Comparing the predictive value of suicide risk screening to the detection of suicide risk using electronic health records in an urban pediatric emergency department

Objective To compare the accuracy, sensitivity and utility of brief screening to predictive modeling for identifying suicide-related outcomes in a pediatric emergency department. Our hypothesis was that predictive modeling would be more accurate and useful compared to brief screening.
Methods This was a retrospective cohort study at an urban pediatric Emergency Department (PED) in the United States. Patients were aged 8 to 18 years old who presented from January 1, 2017 to June 30, 2019. Predictors included positive suicide risk screen and/or demographic, medical and mental health diagnostic Electronic Health Record (EHR) data at the time of visit. The main outcome was subsequent PED visit with suicide-related presenting problem (i.e. ideation or attempt) within a 3-month follow-up period.
Results Of the N=13420 individuals, n=141 had a subsequent suicide-related ED visit. Approximately 63% identified as Black Non-Hispanic. Results showed that a model based only on EHR data performed only slightly worse than the ASQ alone. Combining ASQ screening and EHR data resulted in a 17.4% improvement in sensitivity and 13.4% increase in AUC compared to the ASQ alone. The LASSO models indicated a diagnosis of major depressive disorder in addition to the ASQ improved the AUC by 9%.
Conclusions, Implications, Future Directions Our findings show that EHR data is helpful either in the absence or as an addition to brief suicide screening. To our knowledge, this is the first study to compare brief screening to EHR based predictive modeling and adds to our understanding of how best to identify youth at risk of suicidal behaviors in clinical care settings.