Concussion Exposure and Suicidal Ideation, Planning, and Attempts Among US High School Students

Data from the 2017 and 2019 National YRBSS were combined and cross-sectionally analyzed. A detailed description of the YRBSS sampling methods can be found elsewhere.¹⁹  The YRBSS administers a school-based survey biennially with the goal of monitoring the prevalence of priority health risk behaviors among youth. A 3-stage cluster-sampling design was used to produce a nationally representative sample of public and private school students in grades 9 through 12. Weight factors were applied to each respondent record to adjust for nonresponse and oversampling of certain grades, sex, and racial and ethnic demographics. Data from the YRBSS provide valid measures of health risk behaviors among this cohort.²⁰  Parental permission was obtained at each collection site, and survey participation was voluntary. Respondents provided their information anonymously on computer-scannable answer sheets. An institutional review board at the Centers for Disease Control and Prevention approved all procedures.

A detailed description of questionnaire items, including prompts, response options, and statistical coding, can be found in the Supplemental Table (available online at http://dx.doi.org/10.4085/1062-6050-0117.22.S1). Concussion exposure is a new variable included in the national YRBSS and was consistent across the 2017 and 2019 survey years. The relevant question in the survey is specific to the respondent’s participation in sport or recreational activities. Concussion exposure was the independent variable of interest in our study. To investigate the role of concussion frequency in suicidality, we categorized responses as no concussion history (0), a single concussion (1), and multiple concussions (≥2) in the past 12 months.

The outcome variables for this investigation were 5 questions from the YRBSS regarding mental health and suicidal behaviors and were consistent across the 2017 and 2019 survey years. Self-reported indicators of mental health and suicidality were (1) sadness or hopelessness, (2) suicidal ideation, (3) suicidal planning, (4) suicide attempts, and (5) injurious suicide attempts in the past 12 months. Responses to all suicidality questions were dichotomized as 0 = no (reference) or 1 or more = yes.

Demographic factors (grade, sex, and race and ethnicity) were analyzed as covariates. Sex was dichotomized as male or female. Grade was categorized as 9th, 10th, 11th, or 12th. Race and ethnicity were categorized as Hispanic/Latino, non-Hispanic Black, non-Hispanic White, or other (which included American Indian or Alaska Native, Asian, Native Hawaiian or other Pacific Islander, non-Hispanic, and multiracial). Lifestyle factors known to negatively affect mental health (eg, bullying victimization, sexual orientation),¹⁶  as well as those known to positively affect mental health (eg, physical activity)^21,22  were controlled for in our analyses. Two survey questions about bullying were combined into a single covariate for analyses; students who responded yes to either of these questions were considered to have experienced bully victimization in the past 12 months. Current physical activity guidelines for children²²  were used to dichotomize student responses into those who did (yes) or did not (no; reference) report ≥5 days of physical activity a week for ≥60 minutes.

A total combined sample of 28 442 respondent questionnaires were available for analysis across the 2017 and 2019 survey years. Minimal data were missing for concussion history (9.69%), sadness or hopelessness (1.17%), suicidal ideation (1.59%), suicidal planning (1.68%), and other covariates (0.97%–4.78%); thus, we were left with a final sample of 23 445 students (82.43% of the total sample). However, considerable data were missing for attempted suicide and injurious attempted suicide (21.26%); therefore, a final sample of 17 397 students was analyzed using these outcome measures (61.16% of the total sample). Substantial missing data for these 2 variables presented the potential for selection bias.¹⁹  As such, we conducted attrition analyses using χ² tests and post hoc examination of φ coefficients to observe the degree to which participants differed between the 2017 and 2019 samples across the study variables.

All hypothesis testing was conducted using SAS (version 9.4; SAS Institute). Descriptive statistics were calculated and provided in frequency tables. Inferential statistics consisted of χ² tests to examine bivariate associations between categorical outcomes and predictors. The GENMOD procedure in SAS was applied to evaluate the association between concussion history and (1) sadness or hopelessness, (2) suicidal ideation, (3) suicidal planning, (4) suicidal attempts, and (5) injurious suicidal attempts. In addition, adjusted odds ratios (AORs) with 95% CIs were computed. Weighted generalized linear models, including logistic and probit models, were used to examine binary response data (yes versus no) for each of the 5 associations of concussion history and suicidality. Additional generalized linear models, including log-linear and Poisson regression models, were performed to determine the association between concussion frequency (3 levels: no SRRC history [0 times; reference], 1 SRRC [1 time], and multiple SRRCs [≥2 times]) and suicidality. All statistical models were adjusted for the covariates of age, sex, race and ethnicity, bully victimization, sexual orientation, and physical activity. For each model, an interaction term between concussion exposure and sex was included to explore possible differences between males and females. An a priori α level of .05 was set for all analyses, and Bonferroni correction was applied to adjust for multiple comparisons (P ≤ .005).

Attrition analyses revealed statistical differences across the various study variables between the 2017 and 2019 survey years. However, post hoc examination of φ coefficients indicated no substantial potential for selection bias, as the strengths of associations were considered small.²³  Only race and ethnicity was substantive (φ = 0.164); however, this was still considered a small effect. Thus, rather than sample bias, this suggests that the observed differences may have been due to the large sample size.²³ 

Among respondents, 15.0% reported ≥1 SRRC during the past 12 months, with 5.6% reporting ≥2 SRRCs during the past 12 months. A greater proportion of male students than female students reported ≥1 SRRC during the past 12 months (16.7% versus 13.2%, respectively; Cramer V = 0.048). See Table 1 for complete respondent demographics stratified by SRRC frequency.

A greater proportion of female students than male students reported feeling sad or hopeless (44.3% versus 23.6%, respectively; Cramer V = 0.218), suicidal ideation (23.6% versus 12.5%, respectively; Cramer V = 0.145), suicidal planning (18.8% versus 10.3%, respectively; Cramer V = 0.120), attempting suicide (10.1% versus 5.4%; Cramer V = 0.089), or being injured from an attempted suicide (3.1% versus 1.4%, respectively; Cramer V = 0.056) in the past 12 months. Table 2 shows the demographic characteristics among participants responding yes to depression and suicidality variables.

Similar to previous investigators^13–15  who used the YRBSS, we found significant associations between concussion history and the odds of reporting feelings of depression and suicidal ideation and behaviors (Tables 3 and 4). Those who reported ≥2 SRRCs in the past 12 months were not at significantly greater odds of reporting feelings of depression, suicidal ideation, or suicidal planning than those who reported a single SRRC. However, students who reported ≥2 SRRCs in the past 12 months were at significantly greater odds of reporting suicidal attempts (AOR = 2.03; 95% CI = 1.43, 2.88).

The association between SRRC frequency and suicidal behavior in the past 12 months differed significantly by sex. Among female students, those who reported a single SRRC during the past 12 months were at significantly greater odds of reporting feelings of sadness or hopelessness (AOR = 1.30; 95% CI = 1.15, 1.49), suicidal ideation (AOR = 1.54; 95% CI = 1.32, 1.78), suicidal planning (AOR = 1.46; 95% CI = 1.25, 1.71), attempting suicide (AOR = 1.54; 95% CI = 1.25, 1.89), or being injured from an attempted suicide (AOR = 1.90; 95% CI = 1.38, 2.60) than those who did not report an SRRC, but the strengths of these associations did not significantly increase with multiple (≥2) SRRCs (Tables 3 and 4). Among male students, the odds of reporting suicidal behaviors increased significantly with a single concussion for feeling sad or hopeless (AOR = 1.27; 95% CI = 1.10, 1.47), suicide attempts (AOR = 1.67; 95% CI = 1.26, 2.21), or injurious suicide attempts (AOR = 2.79; 95% CI = 1.74, 4.47; Tables 3 and 4). Furthermore, males who reported ≥2 SRRCs during the past 12 months were at significantly greater odds of reporting attempting suicide (AOR = 1.99; 95% CI = 1.39, 2.83) than males who reported a single SRRC event during the past 12 months (Table 4).

Our findings contribute to a growing body of literature regarding the association between concussion and suicidal behaviors among adolescents. To our knowledge, we are the first to examine the relationship between concussion frequency and suicidality in a nationally representative sample of US high school students. These results suggest that high school students who experienced a recent SRRC were at greater odds of reporting depressive symptoms and suicidal ideation, planning, and attempts when compared with students who did not experience a recent SRRC. Furthermore, the odds of reporting such risk behaviors (specifically, suicide attempts) may be compounded by increased SRRC exposure. Also, the association between concussion frequency and suicidality may differ between males and females.

Prior YRBSS investigators^14,15,24,25  have demonstrated that individuals experiencing ≥1 concussion in the last 12 months were at greater odds of reporting increased feelings of sadness or hopelessness (AORs = 1.48–1.87), as well as suicidal ideation (AORs = 1.26–1.95), planning (AORs = 1.27–1.97), and attempts (AORs = 1.33–3.10). Importantly, our study builds upon this initial research by exploring the possible association between adolescent concussion and suicidality, stratified by the number of self-reported SRRCs during the preceding 12 months. The odds of reporting suicidality may increase along with the number of concussive injuries, suggesting a potentially compounding relationship between concussion and adolescent suicidality within a 12-month period. Specifically, high school students who reported ≥2 concussive injuries exhibited 2-fold greater odds of reporting attempted suicide when compared with adolescents who reported a single concussive event in the preceding 12 months.

Our findings corroborate those of investigators²⁶  who demonstrated that a history of multiple concussions and incomplete physiological recovery before subsequent injury was associated with longer and more complicated recovery profiles among adults. Furthermore, repeated head trauma in athletes and military service members has been linked to chronic and debilitating alterations in mental and psychological health.^27,28  Repeated concussive events may lead to an exaggerated neuroinflammatory response, thereby impairing recovery and disrupting neurologic function.²⁹  Increased neuroinflammation is also associated with severe psychiatric conditions such as major depressive disorder,³⁰  suicidality,³¹  and posttraumatic stress disorder.²⁷ 

A secondary aim of our work was to assess differences in suicidality among male and female adolescents with a history of SRRC. Consistent with prior authors,³²  we observed that a greater proportion of females than males reported feelings of sadness or hopelessness (44.3% versus 23.6%, respectively), as well as suicidal ideation (23.6% versus 12.5%, respectively), planning (18.8% versus 10.3%, respectively), and attempts (10.1% versus 5.4%, respectively) during the preceding 12 months, regardless of concussion status. We failed to detect any sex differences when factoring in concussion status (ie, no history of SRRC versus a history of SRRC). However, when factoring in concussion frequency, unique sex-based reporting differences emerged. Specifically, increased odds of reporting suicidal behaviors were similar across females who reported 1 or multiple concussions when compared with females who reported no concussion history during the preceding 12 months. In contrast, although the odds of reporting suicidal behaviors also increased significantly among males who reported a single concussive event, these findings were seen primarily for active (attempts, injurious attempts) versus passive (ideation, planning) behaviors. More concerning was that males who reported multiple concussions were at 2-fold greater odds of reporting suicidal attempts than males reporting a single SRRC in the preceding 12 months.

Together, these results suggest that males may engage with suicidal behaviors in a more impulsive manner. Though these outcomes may reflect trends in the broader adolescent risk-behavior literature (ie, females were more likely to report suicidal behaviors),³³  this evidence likely does not fully explain the sex differences we demonstrated. Current research³⁴  on concussion and adolescent mental health indicates that females may be more prone to disturbances in psychological well-being after concussion when compared with males. In addition, other investigators³⁵  who used more objective psychophysiological measures (eg, electroencephalogram) identified blunted emotional processing in males, particularly in the context of multiple concussions. Mechanisms for sex-based differences remain largely unknown, yet our study provides the first evidence for potential sex differences in the effects of multiple concussions on short-term adolescent suicidal behavior.

Removal from everyday activities and the need for special accommodations may lead to an increased social burden and feelings of isolation after injury.^36,37  Additionally, persisting neurologic imbalances may hamper effective emotional regulation and coping strategies.³⁸  If left unattended, these factors may predispose individuals to severe psychiatric conditions (ie, suicidal thoughts and behaviors) and may be further exacerbated with repeated exposures. This increases the clinical burden and further complicates the management of patients with concussive brain injuries. Notably, these findings further support the need for comprehensive multidimensional evaluation protocols, including follow-up assessments of psycho-affective health incorporating domains of emotional regulation.

Though our results strengthen the extant knowledge regarding the association between concussion and adolescent suicidality, several limitations should be considered. Foremost, because of the YRBSS’s cross-sectional nature, we were unable to establish the temporal sequence of concussive events and adolescent risk behaviors during the 12 months of each survey. This limitation prevented us from inferring a causal relation between concussion and adolescent suicidality. Second, because of the inherent nature of the YRBSS database, we were able to investigate only nonfatal suicide behaviors. Therefore, these results cannot be generalized to include the overall suicide risk. Additionally, despite the emphasis on anonymity of respondents, YRBSS data are self-reported; thus, it is not possible to account for underreporting or overreporting of concussive events and risk behaviors. For example, students may exhibit a tendency to answer in a socially desirable manner (ie, they may report fewer concussive events or suicidal behaviors than experienced). Third, the YRBSS sampling method accounts only for youth who attend school; hence, our study did not reflect all youth in this age group (ie, sample bias). Furthermore, though a strength of our work was the evaluation of several mental health measures, we lacked access to clinical diagnoses and validated psychological inventories, increasing the potential for response bias. Lastly, whether the sex-based differences we observed were due to sociocultural or other biomechanical factors remains unknown. Future longitudinal examinations are necessary to better understand the possible cumulative effects of concussion on adolescent mental health and promote the development of earlier prevention strategies aimed at mitigating the link between concussive injury and potentially fatal risk behaviors. As such, schools would benefit from implementing regular mental health assessments and interventions to help students avoid feelings of depression and suicidality after concussion.

We found that adolescents who reported concussion were at increased odds of reporting feelings of depression as well as suicidal ideation, planning, and attempts during the preceding 12 months of the YRBSS survey. More importantly, a larger number of concussions incurred in the previous 12 months was associated with greater odds of suicidal attempts and may differentially affect male and female adolescents. Overall, these results highlight the importance and exigency of clinical evaluation and close monitoring of mental health in youth, particularly those who sustain multiple concussions within a short period of time.

The findings and conclusions presented in this manuscript are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Supplemental Table.

Available online at http://dx.doi.org/10.4085/1062-6050-0117.22.S1