Burden of drug use disorders in the United States from 1990 to 2021 and its projection until 2035: results from the GBD study

Background Drug use disorders (DUDs) have emerged as one of the most significant public health crises, exerting a substantial influence on both community health and socio-economic progress. The United States (US) also suffers a heavy burden, it is necessary to figure out the situation from multiple perspectives and take effective measures to deal with it. Therefore, using the data from the Global Burden of Diseases, Injuries, and Risk Factors (GBD) 2021, we evaluated this topic. Methods Annual data on DUDs-related burden were collected from the GBD study 2021. We calculated the indicator of estimated annual percentage change (EAPC) to evaluate the changing trend of burden. The Bayesian model for age-period-cohort was introduced to forecast the burden. Results In 2021, the number and age-standardized rate of prevalence were particularly prominent, with 12,146.95 thousand and 3821.43 per 100,000, respectively. Higher burden was also observed in males, 15–45 years old populations, and opioid use disorders subtype. From 1990 to 2021, the DUDs-related burden increased in the US and all states, especially in West Virginia; and the national death-related burden with the highest increase (EAPC = 7.96). Other significant inverse associations were seen between EAPC, age-standardized rates, and socio-demographic index (SDI). Moreover, in the next 14 years, the projected DUDs burden remains exigent. Conclusions The burden of DUDs in the US is heavy and has been enlarging. This study proposes that greater attention should be paid to the strategies in males, the younger population, opioid use disorders, and low-SDI states implemented by decision-makers to achieve goals such as reducing burden. Supplementary Information The online version contains supplementary material available at 10.1186/s12889-024-19142-0.


Background
Drug use disorders (DUDs), refer to the forced and endless use of certain drugs with dependency characteristics in the pursuit of special psychological effects rather than medical purposes, can cause serious psychological, physiological consequences, and social problems [1][2][3], including cognitive impairment, suicidal tendencies, decreased quality of life, and risk of infectious diseases [4][5][6][7].DUDs continue to be a heavy burden globally, according to the Global Burden of Disease, Injuries, and Risk Factors Study (GBD) 2019, DUDs were in the top 20 leading causes of disability-adjusted life years (DALYs) in the 10-49-year age groups in 2019 [8].Meanwhile, the World Drug Report 2023 indicated that, in 2021, over 296 million people worldwide used drugs; and the number of people suffering from DUDs has soared to 39.5 million, an increase of 45% in the past 10 years [9].However, only one-fifth of people with drug-related disorders received medication treatment, and the gap in access to treatment among regions continued to widen [9].
Notably, the United States (US) is one of the countries with a heavy DUDs-related burden [1,8].According to the GBD study, in 2019, over half of the death cases due to DUDs worldwide occurred in the US [1,8].The GBD database has the advantage of systematic analysis and integration of global disease and health data, and the adoption of the GBD data can provide policymakers, researchers, and the public with comprehensive insights into the status of DUD in the US.Therefore, in this study, we retrieved detailed data on the latest burden of DUDs from the GBD study (2021) to fully investigate the magnitude and temporal trends of burden due to DUDs in the US from 1990 to 2021 by age, sex, state, and drug categories; and attempted to predict the burden in the next 14 years.Our results can be helpful for the country to develop more effective population-specific policies and methods.

Data collection and case definition
Similar to the GBD study 2019, the data collection, processing, and overall analysis methods of the GBD study (2021) have been reported in detail previously [10][11][12].We collected the data on DUDs-related burden from the Institute for Health Metrics and Evaluation (IHME, https:// vizhub.healt hdata.org/ gbd-resul ts/).We obtained the data on DUDs-related burden about the numbers (in thousands), rates (per 100,000 population), and agestandardized rates (per 100,000 population) of incidence, prevalence, deaths, and DALYs by age, sex, and state from 1990 to 2021.Meanwhile, the information about the socio-demographic index (SDI, an indicator that was calculated based on the fertility rate, education, and income) for the states was also collected [10][11][12].
In the GBD study (2021), the DUDs were defined based on the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV-TR) or the International Classification of Diseases (ICD-10) diagnostic criteria, including opioid use disorders, cocaine use disorders, cannabis use disorders, amphetamine use disorders, and other DUDs [10,11].Other DUDs included hallucinogen dependence, inhalant or solvent dependence, sedative dependence, tranquiliser dependence, and other medicines, drugs, substance dependence [10].

Evaluation of DUDs-related burden
The prevalence, incidence, mortality, and burden of DUDs (by age, sex, year, and subtype) were generally evaluated in the following ways: Firstly, based on the input raw data of vital registration, verbal autopsy, and surveillance databases, a cause of death database was generated through standardization, ICD mapping, age-sex division, garbage code redistribution, and noise reduction [10,11].Secondly, based on this cause of death database, the estimation of DUDs-related burden followed the general Cause of Death Ensemble model (CODEm) strategy [10,11].The CODEm strategy mainly included two models: linear mixed-effects regression (LMER) models and spatiotemporal Gaussian process regression (ST-GPR) [10,11].With the progress of the CODEm strategy (covariates adjusted), the mortality rate of DUDs and the years of life lost (YLLs) were primarily evaluated [10,11].In the GBD 2021 study, the covariate included additional data and increased time smoothing, which improved the stability of the results, especially for the estimates of the US and Western Europe [10,11].Thirdly, based on the cause of death database and systematic reviews of epidemiological survey data, using the Bayesian meta-regression method DisMod-MR 2.1, the prevalence, incidence, and years of life lived with disability (YLDs, fully considering the disability weights and comorbidity) was further generated by age, sex, year, and subtype [10,11].DisMod-MR 2.1 is a Bayesian meta-regression tool, which gathers data points across multiple sources and accounts for recognized variability factors such as disparities in case definitions and sampling methods to generate internal consistent estimates [10,11].And lastly, the DALYs (by age, sex, year, and subtype) were obtained by adding YLLs and YLDs [10,11].Based on the global standard population of the GBD study (2021), age-standardized rates were also estimated [13].

Statistical analysis
Data were described as absolute numbers with 95% uncertainty intervals (UIs) by age, sex, year, state, and drug categories.95% UI was generated by the 2.5th and 97.5th percentiles of the ordered estimate values for the CODEm process of 1,000 draws [11].When assessing the temporal trends, we introduced the indicator of estimated annual percentage change (EAPC) [14][15][16].The EAPC was calculated based on a regression model, that is, [ln (age-standardized rate) = α + β × (calendar year) + ε].The EAPC and its 95% confidence interval (CI) were calculated from the model of [100 × (exp (β)-1)] [ [14][15][16].The trends were recognized as a decrease when the upper boundary of 95% CI of EAPC was < 0; while if the lower boundary of 95% CI of EAPC was > 0, the upward trends of burden were defined; otherwise, the trends represented stable [14][15][16].
Meanwhile, using the Pearson or Spearman rank test, associations between EAPC, DUDs-related burden, and SDI were determined, and the expected associations between them were estimated using the locally weighted regression (LOESS) model [17].Additionally, we implemented the Bayesian model for age-period-cohort (BAPC) to forecast the burden until 2030 [18][19][20].The BAPC model is expressed as where λ ij denotes the count of cases, μ denotes the intercept, and α i , β j , and γ k signify the effect of age, period, and cohort, respectively [18][19][20].To account for overdispersion, we utilized the BAPC model, which was implemented using the INLA and BAPC packages in R software [18][19][20].The predicted population was obtained from the World Population Prospects 2022 [21].
All statistical analyses and visualization of results were conducted using the R software (Version 4.0.3;https:// www.R-proje ct.org/), and the two-tailed P value < 0.05 was considered statistically significant.

Burden and temporal trends of DUDs in the US by age and sex
In 2021, the DUDs-related burden (both numbers and age-standardized rates) in males was higher than that in females in both the US and the world (Additional File 1: Tables S1-S8; Fig. 2; and Additional File 2: Figure S1).From 1990 to 2021, in the US, the ASIR, ASPR, ASMR, and ASDR increased in both sexes, especially in females (Additional File 1: Tables S1-S4; Additional File 2: Figure S1).However, at the global level, in both sexes, the ASIR and ASPR decreased, while the ASMR and ASDR remained stable, respectively (Additional File 1: Tables S5-S8).
In 2021, similar to the world, more than half of the burden (numbers and age-specific rates) of DUDs was generally concentrated in the population 15-45 years old in the US in both sexes (Additional File 1: Tables S1-S8; Fig. 2; and Additional File 2: Figures S2-S3).Notably, the burden of deaths was still heavy in the population 45-65 years old in both sexes at the US and global levels (Additional File 1: Tables S3 and S7; Fig. 2; and Additional File 2: Figures S2-S3).From 1990 to 2021, in the US, the age-specific rates of burden increased in both sexes in all age-specific groups, except for the < 15 age-specific group (EAPC for ASIR = -0.27,95%CI: -0.35, -0.18; EAPC for ASPR = -0.27,95%CI: -0.36, -0.18; and EAPC for ASDR = -0.28,95%CI: -0.37, -0.19, respectively) (Additional File 1: Tables S1-S4; Fig. 3; and Additional File 2: Figure S3).However, at the global level, in both sexes, the age-specific rates of incidence increased in the > 85 age-specific groups, while the upward trends of age-specific rates of prevalence, deaths, and DALYs were generally observed in the 45-60 and > 90 age-specific groups (Additional File 1: Tables S5-S8; Fig. 3; and Additional File 2: Figure S3).

Burden and temporal trends of DUDs in the US by drug categories
For the subtypes of DUDs, in 2021, opioid use disorders generally occupied the highest burden in both sexes in the US (except for the incidence burden); the numbers   S1-S4; Fig. 2; and Additional File 2: Figure S1).Similar results could be found at the global level, except for the incidence and prevalence burden (Additional File 1: Tables S5-S8).Moreover, the burden of opioid use disorders accounted for a higher proportion among young people (< 30 years old) (Fig. 2).

Discussion
This study disclosed the magnitude and temporal trends of DUDs-related burden during the past three decades in the US based on the lasted GBD 2021 and also conducted the projection until 2035.It revealed that the absolute number of DUDs in the US accounted for a relatively high proportion of the global numbers in 2021, especially the number of deaths (accounting for 51.6%).As measured by trends, the burden of DUDs in the US increased significantly from 1990 to 2021.However, during the same period, a slight increase was only observed in ASMR worldwide.DUDs have become a serious disease burden in the US, and more targeted health policies and projects should be timely taken to alleviate this burden.
From the perspective of drug categories, opioid use disorders had the heaviest burden in the US in 2021.Especially, ASPR, ASMR, and ASDR of opioid use disorders accounted for over 50% of all types of DUDs.Given that the number of deaths from opioid overdoses reached a historic high in 2016, the US Department of Health and Human Services (HHS) declared the opioid crisis a public health emergency in 2017 [22,23].As of 2020, 37.309 million Americans aged 12 and above had used illegal drugs in the past 30 days, of which 24.7% had opioid use disorders [9].Our study showed that in 2021, opioids accounted for more than 78% of the deaths attributed to DUDs in the US, exceeding half of the global death toll.In the same year, the ASMR of opioid use disorders in the US was 12.92 times higher than in the world.Opioids remain the most lethal drug, notably, strategies have been implemented to control the burden caused by opioid use disorders.Buprenorphine, methadone, and naltrexone have been approved by the FDA respectively for the treatment of opioid dependence, they are proven to be safe and effective when combined with counseling and behavioral therapies, known as medication-assisted treatment (MAT) [24].The FDA subsequently approved the first over-the-counter (OTC) naloxone nasal spray and the first namefene hydrochloride nasal spray in March and May 2023, respectively [25].These drugs belong to opioid receptor antagonists and can be used to treat acute opioid overdose, thereby reducing opioid overdose mortality.Our research findings also suggest that the implementation of these strategies has achieved certain effects, the overall ASIR will decrease in the next  of psychotropic drugs [26].There are also reports showing that during the previous year, males in the US were 2.33 and 2.25 times more likely than females to suffer from DUDs and drug dependence, respectively [27,28].In addition, the number of death cases caused by DUDs among females has been consistently high from age groups of 25 to 59, and the peak age group appears later than males.This may be related to females who use psychotropic drugs later and are more prone to mental disorders compared to males, or there are more obstacles for females in accessing medication, leading to insufficient medication treatment [9,29,30].The comprehensive women-centered treatment method includes treating the whole person and the mother-child dyad [31].This may consist of a variety of intervention measures and services, such as childcare and parenting education; obstetric and gynecological care; general medical care; interventions and services for comorbid mental disorders; social support, including transportation, housing, and occupational rehabilitation; and legal aid [31].However, females with DUDs have to face more pressure and barriers when receiving the treatment, they may endure more social stigmatization, fear legal sanctions, and possibly even lose custody of their children [9,32].More practical and effective strategies for women should also be developed and implemented to alleviate or even relieve burdens.
Regardless of gender, the burden of DUDs is highest in the age-specific groups of 15-45 years and showed a decreasing trend with age, which was consistent with the world pattern.This indicated that in young adults DUDs were still very serious, especially in youthfulness.Studies have estimated that the past year and lifetime prevalence of DUDs among 18-29-year-olds in the US is 8.3% and 14.2%, respectively [33].The earlier the use of psychoactive drugs, the greater the lifelong risk of DUDs [34,35].Among users who try psychotropic drugs before the age of 13, 70% develop DUDs within the next 7 years, while for those after the age of 17, the proportion drops to 27% [36].These adolescents and children can suffer physical, sexual, and emotional abuse, and be in a state of poverty, homelessness, famine, and gender discrimination [36,37].Due to the neurotoxicity of psychoactive drugs on the developing brain, contact with the drugs should be immediately interrupted to reduce the damage [3].Meanwhile, treatment services for DUDs should be provided promptly to accurately identify and meet the needs of adolescents and children [38].In recent years, treatments including but not limited to cognitive behavioral therapies, vocational training, psychoeducation, and motivational enhancement therapies, have been applied across service settings in the US [3,[39][40][41].These approaches have achieved certain benefits in treating adolescent DUDs, as shown in this study, age-specific rates of incidence, prevalence, and DALYs have all decreased in the 10-14 age group.
Our results suggested that in 2021, the EAPCs for DUDs in various states in the US were negatively correlated with the overall SDI.Especially in West Virginia, it has lower SDI but in most cases higher EAPC, indicating a heavy drug abuse burden.According to reports, West Virginia has the highest per capita opioid overdose mortality rate and the highest case rate of neonatal opioid withdrawal (NOW) syndrome in the US [36].Since the early 2010s, several states in the US have begun strengthening prescription drug monitoring programs (PDMP), with dramatic results, for example, Kentucky reduced its number of opioid prescriptions per capita by 85% in 2015 compared with 2010, Florida saw a 50% drop in oxycodone overdose deaths in 2012 compared with 2010, national opioid prescriptions fell 16% annually, and the rate of Americans receiving MAT increased 13% annually [36].It should also be noted that the burden of DUDs usually increases with increasing SDI values [1].People from higher SDI areas have relatively higher socio-economic status, they may have more propensity to initiate drugs [31].However, populations in lower SDI areas may face more socio-economic disadvantages such as poverty, conflict, and lack of education and employment opportunities, making them highly susceptible to mental health issues and DUDs, as well as limited access to health promotion, prevention, and drug treatment services.Therefore, the latter pays a higher price and is highly likely to suffer a heavy burden [31].
Despite previous reports on the limitations of GBD [10][11][12], it is still necessary to clarify the limitations of this work.Firstly, in the GBD study (2021), the DUDs were defined based on the DSM-IV-TR or the ICD-10 diagnostic criteria, if the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) was used, the estimation of DUDs may be changed.Secondly, the limited detailed information obtained from the GBD database regarding the US and its states in this study may limit the analysis of the burden of DUDs, although GBD study (2021) made improvements to the model to improve estimation accuracy.In addition, due to the lack of estimated deaths caused by cannabis use disorders in the GBD 2021 data, the total number of deaths from DUDs may be underestimated.Last but not least, this study evaluated the changing trends and predicted the burden of DUDs based on the GBD study 2021, but due to the information lag in this database (only data information from 1990 to 2021 could be obtained currently), the prediction results may not be accurate enough.However, our results still have significant public health implications for controlling the DUDs-related burden in the US.

Fig. 1 Fig. 3
Fig. 1 Age-standardized rates of drug use disorders-related burden in 2021 and the temporal trends from 1990 to 2021 by states in the US.A ASIR in 2021; B EAPC in ASIR from 1990 to 2021; C ASPR in 2021; D EAPC in ASPR from 1990 to 2021; E ASMR in 2021; F EAPC in ASMR from 1990 to 2021; G ASDR in 2021; and H EAPC in ASDR from 1990 to 2021.ASIR: age-standardized incidence rate; ASPR: age-standardized prevalence rate; ASMR: age-standardized mortality rate; ASDR: age-standardized DALYs rate; EAPC: estimated annual percentage change

Fig. 4
Fig. 4 The influence factors of the EAPCs of drug use disorders-related burden in the US.A ASIR in 1990 and EAPC in ASIR; B SDI in 2021 and EAPC in ASIR; C ASPR in 1990 and EAPC in ASPR; D SDI in 2021 and EAPC in ASPR; E ASMR in 1990 and EAPC in ASMR; F SDI in 2021 and EAPC in ASMR; G ASDR in 1990 and EAPC in ASDR; H SDI in 2021 and EAPC in ASDR.The circle represented the state, and the size of the circle represented the number.The r or ρ indices and P values were evaluated by Pearson or Spearman rank analysis.The blue line and its shade were fitted by LOESS.ASIR: age-standardized incidence rate; ASPR: age-standardized prevalence rate; ASMR: age-standardized mortality rate; ASDR: age-standardized DALYs rate; EAPC: estimated annual percentage change; SDI: socio-demographic index