The heart and science of medicine. UVMHealth.org/MedicalGroup Suicide Risk Assessments in Hospitals Using Systematic Expert Risk Assessment for Suicide (SERAS) Robert R. Althoff, MD, PhD Associate Professor of Psychiatry, Pediatrics, & Psychological Science Director of Adirondack Division of Psychiatry Larner College of Medicine at the University of Vermont VT Suicide Prevention Symposium June 5, 2018
Conflict of Interest Disclosure • Co-Inventor of SERAS (Systematic Expert Risk Assessment for Suicide). • Equity stake in WISER Systems, LLC – holds license from UVM to commercialize SERAS.
The Team William Cats-Baril, PhD William Hudenko, PhD Associate Professor, UVM Grossman School of Assistant Professor, Dartmouth College. CEO Business; Founding Director of MBA on Incente llc, Winner of the 2016 Dartmouth Sustainable Entrepreneurship Entrepreneurial Forum StartUp Competition. Isabelle Desjardins, MD Michael Goedde, MD, MS General and Geriatric Psychiatrist, Associate Professor, UVM-COM; Chief Medical Officer, UVM-COM graduate and training in psychiatry University of Vermont Medical Center and research methods. John Helzer, MD Sanchit Maruti, MD Professor Emeritus, UVM; 40 years of research Addiction Psychiatrist, Assistant Professor, UVM- experience and 15 years of eHealth approaches to COM; Medical Director, UVM-Medical Center alcohol assessment, smoking, ECT, and pain Suboxone Program. management Donna Rizzo, PhD Robert Althoff, MD, PhD Professor and Dorothean Chair, UVM College of Child and Adolescent Psychiatrist; Associate Engineering and Mathematical Sciences. Expert Professor, UVM-COM; Director, Adirondack in complex systems. Division of Psychiatry, New York 3
Epidemiology of Suicide in Hospitals • 2 nd most frequently reported of the five serious events reported to the Joint Commission since 1995 • Many patients who kill themselves in general hospital inpatient units are “unknown at risk” for suicide – No psychiatric history or – No history of suicide attempt • The Sentinel Event Database includes 827 reports of inpatient suicides: The Joint Commission, Sentinel Event Alert, Issue 46, November 17, 2010 Bostwick JM, Rackley, SJ: Completed Suicide in Medical/Surgical Patients: Who is At Risk? Current Psychiatry Reports, 2007;9:242‐246
Joint Commission Sentinel Alert #56 • Detecting and treating suicide ideation in all settings; • Most individuals who die by suicide receive health care services in the year prior to death, usually for reasons unrelated to suicide or mental health. • National Patient Safety Goal 15.01.01 requires General hospitals to identify individuals at risk for suicide. • …requires…“general hospitals treating individuals for emotional or behavioral disorders, to conduct a risk assessment that identifies specific individual characteristics and environmental features that may increase the risk of suicide and to address safety needs such as placing a patient under constant observation if the patient exhibits warning signs.” 5
Detection Tools ED‐SAFE Patient PHQ2/PHQ9 SBQ‐R/ASQ Safety Screener • Detects Major • PHQ2 and • Assesses Depression portions of C‐ suicide related SSRS thoughts and behaviors • Holds predictive • Differentiates validity for suicidal from lifetime suicide nonsuicidal risk patients overall 6
Secondary Screening ED‐SAFE Patient Safety Secondary C‐SSRS SAFE‐T Pocket Card Screener for EDs • Assess dimensions of suicide risk to help the clinician come to a risk assessment. 7
The Need • Tools to specifically assess imminent/near term risk • Tools that use accumulated risk factor information and also account for expert weighting of risk factors and clinical data Brown GK. (2002). A review of suicide assessment measures for intervention research in adults and older adults. Technical report submitted to NIMH under Contract No. 263- MH914950. 8
Hypothesis We can replicate the judgment of Board-Certified Psychiatrists in assessing the near-term risk of suicide (next 72-hours) and the associated expert intervention recommendations 9
Phase 1: Model Development • Convened Panel of National Experts • Comprehensive Literature Review • Discussed important components of an ideal suicide risk Expert assessment tool Panel • Revised and Added Cases to cover full range of suicide risk • Reviewed Cases • Rated patient profiles based on likelihood of Acute suicide Case Review/Revision in the hospital setting in the next 72 hours • Utilized Nominal Group Technique (NGT) to identify risk variables, ranges and weights on those variables and discussed aggregation methods (compensatory/non‐compensatory, etc.) • Compared the experts’ rankings and addressed inconsistencies Group and disagreements Consensus • Created a preliminary model that included critical features that Process should be in the computerized screening
Suicide Risk Assessment Questions 11
Discovery Sample • Testing of algorithm against expert consensus in the ED patient population • Psychiatrists examining patients in ED were trained using the same cases as the Expert Panel 12
S Discovery Samplea2 Mean Age: 42.6 [SD 17 years] Male: 48% Primary Psych Complaint: 8% %
Categories of Risk (RISK) High Moderate Low Minimal Even considering The combination of risk Although there may be The risk of suicide in mitigating protective factors and mitigating suicide risk factors the following 72 hours factors, the acute risk protective factors yield present, the is minimal. of suicide – in the a risk of suicide in the combination of risk following 72 hours‐ is following 72 hours that factors and mitigating high is only moderate protective factors yield a relatively low risk of suicide in the following 72 hours Suicide Possible and Suicide Possible, but Suicide Very Unlikely Likely Unlikely
Categories of Interventions (INT) TYPE OF ROUTINE SPECIALIZED HIGHLY SECURED INTERVENTION SPECIALIZED Psychiatric No Psychiatry or Psychiatry or Psychiatry or Assessment Crisis Consultation Crisis Consultation Crisis Consultation Environment of No No Environment of Environment of Care Modification care free of care free of harmful hazards harmful hazards Level of Routine Routine Routine Constant Observation Observation After Care Routine Yes Yes Yes Referral Education re Yes Yes Yes Yes Suicide Prevention Resources at Discharge
Neural Neural Network Model ‐ RISK Model - RISK Classification Errors: Psych Chief Complaint in 100% of cases
Ne Neural Network Model ‐ INT r Model - INT Classification Errors: Psych Chief Complaint in 100% of cases
Replication and Extension Samples ED Med-Surg Inpatient Psychiatry 18
Partnership 19
Partnership 20
EMR Integration 21
EMR Integration 22
Conclusions Novel Approach to Suicide Risk Assessment in hospitals – Expert Systems Approach/ Neural Network mathematical model – User Friendly – iPad based and patient self-administered – Time Efficient – Able to replicate the Risk Assessment and Intervention recommendations of expertly-trained clinicians for ED and Med- Surg populations – Only predicts expert assessment of near-term risk (72 hours) – Risk assessment is contextual to the environment of care 23
Next Steps - Have just completed a new trial of 480 participants in emergency room setting who received SERAS, C-SSRS, expert risk assessment, and 72 hour follow up - Finalizing data analysis now - Can report that SERAS did again detect several individuals who had not reported suicidal ideation in any other format - We see SERAS as an easy to use general tool that can direct the need for more refined assessment - Integration with social media - Enhancement with VOI to include REACH – a mechanism for incorporating social supports 24
Acknowledgements • Funding: – Fletcher Allen Foundation – University of Vermont Medical Group (UVMMG) Research and Education Committee – SPARK-VT – National Institute of Mental Health • Contributors: – University of Vermont EMRAP program – Jeffords Quality Institute – Ms. Diantha Howard – Ms. Abigail Wager – Ms. Chelsea Manning – Conor Carpenter, MD – Judy Lewis, MD, Isabel Norian, MD, Anne Rich, MD, Tobey Horn, MD for their contribution to the Expert Assessments 25
Recommend
More recommend
