Predicting future hospital utilization for mental health conditions

Area of Science:

• Health Services Research
• Mental Health Services
• Healthcare Administration

Background:

• Predicting future healthcare utilization is crucial for effective resource allocation.
• Mental health hospitalizations represent a significant cost and require careful management.
• Administrative data offers a readily available resource for predictive modeling.

Purpose of the Study:

• To develop a predictive model for mental health inpatient days.
• Utilize administrative variables for forecasting post-discharge utilization.
• Identify individuals at high risk for readmission or prolonged stays.

Main Methods:

• A regression model was developed using administrative data.
• Included variables such as insurance, preindex utilization, and index hospitalization details.
• Data collected from 766 adult members post-mental health hospitalization.

Main Results:

• Five key predictors identified: Medicare coverage, prior inpatient days, index length of stay, depression diagnosis, and outpatient visit frequency.
• These factors significantly predicted mental health inpatient days in the year post-discharge.
• The model demonstrates the utility of administrative data in forecasting utilization.

Conclusions:

• Administrative data can accurately predict future mental health inpatient utilization.
• Enables proactive identification of members needing additional services and interventions.
• Facilitates targeted disease management strategies for mental health conditions.