Predicting Suicide Attempts and Suicide Deaths Using Electronic Health Records
New model substantially outperforms existing suicide risk tools
Researchers have developed a new prediction model that substantially outperforms existing self-report tools used to predict an individual’s risk of a suicide attempt or death by suicide.
Predicting Suicide Attempts and Suicide Deaths Using Electronic Health Records
New model substantially outperforms existing suicide risk tools
• Science Update
Suicide accounted for nearly 45,000 deaths in the United States in 2016. Unfortunately, tools currently used to predict an individual’s risk of a suicide attempt or dying by suicide, such as brief self-report measures, have only moderate accuracy. Now, researchers have developed a new prediction model that substantially outperforms existing self-report tools. The study, supported by the National Institute of Mental Health (NIMH), was published online on May 24, 2018, in the American Journal of Psychiatry.
Research has shown that half of the people who die by suicide, and two-thirds of people who attempt suicide, received a mental health diagnosis or treatment in the previous year. These statistics suggest an opportunity for doctors to identify and assist those who are at risk for suicide before they act.
Lead author Gregory Simon, M.D., M.P.H., a senior investigator at the Kaiser Permanente Washington Health Research Institute, and colleagues, set out to develop an improved way to predict suicide attempts and suicide deaths in the 90 days following a mental health diagnosis. The model used data from electronic health records (EHRs) provided by seven major health systems, including the Henry Ford Health System in Detroit, the HealthPartners Institute in Minneapolis, and Kaiser Permanente regions of Colorado, Hawaii, Oregon, California, and Washington.
“By leveraging existing electronic health record data and advancements in statistical modeling, it is possible to significantly improve the prediction of death by suicide and suicide attempts over conventional self-report methods,” said Michael Freed, Ph.D., chief of the Services Research and Clinical Epidemiology Branch in the NIMH Division of Services and Intervention Research.“ Valid and reliable suicide risk prediction models hold tremendous promise to reduce death by suicide, especially when integrated with evidence supported approaches to suicide prevention.”
Anonymized data from the EHRs of almost three million patients who had a mental health diagnosis recorded at either a primary care or a mental health clinic visit between January 1, 2009, and June 30, 2015, were included in the study. The prediction models used information typically available in EHRs or insurance claims: mental health diagnoses recorded during the past five years, mental health medication prescriptions filled during the last five years, and the use of acute-care (inpatient and emergency department) mental health services in the past five years. The researchers created the model using 65 percent of the EHR data and used the remaining data to test its accuracy.
The researchers found that of all the variables included in the model—mental health diagnoses, substance use diagnoses, use of mental health emergency and inpatient care, history of self-harm, and scores on the Patient Health Questionnaire (a standardized depression questionnaire)—were the strongest predictors of suicide attempt and death.
“This prediction model was more accurate than previous models using health records. For example, people with risk scores in the highest five percent accounted for almost half of suicide attempts compared to about one third with previous models,” said Dr. Simon.
The researchers also found that the overall accuracy of the new prediction model exceeded that seen in models predicting other types of health issues, such as rehospitalization for heart failure, in-hospital mortality from sepsis, and high emergency department utilization. The predictive value of the new model was equal to, or better than, widely accepted tools for prediction of major medical outcomes such as stroke in atrial fibrillation and cardiovascular events.
Although prediction models cannot replace clinical judgment, this new tool may help practitioners make better-informed clinical decisions.
“We believe these risk prediction tools are now accurate enough to help clinicians identify people at high risk and to help health systems reach out to people at risk who miss or cancel appointments,” said Dr. Simon.
Reference
Simon, G. E., Johnson, E., Lawrence, J. M., Rossom, R. C., Ahmedani, B., Lynch F. L., … Shortreed, S. M. (in press). Predicting suicide attempts and suicide deaths following outpatient visits using electronic health records. The American Journal of Psychiatry. OnlineFirst May 24, 2018.
Grants: NIMH: MH092201
No hay comentarios:
Publicar un comentario