Mental Health Service Use Among Adults With Suicide Ideation, Plans, or Attempts: Results From a National Survey

In 2013, over 40,000 people in the United States died by suicide (1), and it is estimated that an additional one million adults make a nonfatal attempt each year (2). Suicide ideation is even more prevalent, with approximately nine million U.S. adults experiencing serious thoughts of suicide annually (2). Because efficacious interventions exist to reduce suicide risk, connecting individuals at elevated suicide risk with care has been identified as a critical suicide prevention strategy (3–6). Nonetheless, successful engagement of this population into mental health treatment has remained a challenge (7,8).

Studies have revealed low rates of service utilization by adolescents and adults at risk of suicide. Results from a national study indicated that only 32.7% of U.S. adolescents with past-year suicide ideation and 43.1% with a past-year suicide attempt reported contact with mental health services in the previous year (9). Similarly, studies have repeatedly shown that less than half of adults with suicide ideation report utilizing mental health services (10–14). These low rates of service use among high-risk individuals are particularly concerning in light of data revealing that only 31% of suicide decedents were in mental health treatment at the time of their death (15). Thus, even though there are efficacious interventions, individuals at risk of suicide often do not connect with these services, representing a lost opportunity to prevent a tragic outcome.

This problem of low treatment engagement has likely persisted, in part, because of a dearth of research on mental health service use among suicidal individuals across the life span. Although several researchers have examined correlates of service use among suicidal adolescents (16), there are considerably fewer data on adults. Delineating characteristics that differentiate service users from nonusers has the potential to inform the development of public health campaigns by highlighting key modifiable barriers to service use. Therefore, a critical first step toward increasing treatment engagement in this high-risk population is to identify the factors associated with mental health service use among adults with suicide ideation, plans, or attempts. By advancing the science of treatment engagement, progress can be made in realizing a decrease in suicide rates (7).

Because there are likely a number of distinct factors that converge to determine whether a suicidal individual utilizes mental health services, it is helpful to organize these factors within the framework of a conceptual model. For this study, the behavioral model of health service use was used to inform the selection of potential correlates of service use (17). This model posits that factors that affect service use fall under one of three primary domains: predisposing factors (for example, age and gender), enabling factors (for example, income and health insurance), and need factors (for example, psychological distress).

Using this model as a guide, this study’s primary aim was to determine factors associated with mental health service use among adults at elevated risk of suicide by using data from a nationally representative survey. This study extends previous work (10,18,19) by examining service use among U.S. adults with experiences along the continuum of suicidality (that is, ideation, plans, and attempts) and by delineating theoretically informed sociodemographic and clinical characteristics that differentiate service users from nonusers.

Methods

Participants and Procedures

The National Survey on Drug Use and Health (NSDUH) is an annual U.S. survey conducted by the Substance Abuse and Mental Health Services Administration (SAMHSA) to characterize substance use and mental health at both the national and the state levels. Data from the most recent survey (2013) were used for the analyses reported here (20). Employing a multistage area probability design, the public use file sample includes 55,160 noninstitutionalized U.S. civilians age 12 and over living in the 50 states and the District of Columbia. Youths and young adults were oversampled, and active duty military personnel and individuals living in an institutional setting (for example, prisons, nursing homes, and hospitals) were excluded. For these analyses, only adults (age ≥18) reporting past-year suicide ideation (N=2,126), plans (N=690), or attempts (N=345) were included. For comparison, findings from individuals not reporting past-year suicidality are also presented (N=35,106).

Information regarding survey methodology has been reported elsewhere (20). In brief, trained researchers from the Research Triangle Institute (RTI) conducted in-person interviews by using computer and audio technologies. Participants were compensated $30. The RTI Institutional Review Board (IRB) approved the data collection procedures, and the Florida State University IRB approved the secondary analyses presented here.

Measures

Sociodemographic characteristics.

Demographic characteristics assessed included sex, race-ethnicity, age, insurance status, marital status, poverty status (<100% of the federal poverty level), and veteran status.

Suicide ideation, plans, and attempts.

Respondents were categorized as having past-year suicide ideation if they answered “yes” to the following question: “At any time in the past 12 months, that is from [date] up to and including today, did you seriously think about trying to kill yourself?” Only participants reporting past-year suicide ideation were asked follow-up questions about whether they had made any past-year plans or attempts. Respondents who denied past-year suicide ideation were categorized as having no past-year suicidality.

Serious psychological distress.

As an additional proxy of mental illness, serious psychological distress was evaluated via the Kessler-6 (K6) (21), with the past year used as the time frame of interest. The K6 consists of six items scored on a 5-point Likert scale (0, none of the time, to 4, all of the time). Total scores range from 0 to 24, with higher scores indicating greater severity of distress. Consistent with recommendations (21,22), individuals with K6 composite scores ≥13 were categorized as having experienced past-year serious psychological distress. The scale has established reliability (α=.89) and validity (for example, predicting DSM diagnoses) (22,23).

General health.

Self-perceived general health was assessed by using an item from the CDC’s Health-Related Quality of Life scale: “Would you say your health in general is excellent, very good, good, fair, or poor?” For logistic regression analyses, responses of “excellent,” “very good,” and “good” were coded as good self-perceived general health and “fair” and “poor” were coded as poor self-perceived general health. This single item is a parsimonious and robust predictor of functional decline and mortality (24,25).

Depression.

A past-year probable diagnosis of a major depressive episode was established by triangulating question responses from the structured interview that map onto DSM-IV criteria (26). This approach has moderate reliability (κ=.52) (27).

Mental health service use.

Participants were asked about mental health service use for problems related to “emotions, nerves or mental health,” excluding treatment specific to alcohol or drug use. We categorized service use into three groups: inpatient, outpatient, and psychiatric prescription medication.

Data Analytic Strategy

All analyses employed weights to account for the complex survey design and ensure that findings are representative of the target population. Because of confidentiality concerns, the public use data released by SAMHSA do not include all cases captured by the original survey; thus minor variations in population-based estimates between SAMHSA-released reports and the results reported here were expected. Descriptive statistics were used to describe the population’s demographic characteristics, clinical characteristics, and service utilization rates. Logistic regression modeling was utilized to examine factors associated with service nonuse among four groups: those with past-year suicide ideation, suicide plans, or suicide attempts and those with no suicidality. Data were analyzed with Stata SE, version 12.0.

Results

Data on characteristics of suicidal individuals are presented in Table 1, stratified by those reporting past-year suicide ideation, plans, or attempts. Logistic regression analyses examining correlates of service use for individuals with past-year suicide ideation, plans, and attempts are presented in Tables 2, 3, and 4, respectively. For comparison, correlates of service use among individuals not reporting past-year suicidality are shown in Table 5.

Past-Year Suicide Ideation

Weighted analyses indicated the following demographic characteristics of adults reporting past-year ideation (Table 1): age 50 or older, 30.8%; female, 53.1%; non-Hispanic white, 69.1%; veteran, 7.3%; married, 33.9%; living in poverty, 23.7%; and covered by health insurance, 77.0%. In terms of mental and general health, 60.0% reported past-year serious psychological distress, 24.1% reported fair or poor past-year general health, and 47.9% reported a past-year major depressive episode. Regarding service use, 52.4% reported past-year contact with any type of mental health services (inpatient stay, 7.4%; outpatient treatment, 32.8%; and psychiatric prescription medication, 45.7%). Past-year service use was more common among females (odds ratio [OR]=1.60), non-Hispanic whites (OR=2.11), individuals with health insurance (OR=1.83), and those with past-year concomitant clinical problems, including serious psychological distress (OR=2.22), poor or fair general health (OR=1.90), and a major depressive episode (OR=3.14) (Table 2).

Past-Year Suicide Plan

Demographic characteristics of adults reporting a past-year suicide plan were as follows (Table 1): age 50 or older, 19.4%; female, 57.6%; non-Hispanic white, 69.5%; veteran, 7.3%; married, 25.2%; living in poverty, 31.7%; and covered by health insurance, 69.8%. In terms of mental and general health, 77.6% reported past-year serious psychological distress, 30.1% reported fair or poor past-year general health, and 60.1% reported a past-year major depressive episode. For service use, 58.4% reported past-year contact with any type of mental health services (inpatient stay, 17.8%; outpatient treatment, 44.5%; and psychiatric prescription medication, 52.1%). Those more likely to report past-year service use were non-Hispanic whites (OR=2.39) and individuals with a more severe clinical picture, including past-year serious psychological distress (OR=4.07), poor or fair general health (OR=1.72), and a major depressive episode (OR=4.97) (Table 3).

Past-Year Suicide Attempt

Demographic characteristics of adults reporting a past-year suicide attempt were as follows (Table 1): age 50 or older, 20.1%; female, 60.0%; non-Hispanic white, 57.8%; veteran, 5.9%; married, 21.3%; living in poverty, 43.2%; and covered by health insurance, 66.1%. With regard to indices of mental and general health, 73.7% reported past-year serious psychological distress, 31.8% reported fair or poor past-year general health, and 50.2% reported a past-year major depressive episode. For service use, 53.9% reported past-year contact with any type of mental health services (inpatient stay, 26.3%; outpatient treatment, 43.1%; and psychiatric prescription medication, 50.7%). Those more likely to report past-year service use were non-Hispanic whites (OR=2.95) individuals with past-year serious psychological distress (OR=3.35) and individuals with a past-year major depressive episode (OR=5.77) (Table 4).

Differences Between Suicidal and Nonsuicidal Individuals

For adults reporting no past-year suicidality, all variables examined were significantly associated with service use (Table 5). Comparison of these findings with those for individuals reporting suicidality resulted in several noteworthy findings. First, in the nonsuicidal group, veterans were less likely than nonveterans (OR=.75) to have utilized mental health services; however, no significant differences were found between veterans and nonveterans reporting ideation, plans, or attempts. This finding suggests that among veterans, clinical severity may facilitate treatment use. This is promising because interventions tested specifically among military personnel, including cognitive-behavioral therapy, have demonstrated efficacy for reducing suicide risk (28). Second, among nonsuicidal individuals, those who were married were less likely than those who were never married, divorced, separated, or widowed to report utilizing services, suggesting that marriage (social support) may buffer against use of professional services. Conversely, among the suicidal groups, no differences in service use emerged by marital status. Finally, whereas nonsuicidal individuals with health insurance were more likely to have received services, this association was not found for individuals with a plan or attempt. It is nonetheless notable that suicidal individuals in this study reported slightly higher rates of uninsured status than the general population (29), which speaks to the importance of adequate insurance coverage to ensure that suicidal individuals receive care.

Discussion

Findings indicate that suicidal adults in the United States reported low use of mental health care services. Patterns of service use were similar for individuals with past-year suicide ideation, plans, or attempts, with slightly more than 50% of individuals across these groups reporting past-year contact with any type of mental health services (inpatient, outpatient, or psychiatric prescription medication). These findings are consistent with studies of service use among adolescents (9,16) and older adults (12), as well as a previous investigation of adults that aggregated prevalence rates across countries (10). This low rate of service use underscores the need to increase treatment engagement among suicidal adults in the United States.

Factors that differentiated suicidal service users from nonusers were examined in terms of the domains outlined by the behavioral model of health service use. In terms of noteworthy predisposing factors, across groups, non-Hispanic whites were significantly more likely to report past-year service use. These findings align with previous reports of younger suicidal individuals (30,31) but are in contrast with findings from other studies that included individuals with psychiatric disorders from a range of age groups (32). It is possible that suicidal adults from minority groups may be less likely to engage in care because of distrust of providers or stigma concerns, which have been shown to be relevant in adolescent populations but have yet to be formally explored among adults (33). In terms of gender, differences emerged only for the suicide ideation group, with females being more likely than males to utilize services. This finding suggests that with increased clinical severity, gender-related factors may be less relevant in decisions to connect with services. This finding is notable, but it does not eliminate the need to engage males in treatment, because men, particularly older men, are more likely to use highly lethal means (for example, firearms) in a suicide attempt (34). Consequently, early intervention (that is, at or before the time of ideation) may be especially important in preventing the onset of a plan or attempt.

Of interest, within the domain of enabling factors, no differences in service use across groups emerged on the basis of poverty or marital or health insurance status, with one exception: those with health insurance in the ideation group were more likely to use services than those without insurance.

Finally, a number of correlates of service use were identified in the domain of need factors—a domain of particular interest because these characteristics tend to be more readily modifiable. Across groups, a clinical picture marked by greater severity was associated with elevated service use. Adults with concomitant serious psychological distress or a past-year major depressive episode were more likely to report any contact with mental health services in the past year. This confluence of suicide ideation and serious psychological distress or major depressive episode represents a particularly severe clinical picture. These findings are promising in that those with the highest risk of suicide appear to have been more likely to receive care. Nonetheless, it is important to view this within the context of past research demonstrating that severe suicide ideation can be associated with help negation and a refusal to engage in treatment (35,36). Unfortunately, because the severity of ideation was not assessed in this survey, this association cannot be further delineated. Worse general health was also associated with greater service use. It is possible that those with worse general health may be more likely to present to nonpsychiatric clinicians (for example, primary care settings and emergency departments), creating opportunities for referral and reinforcement of the importance of maintaining specialist psychiatric care.

Study limitations must be noted. First, the survey assessed past-year suicidality and past-year service use; however, temporality between these variables was not captured. This distinction is important because a previous study found that among adolescents treatment began before the onset of suicidal behaviors in more than half of cases (37). Second, we do not have data on the time frame, severity, or chronicity of individuals’ suicide ideation, plans, or attempts. Incorporation of a more nuanced, psychometrically sound assessment in future iterations may reveal differential patterns of service use for varying levels of clinical severity. Finally, the survey did not collect data on treatment adherence and attrition—arguably more meaningful outcomes than contact with services alone.

Moving forward, efforts must be made to develop and rigorously test programs targeted at increasing service use among suicidal adults. Given the correlates of service use identified by this study, effective approaches may involve improving access to care for those without insurance, implementing screening and referral procedures in primary care and emergency department settings, and crafting interventions that are culturally sensitive and acceptable to males. Nonetheless, further research is warranted to better understand how each of these correlates, along with other factors such as stigma, may serve as a facilitator or barrier to care (38). Prospective studies of treatment engagement, along with assessments designed to address the aforementioned limitations, may be especially useful. Further, researchers have suggested that motivational interviewing may be an effective intervention to engage suicidal individuals in treatment (7,39,40), and results from a recent pilot randomized controlled trial suggest that this approach may have utility with college students at elevated risk of suicide (41). Finally, although the focus of this article has been on professional mental health services, it is also noteworthy that some individuals may disclose thoughts of suicide to family or friends rather than to professionals (42). In this regard, nonprofessional gatekeepers may be critical in linking at-risk individuals to professional care (43). When suicidal individuals are connected to care, it is important to conduct safety planning, with an emphasis on means safety/restriction and frequent follow-up (44,45).

Conclusions

Linking suicidal individuals to effective mental health treatment is a considerable challenge, but if effectuated, can be lifesaving. To our knowledge, this is the first study describing the prevalence and correlates of mental health service use among a nationally representative sample of U.S. adults reporting suicide ideation, plans, or attempts. The similarity of these findings to those of previous studies of other demographic groups raises concerns. Approximately half of individuals reporting past-year suicide ideation, plans, or attempts reported not receiving mental health care services during that same time frame. Our findings also point to several subgroups of suicidal individuals for whom treatment engagement interventions may be particularly needed, such as black and Hispanic groups and men. This study represents an initial step in elucidating correlates of service use that may ultimately be leveraged by researchers, clinicians, and policy makers to create effective treatment engagement interventions.