Trends in Prevalence of Cigarette Smoking Among US Adults With Major Depression or Substance Use Disorders, 2006-2019

Key Points

Question  Between 2006 and 2019, was there a change in prevalence of cigarette smoking among US adults with major depressive episode (MDE), substance use disorders (SUD), or both?

Findings  In this serial cross-sectional study that included 558 960 adult participants, there was a statistically significant decline in the prevalence of self-reported cigarette smoking from 2006 to 2019 among those with MDE (37.3% to 24.2%), SUD (46.5% to 35.8%), or both (50.7% to 37.0%).

Meaning  Among US adults with major depressive episode, substance use disorder, or both, there were significant declines in the prevalence of cigarette smoking between 2006 and 2019.

Importance  Tobacco use is highly concentrated in persons with mental illness.

Objectives  To assess trends in past-month prevalence of cigarette smoking among adults with vs without past-year depression, substance use disorders (SUDs), or both, using nationally representative data.

Design, Setting, and Participants  Exploratory, serial, cross-sectional study based on data from 558 960 individuals aged 18 years or older who participated in the 2006-2019 US National Surveys on Drug Use and Health.

Exposure  Past-year major depressive episode (MDE) and SUD using Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition, Text Revision) criteria.

Main Outcomes and Measures  Past-month self-reported cigarette use, adjusted for sociodemographic characteristics.

Results  Of the sampled 558 960 adults, 41.4% (unweighted) were aged 18 to 25 years, 29.8% (unweighted) were aged 26 to 49 years, and 53.4% (unweighted) were women. From 2006 to 2019, the past-month self-reported cigarette smoking prevalence declined significantly among adults with MDE from 37.3% to 24.2% for an average annual percent change of −3.2 (95% CI, −3.5 to −2.8; P < .001), adults with SUD from 46.5% to 35.8% for an average annual percent change of −1.7 (95% CI, −2.8 to −0.6; P = .002), and adults with co-occurring MDE and SUD from 50.7% to 37.0% for an annual average annual percent change of −2.1 (95% CI, −3.1 to −1.2; P < .001). The prevalence declined significantly for each examined age, sex, and racial and ethnic subgroup with MDE and with SUD (all P < .05), except for no significant changes in American Indian or Alaska Native adults with MDE (P = .98) or with SUD (P = .46). Differences in prevalence of cigarette smoking between adults with vs without MDE declined significantly for adults overall from 11.5% to 6.6%, for an average annual percent change of −3.4 (95% CI, −4.1 to −2.7; P < .001); significant average annual percent change declines were also seen for men (−5.1 [95% CI, −7.2 to −2.9]; P < .001); for women (−2.7 [95% CI, −3.9 to −1.5]; P < .001); for those aged 18 through 25 years (−5.2 [95% CI, −7.6 to −2.8]; P < .001); for those aged 50 years or older (−4.7 [95% CI, −8.0 to −1.2]; P = .01); for Hispanic individuals (−4.4 [95% CI, −8.0 to −0.5]; P = .03), and for White individuals (−3.6 [95% CI, −4.5 to −2.7]; P < .001). For American Indian or Alaska Native adults, prevalence did not significantly differ between those with vs without MDE during 2006-2012 but was significantly higher for those with MDE during 2013-2019 (difference, 11.3%; 95% CI, 0.9 to 21.7; P = .04). Differences among those with vs without SUD declined for women for an average annual percent change of −1.8 (95% CI, −2.8 to −0.9; P = .001).

Conclusions and Relevance  In this exploratory, serial, cross-sectional study, there were significant reductions in the prevalence of self-reported cigarette smoking among US adults with major depressive episode, substance use disorder, or both, between 2006 and 2019. However, continued efforts are needed to reduce the prevalence further.

Prevalence of tobacco use, the leading preventable cause of disease, disability, and death in the US, has been declining.¹ However, subpopulation differences persist, with higher smoking prevalence and lesser declines over time among people with mental illness.^2-6 Population-based studies using data through 2014 showed that declines in smoking prevalence seen in general populations were not observed in those with mental illness.^2-4,6 In a study conducted in the years 1991 and 1992, past-month cigarette smoking was reported by 41.0% of individuals with psychiatric conditions but by only 22.5% of individuals without psychiatric conditions.² In another study conducted in 2001-2002, nicotine-dependent adults with comorbid psychiatric disorders consumed 34.2% of all cigarettes smoked in the US but represented only 7.1% of the population.³ Prevalence of past-month cigarette smoking remained unchanged among adults with mental illness during 2004-2011⁴ and among adolescents or adults with a drug use disorder during 2002-2014 but declined among their counterparts without mental illness.⁶

A randomized clinical trial of smoking cessation interventions involving people with psychiatric illnesses demonstrated the efficacy and safety of smoking cessation medications (varenicline, bupropion, and nicotine patch) in adults with psychiatric disorders.⁷ Furthermore, quitting smoking did not jeopardize the success of behavioral health care and may have improved mental and physical health over the long-term.^8-10 Smoking cessation among patients with psychiatric disorders was associated with decreases in anxiety, depression, and stress,⁸ lowered the likelihood of a new-onset substance use disorder (SUD),⁹ and improved mood and quality of life.^8,10

This exploratory study was conducted to examine trends in past-month self-reported prevalence of cigarette smoking among adults with and without major depressive episode (MDE), SUD, or both. In addition, because cigarette smoking has varied by age, sex, and race and ethnicity,¹¹ as have past-year quit attempts, successful quitting, and responses to cessation medications or financial incentives,^12-15 this study also examined trends in past-month cigarette smoking within age, sex, and race and ethnicity subgroups of adults with and without MDE and SUD.

Data were from individuals aged 18 years or older who participated in the 2006-2019 National Surveys on Drug Use and Health (NSDUH), providing nationally representative data on cigarette smoking, tobacco use, MDE, and SUD (alcohol or drug use disorders) among US civilian, noninstitutionalized adult populations.¹¹ The NSDUH data collection protocol was approved by the institutional review board at the Research Triangle Institute International. Data were collected by interviewers in personal visits to households and noninstitutional group quarters. Each participant provided verbal informed consent.^11,16 Audio computer-assisted self-administered interviewing was used, providing respondents with a private, confidential way to record answers. Race and ethnicity were determined according to respondents’ self-classification of racial and ethnic origin and identification based on classifications developed by the US Census Bureau. Additional details about NSDUH methods are available.¹⁶

The surveys asked respondents about lifetime, past-year, and past-month self-reported use of any tobacco (cigarettes, cigars, chewing tobacco, pipe tobacco, or snuff), alcohol, cannabis, cocaine, heroin, methamphetamine (specifically assessed since 2015), inhalants, and hallucinogens and misuse of prescription opioids, sedatives or tranquilizers, and stimulants. Using diagnostic criteria specified in the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition, Text Revision) (DSM-IV-TR),^11,16,17 The NSDUH estimated prevalence of past-year specific SUD (alcohol, cannabis, cocaine, heroin, methamphetamine [since 2015], inhalants, hallucinogens, and prescription opioids, tranquilizers or sedatives, and stimulants). The surveys also assessed past-year MDE using DSM-IV-TR diagnostic criteria.^11,16,17 These measures of self-reported substance use, use disorders, and MDE have demonstrated good validity and reliability.^18,19Additionally, the survey collected sociodemographic characteristics (eg, age, sex, race and ethnicity, educational attainment, employment status, family income, marital status, health insurance, county type). Race and ethnicity included non-Hispanic American Indian or Alaska Native (hereafter, American Indian or Alaska Native), non-Hispanic Asian (hereafter, Asian), non-Hispanic Black (hereafter, Black), Hispanic, and non-Hispanic White (hereafter White) individuals.

Past-month and lifetime prevalence of self-reported cigarette smoking and any tobacco use were examined among US adults by sociodemographic characteristics and past-year psychiatric conditions based on the 2015-2019 NSDUH data. Bivariable logistic regression analyses were conducted to test whether and how prevalence varied by sociodemographic characteristics and psychiatric conditions. Due to partial methodology changes in the 2015 NSDUH, 7 drug use disorders (ie, inhalants, methamphetamine, hallucinogens, and prescription opioid, tranquilizers, sedatives, and stimulants) started new baseline trends since 2015. Sensitivity analyses based on the 2015-2019 data were conducted to examine whether the corresponding outcomes differed by using drug use disorder as assessed by cannabis, cocaine, or heroin use disorder (3 drug use disorders) or by using 10 drug use disorders (these 3 drug use disorders and the other 7 drug use disorders).

Multivariable logistic regression models were applied to examine adjusted past-month and lifetime prevalence and prevalence difference²⁰ of cigarette smoking and tobacco use by past-year MDE, SUD, and co-occurring MDE and SUD in each year spanning 2006 through 2019 overall. For adults with MDE or SUD, prevalence and prevalence differences were further examined by sex, age, and race and ethnicity. For adults with co-occurring MDE and SUD, the sample sizes were too limited to conduct these subgroup analyses. Each reported prevalence estimate for a specific year was adjusted for age, sex, race and ethnicity, education, family income, health insurance, employment status, marital status, and county type. Multicollinearity and potential interaction effects were examined and were not found in final models. Missing data were logically assigned or imputed using predictive mean neighborhoods, a combination of a model-assisted imputation method and a random nearest-neighbor hot-deck.^21,22

All aforementioned NSDUH analyses used SUDAAN software release 11.0.3 (RTI International) to account for NSDUH’s complex sample design and sample weights. Based on adjusted prevalence results for each study year, Joinpoint Regression Program version 4.8.01 (National Cancer Institute) was used to test for significant changes in nonlinear trends using Bayesian Information Criterion and to estimate average annual percentage changes from 2006 to 2019, which are valid even if the Joinpoint models indicate changes in trends during this study period.²³ For each analysis, P < .05 (2-tailed) was considered statistically significant. Because of the potential for type I error due to multiple comparisons, findings for these analyses should be interpreted as exploratory.

For the 2006-2019 NSDUH, the annual mean weighted screening response rate was 82.9%, and the annual mean weighted interview response rate was 71.4%. Among the 558 960 adult participants, item response rates for use of tobacco, alcohol, and drugs were extremely high. For example, missing data for cigarette use recency (ie, never, lifetime, past-year, past-month) were 0.33% (1830 of 558 960), which were logically assigned (18 missing responses) or imputed using predictive mean neighborhoods (1812 missing responses). Missing data on MDE were 1.0% and were excluded in our analyses.

Table 1 shows the number and unweighted percentages of participants. Among the sampled 558 960 adults, 298 401 were women (53.4%), 231 677 aged 18 through 25 years (41.4%), 99 969 fifty years or older (17.9%), 8234 American Indian or Alaska Native adults (1.5%), 69 532 Black adults (12.4%), 91 025 Hispanic adults (16.3%), and 347 025 White adults (62.1%). Among the US adult population in 2019, 65.8% had lifetime any tobacco use (95% CI, 65.1%-66.6%), 59.5% had lifetime cigarette smoking (95% CI, 58.8%-60.2%), 23.0% had past-month tobacco use (95% CI, 22.4%-23.6%), 18.2% had past-month cigarette smoking (95% CI, 17.7%-18.7%), 7.9% had past-year MDE (95% CI, 7.5%-8.3%), 7.7% had past-year SUD (95% CI, 7.3%-8.2%), and 1.8% had past-year co-occurring MDE and SUD (95% CI, 1.6%-2.0%).

Past-month prevalence of self-reported cigarette smoking and any tobacco use were statistically significantly higher in 2015, 2016, and 2017 than in 2019 among adults aged 18 through 49 years than among those aged 50 years or older; among men than among women; among Black adults, American Indian or Alaska Native adults, White adults, and multiracial adults than among Hispanic adults; and among those without a college degree than among those with a college degree. Prevalence was also higher among those with an annual family income less than $75 000 than among those whose family income was $75 000 or more, among those unemployed than among those with full-time employment, among those uninsured or with Medicaid only than among those with private health insurance only, among those divorced or separated or never married than among those married, and among those residing in small metropolitan areas or nonmetropolitan areas than among those residing in large metropolitan areas (Table 2).

In particular, past-month prevalence of cigarette smoking was significantly higher among those aged 26 through 34 years (26.6% [95% CI, 26.0%-27%]) than among those aged 50 years or older (15.0% [95% CI, 14.6%-15.4%]; unadjusted prevalence ratio [PR], 1.8 [95% CI, 1.7-1.8]) and was higher among men (22.1% [95% CI, 21.7%-22.5%]) than among women (17.3% [95% CI, 17.0%-17.7%], unadjusted PR, 1.3 [95% CI, 1.2-1.3]). Compared with past-month prevalence of cigarette smoking among Hispanic adults of 15.4% (95% CI, 14.9%-16.0%), higher prevalences were found among American Indian or Alaska Native adults, who had a prevalence of 34.9% (95% CI, 31.4%-38.5%; unadjusted PR, 2.3 [95% CI, 2.0-2.5]); Black adults, who had a prevalence of 21.2% (95% CI, 20.4%-22.1%; unadjusted PR, 1.4 [95% CI, 1.3-1.5]); and White adults, who had a prevalence of 20.8% (95% CI, 20.8%-21.2%; unadjusted PR, 1.4 [95% CI, 1.3-1.4]). However, the past-month prevalence of smoking was lower among Asian American adults (9.1% [95% CI, 8.3%-10.1%]; unadjusted PR, 0.6 [95% CI, 0.5-0.7]).

Past-month prevalence of cigarette smoking and any tobacco use were consistently higher among those with MDE, alcohol use disorder, drug use disorder, SUD, or co-occurring MDE and SUD than among those without the corresponding conditions (Table 2). For example, past-month prevalence of cigarette smoking was significantly higher among those with MDE (30.5% [95% CI, 29.4%-31.5%]) than among those without (18.7% [95% CI, 18.4%-19.0%]; unadjusted PR, 1.6 [95% CI = 1.6-1.7]) and higher among those with alcohol use disorder (42.1% [95% CI, 41.0%-43.2%]) than among those without (17.9% [95% CI, 18.2%-18.5%]; unadjusted PR, 2.3 [95% CI, 2.3-2.4]).

Past-month prevalence of cigarette smoking was significantly higher among those with drug use disorder assessed by 3 disorders (58.6% [95% CI, 56.9%-60.2%]) than among those without (18.8% [95% CI, 18.5%-19.1%]; unadjusted PR, 3.1 [95% CI, 3.0-3.2]) and was higher among those with drug use disorder assessed by 10 disorders (58.0% [95% CI, 56.5%-59.4%]) than among those without (18.5% [95% CI, 18.2%-18.8%]; unadjusted PR, 3.1 [95% CI, 3.0-3.2]; (Table 2). It was significantly higher among those with SUD including drug use disorder assessed by 3 disorders (44.6% [95% CI, 43.6%-45.6%]) than among those without (17.7% [95% CI, 17.4%-18.0%]; unadjusted PR, 2.5 [95% CI, 2.5-2.6]) and higher for those with SUD including drug use disorder assessed by 10 disorders (45.5% [95% CI, 44.5%-46.4%]) than for those without (17.4% [95% CI, 17.1%-17.7%]; unadjusted PR, 2.6 [95% CI, 2.6-2.7]). It was significantly higher among those with co-occurring MDE and SUD including drug use disorder assessed by 3 disorders (50.5% [95% CI, 48.4%-52.5%]) than among those without (17.1% [95% CI, 16.8%-17.4%]; unadjusted PR, 3.0 [95% CI, 2.8-3.1]) and higher for those with co-occurring MDE and SUD including drug use disorder assessed by 10 disorders (50.4% [95% CI, 48.4%-52.5%]) than for those without (16.9% [95% CI, 16.5%-17.2%]; unadjusted PR, 3.0 [95% CI, 2.9-3.1]).

In sensitivity analyses, corresponding outcomes did not significantly differ based on drug use disorder assessed by 3 disorders only (cannabis, cocaine, and heroin) vs by 10 disorders. These results confirmed the adequacy of using the drug use disorder measure of 3 disorders only in subsequent analyses.

From 2006 to 2019, adjusted past-month prevalence of cigarette smoking declined from 37.3% to 24.2% among adults with MDE for an average annual percentage change of −3.2 (95% CI, −3.5 to −2.8; P < .001; Figure 1 and eTable 1 in the Supplement) and from 25.8% to 17.6% among adults without MDE for an average annual percentage change of −2.9 (95% CI, −3.2 to −2.7; P < .001). The differences in the prevalence of past-month cigarette smoking among those with vs without MDE declined from 11.5% to 6.6% for an average annual percentage change of −3.4 (95% CI, −4.1 to −2.7; P < .001).

From 2006 to 2019, adjusted past-month prevalence of cigarette smoking declined from 46.5% to 35.8% among adults with SUD for an average annual percentage change of −1.7 (95% CI, −2.8 to −0.6; P = .002; Figure 1 and eTable 1 in the Supplement) and from 24.6% to 16.8% among adults without SUD for an average annual percentage change of −3.0 (95% CI, −3.2 to −2.8; P < .001). There were no significant changes in differences among those with vs without SUD for an average annual percentage change of −0.7 (95% CI, −3.4 to 2.0; P = .62).

From 2006 to 2019, adjusted past-month prevalence of cigarette smoking declined from 50.7% to 37.0% among adults with co-occurring MDE and SUD for an average annual percentage change of −2.1 (95% CI, −3.1 to −1.2; P < .001; Figure 1 and eTable 1 in the Supplement) and from 23.9% to 16.3% among adults without MDE and SUD for an average annual percentage change of −3.0 (95% CI, −3.2 to −2.8; P < .001). There were no significant changes in differences among those with vs without co-occurring MDE and SUD for an average annual percentage change of −1.8 (95% CI, −3.5 to 0.0; P = .06).

During 2006-2019, the adjusted past-month cigarette smoking declined for all examined subgroups with MDE, except for American Indian or Alaska Native adults who remained unchanged (Figure 2, eTable 2 in the Supplement). From 2006-2019 (Figure 3; eTable 2 in the Supplement), differences in adjusted past-month cigarette smoking among those with vs without past-year MDE declined for men for an average annual percentage change of −5.1 (95% CI, −7.2 to −2.9; P < .001), women for an average annual percentage change of −2.7 (95% CI, −3.9 to −1.5; P < .001), those aged 18 through 25 years for an average annual percentage change of −5.2 (95% CI, −7.6 to −2.8; P < .001), those 50 years or older for an average annual percentage change of −4.7 (95% CI, −8.0 to −1.2; P = .01), Hispanic adults for an average annual percentage change of −4.4 (95% CI, −8.0 to −0.5; P = .03), and White adults for an average annual percentage change of −3.6 (95% CI, −4.5 to −2.7; P < .001). However, among those with vs without MDE, there were no significant changes in differences for those aged 26 through 34 years, those aged 35 through 49 years, Black adults, and Asian American adults. For American Indian or Alaska Native adults, prevalence did not significantly differ between those with vs without MDE during 2006-2012, but it was significantly higher for those with MDE during 2013-2019 (difference, 11.3%; 95% CI, 0.9% to 21.7%; P = .04).

During 2006-2019, the adjusted past-month cigarette smoking declined in all examined age, sex, and racial and ethnic subgroups with SUD, except for American Indian or Alaska Native adults who remained unchanged (Figure 2; eTable 3 in the Supplement). From 2006 to 2019 (Figure 3; eTable 3 in the Supplement), differences in adjusted past-month cigarette smoking among those with vs without past-year SUD declined for women for an average annual percentage change of −1.8 (95% CI, −2.8 to −0.9; P = .001), but there were no significant changes in differences for the other examined subgroups.

From 2006 to 2019, differences in lifetime cigarette smoking increased among adults with vs without SUD for an average annual percentage change of 2.3 (95% CI, 2.0-2.7, P < .001) and among adults with vs without co-occurring MDE and SUD for an average annual percentage change of 2.5 (95% CI, 2.0-3.1; P < .001), but there were no significant changes in differences among adults with vs without MDE (eTable 4 in the Supplement). Results of lifetime and past-month tobacco use are similar to the reported corresponding cigarette smoking results (eTables 4 and 5 in the Supplement).

Based on nationally representative US data, this study found that from 2006 to 2019, after controlling for sociodemographic characteristics, past-month self-reported cigarette smoking significantly declined among adults with MDE, SUD, and co-occurring MDE and SUD, overall and for each examined age, sex, and racial and ethnic subgroup with MDE or SUD, except for American Indian or Alaska Native adults, whose prevalence remained unchanged. These reductions in smoking indicate the successes of public health tobacco control efforts in the general population as well as in those with MDE, SUD, or both. Furthermore, differences in cigarette smoking among adults with vs without MDE declined overall and for most subgroups. Differences among those with vs without SUD declined for women.

Unlike earlier results showing no declines in cigarette smoking among those with mental illness,^4,6 findings from this study are consistent with the observation²⁴ that tobacco cessation is achievable among adults with psychiatric disorders who use cigarettes or other tobacco products. Increasing evidence suggests that for patients with nicotine dependence, treatment for other psychiatric conditions has greater long-term success when smoking cessation is a part of comprehensive care.^8,24

Results from this study are consistent with the increases in US behavioral health care facilities offering screening for tobacco use, tobacco cessation counseling, nicotine replacement therapy, and nonnicotine cessation medications.^25,26 Moreover, reductions in the differences in the prevalence of cigarette smoking among adults with vs without psychiatric conditions are also consistent with changes in social norms regarding smoking²⁷ and the effectiveness and safety of pharmacotherapy for people with nicotine dependence and other psychiatric conditions.²⁸

Despite improvements, high prevalence of smoking persists among adults with MDE, SUD, or both, highlighting the ongoing need for targeted efforts. For example, American Indian or Alaska Native persons have higher prevalence of cigarette smoking and lower quitting rates than persons in other US racial and ethnic groups.^15,29,30 Sociostructural disadvantages, poverty, trauma, unemployment, tobacco industry’s targeted marketing, and the lack of clinicians to address psychiatric conditions and smoking may be contributing factors.^29-32 Furthermore, American Indian or Alaska Native persons were less likely to use cessation aids during past-year attempts than persons in other racial and ethnic groups.³³

This study has several limitations. First, the cross-sectional nature of NSDUH data precludes drawing causal relationships. Second, NSDUH is a self-report survey and is subject to recall and social-desirability bias. Third, the data predate the COVID-19 pandemic, so it is unknown whether the findings would be applicable to these populations during the pandemic. Fourth, MDE is the only non-SUD mental illness measure in NSDUH data directly assessed based on DSM criteria. It is unknown whether similar results would be found for psychotic disorders and other mental disorders. Fifth, for drug use disorders during 2006-2019, this study assessed 3 drug use disorders rather than 10 drug use disorders. However, sensitivity analyses show that using 3 vs 10 drug use disorders did not significantly change outcomes, perhaps because of correlations across drug use disorders.³⁴ Sixth, this study may have underestimated prevalence of MDE, SUD, co-occurring MDE and SUD, and cigarette smoking because the NSDUH excluded people experiencing homelessness not living in shelters and institutionalized (eg, jail or prison) populations who often have more substance use and psychiatric conditions than the general population.^35,36 Seventh, this study could not examine nicotine vaping, which was not assessed in the 2006-2019 NSDUH, but vaping may be used as a tobacco harm reduction approach among people who are unable to quit using conventional smoking cessation approaches.³⁷ Because most patients with opioid use disorder believed that smoking and vaping increased their vulnerability to COVID-19,³⁸ more research is needed to continue to monitor national trends in differences in prevalence of tobacco use and nicotine vaping among adults with vs without psychiatric conditions (including SUD) during the COVID-19 pandemic.

In this exploratory, serial, cross-sectional study, there were significant reductions in the prevalence of self-reported cigarette smoking among US adults with major depressive episode, substance use disorder, or both, between 2006 and 2019. However, continued efforts are needed to reduce the prevalence further.

Corresponding Author: Wilson M. Compton, MD, MPE, National Institute on Drug Abuse, National Institutes of Health, 301 N Stonestreet Ave, 3WFN Room 09D18, MSC 6025, Bethesda, MD 20892-6025 (wcompton@nida.nih.gov).

Accepted for Publication: March 14, 2022.

Author Contributions: Dr Han had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Han, Volkow, Tipperman, Einstein, Compton.

Acquisition, analysis, or interpretation of data: Volkow, Blanco, Compton.

Drafting of the manuscript: Han, Compton.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Han.

Administrative, technical, or material support: Volkow, Tipperman, Einstein.

Supervision: Volkow.

Conflict of Interest Disclosures: Dr Compton reported owning stock in General Electric, 3M Co, and Pfizer Inc. No other disclosures were reported.

Funding/Support: This study was sponsored by the National Institute on Drug Abuse of the National Institutes of Health and the Substance Abuse and Mental Health Services Administration.

Role of the Funder/Sponsor: The sponsors supported the authors who were responsible for preparation, review, and approval of the manuscript and the decision to submit the manuscript for publication. The sponsors had no role in design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation of the manuscript; and decision to submit the manuscript for publication. The sponsors reviewed and approved the manuscript.

Disclaimers: The findings and conclusions of this study are those of the authors and do not necessarily reflect the views of the National Institute on Drug Abuse of the National Institutes of Health, the Substance Abuse and Mental Health Services Administration, or the US Department of Health and Human Services.