UCLA team receives award for leveraging EHR data to predict suicidality after medical hospitalization

A UCLA team of psychiatry residents have received the Dlin/Fischer Clinical Research Award from the Academy of Consult Liaison Psychiatry for their abstract on predicting suicidality after medical hospitalization.

The team examined how recent national suicide prevention initiatives have called for the development of precision algorithms to help identify individuals at high-risk of suicide. Precision approaches utilize the rich information available in electronic health records (EHR) to model profiles of patients with rare outcomes. Algorithms resulting from precision approaches have demonstrated efficacy in identifying patients at high risk of suicide. Further, they hold promise for enhancing early intervention and prevention.

According to the study, accurate precision approaches have thus far not been developed for modeling suicide risk in patients with co-existing serious mental and non-psychiatric medical illness. The UCLA team addressed this unmet need by leveraging EHR data spanning ten years to develop suicide risk profiles in medically ill patients. The method applied a precision approach to a 2006-2016 database comprising 206,129 encounters and 16,552 medical admissions of adult patients at UCLA with serious mental illness (major depressive disorder, bipolar disorder, or psychotic disorder). 

The model identified 107 of 108 rehospitalizations for suicide attempt and 312 of 338 rehospitalizations for suicidal ideation. It also identified three pathways of increased risk of suicide attempt, suicidal ideation, and any suicidality.

Their findings suggest they have a highly accurate and clinically interpretable model that predicts readmission for suicidality after medical hospitalization among patients with serious mental illness.

The research was led by Dr. Juliet Edgcomb, chief resident of the Psychiatry Residency Training Program at the Semel Institute for Neuroscience & Human Behavior and Resnick Neuropsychiatric Hospital, and Dr. John Brooks, associate professor of psychiatry, who served as senior faculty on the project. Other contributors included Dr. Gerhard Hellemann, associate professor of psychiatry, and Dr. Trevor Shaddox, a PGY-2 Psychiatry Research-track Resident.

This research was supported by EHR data from the CTSI Integrated Clinical and Research Data Repository (xDR). Visit the xDR page for more information on how to access data or contact patientdata@mednet.ucla.edu.

Image source: Pexels

Image caption: Accurate precision approaches have not been developed for modeling suicide risk in patients with co-existing serious mental and non-psychiatric medical illness. A UCLA team addressed this unmet need by leveraging EHR data that spans ten years to develop suicide risk profiles in medically ill patients.