There is a well-recognized association between traumatic brain injury (TBI) and suicidal thoughts and behaviors^{1, 2} however, few studies have provided quantifiable estimates to support the mechanistic hypotheses underpinning this relationship.³ Research has supported the idea that TBI is associated with a range of psychiatric disorders observed following TBI which may impact suicidal behaviors.^{4, 5}

Negative physical and psychiatric health outcomes, including suicidal ideation, attempts, and suicide, are a major public health concern, particularly among individuals who sustain a TBI. Systematic reviews report an increased suicide risk among people with a history of TBI.^6-8 Measuring definitive rates of suicidal ideation/attempts among those with a history of TBI is challenging due to variable follow-up periods post-TBI, reliance on retrospective data (e.g., existing medical records), and differing strategies for measuring outcomes. One study of 559 adults, who denied prior suicide attempts and sustained complicated mild to severe TBI, reported that 20% of the study population experienced suicidal ideation during their first year post-TBI,⁹ a rate at least double those of the general population (2.1% to 10%) based on a systematic review of epidemilogical studies.¹⁰ Prevalence of past year suicide attempts among a similar cohort receiving care within the TBI Model Systems, a program sponsored by the National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR), Administration for Community Living, U.S. Department of Health and Human Services, was approximately 1-2%,¹¹ while the prevalence was 0.6% of adults aged 18 years or older in the general population.¹² Although research has identified several risk factors and predictors of suicidal ideation/attempts among those with TBI, few studies have examined the mechanisms that underlie such associations.

TBI is also significantly associated with high rates of post-injury substance use,^13-17 psychiatric disorders,^18-20 post-traumatic stress disorder,²¹ and sleep disturbances;^22-24 each of these conditions may have a bidirectional relationship with suicidal ideation/attempts. Most studies examining predictors of suicidal ideation/attempts among those with TBI employ some type of multiple regression to identify variables that contribute independently to poor suicide-related outcomes. During the first year post-TBI, documented predictors of suicidal ideation include history of prior suicide attempt, history of bipolar disorder or manic depression, and having less than a high school education.⁹ Significant predictors of a suicide attempt post-TBI include post-injury substance use, hopelessness, and post-injury psychiatric history, or emotional distress.^{25, 26} Closer examination of the inter-relationships among these variables using mediation analyses have not been conducted and could provide greater illumination of the relationship between TBI and suicidal thoughts and behaviors by way of potential mediator variable(s) (e.g., substance use, psychiatric disorders/emotional distress, sleep disturbances). Findings from mediation analyses can identify modifiable factors that could have practical implications for public health and clinical practice.

We used a mediation approach to investigate indirect effects of conditions known to be associated with both TBI and suicidal ideation/attempt among patients with and without a diagnosed nonfatal TBI, within 1 year of the TBI. The purpose of testing for mediation was to identify potentially modifiable conditions that could improve long-term outcomes of TBI, particularly suicidal ideation/attempts. To our knowledge, this is the first study to examine the mediation effects of opioid use disorders (OUD), sleep disorder, psychiatric disorder, and substance use disorders other than OUD on the association between TBI and suicide.

The descriptive statistics of the two samples are reported in Table 1. In the Medicaid sample, among those who had a TBI and a suicidal ideation/attempt there was a higher percentage of females than males, those aged 10-17 years compared with other age groups, and who were white compared with other races. The majority (92.81%) of those with a TBI and suicidal ideation/attempt had a psychiatric disorder present within the study period. While only 14.75% had a documented opioid use disorder, 43.71% had presence of substance use disorders other than OUD; 16.77% had a sleep disorder. In the private health insurance sample, among those who had a TBI and a suicidal ideation/attempt there was a higher percentage of females than males and those aged 10-17 years compared with other age groups. Less than half of the private health insurance sample who had a TBI and suicidal ideation/attempt had opioid use disorder (6.12%), sleep disorder (1.65%), psychiatric disorder (15.98%), or substance use disorders other than OUD (41.19%).

In both the Medicaid and private health insurance samples, patients with a TBI had at least 2 times higher odds of suicidal ideation/attempt in the preceding 12 months. Males had lower odds than females of suicidal ideation/attempt (Table 2). In the Medicaid sample, those aged 55-64 years had lower odds than other age groups of suicidal ideation/attempt. The same is true among the private health insurance sample, although the odds ratios are much higher for those aged 10-17 and 18-24 years. Both the Medicaid and private health insurance samples showed that individuals with a CCI of 1 or higher had higher odds of a suicidal ideation/attempt than those without comorbidities.

Table 3 contains the estimates of the direct association of TBI on suicidal ideation/attempt while controlling for opioid use disorder, sleep disorder, psychiatric disorder, substance use disorders other than OUD, sex, age, race/ethnicity, plan type, BOE, inpatient/outpatient source, and CCI score. When controlling for the mediators, those with a TBI had 1.49 and 1.73 times higher odds to have a suicidal ideation/attempt in the Medicaid and private health insurance samples, respectively. In both samples those with sleep disorder, psychiatric disorder, and substance use disorders other than OUD had higher odds of suicidal ideation/attempts.

Table 4 presents the mediation analysis results. In the Medicaid sample psychiatric disorders partially mediated the association between TBI and suicide (indirect AOR = 1.16; CI = 1.15-1.17), as did substance use disorders other than OUD (AOR = 1.03; CI = 1.03-1.03). Psychiatric disorders mediated 22.40% of the total effect, while opioid use disorder mediated 7.47%. Similarly, in the private health insurance sample psychiatric disorders partially mediated the association between TBI and suicide (indirect AOR = 1.03; CI = 1.03-1.03), as did substance use disorders other than OUD (AOR = 1.01; CI = 1.01-1.01). Psychiatric disorders mediated 3.97% of the total effect and substance use disorders other than OUD mediated 2.08% of the total effect. Among those in the private health insurance sleep disorder mediated 1.03% of the total effect.

In this study of more than 600,000 Medicaid and privately insured patients over a 12-month period, those with a documented TBI were over 2 times more likely to have documented suicidal ideation/attempt compared with those without a TBI. The association between TBI and suicide was revealed to take place, in part, with mediators, with 22.2% of the association mediated by psychiatric disorders and 7.46% mediated by substance use disorders other than OUD in patients with Medicaid; and among those with private health insurance, 4.18% of the association was mediated by psychiatric disorders and 2.17% by substance use disorders other than OUD. After controlling for mediators, patients in both the Medicaid and private health insurance sample were at least 1.49-1.73 times more likely to have suicidal ideation/attempt when compared to patients without TBI, respectively. While results trended in the same direction for both Medicaid and privately insured patients, we observed a general increase in the percentage of mediated associations among patients with Medicaid insurance. Given differences in socioeconomic status between individuals with Medicaid and private insurance, this finding may reflect unique contributions of socioeconomic status to the relationship between TBI, suicidal ideation/attempt, and mediators. Further research integrating data sources with additional socioeconomic indicators would be helpful to better understand this potential relationship. Taken together, our results emphasize the important direct associations between TBI and suicidal ideation/attempt. While psychiatric disorders, substance use disorders other than OUD, and sleep disorders contribute indirect effects to this relationship (1.25–22.2%), direct effects from TBI had the largest impact on suicidal ideation/attempt, indicating that close monitoring of patients with TBI for suicidal ideation/attempt is important, even in the absence of psychiatric, substance use, or sleep comorbidities.

Previous studies have posited that post-TBI psychiatric disorders may act as mediators between TBI and suicide;³³ our study affirms that psychiatric disorders partially mediate the relationship between TBI and suicide. Our study echoes previous findings that depression,^{3, 34} PTSD,³⁵ and sleep quality³⁶ mediate associations between TBI and suicide. Fonda et al.’s study of TBI and attempted suicide among U.S. veterans of the Iraq and Afghanistan wars found that 83% of the association between attempted suicide and TBI was mediated through comorbid psychiatric conditions and 51.4% through substance use disorder.³⁵ In contrast, our study found that psychiatric disorders only mediated 22.2% and substance use disorders other than OUD mediated 7.46% of the TBI-suicide association in our Medicaid sample and 4.18% mediated by psychiatric disorders and 2.17% mediated by substance use disorders other than OUD in patients with private health insurance. This difference in mediation effect may be due to differences in study populations (e.g., general public versus veterans) and outcomes, as we included both suicidal ideation/attempts.

Our study findings have important implications for clinical practice. Identifying treatable conditions that increase risk for suicide among patients with a TBI—and quantifying their impact—provides actionable information for suicide prevention efforts. Treatment of opioid use disorders, sleep disorders, psychiatric disorders, and substance use disorders other than OUD among patients with TBI may minimize risk of suicidal ideation/attempt. Other suicide prevention strategies (e.g., reducing access to lethal means, ensuring coverage of psychiatric health conditions by health insurance policies, ensuring continuity of care among those at risk within health systems) may also help address issues underlying some of the risk conferred by conditions associated with TBI.³⁷ However, in our study, mediators did not address the majority of the relationship between TBI and suicidal ideation/attempt, suggesting there are other factors that contribute to suicide risk among patients with a TBI that are not yet fully understood. Further, there is a direct relationship between TBI and suicidal ideation that may not be improved by treatment of mediators like psychiatric disorders or substance use disorders other than OUD. The mechanism for this direct relationship between TBI and suicide is not yet understood; further studies, including those that examine biomarkers and conduct neuroimaging, are needed to elucidate this relationship. Overall, our findings reinforce the importance of primary prevention of TBI and the need for close monitoring of patients with a TBI for risk of suicide in the 12 months following injury.

There are several limitations to this study. First, patients with mediators that occurred outside of the 3 months that were controlled for may be included in the study. Patients may have experienced one of the mediators prior to the TBI and outside the study timeframe, which may result in results that are overestimated. Second, survey weights were not used and therefore these results cannot be generalized to the national population. Third, the analysis included all cases with ICD 10-CM codes related to intentional self-harm, but these codes do not differentiate between events that were suicide attempts and events in which the self-harm was intentional but there was no intent to die.³⁸ However, there is some research that suggests that most ED visits for self-harm are suicidal in nature.³⁹ Fourth, our study is reliant on medical coding to capture clinical diagnoses and may underrepresent cases of TBI, suicidal ideation/attempt, sleep disorders, psychiatric disorders, and substance use disorders; given the known associations between TBI and other disorders, healthcare providers may be more likely to assess suicidal behaviors and sub-clinical sleep, psychiatric, and substance use disorders in patients with a known TBI diagnosis, which may bias our results away from the null. Additionally, at the time of this analysis, we were unable to stratify TBI encounters by injury severity due to a lack of uniform criteria for injury severity established for ICD-10-CM coding. Finally, we did not conduct analyses looking at suicidal ideation/attempts separately, due to the small size of the sample.

Sleep disorders, psychiatric disorders, and substance use disorders other than OUD partially mediate the relationship between TBI and suicidal ideation/attempt. However, the direct effect between TBI and suicidal behaviors accounts for the largest associations. These results suggest that closely monitoring patients with TBI for suicidal ideation/attempts, even in the absence of psychiatric, substance use, or sleep comorbidities, could help prevent suicides. Further research is needed to understand the direct relationship between TBI and suicidal ideation/attempt.