More Accurate Detection of Self-Injurious Behavior Needed for Minoritized Youth

More accurate identification of self-injurious thoughts and behaviors (SITB) could ensure more equitable care in efforts to prevent youth suicide. According to a study published today in JAMA Network Open, chief concerns and diagnostic codes have done a disservice to minoritized youths who experience suicide-related emergencies, underpinning the need to combat algorithmic bias, improve strategies for suicide prevention, and better evaluate youths’ needs in these situations.¹

Suicide is the top cause of death in young people | Image Credit: © Cavan - stock.adobe.com

Currently, data from Mental Health America show that well over 13% of youth (n = 3.4 million) in the US experience serious suicidal thoughts, and over 20% have experienced 1 or more major depressive episodes over the last year. These figures demonstrate the severity of mental health (MH) issues in today’s youth that require further attention.

According to their database, over 50% of youth who have major depression did not receive MH treatment from a medical professional, counselor, or other MH provider, largely because they felt they should handle their MH themselves or they were not aware of how or where to find treatment. Furthermore, youth report fears about MH stigma and the prospect of their information not being kept private.² The current authors note how suicide is the most prevalent cause of death that affects younger populations; coupled with these statistics from Mental Health America, it remains apparent how serious and under the radar MH issues can be for those who are struggling.¹

To evaluate the state of SITB-detection methods, the researchers conducted a retrospective, cross-sectional study to evaluate electronic medical records (EMRs) from visits to the emergency department (ED) in a large Southern California health system. They also analyzed how their system algorithms performed across varying demographics. Individuals aged 6 to 17 years, who were admitted with at least 1 MH-related ED visit, were included beginning in October 2017 until 2019. The presence of at least 1 International Statistical Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) code was used to define SITB, and case surveillance (CS, which relied on chief concern and codes alone) and adjusted CS (aCS, which factored in additional characteristics such as medications, prior acute care visits, and more) were features used to classify patients.

A total of 2702 youths were identified, 1384 of which identified as female (51.2%); 131 who were Asian (4.8%); 266, Black (9.8%); 719, Hispanic (26.6%); and 1319, White (48.8%). The remaining 233 identified as belonging to “other race.” From this group there were 898 children (aged 6-12 years) and 1804 adolescents (aged 13-17 years).

Almost 50% of ED visits for children were SITB related (n = 1286), while other visits were related to attention-deficit/hyperactivity disorder (ADHD; n = 272; 30.3%), self-injury or suicide (n = 266; 29.6%), anxiety disorders (n = 207; 23.1%), depressive disorders (n = 196; 21.8%), or another MH-related symptom (n = 193; 21.5%). For adolescents, depressive disorders were the most prevalent reason for visiting (n =747; 41.4%), followed by self-injury or suicide (n = 684; 37.9%), anxiety disorders (n = 561; 31.1%), ADHD (n = 306; 17%), and disorders related to substance use (n = 299; 12.7%). The researchers noted that suicide attempts occurred more in adolescents vs children (9.3% vs 6.5%; P = .01). Similarly, adolescents reported chief concerns related to suicide more often (30% vs 8.4%; P = .001).

When it came to detecting SITB, the analysis revealed that the youngest group of children (aged 6-9 years) experienced the worst accuracy (81.2%; 95% CI, 75.3%-86.3%) and that accuracy improved in older patients (aged 10-12 years: 84.6%; 95% CI, 81.7%-87.2%]; aged 13-17 years: 92.4%; 95% CI, 90.5%-94%). The researchers also mentioned that chief concern and SITB detection grew more sensitive with age when ICD-10-CM codes were used.

Overall, aCS (which used all available structured data) demonstrated superiority over CS (area under the receiver operating characteristic curve [AUROC], 0.975; 95% CI, 0.968-0.980 vs 0.894; 95% CI, 0.882-0.905; P < .001). AUROC was observably lower for preadolescents (0.841), as well as Black (0.859), Hispanic (0.861), and male (0.869) youths compared with adolescents (0.925) and female youths (0.923) and White youths (0.901).

“Several factors, such as underlying psychopathology, reimbursement practices, stigma, and clinician biases, may explain why youths do not receive suicide-related codes and chief concerns,” the authors wrote.

Considering these results, the researchers see an opportunity to improve early detection of suicidal thinking.

“Phenotypes impact policymaking around potential treatments, allocation of resources, and, consequently, the health of populations. Our findings support the need to establish best practices for evaluating phenotype definitions of SITB in youths and to ground these practices in algorithmic fairness metrics… Shifts in clinician practices could increase sensitivity to detect suicide-related emergencies with diagnostic codes and chief concerns. In the interim, without accurate and equitable phenotype definitions, suicide prevention strategies will inadequately target the populations they aim to serve,” they concluded.

References

1. Valtuille Z, Trebossen V, Ouldali N, et al. Pediatric hospitalizations and emergency department visits related to mental health conditions and self-harm. JAMA Netw Open. 2024;7(10):e2441874. doi:10.1001/jamanetworkopen.2024.41874

2. Youth Ranking 2024. Mental Health America. Accessed October 29, 2024. https://mhanational.org/issues/2024/mental-health-america-youth-data#five