Youth Suicide

Journal of Adolescent Health 70 (2022) 83e90

www.jahonline.org

Original article

Preventing Adolescent and Young Adult Suicide: Do States With Greater Mental Health Treatment Capacity Have Lower Suicide Rates?

Evan V. Goldstein, Ph.D., M.P.P. a, Laura C. Prater, Ph.D., M.P.H., M.H.A. b, and Thomas M. Wickizer, Ph.D., M.P.H. c,* a Division of Health System Innovation & Research, Department of Population Health Sciences, University of Utah School of Medicine, Salt Lake City, Utah b Firearm Injury Prevention & Research Program, Harborview Medical Center, The University of Washington, Seattle, Washington c Division of Health Services Management & Policy, The Ohio State University College of Public Health, Columbus, Ohio

Article history: Received December 30, 2020; Accepted June 17, 2021 Keywords: Gun violence; Suicide prevention; Adolescent suicide; Firearm suicide; Mental health

A B S T R A C T IMPLICATIONS AND

Purpose: Youth suicide is increasing at a significant rate and is the second leading cause of death for adolescents. There is an urgent public health need to address the youth suicide. The objective of this study is to determine whether adolescents and young adults residing in states with greater mental health treatment capacity exhibited lower suicide rates than states with less treatment capacity. Methods: We conducted a state-level analysis of mental health treatment capacity and suicide outcomes for adolescents and young adults aged 10e24 spanning 2002e2017 using data from Centers for Disease Control and Prevention, U.S. Bureau of Labor Statistics, Federal Bureau of Investigation, and other sources. Multivariable linear fixed-effects regression models tested the relationships among mental health treatment capacity and the total suicide, firearm suicide, and nonfirearm suicide rates per 100,000 persons aged 10e24. Results: We found a statistically significant inverse relationship between nonfirearm suicide and mental health treatment capacity (p ¼ .015). On average, a 10% increase in a state’s mental health workforce capacity was associated with a 1.35% relative reduction in the nonfirearm suicide rate for persons aged 10e24. There was no significant relationship between mental health treatment capacity and firearm suicide. Conclusions: Greater mental health treatment appears to have a protective effect of modest magnitude against nonfirearm suicide among adolescents and young adults. Our findings under- score the importance of state-level efforts to improve mental health interventions and promote mental health awareness. However, firearm regulations may provide greater protective effects against this most lethal method of firearm suicide.

� 2021 Society for Adolescent Health and Medicine. All rights reserved.

Conflicts of interest: The authors have no conflicts of interest to disclose. * Address correspondence to: Thomas M. Wickizer, Ph.D., M.P.H., Division of

Health Services Management & Policy, 1841 Neil Avenue, Columbus, Ohio 43210. E-mail address: [email protected] (T.M. Wickizer).

1054-139X/� 2021 Society for Adolescent Health and Medicine. All rights reserved. https://doi.org/10.1016/j.jadohealth.2021.06.020

CONTRIBUTION

The increase in youth sui- cide requires the develop- ment of more effective interventions. This study elucidates differences be- tween nonfirearm and firearm suicide to under- stand different prevention pathways. Mental health treatment capacity is important for nonfirearm suicide prevention, while firearm suicide prevention may be best addressed through firearm safety and storage policies.

The U.S. is in the midst of a suicide epidemic taking the lives of almost 50,000 Americans each year, with rates increasing in every state from 1999 to 2016 [1]. Although suicide is the 10th leading cause of death in the U.S. overall, it is the second leading

E.V. Goldstein et al. / Journal of Adolescent Health 70 (2022) 83e9084

cause of death among adolescents and young adults aged 10e24 [1,2]. Adolescent and young adult suicides are increasing at a faster rate among young females compared to young males [3]; the firearm suicide rate among persons aged 14e24 increased by 54% since 2004 [4]. Among young adults aged 15e24, firearms, suffocation/hanging, and poisoning are the most frequently used suicide methods, respectively [5,6]. For persons younger than 15, suffocation is the most frequently used method, followed by firearms and poisoning [5]. For females aged 15%e24%, 45% of suicides were attributed to a firearm injury, 28% to suffocation, and 17% to poisoning [4]. Among males in the same age range, 53% of suicides were attributed to a firearm injury, 34% to suf- focation, and 8% to poisoning [4]. Beyond individual mortality, adolescent and young adult suicides have devastating social consequences. Among high-income countries in 2014, adoles- cent and young adult suicides accounted for an estimated loss of 406,730 years of potential life, 77% of which was attributable to the U.S., the country with the most significant adolescent and young adult suicide problem [7].

Researchers have found individual and household factors associated with the risk of suicide among adolescents and young adults. At the household level, family discord and parental divorce are associated with increased risk of adoles- cent suicide [3]. Access to firearms in the home is associated with higher suicide rates [8], while greater social support and public welfare expenditures appear to have a protective effect [9]. Suicide rates are higher among male youth compared to female youth, but suicidal ideation is more common among female youth [10]. Male adolescents are also more likely than females to use firearms in lethal suicide attempts [3]. White adolescents have experienced higher suicide rates than nonwhite adolescents [10], despite recent rising rates among black youth [11]. Those who report same-sex sexual orientation are also at greater risk for suicide [12]. Mental illness, especially depression, has been associated with increased risk of adoles- cent suicide [13]. Unfortunately, many suicides occur prior to uncovering mental illness [14], making the suicide attempt the first sign of distress.

With firearms accounting for so many U.S. suicide deaths, many public policy efforts have focused on limiting access to firearms through state-level regulatory restrictions. In general, the academic literature demonstrates that stricter firearm laws, such as policies aimed at regulating the supply of firearms through background checks and mandatory waiting periods before firearm issue, are associated with lower firearm fatality rates [15,16]. Evidence predating the recent spike in adolescent suicides demonstrated that child access prevention laws reduced the rate of youth firearm suicide, and offered some protective effect on firearm suicide for older members within the household by limiting access to firearms [17]. Meaningful firearm safety and control policies remain controversial and difficult to enact even at the state level, despite states’ authority to do so [18].

Beyond firearm regulation, much public attention has focused on mental health treatment interventions to reduce youth sui- cide. However, studies examining the effectiveness of these in- terventions have been limited by power issues and small sample sizes [19]. Substance abuse, interpersonal trauma, and mental illness are known risk factors strongly linked to suicide attempts among younger persons [13,20], but studies have shown promise that primary care-based interventions, adequate outpatient care, and access to ongoing mental healthcare may reduce youth

suicide [21,22]. Prior research also suggests the assessmen- t/restriction of lethal means (i.e., firearms, medications) and counseling by clinicians can reduce lethal suicide attempts among adults and may improve opportunities to detect and treat mental health conditions [23e26], but knowledge is more limited for youth.

Mental healthcare shortages are well-documented across the U.S., and many families find it difficult to access child or adolescent mental health clinicians. Prior research [27] has demonstrated an association between access to mental health- care and reduced risk of suicide among persons of all ages, including one recent study suggesting that living in a federally designated mental health professional shortage area was corre- lated with suicide death [28]. But less is known about the pro- tective effects of mental health services for suicide among adolescents and young adults. To our knowledge, there has not been a comprehensive state-level analysis of mental health treatment capacity and suicide rates among adolescents and young adults.

The severity of youth suicide in the U.S., and the fact that states have significant power to fund and design their mental healthcare systems and enact firearm safety and control pol- icies, prompted this state-level analysis examining the rela- tionship between mental health treatment capacity and suicide. Using data from 2002 to 2017, we examined whether states with greater mental health treatment capacity have lower sui- cide rates among adolescents (aged 10e19) and young adults (aged 20e24), including both firearm and nonfirearm suicide rates, compared with states having less treatment capacity.

Methods

Data and study design

Our analysis merged data from multiple sources. The pri- mary data sources were the Centers for Disease Control and Prevention (CDC) Web-based Injury Statistics Query and Reporting System (WISQARS) [4], an interactive database that compiles information on fatal and nonfatal injury and violent death in the U.S., and the Bureau of Labor Statistics (BLS) Occupational Employment Statistics program, which produces state-level, longitudinal employment data for nearly 800 occu- pations. Additional data sources included the American Com- munity Survey and Current Population Survey from IPUMS, Kaiser Family Foundation State Health Facts database, Federal Bureau of Investigation National Instant Criminal Background Check System, U.S. Census Bureau Historical Poverty Tables, and the Urban Institute State and Local Finance initiative.

We performed a state-level, time-series cross-sectional analysis that took advantage of natural variation between states and over time in our variables of interest. The state-year was the unit of analysis, which is appropriate because states have authority over the funding, design, and regulation of their mental healthcare systems, as well as firearm safety and control regulation. The final analytic file contained 186 observations spanning four time periods: t ¼ 2002, 2007, 2012, and 2017.

Dependent variables

Our first dependent variable measured total intentional suicide among adolescents (aged 10e19) and young adults (aged 20e24).

E.V. Goldstein et al. / Journal of Adolescent Health 70 (2022) 83e90 85

Our second and third dependent variables measured intentional firearm and nonfirearm suicide, respectively, among adolescents and young adults. Because the CDC recognizes suicide as a leading cause of death among persons up to age 24 [1], we extracted crude rates of annual suicides per 100,000 persons aged 10e24 from the WISQARS Fatal Injury database for the three dependent variables [4]. CDC data restrictions prevent the analysis of state-level suicide rates involving less than 10 decedents. For this reason, for each dependent variable, we combined the annual suicide rates for each time period and its preceding year to develop an average rate for the 2-year period. Following this process, we excluded state- years from the analysis if the 2-year average suicide rate for any dependent variable was still constructed from less than 10 de- cedents. Excluded state-years were Connecticut (2012), Delaware (2002, 2007), Hawaii (2002, 2007, 2017), Massachusetts (2012), New Hampshire (2012), Rhode Island (all years), Vermont (2002, 2007), and Washington, DC (all years), representing 8.8% of all possible state-years.

Independent variables

We had one independent variable: mental health treatment capacity, measured as the annual mental health workforce size for each state-year. To construct this measure, we extracted occupa- tional (OCC) codes gathered from the Occupational Employment Statistics database [29]. OCC codes 19-3031 (clinical, counseling, and school psychologists), 21-1011 (substance abuse and

Table 1 Characteristics of the analytic sample: 2002e2017

2002 2007

Total suicide rate per 100,000 (2-year averages)

8.5 (3.2) 8.6

Firearm suicide rate per 100,000 (2-year averages)

4.8 (2.4) 4.3

Nonfirearm suicide rate per 100,000 (2-year averages)

3.7 (1.3) 4.4

Mental health practitioners per state, in 1,000s

10.6 (10.5) 12.4

Annual FBI firearm background checks per state, in 100,000s

1.8 (1.5) 2.4

State population, in 100,000s 61.6 (64.6) 64.5 Race (% of population) White 80.6% (10.1) 79.3% Black 10.3% (9.7) 10.4%

Male (% of population) 49.0% (.8) 49.4% Adult population with high

school diploma (%) 83.9% (4.1) 86.0%

Population reporting divorced marital status (%)

7.9% (1.2) 8.0%

Per capita public expenditure on parks, recreation, and libraries

$175.3 (63.7) $185.2

State unemployment rate 5.2% (1.0) 4.6% Population living below

poverty (%) 11.8% (3.2) 11.9%

Affordable Care Act Medicaid expansion Expansion not yet adopted 46 0% 46 State adopted expansion 0 100% 0 Observations 46 46

Authors’ analysis of data from the Web-based Injury Statistics Query and Reporting S Population Survey from IPUMS CPS, Federal Bureau of Investigation (FBI) National Insta initiative, U.S. Census Bureau, and Kaiser Family Foundation, 2002e2017. For each var included in the analytic. Standard deviations are shown in parentheses for continuou States could enact the Affordable Care Act Medicaid expansion beginning in 2014.

behavioral disorder counselors), 21-1014 (mental health coun- selors), 21-1022 (medical and public health social workers), 29- 1066 (psychiatrists), 31-1013 (psychiatric aides), and 21-1023 (mental health and substance abuse social workers) were used to construct the variable for each state-year in the analytic sample.

Covariates

We included covariates in our statistical models to adjust for potential confounding factors. We used data from the Bureau of Labor Statistics, U.S. Census Bureau, and American Community Survey to adjust for state-level, temporal differences in unem- ployment rate, poverty rate, and educational attainment, race, and gender compositions. Our models adjusted for the total population of each state across time to account for population-to size-related variation in mental health workforce capacity. We included data from the Current Population Survey to adjust for the percentage of people in each state-year who reported “divorced” for their marital status. To adjust for state-level dif- ferences in the availability of social support resources, we used data from the Urban Institute to construct a proxy measure of the per capita public expenditure on parks, recreation, and libraries. Because the Affordable Care Act Medicaid expansion may have been associated with reductions in suicide by improving access to healthcare [30], we included data from Kaiser Family Foun- dation to adjust for whether states enacted the Affordable Care Act Medicaid expansion. Finally, we included dummy variables

2012 2017

(3.8) 10.1 (3.6) 12.5 (5.1)

(2.3) 5.0 (2.6) 6.3 (3.4)

(1.8) 5.1 (1.9) 6.2 (2.3)

(13.1) 12.9 (14.5) 14.8 (15.8)

(2.6) 4.1 (4.5) 5.1 (7.3)

(67.6) 65.3 (72.0) 67.1 (73.9)

(10.4) 77.2% (13.0) 77.4% (10.7) (9.6) 10.8% (9.9) 10.8% (9.7) (.8) 49.4% (.8) 49.4% (.8) (3.7) 87.9% (3.2) 89.5% (2.7)

(1.1) 8.5% (1.3) 8.6% (1.4)

(73.2) $175.4 (61.6) $179.4 (76.4)

(.9) 7.7% (1.6) 4.4% (.9) (2.9) 14.5% (3.3) 12.1% (2.9)

0% 46 0% 19 39.6% 100% 0 100% 29 60.4%

46 48

ystem (WISQARS) Fatal Injury system, American Community Survey and Current nt Criminal Background Check System, the Urban Institute State and Local Finance iable, unadjusted average percentages or counts per year are shown for the states s variables, and percentages are shown in parentheses for categorical variables.

Table 2 National suicide and crude death rates by year and age group: 2002e2017

Panel A: ages 10e19

Year Total suicide rate per 100,000, ages 10e19

Firearm suicide rate per 100,000, ages 10e19

Nonfirearm suicide rate per 100,000, ages 10e19

Crude death rate per 100,000, ages 10e19 (all causes)

2002 4.23 1.98 2.25 42.85 2007 3.87 1.59 2.28 39.00 2012 4.97 2.05 2.92 30.79 2017 7.18 3.09 4.09 33.65

Panel B: ages 20e24

Year Total suicide rate per 100,000, ages 20e24

Firearm suicide rate per 100,000, ages 20e24

Nonfirearm suicide rate per 100,000, ages 20e24

Crude death rate per 100,000, ages 20e24 (all causes)

2002 12.33 6.65 5.68 95.01 2007 12.62 6.03 6.59 98.13 2012 13.68 6.47 7.21 84.61 2017 17.04 8.38 8.66 95.57

National suicide rates were obtained from the CDC Web-based Injury Statistics Query and Reporting System (WISQARS) Fatal Injury Reports and crude death data were obtained from the CDC WONDER database. Crude rates per 100,000 shown. CDC ¼ Centers for Disease Control and Prevention.

E.V. Goldstein et al. / Journal of Adolescent Health 70 (2022) 83e9086

for each time period and each state to adjust for secular time trends and unmeasured, time-invariant state-level policies and characteristics.

Firearm availability is associated with suicide [31], but we were unable to directly control for it. Consistent measures of firearm availability are not available at the state level for all states. As recommended elsewhere [32], we included the mea- sure of annual federal firearm background checks from the Federal Bureau of Investigation National Instant Criminal Back- ground Check System as a proxy for gun ownership rates in the

Table 3 Highest and lowest total suicide rates per 100,000, by state and year: 2002e2017

2002

States with highest ratesa Total suicide rate per 100,000

Alaska 19.78 Wyoming 17.84 South Dakota 14.19 Idaho 12.56 Montana 12.33 New Mexico 11.64 Colorado 11.37 Arizona 10.96 Kansas 10.41 New Hampshire 10.39

2017

States with highest ratesa Total suicide rate per 100,000

Alaska 32.81 Montana 23.48 South Dakota 23.12 Wyoming 19.72 North Dakota 18.81 New Mexico 18.63 Colorado 16.99 Oklahoma 16.60 Utah 16.37 Idaho 15.99

Authors’ analysis of data from the CDC Web-based Injury Statistics Query and Reportin average of crude firearm suicide rates for individuals aged 10e24 for each study time observations were available in this analysis. For 2002 and 2017, we compared the av highest and lowest suicide rates using bivariate t-tests and Mann-Whitney U-tests. p

a The states with the highest suicide rates had significantly greater federal firearm

statistical models for total and firearm suicide rates, but not in the nonfirearm suicide rate model.

Analysis

We tested multivariable linear fixed-effects regression models to examine the relationships between mental health treatment capacity and suicide rates. Robust standard errors were clustered at the state level to correct for problems poten- tially caused by heteroscedasticity or serial correlation. To

States with lowest rates Total suicide rate per 100,000

New Jersey 3.58 Massachusetts 4.41 California 4.42 New York 4.71 Connecticut 4.84 South Carolina 6.1 Illinois 6.31 Florida 6.54 Maryland 6.69 Virginia 6.71

States with lowest rates Total suicide rate per 100,000

New Jersey 5.54 New York 5.90 Connecticut 6.50 Massachusetts 6.56 California 6.83 Maryland 6.87 Delaware 7.58 Florida 8.17 Illinois 8.23 North Carolina 9.56

g System (WISQARS) Fatal Injury Reports. The total suicide variable is the 2-year period (and its preceding year), as described in the manuscript. Not all state-year erage federal firearm background checks per capita between the states with the < .01 using both tests. background checks per capita than the states with the lowest suicide rates.

Table 4 Estimating the effects of greater mental health treatment capacity on suicides per 100,000 persons aged 10e24

1 2 3

Outcome: total suicides/100,000 persons

Outcome: firearm suicides/100,000 persons

Outcome: nonfirearm suicides/100,000 persons

Mental health practitioners per state, in 1,000s �.073 �.021 �.052* .106 .521 .015

Annual FBI firearm background checks, in 100,000s .025 .022 .430 .221

State population, in 100,000s �.009 �.007 �.003 .746 .706 .984

Race (%) White �.032 �.12 .083

.734 .136 .112 Black �.643* �.518þ �.131

.035 .077 .307 Male population (%) 1.192 .324 .866þ

.162 .541 .069 Adult population with high school diploma (%) �.453** �.14 �.299**

.005 .222 <.001 Population reporting divorced marital status (%) �.294 .047 �.335*

.127 .693 .018 Per capita public expenditure on parks, recreation, and libraries .005 .003 .002

.361 .432 .432 Unemployment rate (state) .086 .062 .01

.653 .546 .931 Population living below poverty (%) .074 .026 .043

.544 .786 .376 Affordable Care Act Medicaid expansion Expansion not yet adopted Reference Reference Reference State adopted the expansion �.459 �.432 �.066

.448 .342 .849 Year 2002 Reference Reference Reference 2007 .843 �.339 1.138**

.177 .348 <.001 2012 3.450** .583 2.847**

<.001 .268 <.001 2017 7.272** 2.505** 4.666**

<.001 .004 <.001 Constant .322 15.327 �15.783

.933 .543 .503 Observations 186 186 186 Adjusted R2 .69 .52 .69

p values are shown in italics below each coefficient. State fixed-effects (FE) coefficients not shown. Authors’ analysis of data from the Web-based Injury Statistics Query and Reporting System (WISQARS) Fatal Injury system, American Community Survey from IPUMS USA, Current Population Survey from IPUMS CPS, Federal Bureau of Investigation (FBI) National Instant Criminal Background Check System, U.S. Census Bureau, Kaiser Family Foundation, and the Urban Institute State and Local Finance initiative, 2002e2017. FBI ¼ Federal Bureau of Investigation. þp < .10; *p < .05; **p < .01.

E.V. Goldstein et al. / Journal of Adolescent Health 70 (2022) 83e90 87

facilitate the interpretation of our findings, we generated regression-adjusted annual probabilities of suicide for each study year, while keeping other covariates at their observed values (i.e., estimating average marginal effects). We established an a priori two-sided statistical significance level of .05. Analyses were conducted using Stata version 15.1 (College Station, TX). Insti- tutional Review Board approval was not necessary for this state- level study.

Results

On average, the total suicide rate among individuals aged 10e 24 in the states included in this analysis increased 47.1% from 2002 to 2017 (Table 1). The average firearm and nonfirearm suicide rates grew by 31.3% and 67.6%, respectively, over the same time period. Mental health treatment capacity, as measured by our mental health workforce variable, grew by 28.6% on a per

capita basis. Table 2 shows that the total suicide rate from 2002 to 2017 grew more among adolescents aged 10e19 (69.8% in- crease) than young adults aged 20e24 (38.2% increase). From 2002 to 2017, the firearm and nonfirearm suicide rates increased by 56.1% and 81.8%, respectively, among 10- to 19-year olds and by 26.2% and 52.5%, respectively, among 20- to 24-year olds. In 2002, 9.9% of all deaths among individuals aged 10e19 were suicides. By 2017, approximately 21.3% of all deaths among per- sons aged 10e19 and 17.8% of all deaths among persons aged 20e 24 were suicides.

Table 3 demonstrates the between-state variation in total suicide rates over the study period, listing states with the highest and lowest total suicide rates at the beginning and end of our study. Among the states included in our analysis, Alaska, Wyoming, Montana, and South Dakota experienced the highest adolescent and young adult suicide rates, on average, from 2002 to 2017, and the rates increased in all four states from 2002 to

0.0000%

0.0020%

0.0040%

0.0060%

0.0080%

0.0100%

0.0120%

0.0140%

0.0160%

2002 2007 2012 2017

P ro

b a b il

it y o

f su

ic id

e

Any suicide Firearm suicide Non-firearm suicide

Figure 1. Adjusted probability of suicide, by method of suicide: 2002e2017. This figure shows the regression-adjusted probability of any suicide, firearm suicide, and nonfirearm suicide for the years 2002, 2007, 2012, and 2017. These probabilities were calculated for the entire estimation sample for each year, keeping all other covariates at their observed values (i.e., using average marginal effects).

E.V. Goldstein et al. / Journal of Adolescent Health 70 (2022) 83e9088

2017. In contrast, New Jersey, New York, Massachusetts, Califor- nia, and Connecticut were consistently ranked among the states experiencing the lowest suicide rates from 2002 to 2017, although the suicide rates also increased in these states from 2002 to 2017. Bivariate analyses comparing the states with the lowest and highest suicide rates in 2017 also showed that the 10 states with the highest suicide rates likely had significantly greater firearm availability, as measured by the annual federal firearm background checks per capita. In 2017, the states with the highest suicide rates had an average of .096 federal firearm background checks per capita, compared to an average of .047 federal firearm background checks per capita in the states with the lowest suicide rates (p < .001).

Table 4 shows the results of our multivariable analysis. We found an inverse relationship between the state-level mental health workforce capacity and the total suicide rate (b ¼ �.073, p ¼ .106). Although the finding was not statistically significant at the .05 significance level, the result implies that, on average, a 10% relative increase in the mental health workforce capacity in a state would be independently associated with a .923% relative reduction in the total suicide rate for persons aged 10e24 (p ¼ .106). We found a statistically significant, inverse relationship between the mental health workforce capacity and the nonfirearm suicide rate (b¼ �.052, p ¼ .015). This result implies that, on average, a 10% relative increase in the mental health workforce capacity in a state would be independently associated with a 1.35% relative reduction in the nonfirearm suicide rate for persons aged 10e24 (p ¼ .015). There was no statistically significant relationship between state-level mental health workforce capacity and the firearm suicide rate.

Figure 1 shows the adjusted probability of suicide in a given year for persons aged 10e24 over the study period, as observed in our estimation sample. The adjusted probability of a person aged 10e24 dying by any method of suicide in a year increased from .0071% in 2002% to .0143% in 2017da 101.4% relative increase. Although the adjusted probability of firearm suicide increased only modestly over time, the adjusted probability of a persons aged 10e24 dying by nonfirearm

suicide in a year increased considerably from .0027% in 2002 to .0074% in 2017.

There was a negative relationship between the percentage of a state’s population reporting divorced marital status and the nonfirearm suicide rate (Table 4; p ¼ .018). The percentage of a state’s adult population with a high school diploma was also inversely related with the total (p ¼ .005) and nonfirearm suicide rates (p < .001).

Discussion

Our findings suggest that greater mental health treatment capacity at the state level has a statistically significant protective effect of modest magnitude against nonfirearm suicide among adolescents and young adults aged 10e24, though no protective effect against firearm suicide. Our findings have relevance for policy considerations and for the development of interventions aimed at reducing youth suicide incidence.

Substance abuse, interpersonal trauma, and mental illness are strongly linked to suicide attempts among younger persons [13,20,33,34]. The high case-fatality rate of firearm suicide [35] may dampen the ability of mental health practitioners to diag- nosis a mental illness or successfully intervene when necessary, yet only 7% of those who make a nonfatal suicide attempt go on to die from a future attempt [36]. For younger persons who will attemptdor have attempteddsuicide using less lethal means, risk factors for suicide may be more sensitive to greater mental illness detection efforts, and improving access to mental health treatment when needed may help prevent nonfirearm suicide attempts.

Previous studies have shown promise that adequate outpa- tient care, primary care-based interventions such as improved screening for suicide risk factors and access to cognitive behav- ioral therapy, and access to mental healthcare after presenting in an emergency department following a suicide attempt may reduce youth suicide [21,22,34,37,38]. Lethal means assessment/ restriction has also shown promise among youth with programs such as SafetyCheck [39]. As index suicide attempts (IA) have been shown to be more lethal for youth and young adults across

E.V. Goldstein et al. / Journal of Adolescent Health 70 (2022) 83e90 89

all methods [3,40], prevention efforts should start prior to an IA and consider an approach that addresses both identifying serious suicidal ideation and restricting access to lethal means.

Our findings may therefore support state-level efforts to improve mental health treatment capacity and promote mental health awareness. For example, states can enhance school-based mental health services capacity, which has been shown to help reduce depressive episodes and suicide risk among adolescents [41]. Schools often access funds for school-based mental health and substance abuse services through state sources, including Medicaid benefits (e.g., Early and Periodic Screening, Diagnostic and Treatment) and Medicaid waiver programs; through the state-level allocation of funds from the Every Student Succeeds Act (2015); and through state applications to the federal School- Based Mental Health Services Grant Program.

States can also raise awareness about youth mental health issues by promoting mental health literacy programs like Mental Health First Aid (MHFA), which provides training on common mental health conditions and how to refer youth for care. Since 2015, 20 states have prioritized MHFA by enacting policies to fund training, require certification for public sector employees, and establish state-wide mental health training requirements. California and Pennsylvania led the U.S. in funding MHFA trainings in 2014, and Texas allocated $5 million to train youth educators in MHFA [42]. Prior evaluations have shown that the MHFA program may help reduce unmet need for behavioral healthcare in rural areas [43]. The Youth MHFA program has also helped participants (e.g., neighbors and teachers) become more aware of mental health resources, accepting of young persons with mental health conditions, and willing to help in times of need [44]. States can also fund mental health awareness campaigns using social media, such as California’s Each Mind Matters Campaign, which have improved positive beliefs about the possibility of recovery from mental illness [45].

Our findings do not suggest that greater mental health treatment capacity will systematically reduce firearm suicide among adolescents and young adults. The risk of firearm suicide may be less about diagnosing a mental illness and more about the potential impulsivity of those who attempt suicide with firearms [46,47] and the lethality of firearm suicide [48], which together often prevent intervention from health professionals. Prior research suggests that the adoption of stricter firearm safety and control policies will likely yield greater protective effects against firearm suicide [27]. Measures often discussed by policymakers include mandatory waiting periods before firearm issue and child access prevention laws, which are shown to reduce youth firearm suicide [49,50]. However, policy interven- tion to improve firearm control is often overwhelmed by pre- vailing political forces, even though large majorities of Americansdincluding both firearm owners and nonfirearm ownersdsupport a range of regulatory measures to strengthen firearm safety laws [51]. As described in Table 3, states with the highest suicide rates had significantly greater federal firearm background checks per capita, a proxy for gun ownership. Yet evidence-based policies shown to reduce firearm suicide appear to be absent in states with the highest suicide rates [17].

Our findings also suggest that higher rates of high school completion were significantly associated with lower suicide rates, consistent with other studies [52]. These findings may support the idea that investments in education are important for preventing suicide among adolescents and young adults. At the

individual level, suicide risk tends to increase with poor school performance and dropout [53,54], though the link between educational attainment and suicide is less certain. However, when considering education as a measure of aggregate human capital in the context of other related socioeconomic factors, education may have a protective effect [55]. Improving funding for K-12 public education in states with high suicide rates, encouraging other investments in human capital development, and providing opportunities for family counseling as part of schooling [56] should be explored as population-level suicide prevention strategies.

Limitations

This study has several limitations. First, we used a non- randomized, retrospective study design, which imposes limits on causal inference. Second, because we conducted a state-level analysis, readers should refrain from making inferences about individual behavior. Third, without more granular data (e.g., in- dividual or county level), we could not perform a multivariable analysis within each state longitudinally. Fourth, as described earlier, CDC data restrictions prevented us from constructing our dependent variables for all state-years. For this reason, the generalizability of our results is potentially limited to the states included in our analytic sample. Fifth, for the same reasons of insufficient data and data restrictions, we could not conduct subgroup analyses by sex or age (e.g., only persons <18 years). It would be important to explore how sex may have moderated our findings in future studies using different data. Sixth, due to Bu- reau of Labor Statistics data limitations, we could not identify and include other types of providersdsuch as mental health nurse practitioners or adolescent behavioral health physiciansdin our measure of mental health treatment capacity. Nonmental health practitioners may provide mental health screening or other services to adolescents and young adults. This limitation also prevented us from identifying and constructing a measure of only school-based mental healthcare providers. Finally, we could not directly control for firearm availability, an important pre- dictor of youth suicide. Consistent with other studies we included a proxy measure of the annual number of federal firearm background checks performed in each state [32]. How- ever, federal background checks do not capture private firearm purchases, hence this variable is an incomplete proxy measure for firearm availability.

Conclusions

Increasing the mental health workforce and the availability of mental health services at the state level appears to be important for nonfirearm suicide prevention. In contrast, mental health treatment capacity appears to have little effect on the more lethal method of firearm suicide. Mental health dis- orders may go undiagnosed among youth who die by all methods of suicide. However, given that suicide can be an impulsive act [46] and suicide attempts using a firearm are nearly always fatal, preventing firearm suicide directly may be best addressed through the enactment of evidence-based firearm safety and storage regulations by state-level policy- makers. Population-level investments in human capital devel- opment may also promote future well-being and protect young persons from suicide.

E.V. Goldstein et al. / Journal of Adolescent Health 70 (2022) 83e9090

Funding Sources

Dr. Prater receives research funding support from the State of Washington.

References

[1] Centers for Disease Control and Prevention. Suicide & self-inflicted injury. 2017. Available at: https://www.cdc.gov/nchs/fastats/suicide.htm. Accessed December 1, 2019.

[2] Heron M. Deaths: Leading causes for 2016. Natl Vital Stat Rep 2018;67:1e 77.

[3] Ruch DA, Sheftall AH, Schlagbaum P, et al. Trends in suicide among youth aged 10 to 19 years in the United States, 1975 to 2016. JAMA Netw Open 2019;2:e193886.

[4] Centers for Disease Control and Prevention. Web-based injury statistics query and reporting system (WISQARS) [Internet]. Available at: https:// www.cdc.gov/injury/wisqars/index.html. Accessed December 15, 2020.

[5] Choi NG, DiNitto DM, Marti CN. Youth firearm suicide: Precipitating/risk factors and gun access. Child Youth Serv Rev 2017;83:9e16.

[6] Centers for Disease Control and Preventation. 10 leading causes of death by age group, United States e 2017. 2017. Available at: https://www.cdc.gov/ nchs/fastats/leading-causes-of-death.htm. Accessed December 1, 2019.

[7] Doran CM, Kinchin I. Economic and epidemiological impact of youth sui- cide in countries with the highest human development index. PLoS One 2020;15:e0232940.

[8] Knopov A, Sherman RJ, Raifman JR, et al. Household gun ownership and youth suicide rates at the state level, 2005e2015. Am J Prev Med 2019;56:335e42.

[9] Minoiu C, Andrés AR. The effect of public spending on suicide: Evidence from U.S. state data. J Socio Econ 2008;37:237e61.

[10] Cash SJ, Bridge JA. Epidemiology of youth suicide and suicidal behavior. Curr Opin Pediatr 2009;21:613e9.

[11] Lindsey MA, Sheftall AH, Xiao Y, Joe S. Trends of suicidal behaviors among high school students in the United States: 1991-2017. Pediatrics 2019;144: e20191187.

[12] Russell ST, Fish JN. Mental health in lesbian, gay, bisexual, and transgender (LGBT) youth. Annu Rev Clin Psychol 2016;12:465e87.

[13] Nanayakkara S, Misch D, Chang L, Henry D. Depression and exposure to suicide predict suicide attempt. Depress Anxiety 2013;30:991e6.

[14] Stone DM, Simon TR, Fowler KA, et al. Trends in state suicide rates 1999- 2016. Morb Mortal Wkly Rep 2018;67:617e24.

[15] Fleegler EW, Lee LK, Monuteaux MC, et al. Firearm legislation and firearm- related fatalities in the United States. JAMA Intern Med 2013;173:732e40.

[16] Santaella-Tenorio J, Cerdá M, Villaveces A, Galea S. What do we know about the association between firearm legislation and firearm-related in- juries? Epidemiol Rev 2016;38:140e57.

[17] Smart R, Morral A, Smucker S, et al. The science of gun policy. 2nd ed. Santa Monica: RAND Corporation; 2020.

[18] Goldstein EV, Prater LC, Bose-Brill S, Wickizer TM. The firearm suicide crisis: Physicians can make a difference. Ann Fam Med 2020;18:265e8.

[19] Robinson J, Bailey E, Witt K, et al. What works in youth suicide prevention? A systematic review and meta-analysis. EClinicalMedicine 2018;4-5:52e91.

[20] Joshi K, Billick SB. Biopsychosocial causes of suicide and suicide prevention outcome studies in juvenile detention facilities: A review. Psychiatr Q 2017;88:141e53.

[21] Asarnow JR, Baraff LJ, Berk M, et al. An emergency department intervention for linking pediatric suicidal patients to follow-up mental health treat- ment. Psychiatr Serv 2011;62:1303e9.

[22] Campo JV. Youth suicide prevention: Does access to care matter? Curr Opin Pediatr 2009;21:628e34.

[23] Boggs JM, Beck A, Ritzwoller DP, et al. A quasi-experimental analysis of lethal means assessment and risk for subsequent suicide attempts and deaths. J Gen Intern Med 2020;35:1709e14.

[24] Mann JJ, Apter A, Bertolote J, et al. Suicide prevention strategies: A sys- tematic review. J Am Med Assoc 2005;294:2064e74.

[25] Yip PSF, Caine E, Yousuf S, et al. Means restriction for suicide prevention. Lancet 2012;379:2393e9.

[26] Daigle MS. Suicide prevention through means restriction: Assessing the risk of substitution. A critical review and synthesis. Accid Anal Prev 2005; 37:625e32.

[27] Goldstein EV, Prater LC, Wickizer TM. Behavioral health care and firearm suicide: Do states with greater treatment capacity have lower suicide rates? Health Aff (Millwood) 2019;38:1711e8.

[28] Johnson KF, Brookover DL. Counselors’ role in decreasing suicide in mental health professional shortage areas in the United States. J Ment Health Couns 2020;42:170e86.

[29] Bureau of Labor Statistics. Occupational Employment Statistics [Internet]. Available at: https://www.bls.gov/oes/. Accessed December 15, 2020.

[30] Borgschulte M, Vogler J. Did the ACA Medicaid expansion save lives? J Health Econ 2020;72:102333.

[31] Anglemyer A, Horvath T, Rutherford G. The accessibility of firearms and risk for suicide and homicide victimization among household members: A systematic review and meta-analysis. Ann Intern Med 2014;160: 101e10.

[32] Lang M. Firearm background checks and suicide. Econ J 2013;123:1085e 99.

[33] Nock MK, Green JG, Hwang I, et al. Prevalence, correlates, and treatment of lifetime suicidal behavior among adolescents. JAMA Psychiatry 2013;70: 300.

[34] Cavanagh JTO, Carson AJ, Sharpe M, Lawrie SM. Psychological autopsy studies of suicide: A systematic review. Psychol Med 2003;33:395e405.

[35] Spicer RS, Miller TR. Suicide acts in 8 states: Incidence and case fatality rates by demographics and method. Am J Public Health 2000;90:1885e 91.

[36] Owens D, Horrocks J, House A. Fatal and non-fatal repetition of self-harm. Br J Psychiatry 2002;181:193e9.

[37] Bridge JA, Horowitz LM, Fontanella CA, et al. Prioritizing research to reduce youth suicide and suicidal behavior. Am J Prev Med 2014;47(3 Suppl 2): S229e34.

[38] Fontanella CA, Warner LA, Steelesmith DL, et al. Association of timely outpatient mental health services for youths after psychiatric hospi- talization with risk of death by suicide. JAMA Netw Open 2020;3: e2012887.

[39] Wolk CB, Jager-Hyman S, Marcus SC, et al. Developing implementation strategies for firearm safety promotion in paediatric primary care for suicide prevention in two large US health systems: A study protocol for a mixed-methods implementation study. BMJ Open 2017;7: e014407.

[40] McKean AJS, Pabbati CP, Geske JR, Bostwick JM. Rethinking lethality in youth suicide attempts: First suicide attempt outcomes in youth ages 10 to 24. J Am Acad Child Adolesc Psychiatry 2018;57:786e91.

[41] Paschall MJ, Bersamin M. School-based health centers, depression, and suicide risk among adolescents. Am J Prev Med 2018;54:44e50.

[42] National Council for Behavioral Health. Mental health first aid policy handbook. 2019. Available at: https://www.thenationalcouncil.org/wp- content/uploads/2019/03/031219_NCBH_MHFAPolicyHandbook_v6.pdf? daf¼375ateTbd56. Accessed December 15, 2020.

[43] Talbot JA, Ziller EC, Szlosek DA. Mental health first aid in rural com- munities: Appropriateness and outcomes. J Rural Health 2017;33:82e 91.

[44] Noltemeyer A, Huang H, Meehan C, et al. Youth mental health first aid: Initial outcomes of a statewide rollout in Ohio. J Appl Sch Psychol 2020;36: 1e19.

[45] Collins RL, Wong EC, Breslau J, et al. Social marketing of mental health treatment: California’s mental illness stigma reduction campaign. Am J Public Health 2019;109:S228e35.

[46] Simon TR, Swann AC, Powell KE, et al. Characteristics of impulsive suicide attempts and attempters. Suicide Life Threat Behav 2002;32:49e59.

[47] Peterson LG, Peterson M, O’Shanick GJ, Swann A. Self-inflicted gunshot wounds: Lethality of method versus intent. Am J Psychiatry 1985;142: 228e31.

[48] Conner A, Azrael D, Miller M. Suicide case-fatality rates in the United States, 2007 to 2014 a nationwide population-based study. Ann Intern Med 2019;171:885e95.

[49] Webster DW, Vernick JS, Zeoli AM, Manganello JA. Association between youth-focused firearm laws and youth suicides. J Am Med Assoc 2004;292: 594e601.

[50] Gius M. The impact of minimum age and child access prevention laws on firearm-related youth suicides and unintentional deaths. Soc Sci J 2015;52: 168e75.

[51] Barry CL, Stone EM, Crifasi CK, et al. Trends in public opinion on us gun laws: Majorities of gun owners and nonegun owners support a range of measures. Health Aff (Millwood) 2019;38:1727e34.

[52] Fontanella CA, Saman DM, Campo JV, et al. Mapping suicide mortality in Ohio: A spatial epidemiological analysis of suicide clusters and area level correlates. Prev Med 2018;106:177e84.

[53] Kosidou K, Dalman C, Fredlund P, et al. School performance and the risk of suicide attempts in young adults: A longitudinal population-based study. Psychol Med 2014;44:1235e43.

[54] Daniel SS, Walsh AK, Goldston DB, et al. Suicidality, school dropout, and reading problems among adolescents. J Learn Disabil 2006;39:507e14.

[55] Kroll-Desrosiers AR, Crawford SL, Moore Simas TA, et al. Improving preg- nancy outcomes through maternity care coordination: A systematic re- view. Womens Health Issues 2016;26:87e99.

[56] Stormshak EA, Connell AM, Véronneau MH, et al. An ecological approach to promoting early adolescent mental health and social adaptation: Family-centered intervention in public middle schools. Child Dev 2011;82: 209e25.