Trends of Suicidal Behaviors Among High School Students in the United States: 1991–2017

The Centers for Disease Control and Prevention (CDC) developed the Youth Risk Behavior Survey (YRBS) to monitor priority health-risk behaviors among youth in the United States.¹⁶  The national school-based YRBS is a cross-sectional survey conducted every 2 years since 1991. Each survey cycle uses a 3-stage cluster sampling design to produce a nationally representative sample of ninth- through 12th-grade students attending public and private schools in the 50 states and the District of Columbia. CDC investigators applied a weighting factor to adjust for nonresponse and the oversampling of black and Hispanic or Latino students. The Institutional Review Board at the CDC approved the protocol of national YRBS.¹⁶ 

Student participation in the survey is anonymous and voluntary through a self-administered computer-scannable questionnaire. A parent or legal guardian provides permission for youth to complete the survey. During 1991–2017, national YRBS sample sizes ranged from 10 904 to 16 410. School response rates ranged from 75% to 81%; student response rates ranged from 81% to 90%, and overall response rates for the surveys ranged from 60% to 71%. More details regarding sampling strategies and the psychometric properties of the YRBS questionnaire are reported elsewhere.¹⁷  The analytic sample in this study consisted of 198 540 adolescents (without missing values on sex and race and ethnicity) from the combined national YRBS data set from the years 1991 through 2017.

The YRBS asked 2 self-report questions about race and Hispanic heritage. Consistent with previous research,¹⁸  the 6 categories for race and ethnicity were non-Hispanic white, non-Hispanic black, non-Hispanic Asian or PI, Hispanic, American Indian/Alaska Native (AI/AN), and non-Hispanic multiple races. Native Hawaiian or other Pacific Islander is collapsed into Asian American or PI.

Four questions assessed suicidal thoughts and behaviors: (1) “During the past 12 months, did you ever seriously consider attempting suicide?” (suicide ideation), (2) “During the past 12 months, did you make a plan about how you would attempt suicide?” (suicide plan), (3) “During the past 12 months, how many times did you actually attempt suicide?” (suicide attempts; any response ≥1 was considered a “yes” response), and (4) “Did any attempt result in an injury, poisoning, or overdose that had to be treated by a doctor or nurse?” (injury by attempt). Response options were dichotomized into “yes” or “no.” These questions demonstrated substantial reliability in previous studies.¹⁹ 

Logistic regression analyses were used to assess trends by different sex and racial and ethnic groups, with each of the 4 suicidal behavior questions serving as dichotomous outcomes. Survey year was treated as a continuous variable and was used to examine the linear trend. Quadratic terms over and above the standardized linear terms of survey year were used to examine the quadratic trend of responses. Secular trends in the prevalence of suicidal ideation, suicide plan, suicide attempts, and injury by attempt were examined by using multivariate logistic regression models that included linear time variables and then both linear and nonlinear (quadratic) time variables while controlling for the grade levels. Student grade levels were controlled to limit the effects of age. Trends that include significant quadratic and linear components demonstrated nonlinear variation in addition to an overall increase or decrease over time. Regression odds ratios (ORs), as well as 95% confidence intervals (CIs), were reported. Statistical significance was set to P < .05 with Wald χ² tests using design-adjusted coefficient variance-covariance matrices. Statistical significance was set at 2-sided α = .05. Trends analyses were conducted by using “SURVERYLOGISTIC” process in SAS software version 9.4 (SAS Institute, Inc, Cary, NC) to adjust for the complex sampling design. The national YRBS data are weighted to the sample size. Thus, there is no need for additional adjustments to the weights.²⁰  Both linear and nonlinear (quadratic) time variables were estimated while controlling for grade levels to explore nonlinear trends of suicidal behaviors across time, which was also suggested by previous literature.²¹  Linear trend is depicted with a straight line, whereas a quadratic trend is depicted with a curve with 1 bend. A significant quadratic trend indicates a nonlinear change in the data over time. Trends with both significant quadratic and linear components reveal nonlinear variation in addition to an overall increase or decrease over time. According to the missing data analysis, missing patterns were assumed as not missing at random. Hence, missing values were handled by using the Taylor series variance estimation with the “NOMCAR” option in the “SURVERYLOGISTIC” process.

The sample sizes by sex and race and ethnicity are shown in Table 1. Of the 198 540 adolescents in the sample, there were slightly more boys than girls (51% boys; 49% girls), and the average age for both groups was ∼16 years (male mean age: 16.11 years; female mean age: 16.02 years). Figure 1 portrays changes in the distribution of suicidal behaviors by survey year and race and ethnicity. From 1991 to 2017, the weighted overall prevalence rates of suicidal ideation, plan, attempt, and injury by attempt were 18.8%, 14.7%, 7.9%, and 2.5%, respectively. ORs for the linear trends of suicidal thoughts and behaviors by sex and race and ethnicity groups are shown in Tables 2 and 3, respectively. Across all sex and racial and ethnic groups (except non-Hispanic multiple-race group), significant linear decreases were present in self-reported suicidal ideation and suicide plan. Among AI/AN adolescents, there was a noticeable increase in suicidal behaviors during 1995 and 1997.

For suicide attempts, however, there was a statistically significant annual decline of suicide attempts among adolescent girls (OR, 0.98; 95% CI, 0.98–0.99; P < .001), whereas no significant trend was observed for boys (see Table 2). Results for trends of suicide attempts across race and ethnicity indicated adolescents who identified as white, Hispanic, Asian American or PI, and AI/AN had a significant linear decrease; however, black adolescents had a significantly linear increase in self-reported suicide attempts (OR, 1.02; 95% CI, 1.01–1.04; P < .001). Linear trends for injury by attempt were nonsignificant among adolescents from both sex groups and all races and ethnicities. There were no significant trends across any suicidal outcomes among individuals in the non-Hispanic multiple-race group.

Subgroup differences were observed after stratifying by sex and race and ethnicity (Tables 4, 5, and 6). There were significant linear decreases in suicidal ideation and suicide plan across the years among all adolescent subgroups, except for nonsignificant trends found for black girls and AI/AN boys for suicide plan only, as well as AI/AN girls, non-Hispanic multiple-race boys and girls for suicidal ideation and plan. Additionally, there were significant quadratic trends in suicidal ideation and suicide plan among adolescents who were white girls and boys, black girls and boys, Hispanic girls and boys, and Asian American girls and boys (suicidal ideation only), suggesting a nonlinear relationship with a downward but concave-up trend over time among these groups (P < .001).

When examining trends for self-reported suicide attempts, girls that classified as white, Hispanic, and Asian American or PI had a significant linear decrease (see Table 4). Also, marginally significant quadratic trends in suicide attempts were observed among white girls (OR, 1.00; 95% CI, 1.00–1.01; P = .049) and Hispanic girls (OR, 1.00; 95% CI, 0.99–1.00; P = .048), suggesting accelerations in the decreasing rates of attempting suicide over time among these 2 groups. Boys in these racial and ethnic groups, together with AI/AN and non-Hispanic multiple-race boys, had nonsignificant changes in suicide attempts (see Table 5).

However, among black adolescents, there were significant and positive linear trends found for suicide attempts among both girls (OR, 1.02; 95% CI, 1.01–1.03; P = .003) and boys (OR, 1.04; 95% CI, 1.02–1.07; P < .001). The quadratic trend examining the rate of suicide attempts among black girls was also significant (Table 4), indicating an acceleration in the increasing rate of self-reported suicide attempts in this group (OR, 1.01; 95% CI, 1.00–1.01; P = .01). For injury by attempt (Table 5), there was a significant linear increase for black boys (OR, 1.04; 95% CI, 1.00–1.08; P = .048).

Nationally representative data from 1991 to 2017 reflect findings of concern regarding US high schoolers’ engagement in suicidal behaviors. Our findings reveal that over that span of time, almost 1 in 5 adolescents are thinking about suicide (18.8%), and >1 in 10 has a suicide plan (14.7%). Despite increased attention given to the creation of campaigns to reduce youth suicide in the United States over the last decade,^22–24  these findings suggest that continued concern and attention regarding suicidal behaviors among high school–aged youth is warranted, although more recent data and attention have been given to younger US youth.^6,8 

Our findings also indicate 2 distinct trends regarding suicidal thoughts and behaviors. First, the rates for suicidal ideation and suicide plans have been trending downward across all sex and racial and ethnic groups. This is an important finding because suicidal ideation and suicide plan are key risk factors to engagement in suicidal behaviors (eg, suicide attempt).²⁵  The findings, however, are not as heartening regarding suicide attempts and injury by attempt, particularly for specific sex and racial and ethnic groups. Black adolescents experienced an increase in rates for suicide attempts, and a significant increase was reported for black adolescent boys concerning the rates of injury caused by suicide attempts. Adolescent girls from all racial and ethnic groups experienced a significant downward trend in suicide attempts (white, Hispanic, and Asian American or PI); however, for black adolescent girls, there was a significant increase.

Findings regarding the rising rates of suicide attempts in black youth may be related to the documented disparities in mental health treatment^26–28  and common social etiologic factors disproportionately experienced by black youth.^29–31  Authors of epidemiological studies, since 1990, affirm lower mental health treatment among black adolescents compared with their white counterparts.^26,32–34  Lower rates of treatment could be attributed to black adolescents not wanting to be judged as weak and refusing to acknowledge mental illness symptoms.^35,36  Another explanation could be that black adolescents view mental health treatment with skepticism because of mistrust of providers causing them to prefer to address mental health symptoms with family and peer networks.³⁷  This could lead to unmanaged and unmonitored mental health concerns that escalate to self-harm behaviors. Finally, social etiologic factors disproportionately experienced by black adolescents include racial discrimination,^30,31  adverse childhood experiences (eg, abuse and neglect),²⁹  and poverty,^38,39  all of which have been positively associated with suicidal behaviors in this group.

The significant increase found for black boys for injury by attempt could be explained by the documented change of suicide method in adolescents.^40–43  For example, during this study period, rates of suicide by firearms decreased from 60% to 22%, whereas rates of suicide by hanging and/or suffocation increased from 20% to 60%.⁴⁴  Unfortunately, hanging and/or suffocation is the leading method of suicide in youth.^45–47  It is highly lethal, and items related to hanging and/or suffocation are more accessible than a firearm.^47,48  Our findings regarding injury by attempt suggest black boys may be engaging in more lethal methods of suicide, which ultimately lead to the need for medical care.

AI/AN adolescents saw a notable spike in suicidal behaviors between the years 1995 and 1997, which might be associated with an uptick in substance use rates observed among AI/AN adolescents during that same period.⁴⁹  Namely, between 1995 and 1997, AI/AN adolescents saw increases in marijuana, methamphetamine, inhalant, cocaine, and cigarette use.⁵⁰  Elsewhere, substance use has been strongly associated with increased rates of suicidal behaviors among AI/AN youth during that period,^49,51,52  suggesting a strong possible connection between the increase in substance use and rising suicidal behavior rates for AI/AN adolescents.

Further research is necessary to understand the rise in suicide among older black adolescents. It is critical to develop and test the efficacy of treatment or prevention interventions on a larger scale that have been shown to reduce suicide risk among older black adolescents.¹⁵  Preventive interventions within the black community should invest in key protective factors that help reduce stigma related to mental illness and increase help-seeking behaviors.^53,54  Research specifically designed to examine treatments in clinical and nonclinical settings that ameliorate the symptoms or signs of suicidal behaviors among black adolescents should be considered. Primary care settings, where well visits are most likely to occur for adolescents,^41–43  can play a critical role in screening efforts to identify black adolescents who might be at risk for engagement in suicidal behaviors.⁴⁴ 

This study has several limitations. One important limitation is that the YRBS data are cross-sectional. Therefore, we cannot determine the causality and directionality of the associations found. Second, the results only applied to youth who attend school based on the YRBS sampling design. Thus, they are not representative of all adolescents in this age group. Nationwide, ∼5.8% of individuals drop out of high school, among which Hispanic (9.1%) and black (7.0%) students have higher rates than their white (4.5%) and Asian American (2.0%) counterparts.⁵⁵  Third, the measurements of suicidal thoughts and behaviors are self-reported. Despite the survey questions demonstrating good test-retest reliability,¹⁹  there is potential for some level of under- or overreporting bias.

When examining youth suicidal behavior, white youth have historically had higher rates of suicide attempts and deaths by suicide compared with their black counterparts; however, this study provides some evidence to the contrary. The results suggest that, over time, black youth have experienced an increase in suicide attempts, which is the most prominent risk factor associated with suicide deaths, and for black boys, a significant increase in injury by attempt, which suggests black boys may be engaging in more lethal means when attempting suicide. Examining trends in suicidal ideation and behaviors over time by sex and race and ethnicity allows us to determine where to focus prevention and intervention efforts. Future research should be used to examine the underlying reasons for these observed changes.

AI/AN

American Indian and Alaskan native

CDC

Centers for Disease Control and Prevention

CI

confidence interval

OR

oddsratio

PI

Pacific Islander

YRBS

Youth Risk Behavior Survey