Technology-enabled Suicide Prevention in Primary Care and Other Medical Settings: A Scalable Solution

Title: Technology-Enabled Suicide Prevention in Primary Care and Other Medical Settings: A Scalable Solution

Background: Suicide is the 12th leading cause of death in the United States, resulting in over 48,000 deaths in 2021. Eighty percent of suicide decedents had contact with a primary care physician within 1 year and 44% had contact within 1 month of the suicide. Suicide risk screening in primary care and other medical settings is critical, yet there are many barriers to implementation. Technology through mobile health (mHealth) platforms offer scalable, automated remote patients assessments with real-time data collection for suicide prevention in multiple medical settings, enabling immediate intervention to prevent risk escalation and sentinel events.

Method: A technology-enabled suicide prevention program was implemented across various clinical settings including primary and specialty care, health insurance plan sponsored protocols, and disability claimants. Patients engaged through bulk eligibility files, direct provider referral, and self-sign-up protocols. The mHealth platform features psychoeducation, mood/sleep tracking, and asynchronous assessments for continuous monitoring of new and emerging suicide risk. Three factors determine potential risk: 1) Positive response to suicidality on the PHQ-9, PHQ-2/9, or Edinburgh, 2) Natural language processing (NLP) identifies potential risk, or 3) Increase trending in proprietary risk score (Pardes et al., 2022). Urgent alerts are triggered for high-risk activity, initiating the SAFE-T Protocol and Columbia Suicide Severity Risk Score screen version as well as safety planning initiatives (SPI) and lethal means restriction counseling.

Results: Data from 10,099 mHealth patient users collected between 4/1/22-3/16/23 showed 718 patients with self-harm responses >0 on the PHQ-9/PHQ-2/9 or Edinburgh, triggering 1,127 urgent alerts. An additional 33 patients were identified through NLP. 300 unique patients completed outreach after an urgent alert: 23 evaluated at high risk, 81 at moderate risk, 143 at low risk, and 90 indicating no risk. Among patients identified with a high-risk C-SSRS score (Nf23), a total of 48 urgent alerts were triggered (46 by assessment, 2 by NLP detection).
<p class="MsoNormal">

32% of patients who triggered an NLP alert either did not take an assessment or did not indicate self-harm on the PHQ-9, PHQ-2/9, or Edinburgh prior to the urgent alert. Patients who triggered NLP urgent alerts were at medium or high risk for suicide at more than double the rate of individuals who triggered PHQ-9 urgent alerts.

Conclusion: mHealth asynchronous and automated suicide risk screening and concomitant care team support enable proactive identification, risk stratification, and intervention with evidence-based assessments and SPI across large populations of patients in primary care and other medical settings. Thirty-five percent of patients who completed safety screening assessments were at high or moderate risk of suicide necessitating urgent interventions. Technology-enabled suicide prevention programs provide effective solutions at scale for suicide prevention in diverse medical settings.

References:

Ayer L, Horowitz LM, Colpe L et al., Clinical Pathway for Suicide Risk Screening in Adult Primary Care Settings: Special Recommendations; Journal of the ACLP 2022:63:497-510

Mann JJ, Michel CA, Auerbach RP, Improving Suicide Prevention Through Evidenced-Based Strategites: A Systematic Review; Am J Psychiatry 2021: 178:611-624.