Lit Matters #2: Suicide Risk Assessment in the ED

Suicide risk assessment in the emergency department is more accurate when clinician judgment is supplemented by structured patient self-report and electronic health record data. In psychiatric ED patients, combined models outperformed bedside assessment alone for predicting suicide attempts at both 1 and 6 months.

Suicide Risk Assessment in the ED

• Clinician gestalt limits: Treating clinician assessment had the weakest discrimination for both 1-month and 6-month suicide attempts, a useful reminder that bedside impression alone misses meaningful risk.
• Brief patient self-report: A short tablet-based self-report performed better than clinician assessment for near-term and 6-month prediction, supporting structured questioning over unstructured interview alone.
• EHR risk signal: Machine-learning predictions drawn from data already in the electronic health record also outperformed clinician judgment, showing that chart-level history carries actionable prognostic value.
• Best combined approach: The strongest performance came from pairing patient self-report with an EHR-based risk score rather than choosing one tool over the other. We get into what that means for ED workflow in the episode.
• Meaningful event rate: Suicide attempts after the index ED visit were common in this cohort, with roughly 13% by 1 month and about 22% by 6 months, underscoring why missed risk matters.
• High-risk psychiatric cohort: These findings came from adults presenting with psychiatric complaints and seen by psychiatric services, so the signal is most applicable to behavioral health ED populations rather than undifferentiated screening.