Suicide Prediction and Prevention for People at Risk for Opioid Use Disorder: Supplement to COMPUTE 2.0

Overview

This study integrates the Mental Health Research Network (MHRN) suicide risk models into Opioid Wizard, an electronic health record (EHR) clinical decision support (CDS) to identify and treat patients at high risk of opioid use disorder (OUD)/overdose or diagnosed with OUD, to alert primary care clinicians (PCCs) to patients at elevated risk for suicide and guide them through structured suicide risk assessment. In both intervention and control clinics, suicide risk scores will be calculated for all Opioid Wizard-eligible patients and relevant EHR data to inform analyses will be archived. In intervention clinics, Opioid Wizard will alert PCCs to Opioid Wizard-eligible patients who are at increased risk of suicide and coach them through use of the Columbia Suicide Severity Risk Scale (CSSRS), a structured tool in the EHR that will help PCCs assess immediate suicide risk. Based on the resulting CSSRS score, Opioid Wizard will provide EHR links for risk-based referrals and follow-up recommendations, including care as usual, routine or emergent referral to behavioral health, or transportation to the emergency department (ED) for further assessment. Primary outcome measures include completion of CSSRS assessments for at-risk patients and patient engagement in outpatient mental health care.

When a patient is at elevated risk of suicide, the PCC will be prompted by Opioid Wizard to complete the CSSRS, easily available to all PCCs in the EHR and saved as discrete data elements. Risk-based (depending on CSSRS score) referral and follow-up recommendations for suicide prevention will be given, with specific care recommendations ranging from care as usual (very low risk) to referral to behavioral health for evaluation and safety planning (moderate to high risk) to immediate evaluation in the emergency department and potential inpatient admission (very high risk), building on workflows developed for use by care managers in in our recently completed suicide prevention trial of over 19,000 people at elevated risk of suicide. (28) The MHRN suicide risk models will be programmed into the EHR, a rigorous process that will take approximately 6 months. This process includes building the model in a testing environment in the EHR, testing the model with fictitious patients in a EHR testing environment and conducting chart audits, revising as needed, testing the model in a different EHR testing environment with real patient data, revising as needed, testing the model by running it silently in the EHR production environment and conducting chart audits, and revising as needed. This is followed by testing in the EHR production environment with 5-15 physicians in 1-2 pilot clinics. Prior to the go-live date for the suicide risk calculator, training on use and interpretation of the suicide risk model and the CSSRS will be provided to all PCCs and their rooming staff in intervention clinics. Training for control clinics will be separate and will provide training on the use of the CSSRS.