Increasing burden of opioid overdose mortality in the United States: Years of life lost by age, race, and state from 2019-2021

The United States’ opioid crisis is worsening, but the nationwide burden has not been characterized during the COVID-19 pandemic. In this study we calculate years of life lost to opioid overdose deaths by demographic group and examine trends from 2019 to 2021. Using the Multiple Cause of Death dataset from CDC WONDER, we extracted opioid overdose deaths stratified by race/ethnicity, age, and state, and estimated crude and age-adjusted mortality rates, years of life lost, and reduction in life expectancy at birth. Increasing annually, opioid overdose deaths reached 80,411 in 2021, leading to 3 million years of life lost, and reducing life expectancy by 0.65 years. From 2019 to 2021, opioid overdose death rates increased across all groups. American Indian/Alaska Native and Black/African American men now experience the highest burden, with 1,500 years of life lost per 100,000, life expectancies at birth reduced by almost 1 year, and 46 and 51 deaths per 100,000, respectively. This study highlights the continued growth of the opioid crisis, and the surge in years of life lost coinciding with COVID-19. There is an urgent need for more effective interventions, particularly in light of this epidemic’s changing demographics.


Main text
The opioid epidemic is an urgent public health issue in the United States.With use and misuse increasing since the mid-1990s, the US is now the largest consumer of opioids globally 1 .Fatal opioid overdoses in the US have increased more than tenfold since 1999, reaching a record of 80,411 deaths in 2021 2 .The initial rise in opioid-related deaths is commonly attributed to rising prescription rates following aggressive marketing by the pharmaceutical industry, which was inadequately regulated by the Food and Drug Administration (FDA) [3][4][5][6][7] .In attempts to curb the epidemic, some state and national regulations on opioid prescriptions were implemented 8,9 , and in 2016 the Centers for Disease Control and Prevention (CDC) issued prescription recommendations 10 .While this reduced prescriptions nationally [11][12][13][14] , opioid overdose deaths continue to rise.The crisis is now driven primarily by illicitly manufactured opioids 15,16 .Following the onset of the COVID-19 pandemic, there was a surge in both fatal and non-fatal opioid overdoses in 2020 [17][18][19] .
Opioids are a class of drugs that block pain signals by binding to opioid receptors on neurons throughout the nervous system.Recreationally used opioids can be derived from poppies (e.g., morphine, opium), created semi-synthetically by chemical modification of naturally-occurring compounds (e.g., heroin, oxycodone) or created completely synthetically (e.g., fentanyl, tramadol).Opioids are prescribed to treat moderate to severe pain, either as a short-term or chronic treatment.However, prescription opioids are also used recreationally for euphoric effects, by individuals obtaining them either through prescriptions or other means [20][21][22][23][24][25][26] .An estimated 12.5% of US adults with an opioid prescription (11.5 million) misused it in 2015, and 2% of prescribed opioid users (or up to 12% for long-term chronic pain prescriptions) develop opioid use disorder (OUD) 27,28 .OUD includes addiction, increased tolerance, and withdrawal symptoms when use is stopped, and contributes to health and social problems, such as incomegenerating criminal activity [29][30][31] .Individuals with OUD have 10 times higher mortality than the general population, and experience overdose death rates of about 8 per 1,000 person-years 32,33 .While effective medications exist for OUD, it is estimated that only 28% of people needing treatment receive it (and only 43% receive any intervention) 34,35 .
While opioid overdose mortality rates are alarmingly high, measures of deaths alone do not account for the young age at which most opioid-related deaths occur, compared to other causes of death 36 .Years of life lost is a particularly fitting metric, estimating how many additional years individuals would have lived, had they not died prematurely from the cause in question 37 .The YLL due to the opioid crisis have been estimated for Ontario, Canada between 1990 and 2010 38 , in the US between 2001 and 2016 39 , and for the state of Ohio in 2016 40 .The YLL to opioids in Washington State were compared across racial and ethnic groups, by educational attainment, and urbanicity 41 .There are estimates of the YLL from fatal overdoses involving opioids and other drugs in adolescents 42 and in Black women 43 .Studies have examined the impact of opioid-related deaths on the US life expectancy at birth in 2015 44 , by state and opioid type in 2016 45 , and across demographic groups in 2017 46 .However, these metrics of burden have not been used to evaluate recent impacts of the crisis, or to compare comprehensively across demographic groups.
In this study, we quantify the evolution of the opioid crisis in the years 2019, 2020 and 2021, coinciding with the onset of the COVID-19 pandemic.For each year in question, we estimate the years of life lost to opioid overdose deaths in the United States, in addition to estimating how much this reduces the US population's life expectancy at birth.We compare these metrics of burden across demographic groups, highlighting which groups have seen the largest increases in the last few years.We also examine how the number of deaths, death rates, and age patterns vary over time and by demographic group.

Data
This study uses data from the Multiple Cause of Death database from CDC WONDER, the Wide-ranging ONline Database for Epidemiological Research.For each year in question (2019, 2020 and 2021), we accessed data on the number of deaths, and crude and age-adjusted death rates by cause of death, with the option of stratifying by age, gender, race, Hispanic origin, or state.At the time of writing, 2021 is the most recent year for which data is available on CDC WONDER.All age-adjusted rates reported here used the 2000 Standard Population.In this database, when fewer than 10 death counts are in any given category, the statistics are suppressed to ensure confidentiality.For this reason, we avoid subdividing by more variables than necessary, and when stratifying opioid overdose deaths by age, we use 5-year age groups.The mortality data are based on death certificates of US residents in all counties.Causes of death are categorized by Underlying Cause of Death and by Multiple Cause of Death, using ICD-10 codes (International Classification of Diseases).For opioid overdose deaths, we filter for deaths with both an underlying cause of death code indicating drug poisoning (X40-44: unintentional, X60-64: suicide, X85: homicide, Y10-14: undetermined intent) and a multiple cause of death indicating an opioid (T40.0-T40.4:Opium, Heroin, Natural and Semisynthetic Opioids, Methadone, and Synthetic Opioids, and T40.6:Other and Unspecified Narcotics).This is a standard method of classifying opioid-related overdose deaths 3 .
CDC WONDER reports gender in 2 categories: "male" or "female".We used the option to subdivide into 6 race categories: American Indian/Alaska Native (AI/AN), Asian, Black or African American, Native Hawaiian or Other Pacific Islander (NHOPI), white, and "More than one race".Ethnic origin is reported separately, with 3 categories: Hispanic or Latino, non-Hispanic or Latino, or "Not Stated".We omit individuals with missing ethnic origin ("Not Stated"), who account for only 0.2% of all deaths and 0.6% of opioid overdose deaths in 2021.This leads to 12 remaining combinations of race and Hispanic origin, many of which with very few individuals (such as Hispanic origin and NHOPI race).For this reason, we combine race and ethnic origin, forming 7 groups: one containing all individuals of Hispanic or Latino origin and of any race (of whom 96% identified as "white"), and 6 groups containing individuals of non-Hispanic or Latino origin, grouped by race.
This study also uses 2019 and 2020 US population period life tables from the NVSS, reported by gender and race/ethnicity, using groups similar to those described above 47,48 .Life tables track the mortality experience of a population, and are commonly used in demography and actuarial science [49][50][51] .

Life expectancy and years of life lost
We estimate the impact of the opioid crisis on life expectancy in the US.Life expectancy is the average remaining length of life of individuals at each age, assuming that current age-specific mortality patterns were to remain constant throughout their lifetimes.The reduction in life expectancy due to a cause of death accounts for both the number of deaths, and the age at which individuals are dying.In general, the impact of a particular cause of death on life expectancy can be estimated by first finding what the life expectancy would have been in the absence of the cause (the counterfactual life expectancy).The counterfactual age-dependent mortality rates in the absence of the cause of death in question are estimated by subtracting the number of deaths due to this cause from the total deaths, for each age.From these counterfactual mortality rates, the life expectancy in the absence of the cause can be calculated using standard life table equations 47 .For each age, the difference between real life expectancies (rLE age ) and the estimated counterfactual life expectancies (cLE age ) reveals the impact that the cause in question has on life expectancy:

‫ܧܮ∆‬ ൌ ‫ܧܮܿ‬ െ ‫ܧܮݎ‬
Using this method, we estimate the reduction in life expectancy at birth due to the opioid epidemic, for the entire US population, and by demographic group.
We can calculate years of life lost (YLL) due to the particular cause of death, by multiplying the number of deaths at each age due to this particular cause (N age ) by the counterfactual life expectancy at that age in the absence of the cause of death (cLE age ).Total YLL are obtained by summing over all ages:

‫ܮܮܻ‬ ൌ ሺܰ ൈ ‫ܧܮܿ‬ ሻ
Here, we estimated YLL in the US due to the opioid crisis in the years 2019, 2020 and 2021, separately for each demographic group (race and gender).Due to the CDC WONDER's suppression of counts <10 (see above) we accessed opioid overdose death counts in 5-year age groups, and assumed that the deaths occurred at the midpoint of the age intervals.
Previous studies of the YLL from overdose deaths have estimated the years lost from one individual's death by comparing their age at death to either a specific cutoff age 41 , or to the remaining LE for that age [38][39][40]42,43 . In ths study, we compare the age at death to the estimated counterfactual remaining LE.For major causes of death that have a significant impact on the population's life expectancy, accounting for the resulting reduction in LE avoids underestimating the YLL.
CDC WONDER censors population counts above the age of 85, meaning we have no denominator for calculating death rates for those ages.However, since only 0.1% of opioid overdose deaths occurred at ages 85-100 in 2019-2021, we assume that these deaths do not impact the population's life expectancy.We therefore complete our counterfactual life tables with data from the NVSS life tables, scaled by a multiplicative factor in order to match the values at age 84.For 2019 and 2020 estimates we used the corresponding NVSS values, and for 2021 we used the 2020 NVSS values since 2021 tables were not released at the time of writing.
These results and more can be found online on our interactive web app, which visualizes data on the opioid crisis: https://annehebert.github.io/dashboard.html

Results
Opioid overdose deaths in the United States have been on the rise since the mid-1990s (Figure 1A).In particular, there was a 38% increase in opioid-related deaths from 2019 to 2020, and a 17% increase from 2020 to 2021, reaching 80,411 deaths in 2021 (Table 1).More men die of opioid-related deaths each year than women: in 2021, 70% of deaths (56,757/80,411) were among men (Table 1).In all demographic groups, opioid overdose deaths have substantially increased over time.Of the individuals dying of opioid overdoses in 2021, 53,022 were white, 14,537 Black, 9,921 Hispanic, 932 American Indian/Alaska Native (AI/AN), 533 Asian and 62 Native Hawaiian and Other Pacific Islander (NHOPI), as well as 926 non-Hispanic individuals of more than one race (Table 1).Although opioid overdose deaths are distributed across all age groups, most deaths are in 25-44 year olds.
Both prescription drugs and a variety of illicitly manufactured opioids are responsible for fatal opioid overdoses.The sharp increase in overdose deaths since 2015, and in particular since 2019, is primarily driven by synthetic opioids (Table 1).In 2021, 70,601 overdose deaths involved synthetic opioids (mainly fentanyl 52 ), nearly doubled since 2019 (Table 1).The synthetic opioid overdose death rate increased from 3 deaths per 100,000 people in 2015 to 11 in 2019, and then 21 per 100,000 in 2021 (Figure 1B).The rate of overdose deaths involving heroin rose from 2010 to 2015 (from 1 to 4 per 100,000 people), but has since seen a decrease to 2.8 deaths per 100,000 in 2021.A majority of opioid overdose deaths occur among younger people.In the earlier stages of this crisis, more middle-aged adults were dying than younger adults (Figure 2A).For example, in 2010 a majority of deaths among women occurred around 50 years old, and deaths among men were fairly evenly distributed across ages (Figure 2A).However, in recent years there has been a significant increase in deaths among young adults, bringing the median age of individuals dying of opioid overdoses to 40 years old for men, and 41 for women in 2021.Despite the large difference in the raw number of opioid-related deaths between genders in younger age groups, the percentage of total deaths at each age that involve opioids is more similar between genders (Figure 2B).Nationwide, in 2021, more than 20% of all deaths of women at each age between 20-34 years involved opioids, with the highest contribution being 24% of 26-29 year old deaths.More than 20% of all deaths of men at each age between 23-39 years old involved opioids, with the highest being 27% of 31 year old deaths (Figure 2B).The burden of the opioid epidemic is spread unequally across race and ethnic origin groups.Since the early-2000s, the highest death rates were among American Indian/Alaska Native (AI/AN) and white people (Figure 3A).However, the opioid overdose death rate of Black people surged since 2015, surpassing that of white people in 2020.In 2021, Black and AI/AN men experienced the highest age-adjusted death rates, at 51 and 46 deaths per 100,000 (Figure 3B).In 2021, the highest age-adjusted death rates in women were among AI/AN and Black women (31 and 18 deaths per 100,000, respectively).The life expectancy at birth (LE) of the US population is significantly shortened due to the opioid crisis.We found that LE was reduced by 0.44 years in 2019, 0.58 years in 2020, and 0.65 years in 2021 (Table 2) (see Methods).Subdividing by demographic group, we estimated that in 2019, the LE at birth of white men was reduced by the largest amount (0.70 years) followed by Black and AI/AN men (0.55 and 0.50 years, respectively).Amongst women, the LE at birth of AI/AN women was most reduced (0.39 years), followed by white and Black women (0.36 and 0.24 years, respectively).Each year from 2019 to 2021, there was an increase in the LE reduction in all demographic groups except Asian men.In 2021, the reduction of LE at birth of white men was still the largest (0.98 years) followed by Black and AI/AN men (0.96 years for both) (Table 2).Amongst women in 2021, the LE of AI/AN women was again reduced by the most (0.76 years), followed by white and Black women (0.50 and 0.44 years respectively).
Years of life lost is a metric taking into account the number of people dying at each age, and the difference between their age at death and their life expectancy.We estimate that nearly 3 million years of life were lost to opioid overdoses in the US in 2021 alone (Table 2).Nearly half (1.3 million years) were lost by white men, followed by white women (636,000 years), Hispanic men (308,000 years), and Black men (302,000 years).In total, 352,000 (14%) more years of life were lost to opioids in 2021 compared to 2020, and 634,000 (34%) more years of life were lost in 2020 than in 2019.There was a substantial increase in years of life lost (YLL) across all groups from 2019 to 2021, ranging from a 290% increase (Asian women) to a 38% increase (white women).In fact, despite losing the most years of life each year, white people actually saw the smallest relative increase from 2019 to 2021.The variation in size of the race/ethnic origin groups obscures differences in the YLL burden across the groups.However, the normalized YLL per capita allows for meaningful comparisons (Figure 3C).In 2021, AI/AN and Black men lost the most years per capita to the opioid crisis (1,500 years per 100,000 people), followed by white men (1,300 years per 100,000) (Figure 3C).Amongst women, AI/AN women lost the most years of life per capita in 2021 (1,100 years per 100,000).Amongst both men and women, Asian people lost the fewest years, around 200 and 40 years per 100,000 people, respectively.We only report the years of life lost and the reduction in life expectancy for the 5 largest race and ethnicity groups.There was not enough data from the Native Hawaiian and Other Pacific Islanders group to estimate the years of life lost or the reduction in life expectancy (see Methods).

Years of life lost
There is significant geographic variation in the severity of the opioid epidemic across the United States.In 2021, the most years of life (YLL) per capita were lost in West Virginia, at 2,400 years lost per 100,000 people, followed by Delaware and Tennessee (1,700 and 1,600 years per 100,000, respectively) (Figure 4A).Between 2019 and 2021, the YLL per capita more than tripled in Montana, Alaska, and Hawaii, and more than doubled in 12 additional states.In fact, during that time, the YLL per capita increased in all states except South Dakota, New Jersey, New Hampshire, and Delaware (Figure 4D).Similarly, the age-adjusted opioid overdose death rates more than doubled in 10 states between 2019 and 2021 (Figure 4E).Alaska, Louisiana, Mississippi, and Alabama had the largest relative increase in that period, with 2021 opioid overdose death rates about 2.5 times higher than in 2019, and New Hampshire was the only state in which this death rate decreased.In 2021, the age-adjusted opioid overdose death rate ranged from 77.2 deaths per 100,000 in West Virginia, to 5.7 in South Dakota (Figure 4B).In 2021, the percentage of all deaths of 25-44 year olds due to opioid overdoses was highest at 40.1% in Rhode Island, followed by West Virginia (36.6%) and Massachusetts (36.3%) (Figure 4C).In contrast, in South Dakota, Nebraska, and Hawaii, fewer than 7% of the deaths in 25-44 year olds were caused by opioids.Alabama and Mississippi had the largest relative increase in the percentage of deaths in 25-44 year olds that were due to opioid overdoses from 2019 to 2021, and 11 states had a decrease (Figure 4F).While there is geographic variation in the burden of the opioid crisis, the severity of the crisis worsened significantly across the United States.States in the South and the Midwest experienced the highest relative increases in burden, and New Hampshire was the only state with a decrease in burden across all metrics used here.

Discussion
The opioid crisis in the United States has been intensifying over the last 25 years, reaching 80,411 opioid overdose deaths in 2021.In this study, we estimate the years of life lost to the opioid crisis each year from 2019 to 2021.Here, years of life lost highlights both the large number of opioid overdose deaths, and how many of these deaths are among young people: opioid overdoses are a leading cause of death in young adults nationwide, causing more than 20% of all deaths in 20-39 year olds.We find the years of life lost (YLL) increased annually, reaching 3 million in 2021.We estimate 2 and 2.6 million YLL in 2019 and 2020, respectively, consistent with previous estimates of 1.7 million YLL in 2016 39 .In particular, we found a 34% increase in YLL in 2020 compared to 2019.This surge in fatal opioid overdoses at the onset of the COVID-19 pandemic coincided with disruptions in access to treatment and harm reduction programs, and increased social isolation during lockdown periods [54][55][56] .
Life expectancy in the United States has been stagnating or even declining since 2015 57,58 .While this recent decline is due to a variety of factors, a significant portion is attributed to the opioid crisis 59 since individuals dying of opioid overdoses are much younger than those dying of other leading causes of death 36,60 .In this study, we estimated that in 2021 alone, the opioid crisis reduced the US population's life expectancy (LE) at birth by 0.65 years, compared to 0.44 years in 2019.We found the 2021 reduction in LE at birth varied across demographic groups, from 0.98 years of LE lost for white men to 0.03 years for Asian women.While the burden of this crisis increased across all groups in the last few years, the increase was most pronounced for American Indian/Alaska Native (AI/AN) and Black or African American people, who now have the highest opioid overdose death rates.We also found that AI/AN men and Black men are now losing the most years of life per capita.The rapidly rising opioid overdose death rates among Black and AI/AN people are particularly concerning in light of the discrepancies in access to OUD treatment, resulting in lower initiation and retention rates among people of color 61,62 .These results add to previous work finding that, contrary to early beliefs and media portrayals of the opioid crisis, this is no longer simply a "rural white problem" 63 .
This study provides further evidence that the opioid epidemic is still worsening at the national and state level, despite numerous efforts to turn the tide.While some interventions have been successful, some have had unintended or negative repercussions 64,65 .For example, while prescription drug monitoring programs were found to decrease prescription opioid mortality rates, without addressing existing OUD they may have contributed to the surge of illicitly manufactured opioid use 66 .And, despite the development of proven, effective treatments for OUD, stigmas impede access 67,68 .There is an urgent need for well-researched, multifaceted interventions, including stigma-reduction, in order to stabilize this constantly-evolving, deadly crisis 69 .
Metrics based solely on overdose deaths are not complete measures of the burden of the opioid epidemic.Many people are impacted by this crisis in other ways.Individuals suffering from OUD have reduced quality of life 70 and impact the lives of family and community members.As an example, children in households with OUD face higher risk of behavioral problems and of involvement in the welfare system 71,72 .Additionally, there is significant economic cost from healthcare and reduced workforce participation due to OUD or early death 73,74 .OUD is also associated with increased risk of infection and death from bloodborne pathogens such as HIV and HCV, suggesting that the opioid crisis may be responsible for additional deaths [75][76][77][78] .Future work incorporating other metrics of burden such as nonfatal overdose hospitalization rates or elevated mortality rates from other causes may allow for a more complete picture.
Our YLL estimates are based on a lower bound of opioid overdose deaths.When stratifying deaths by demographic group, categories with fewer than 10 counts are suppressed by CDC WONDER for confidentiality reasons.In particular, smaller demographic groups with fewer total opioid-related deaths (even if death rates are high) have a higher number of suppressed categories, hence their burden is more underestimated than that of larger groups.Racial misclassification on death certificates may also be causing us to underestimate the opioidrelated burden of smaller demographic groups, in particular for AI/AN people 79 .There is also likely underreporting of opioid overdose deaths, varying by state 80,81 .Altogether, this leads us to conclude that our results are underestimates of the annual years of life lost from the opioid crisis.
The continued rapid growth of the opioid epidemic highlights the importance and urgency of implementing more effective interventions to save hundreds of thousands of lives in just the next few years.We found that not only is the opioid crisis leading to several millions of years of life lost each year and significantly reducing life expectancy in the US, but these metrics are worsening each year.This study adds to the literature showing that the opioid epidemic is now most heavily impacting AI/AN and Black communities, and that the narrative of the opioid crisis primarily affecting white people is no longer accurate.An understanding of which are the vulnerable populations is imperative in order to implement effective and well-targeted interventions.

Figure 1 :
Figure 1: Trends in opioid overdose deaths in the United States from 1999 to 2021.A) Total number of overdose deaths each year involving an opioid.B) Annual rates (per 100,000 population) of overdose deaths, by type of opioid involved.Note that some deaths involve more than one type of opioid.Synthetic Opioids include, for example, drugs such fentanyl (and its derivatives) and tramadol, but excludes methadone.Natural and Semisynthetic Opioids include, for example, oxycodone, hydrocodone, and morphine.Heroin is categorized separately although it is also a semisynthetic opioid.Methadone, a synthetic opioid prescribed to treat opioid use disorder, is categorized separately.Other and Unspecified Narcotics include deaths due to opioid or cocaine-like substances, or mixtures thereof, that were not further sub-categorized on the death certificate.

Figure 2 :
Figure 2: Age distributions of opioid overdose deaths in the United States.Age distribution of opioid overdose deaths by gender, in total numbers A) and as percentage of all deaths B).Distributions for the years 2000, 2010, 2019, 2020 and 2021 are shown in subsequently lighter colors.Note the similarity of the 2021, 2020 and 2019 distributions of % deaths in (B).In 2021 and 2020, despite the increase in the number of opioid overdose deaths, they represented a similar percentage of total deaths in the US due to additional COVID-19 deaths 53 .

Figure 3 :
Figure 3: Burden of opioid overdose deaths by race and ethnicity in the United States.A) Opioid overdose death rates from 1999-2021, by race and ethnicity group.Dashed lines are death rates from 1999-2020 with the bridged-race categories, and solid lines are death rates from 2018-2021, with the new single race categories.The rates for Native Hawaiian and Other Pacific Islander and for More Than One Race were only reported separately in the new single race categories, starting in 2018.B) Age-adjusted death rates and C) years of life lost per capita, by race/ethnicity groups and by gender in 2021.

Figure 4 :
Figure 4: Geographic variation in the opioid epidemic in 2021 (top row) and relative change from 2019 (bottom row).A) and D) Years of life lost per capita due to opioid overdose deaths, by state, B) and E) Age-adjusted opioid overdose death rates by state, C) and F) Percentage of all deaths in 25-44 year olds that are opioid-related, by state.Gray indicates states for which years of life lost estimates were not possible in 2019, due to suppression of counts <10 (see Methods).

Table 1 : Deaths involving opioids in 2021, 2020, 2019, 2010 and 2000 in the United States.
The number of opioid overdose deaths each year, stratified by gender, race and ethnic origin, age, or opioid type involved.All individuals of Hispanic origin are aggregated, regardless of race (see Methods).Race categories reported by CDC changed in 2018, but we only include here the race and ethnic origin groups with a number of deaths high enough that we were able to estimate the years of life lost.Note that in many cases more than one type of opioid is involved.

Table 2 : Years of life lost in the US due to opioids
. Years of life lost (top) and reduction in life expectancy (bottom) due to opioid overdose deaths in 2021, 2020 and 2021, by race and ethnicity group and by gender.