Predicting future hospital utilization for mental health conditions.

To develop a model using administrative variables to predict number of days in the hospital for a mental health condition in the year after discharge from a mental health hospitalization. Background, index hospitalization and preindex inpatient, emergency room, and outpatient utilization information were collected for 766 adult members discharged from a mental health hospitalization during a 1-year period. A regression model was developed to predict hospitalized days for a mental health condition in the year after discharge. A regression model was created containing five statistically significant predictors: Medicare insurance coverage, preindex mental health inpatient days, index length of stay, depression diagnosis, and number of mental health outpatient visits with a professional provider. It is possible to predict future mental health inpatient utilization at the time of discharge from a mental health hospitalization using administrative data, thus allowing disease managers to better identify members in greatest need of additional services and interventions.