A new advancement in predicting suicide risk among adolescents in the Emergency Department.

A large-scale study recently published in JAMA Network Open shows that clinical data analysis algorithms are significantly more capable of identifying adolescents at high risk of suicide compared to traditional screening methods.

An alarming context of child and adolescent mental health

Suicide is currently the second leading cause of death among adolescents in the United States, with rates increasing by 62% between 2010 and 2020. Notably, statistics indicate that about 80% of adolescents who died by suicide had contact with the healthcare system in the year prior, and nearly 50% had visited an Emergency Department. However, only 32% had previously received a mental health–related diagnosis, revealing a significant gap in early risk identification.

To address this issue, the Joint Commission has established a National Patient Safety Goal requiring healthcare facilities to conduct suicide risk screening using evidence-based tools.
Comparative study: Direct screening or data-driven algorithm?

A retrospective cohort study was conducted on 19,653 adolescents (aged 10–18 years) presenting to a pediatric Emergency Department in the Northeastern United States. The study aimed to compare the effectiveness of two approaches:

1. Direct screening: Using standardized questionnaires such as ASQ (Ask Suicide-Screening Questions) and C-BSSRS.
2. Risk algorithm: Using data from electronic health records (EHR) over up to four years, including age, sex, insurance status, and ICD-10 diagnostic codes.

The numbers speak: Algorithms clearly outperform

Analysis results from May 2023 to December 2024 highlighted the superiority of technology in preventive medicine:

• Identification capability: The algorithm identified 127% more patients who later engaged in suicidal behavior compared to the screening method.
• Higher sensitivity: The algorithm correctly identified 50.7% of suicide cases, while direct screening achieved only 36.5%.
• Overall performance: Across most evaluation metrics (such as AUROC, sensitivity, and positive predictive value), the algorithm demonstrated statistically significant superiority.

Why are algorithms more effective?

The advantage of the algorithm lies in its ability to leverage the richness of historical clinical data. While direct screening often reflects only the patient’s status at the time of assessment and may result in high false-positive rates, the algorithm analyzes longitudinal data over time.
The study found that patients identified as high-risk by the algorithm had more prior healthcare visits (median of 3 visits compared to 2 in the screening group) and more diagnostic codes recorded. Common diagnoses associated with higher risk included major depressive disorder, anxiety disorders, and a history of suicide attempts.

Implications for the healthcare sector

Implementing suicide risk algorithms not only helps hospitals meet the stringent requirements of the Joint Commission but also offers substantial practical benefits:

• Resource optimization: Enables limited mental health resources to be directed toward patients with the greatest need.
• Reduced administrative burden: Minimizes time-consuming universal screening processes with limited effectiveness.
• Improved accuracy: Surpasses cost-effectiveness thresholds to support evidence-based early intervention strategies.

Conclusion

Although there are limitations, such as the study being conducted at a single institution and reliance on diagnostic coding, this research opens a new direction for identifying suicide risk. The integration of direct clinical assessment and data-driven algorithms is expected to become an optimal strategy for protecting adolescent lives in the future.

Read the full article in JAMA.