely contributed to elevated COVID-19term·key mortalityterm·key among working-age adults, due to increased severity of COVID-19term·key infection ( 38 ). Deaths to working-age adults have social impacts not directly captured in mortalityterm·key statistics, including psychological trauma and economic precarity for surviving children, friends, and family, as well as lost economic productivity and social and political participation ( 39–42 ).

The large spike in early death in the United States during the 2020–2021date COVID-19term·key pandemic was due not only to the emergence of a novel pathogen. Other nations were similarly exposed. Rather, the mortalityterm·key impact of the pandemic in the United States was a product of longstanding social inequities and policy failures that made the United States particularly vulnerable, factors that had also contributed to the US mortality disadvantageterm·dom before the pandemic. US policies throughout the country's history have harmed the health of Black and Native Americans ( 23 ), beginning with the genocide of native populations and the enslavement of people of African descent ( 43 , 44 ) and continuing through long periods of legal segregation, political disenfranchisement, and economic exclusion. These harms have reverberated through the intergenerational transfer of resources, power, and health.

In recent decades, policy failures have adversely affected the health of US residents of all racial/ethnic groups, particularly those without a college degree ( 45 ). Since the 1970s, the United States—like other wealthy countries—has undergone structural economic changes, with increased exposure to trade ( 46 ), automation ( 47 ), and a shift to service sector employment. The United States has failed to protect less-educated workers from the adverse health consequences of these changes. Stagnant minimum wages and losses of collective bargaining protections ( 48 ) have contributed to widening economic inequality. A scant safety net for working-age adults and the absence of universal healthcare have privatized risk, tying health more closely to personal wealth and employment ( 49 ). Additionally, lax regulation of opioids, firearms, environmental pollutants, unhealthy foods, and workplace safety has contributed to elevated US mortalityterm·key, particularly among lower-educated and lower-income people. Increasingly divergent policies at the state level have resulted in widening health gaps across US states ( 50 ). In those geographic areas of the United States where excess mortalityterm·key has increased the most, voters have turned towards policy-makers who have further undermined population health ( 51 , 52 ), e.g. through refusal to expand Medicaid or to implement firearm regulations ( 53 ). Underlying these policy trends, the commitment of a majority of US White residents to a political program of deregulation and divestment from the social determinants of health has hamstrung government efforts to protect the health of all US residents, inflicting the greatest harm on the most vulnerable, incuding Americans of color and White Americans without a college degree ( 54 ).

During COVID-19term·key, these same “fundamental causes”—socioeconomic inequality, a limited safety net, and a politics unaccountable to population health—manifested in the large increase in excess US deathsterm·dom that we report here. Residential segregation, limited ability to work from home, distrust of the medical establishment, decreased access to health care, and a higher prevalence of comorbid conditions contributed to excess risk among low-income and minoritized populations ( 34 , 55 ). Gaps in the safety net forced many low-income people to work despite high infection risk ( 13 ). Housing insecurity increased crowding and transmission at home. The exclusion of immigrant populations from the safety net ( 56 ), anemic workplace protections ( 57 ), and crowded housing ( 58 ) contributed to the large impact of COVID-19term·key, particularly among US Hispanic residents ( 59 ). Failure to mandate that employers give their workers time off to get vaccinated contributed to disparities in uptake ( 60 ). Finally, the politicization of the pandemic—including mask use, physical distancing, and vaccine uptake—has resulted in substantial COVID-19term·key–related mortalityterm·key and contagion, particularly among US White residents and those with less than a college education ( 61 ). The rise in mortalityterm·key among White US residents between 2020 and 2021date was likely a consequence of low vaccine uptake in this population. We note that the rise in US mortalityterm·key during the pandemic included a large rise in deaths due to drug overdoses ( 62 ), reflecting the failure of US policy to adequately address its longstanding drug crisis.

In contrast to the decentralized approach to pandemic control taken in the United States, other countries launched rapid, coordinated national responses; had greater adherence to control measures; outpaced the United States in vaccine uptake; had better baseline health conditions due to universal health care and a stronger safety net; coordinated economic policy that sought to maintain ties between employers and their workers; and avoided long-term interruptions in children's schooling that occurred in the United States. These countries not only weathered the emergency phase of the pandemic better than the United States but were also better poised to protect future population health at the close of 2021date.

Our analysis has limitations. First, the mortalityterm·key data for 2021date are preliminary due to the lag time between the occurrence and registration of deaths. Still, in previous years, by 26 weeks after the end of the year, 99% of US deaths have been recorded ( 63 ). Data for Japan and Ireland had not yet been posted for 2021date and were imputed using 2020 death rates. Second, differential reporting of US race and ethnicity on death certificates and in Census denominators may bias estimates. In particular, American Indian/Alaskan Native “race” and Hispanic “ethnicity” are underreported on death certificates, which may lead to underestimation of deaths for these groups and overestimation for other groups ( 64 , 65 ). Third, the literature on US mortalityterm·key in international context has used a range of country comparison sets and results vary somewhat depending on the countries included and whether weighted or unweighted averages are used. We utilized data on all other wealthy countries with a lengthy time series, excluding former communist countries, which experienced a large mortalityterm·key crisis in the 1990s following the collapse of the Soviet Union. However, our findings are qualitatively similar when using alternate comparison sets used in other studies of excess US deathsterm·dom ( 8 , 11 ) or using unweighted averages of peer countries. Due to data limitations, the composition of our comparison group changed between 1933date and 1960; however, it was consistent after 1960. Fourth, our analysis takes a period approach to estimating the number of missing Americansterm·dom, similar to the literature on international life expectancyterm·key comparisons. Estimating the number of US residents who would be alive today if counterfactual mortalityterm·key rates persisted for their lifetime would require a different, cohort-based approach. Finally, the missing Americansterm·dom in this analysis do not reflect all the US residents lost to premature death—nor do they reflect all deaths that would be averted if the US had the mortalityterm·key rates of international top performers.

The future of COVID-19term·key mortalityterm·key is uncertain, but even a full return to prepandemic conditions would hardly end the US mortalityterm·key crisis. Our study showed that even before the pandemic, the United States experienced more than 600,000 excess US deathsterm·dom each year, with most occurring among Americans under 65 years. How can the United States reduce the number of missing Americansterm·dom? Expanding current US policies with demonstrated health impact and ensuring that all US residents have access to health care and public social welfare benefits when they need them are places to start. For example, research has documented the health benefits of expanded access to Medicaid ( 66 , 67 ), minimum wage and Earned Income Tax Credit laws ( 68 ), unemployment insurance subsidies ( 69 ), nutritional assistance ( 70 ), early childhood education ( 71 ), and firearm ( 72 ) and environmental regulations ( 73 ). Learning from peer nations, the United States could extend social protections and universal health coverage to nonelderly adults and enhance the regulation of pollutants and firearms. Preventing future missing Americansterm·dom will require efforts to address fundamental causes of poor population health in the United States that existed even before the pandemic began ( 44 , 74 ).

Supplementary Material Click here for additional data file.