Implementation of suicide risk prediction models dependent on clinician engagement

Implementation of suicide risk prediction models may be crucial for prevention efforts, provided guidance considers clinician engagement, health system goals, available resources, training and value with respect to existing initiatives.

“The goal of this applied qualitative research was to equip potential suicide risk prediction model adopters to deploy these models in a manner that is patient-centered, supports clinicians, and is sustainable,” Bobbi Jo H. Yarborough, of the Kaiser Permanente Center for Health Research in Oregon, and colleagues wrote in BMC Psychiatry.

Yarborough and colleagues sought to ascertain whether suicide risk prediction models derived from analyzing data from electronic health records is an effective method of suicide prevention.

Their qualitative study included 40 health care professionals (30 clinicians and 10 health care administrators) interviewed from one health system in anticipation of implementing an automated EHR-derived suicide risk prediction model (Kaiser Permanente Northwest), along with two health systems piloting different implementation approaches (Kaiser Permanente Washington, HealthPartners). Interviews were conducted by phone between May 2020 and November 2020, lasted approximately 30 minutes and were recorded.

Interview guides tailored to each locale focused on participants’ expectations for and experiences with suicide risk prediction models in clinical practice, as well as suggestions for improving implementation. Interview prompts and content analysis were guided by Consolidated Framework for Implementation Research constructs.

Results showed that all 40 participants completed the interview process, with 73% of interviewees reporting they felt health systems had a responsibility to monitor suicide risk using EHR and insurance claims data. There was broad agreement among participants that risk models should be implemented by experienced clinicians in mental health departments, while 60% of interviewees had a favorable first impression of the risk model, and 26% recorded mixed feelings. Four respondents felt the model was not helpful or added little value to the process.

Administrators and clinicians found use of the suicide risk prediction model and the two implementation approaches acceptable, while clinicians desired opportunities for early buy-in, implementation decision-making and feedback, and expressed concern about having enough suicide prevention resources for potentially increased demand and their personal liability. Suggestions for making risk model workflows more efficient and less burdensome included consolidating suicide risk information in a dedicated module in the EHR and populating risk assessment scores and text in clinical notes.

“It is important to clarify clinicians’ role expectations and liability, support clinicians with summarized risk information in the EHR and provide the opportunity for clinicians to give and receive feedback on the process of risk model implementation,” Yarborough and colleagues wrote.