National Trends of Mental Health Care Among US Adults Who Attempted Suicide in the Past 12 Months

Key Points

Question  What are the national trends in suicide attempts and mental health service use among adults who attempted suicide in the past 12 months from 2008 to 2019?

Findings  In this cross-sectional analysis of 484 732 National Survey of Drug Use and Health respondents, the rate of suicide attempts among US adults has increased within the past decade. However, past-year receipt of mental health services did not increase significantly among those who made suicide attempts.

Meaning  These findings suggest a need to expand service accessibility and/or acceptability, as well as public and population-wide prevention efforts.

Importance  Although suicide attempts remain the strongest risk factor for future suicide, little is known about recent trends in the prevalence of and risk factors for suicide attempts and past-year use of services among adults who attempted suicide.

Objective  To estimate annual rates of suicide attempts and use of mental health services among US adults from 2008 to 2019.

Design, Setting, and Participants  This US nationally representative cross-sectional study used the National Survey of Drug Use and Health (NSDUH) from 2008 through 2019. Participants included noninstitutionalized US civilians 18 years or older (n = 484 732). The overall annual rates of suicide attempts per 100 000 adults in the general population and national trends from 2008 to 2019 were estimated, with suicide attempts defined as self-reported efforts to kill one’s self in the past 12 months. Subgroup analyses were also performed by demographic characteristics and clinical conditions. The trends in past-year use of mental health services among those who reported past-year suicide attempts were then examined. Data were analyzed from October to December 2021.

Main Outcomes and Measures  Rate of suicide attempts from 2008 to 2019. Multivariate-adjusted logistic regression analyses were used to determine whether adjusting for sociodemographic and clinical factors associated with past-year suicide attempts could account for the change within the study period.

Results  Of 484 732 survey participants, most were 35 years or younger (69.8%), women (51.8%), and non-Hispanic White individuals (65.7%). From 2008 to 2019, the weighted unadjusted suicide attempt rate per 100 000 population increased from 481.2 to 563.9 (odds ratio [OR], 1.17 [95% CI, 1.01-1.36]; P = .04) and remained significant after controlling for sociodemographic characteristics (adjusted OR [aOR], 1.23 [95% CI, 1.05-1.44]; P = .01). Rates of suicide attempt increased particularly among young adults aged 18 to 25 years (aOR, 1.81 [95% CI, 1.52-2.16]; P < .001), women (aOR, 1.33 [95% CI, 1.09-1.62]; P = .005), those who were unemployed (aOR, 2.22 [95% CI, 1.58-3.12]; P < .001) or never married (aOR, 1.60 [95% CI, 1.31-1.96]; P < .001), and individuals who used substances (aOR, 1.44 [95% CI, 1.19-1.75]; P < .001). In multivariate analyses, the temporal trend of increasing suicide attempts remained significant even after controlling for other significant sociodemographic and clinical factors (aOR, 1.36 [95% CI, 1.16-1.60]; P < .001). Several sociodemographic and clinical subgroups remained independently associated with suicide attempts, especially those with serious psychological distress (aOR, 7.51 [95% CI, 6.49-8.68]; P < .001), major depressive episodes (aOR, 2.90 [95% CI, 2.57-3.27]; P < .001), and alcohol use disorder (aOR, 1.81 [95%CI, 1.61-2.04]; P< .001) as well as individuals who reported being divorced or separated (aOR, 1.65 [95% CI, 1.35-2.02]; P < .001) or being unemployed (aOR, 1.47 [95% CI, 1.27-1.70]; P< .001) and those who identified as Black (aOR, 1.41 [95% CI, 1.24-1.60]; P < .001) or American Indian or Alaska Native, Asian, or Native Hawaiian or Other Pacific Islander (aOR, 1.56 [95% CI, 1.26-1.93]; P < .001). Among adults with a suicide attempt, there was no significant change in the likelihood of receiving past-year mental health or substance-related services. During the study period, 34.8% to 45.5% reported needing services but did not receive them, with no significant change from 2008 to 2019.

Conclusions and Relevance  Although suicide attempts appear to be increasing, use of services among those who attempted suicide has not increased, suggesting a need to expand service accessibility and/or acceptability, as well as population-wide prevention efforts.

The number of deaths by suicide has increased by more than 60% in recent decades, from 29 199 in 1999 to 48 344 in 2018,¹ and suicide remains one of the top 10 leading causes of death in the US.^2,3 Recent efforts have been made to address this crisis nationally,^4-7 with the National Action Alliance for Suicide Prevention promulgating a national strategy to improve public recognition of suicide warning signs⁸ and the Suicide Prevention Resource Center developing Zero Suicide, a quality improvement framework for transforming suicide prevention in health care systems.⁹ These initiatives, like others,¹⁰ ultimately rely on fostering contact of individuals with suicidal behaviors with the health care systems.

Previous studies^11-15 have highlighted that a small but growing number of those who have died by suicide have had contact with health services during the previous 12 months, a potential opportunity for preventive intervention. For example, a systematic review and meta-analysis¹⁴ using data published from 2000 to 2017 suggests that 18.3% of individuals who died due to suicide had an inpatient mental health stay, 26.1% received outpatient mental health services, and 25.7% had contact with both inpatient and outpatient mental health services,¹⁴ offering an opportunity for preventive intervention.

Suicide attempts are the single most important risk factor for suicide and the risk factor most likely to precipitate contact with the health care system.^16-18 The rate of suicide in the year after a suicide attempt has been estimated to be 0.8% to 3.0% for men and 0.3% to 1.9% for women and nearly 100 times higher than the rate of death among community controls.^19-22 Because the most effective suicide prevention approaches rely on identifying and treating individuals at high risk for suicide, suicide attempts offer a valuable opportunity to intervene at a critical time to prevent future suicide.^23-25

Although previous studies have focused on health care contact before suicide, less is known about how frequently individuals who attempt suicide have clinical contacts in proximity to their attempt. Because the US does not currently have a population-based surveillance system for suicide attempts, nearly all prior studies examining service use among people who attempt suicide are based on emergency department data and thus involve individuals with access to health care.^26-29 These studies are limited because they do not account for the large percentage of high-risk individuals who never present to care. To our knowledge, only 1 prior study using population-based data from the 1990-1992 to 2001-2003 periods³⁰ has assessed the frequency of health care contact before a suicide attempt and found that 79% seek some form of general medical or mental health treatment in the year before an attempt. However, this study is based on older data and does not examine reasons for not seeking health care. There is thus a need for more recent and inclusive information on suicide attempts in the US.

This study uses nationally representative data from the 2008-2019 National Survey on Drug Use and Health (NSDUH), administered annually by the Substance Abuse and Mental Health Services Administration,³¹ to examine the changing prevalence of and risk factors for suicide attempts in recent years as well as changes in use of mental health services and reasons for nonuse among adults who report a suicide attempt in the past 12 months. We sought to address the following questions. First, what are the prevalence and temporal trends of past-year attempted suicide among US adults? Second, what are the rates and temporal trends in use of mental health services among US adults who attempted suicide in the past year? Third, what are the reasons given for not receiving mental health care among those who attempted suicide and report not receiving care? We focus on both mental health and substance use treatment, which are both of potential clinical importance in suicide prevention efforts.

We used NSDUH data from 2008 to 2019, which collects systematic information about drug use, health, and services use among noninstitutionalized civilians (ie, individuals who are not incarcerated, hospitalized, or in shelters) 12 years or older in the US.³¹ The NSDUH uses a multistage stratified sampling design to produce nationally representative estimates. We limited our samples to adults aged 18 years or older (484 732 unweighted) who were asked if they had tried to kill themselves in the past 12 months. Survey participants completed interviews in their residence in English or Spanish using a combination of computer-assisted self-interview or in-person questionnaires where responses were recorded by the interviewer. Participants received $30.00 for completion of the survey. The annual mean weighted response rate in 2008 to 2019 was 58.2%.^31,32

Our study was exempted from review by the institutional review board at Yale School of Medicine because it was based on publicly available, deidentified data. Study procedures followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. Further details of the survey, including study descriptions, questionnaires, sampling methods, and other technical reports are available on the NSDUH website.³²

Sociodemographic characteristics included age, sex, self-reported race and ethnicity, marital status (ie, married, widowed, divorced/separated, or never married), educational attainment, employment, annual family income, health insurance coverage, and urbanicity based on metropolitan statistical areas.³³ Past-year substance use and substance use disorders (SUDs) were derived from questions that addressed formal diagnostic criteria for substance abuse and dependence as defined in DSM-IV³⁴ and that addressed past-year alcohol, tobacco, cannabis, heroin or pain reliever, stimulant, sedative, hallucinogen, cocaine or crack, and inhalant use and an aggregate indicator of any substance use, abuse, or dependance (excluding alcohol and tobacco).

Past-year mental health problems were addressed with questions concerning serious psychological distress (SPD) and major depressive episodes (MDEs). The SPD is based on the Kessler Psychological Distress Scale, which uses 6 items designed to identify serious impairment from any mental health concern other than SUDs.³⁵ We measured MDE via a series of questions based on DSM-IV criteria for MDE.³⁴

Past-year use of mental health services was assessed by questions concerning receipt of any outpatient mental health services in the past 12 months and whether they were received from a mental health center, a private mental health clinician, a clinician outside a clinic, a medical clinic, a day treatment program, or another setting. The number of days of inpatient mental health treatment and receipt of any prescription medications for mental health problems in the past year were also assessed. Finally, alternative services were assessed, including a visit to an acupuncturist, chiropractor, massage therapist, herbalist, in-person support group, internet-based support group, religious advisor, telephone hotline, or other type of visit. Past-year use of substance-related services was assessed by questions that addressed receipt of treatment in the past 12 months for drug or alcohol use at a hospital, residential rehabilitation facility, outpatient rehabilitation facility, mental health center, emergency department, private physician’s office, self-help group, or another setting.

Finally, the NSDUH asks all adult respondents, regardless of mental health status, whether they needed mental health treatment or counseling at any time in the past 12 months but did not receive it. Among respondents with perceived unmet mental health treatment needs, respondents were asked to identify reasons for not receiving mental health treatment from a list of 11 potential reasons.

Data were analyzed from October to December 2021. First, the overall annual suicide attempt rates per 100 000 adults in the general population from the 2008-2009 to 2018-2019 periods were estimated, and bivariate logistic regression was used to estimate national trends^36,37 for suicide attempts across these years. We then repeated these calculations to evaluate the change in suicide attempts among each sociodemographic subgroup. To determine whether the change in suicide attempts remained significant after controlling for sociodemographic characteristics associated with changing suicide rates during these years, we further calculated adjusted odds ratios (ORs) for the entire sample as well as each sociodemographic subgroup controlling from the other groups. Similarly, we examined the trend in suicide attempts for individuals with past-year SPD, MDE, and substance abuse and dependence by again calculating unadjusted ORs and adjusted ORs (aORs) for suicide attempts across the entire study period.

Multivariate analyses were then used to determine whether sociodemographic and clinical factors associated with past-year suicide attempts could account for the change in suicide attempts between 2008 and 2019. In these analyses, past-year suicide attempt was the dependent variable, and the independent variable was the survey years, with assigning a value of 0 for 2008-2009, a value of 0.2 for 2010-2011, and so on, to a value of 1.0 for 2018-2019.^36,37 Covariates included sociodemographic and clinical factors that had emerged as significantly associated with the changes in the rate of suicide attempts in the bivariate analysis. The following 4 cumulative multivariate analyses were conducted examining the association of past-year suicide attempts with (1) survey years alone; (2) survey years plus sociodemographic factors; (3) survey years and sociodemographic factors plus past-year SUDs; and (4) survey years, sociodemographic factors, and past-year SUDs plus past-year mental health problems. If adjusting for these factors renders the temporal analysis insignificant, it is plausible to conclude that those factors account for the increase in suicide attempts during the study period. Additional details on the model building strategy are explained in the eMethods in the Supplement.

Finally, we examined the change in past-year use of mental health– and substance use–related services among individuals with past-year suicide attempts between 2008 and 2019. We also examined the percentage of individuals with past-year suicide attempts who perceived a need for treatment but did not receive treatment and the associated reasons for not receiving treatment. Again, unadjusted and adjusted ORs and their corresponding 95% CIs were estimated for these analyses.

We used the statistical software Stata, version 16.1 MP/4-Core (StataCorp LLC), for all analyses. We used svy commands to account for multistage, complex survey sampling techniques (ie, unequal probability of selection, clustering, and stratification) used in the data collection and the NSDUH complex survey design. We set a 2-sided P < .05 as the threshold of statistical significance.

In the entire NSDUH sample of individuals aged 18 years or older (weighted sample of 484 732), most were aged 35 years or younger (69.8%); 51.8% were women and 48.2% were men. In terms of self-reported race and ethnicity, 15.1% of the individuals were Hispanic, 11.7% were non-Hispanic Black, 65.7% were non-Hispanic White, and 7.5% were non-Hispanic other (including American Indian or Alaska Native, Asian, Native Hawaiian or Other Pacific Islander, and those who reported 2 or more groups). These variables are important to examine potential racial or ethnic differences in recent trends in suicide attempts to inform suicide prevention strategies. A total of 4342 past-year suicide attempts between 2008 and 2019 were reported. During this period, the weighted unadjusted rate of suicide attempts per 100 000 population increased significantly from 481.2 to 563.9 from 2008 to 2019 (OR, 1.17 [95% CI, 1.01-1.36]; P = .04) (Table 1). After adjustment for sociodemographic differences, this increase remained significant (aOR, 1.23 [95% CI, 1.05-1.44]; P = .01). Significant increases in suicide attempts were observed among people aged 18 to 25 years (aOR, 1.81 [95% CI, 1.52-2.16]; P < .001), women (aOR, 1.33 [95% CI, 1.09-1.62]; P = .005), individuals who never married (aOR, 1.60 [95% CI, 1.31-1.96]; P < .001), those with highest educational attainment of a high school degree or equivalent (aOR, 1.30 [95% CI, 1.01-1.68]; P = .04), those with full-time employment (aOR, 1.58 [95% CI, 1.27-1.98]; P < .001) or unemployed (aOR, 2.22 [95% CI, 1.58-3.12]; P < .001), individuals with family income of at least $75 000 per year (aOR, 1.56 [95% CI, 1.14-2.13]; P = .005), and those with private insurance (aOR, 1.43 [95% CI, 1.06-1.92]; P = .02) or no insurance (aOR, 1.42 [95% CI, 1.06-1.91]; P = .02). The only subgroup with a significant decrease in suicide attempts during the study period consisted of individuals aged 50 to 64 years (aOR, 0.50 [95% CI, 0.28-0.92]; P = .03).

In regard to clinical subgroups (Table 2), after adjustment for sociodemographic characteristics, suicide attempts were found to have significantly increased among individuals using any substance (aOR, 1.44 [95% CI, 1.19-1.75]; P < .001), and specifically those who used alcohol (aOR, 1.24 [95% CI, 1.01-1.52]; P = .04), tobacco (aOR, 1.38 [95% CI, 1.10-1.74]; P = .006), cannabis (aOR, 1.44 [95% CI, 1.16-1.80]; P = .001), or cocaine or crack (aOR, 1.88 [95% CI, 1.27-2.77]; P = .002) and individuals meeting criteria for any substance abuse or dependence and specifically alcohol use disorder (aOR, 1.61 [95% CI, 1.21-2.14]; P = .001), cannabis use disorder (aOR, 1.70 [95% CI, 1.14-2.53]; P = .01), and opioid use disorder (aOR, 1.67 [95% CI, 1.022.76]; P = .04). No significant change in the rate of suicide attempts was seen for individuals with past-year mental health problems, including both SPD and MDE.

In the multivariate model evaluating whether controlling for sociodemographic and clinical factors could account for the increase in suicide attempts from 2008 to 2019 (Table 3), we found the temporal trend of increasing suicide attempts remained significant, even after adjusting for an extensive number of significant factors. When sociodemographic factors and past-year SUDs were added in models 2 and 3, the adjusted time trend for suicide attempts increased in magnitude (aOR, 1.36 [95% CI, 1.16-1.60]; P < .001) in model 3. When mental health problems were added in model 4, the effect size of time declined in magnitude but was still significant (aOR, 1.17 [95% CI, 1.00-1.37]; P = .045). Although no factors for which measures were available fully accounted for the increase in suicide attempts from 2008 to 2019, several sociodemographic and clinical subgroups remained independently associated with suicide attempts, especially SPD (aOR, 7.51 [95% CI, 6.49-8.68]; P < .001), MDE (aOR, 2.90 [95% CI, 2.57-3.27]; P < .001), and alcohol use disorder (aOR, 1.81 [95% CI, 1.61-2.04]; P < .001) as well as individuals who reported being divorced or separated (aOR, 1.65 [95% CI, 1.35-2.02]; P < .001), who reported being unemployed (aOR, 1.47 [95% CI, 1.27-1.70]; P < .001), and who identified as Black (aOR, 1.41 [95% CI, 1.24-1.60]; P < .001) or non-Hispanic other race and ethnicity (aOR, 1.56 [95% CI, 1.26-1.93]; P < .001). The increasing goodness-of-fit across the 4 models is further described in the eFigure in the Supplement.

There were few changes in past-year use of services among the 4372 individuals who reported past-year suicide attempts from 2008 to 2019 (Table 4). Specifically, there were no significant changes in the likelihood of having any outpatient, inpatient, or medication services for mental health reasons and no changes in the use of treatment services for substance use. The only significant change in use of services during the study period examined was an increase in the number of mental health center visits (aOR, 1.70 [95% CI, 1.07-2.70]; P = .03), although this change would no longer be significant after Bonferroni correction for comparisons involving 6 different sources of mental health care (.05/6 = .008).

Finally, among the 4372 individuals with past-year suicide attempts in the NSDUH sample (Table 5), 34.8% (2010-2011) to 45.5% (2018-2019) reported believing that they needed mental health services at some point during the past year but did not receive them, with no significant change from 2008 to 2019. Exploration of the reasons individuals gave for not receiving services revealed a significant increase in the number of individuals who reported not knowing where to go for treatment (aOR, 1.96 [95% CI, 1.16-3.32]; P = .01), and lacking transportation or reporting that services were too far away (aOR, 5.15 [95% CI, 2.35-11.30]; P < .001).

In this nationally representative sample of US adults reporting past-year suicide attempts between 2008 and 2019, we found that the rate of suicide attempts has increased significantly, especially among young adults, women, people who are unemployed or never married, and individuals who use substances or have related use disorders. However, in the multivariate model, the combination of all significant sociodemographic and clinical factors associated with changes in suicide attempts did not account for this increase, which is likely associated with unmeasured factors such as potential contextual or social factors not assessed in the survey. Despite an increase in suicide attempts, we did not find a corresponding increase in use of services among those who attempted suicide, and a large percentage of those reporting suicide attempts indicated that they had needed mental health services but did not receive them in the year of their attempt. Because prior suicide attempts are the single most important risk factor associated with future suicide, suicide prevention strategies must rely on use of services after an attempt. However, this study suggests that many individuals who need help are not receiving these potentially life-saving services.

Compared with older nationally representative studies showing no change in the prevalence of suicide attempts between the 1990-1992 and 2001-2003 periods,^30,38 our findings demonstrate a substantial and alarming increase in suicide attempts, nationally, as well as among specific subgroups. Our findings are consistent with a more recent study showing an increase in suicide attempts between the 2004-2005 and 2012-2013 periods and replicates findings showing a persistent increase in suicide attempts among young adults, women, and individuals with educational attainment of a high school degree.²²

In addition to findings related to recent trends in the prevalence of suicide attempts, this study extends prior studies by examining service use among people who made attempts and demonstrates that only about 40% received services in the year of their attempt, with no significant change in this proportion during the study period, although the trend was toward reduced use of services from the extremes of 2008-2009 to 2018-2019. Service use in our study was significantly less frequent than in a prior study²⁹ conducted on an insured sample encompassing 10 large health care systems, in which 95% of adults who attempted suicide had health care contact in the year preceding their attempt. Our findings suggest that studies conducted entirely in treatment-seeking or insured populations^26-29 may overestimate the percentage of suicide attempters who receive treatment in proximity to their attempt, thereby underscoring the importance of monitoring population-based samples that capture people who attempt suicide who are uninsured and/or who do not seek treatment after their attempt.

Our findings showing a significant increase in suicide attempts without a corresponding increase in service use during the study period are concerning. One would hope that as suicide attempts increase, the percentage of individuals who receive treatment in proximity to their attempt would also increase. Current suicide prevention interventions largely focus on individuals connected to treatment and high-risk individuals who have contact with the health care system. Current interventions include screening and risk assessment, evidence-based treatment of depression, safety planning, means restriction, and follow-up care after attempts or hospitalizations.³⁹

However, our finding that less than half of suicide attempters had clinical contact around the time of their attempt suggest that it is not only important to expand initiatives for high-risk individuals with clinical contact, but also to implement public health–oriented strategies outside the formal treatment system. Our findings identify subgroups with rising rates of suicide attempts among whom targeted interventions may be especially needed, including young adults, individuals who are unemployed or never married, and individuals who use substances. These findings highlight the potential importance of social media interventions, media-reporting guidelines, and initiatives on college campuses to target the rising rates of suicide attempts among young people.^40,41 In addition, unemployment benefit programs and income supports may help curb rising rates among the unemployed.⁴² For individuals who use substances, novel treatment programs that exist outside the formal treatment system, such as in churches,⁴³ and continued expansion of telehealth-based interventions, especially for medications for opioid use disorder in rural areas, may help engage individuals not already in treatment.⁴⁴ Finally, the National Center for Health Statistics maintains annual surveillance data on deaths by suicide, but no national surveillance data are available on annual trends in other types of suicidal behavior, such as attempts, nonsuicidal self-injury, and even suicide ideation.^30,45

In this study, the most common reason that respondents gave for not receiving treatment was that they could not afford the cost, although the frequency with which this reason was given did not increase. This finding, although it included data after the implementation of the Affordable Care Act, underscores the potential importance of further efforts to improve access to insurance and enforcement of mental health parity. These efforts are especially important given that we found the uninsured to have one of the highest rates and greatest increase in suicide attempts during the last 10 years. Notably, the 2 reasons for not receiving treatment that increased during the years of this study—not knowing where to go for treatment and lacking transportation or reporting that services were too far away—do not immediately lend themselves to obvious remedies. However, the recent growth in telehealth services and the establishment of 988 as a national mental health crisis and suicide prevention number have the potential to improve access to treatment for high-risk individuals facing these specific barriers.⁴⁶

This study has several limitations. Most notably, the retrospective, self-reported nature of the data introduces the possibility of recall bias. Second, although the modest response rate in the NSDUH of 58.2% is greater than or similar to the response rates of other nationally representative surveys (eg, National Survey of Children’s Health or National Epidemiologic Survey of Alcohol and Related Conditions Wave III), it introduces the possibility of nonresponse bias. Although the development and application of sampling weights were intended to attenuate this potential bias, such bias may still be present. Third, suicide attempts were assessed with a single question that asked whether respondents tried to kill themselves in the past 12 months. Although this question refers specifically to suicide attempts, we cannot rule out that respondents may have interpreted it as referring to other types of behaviors, such as nonsuicidal self-injury. Fourth, we could not determine the precise timing of treatment contact and whether contact occurred before or after a suicide attempt. Finally, the NSDUH does not survey individuals currently hospitalized with a suicide attempt or homeless or incarcerated adults who have high rates of suicidal behavior.⁴⁷ Thus, it is possible that the suicide attempt rates in this study may underestimate the true prevalence. Exclusion of noninstitutionalized individuals in epidemiological surveys is part of a broader challenge to study suicidal behavior in these high-risk populations.

The findings of this study suggest that suicide attempts have increased in the US in recent decades without a corresponding increase in service use among those who made such attempts. These data further suggest a need to expand service accessibility and/or acceptability of services as well as population-wide suicide prevention efforts.

Accepted for Publication: November 17, 2021.

Published Online: January 19, 2022. doi:10.1001/jamapsychiatry.2021.3958

Corresponding Author: Greg Rhee, PhD, Department of Psychiatry, Yale School of Medicine, 100 York St, Ste 2J, New Haven, CT 06511 (taeho.rhee@yale.edu and tgrhee.research@gmail.com).

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

Concept and design: Rosenheck, Rhee.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Bommersbach, Rhee.

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

Statistical analysis: Rhee.

Administrative, technical, or material support: Rhee.

Supervision: Rosenheck, Rhee.

Conflict of Interest Disclosures: None reported.

Funding/Support: This study was supported in part by grant T32AG019134 from the National Institute on Aging through Yale School of Medicine for the past 3 years (Dr Rhee), grant R21MH117438 from the National Institute of Mental Health (Dr Rhee), grant R21AG070666 from the National Institute on Aging (Dr Rhee), and the Institute for Collaboration on Health, Intervention, and Policy of the University of Connecticut.

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