Patient casemix classification for medicare psychiatric prospective payment

Area of Science:

• Health Services Research
• Psychiatric Care
• Health Economics

Background:

• Medicare's prospective payment system (PPS) requires accurate casemix classification for inpatient psychiatric facilities.
• Existing systems may not fully capture daily resource utilization variations among patients.
• Developing a robust classification is crucial for equitable and efficient reimbursement.

Purpose of the Study:

• To develop and evaluate a casemix classification system for inpatient psychiatric treatment under Medicare.
• To identify patient characteristics that predict daily resource use.
• To compare the proposed system's efficiency and equity against existing models.

Main Methods:

• Survey data from 696 patients across 40 facilities were collected on patient characteristics and daily activities.
• Data were combined with Medicare claims to estimate intensity-adjusted daily costs.
• Classification and Regression Trees (CART) analysis identified key predictors for cost variation.

Main Results:

• CART analysis revealed age, psychiatric diagnosis, daily living deficits, and detox/ECT use as strong cost predictors.
• A 16-group interactive model explained 40% of daily cost variation, outperforming a noninteractive model (32% explanation).
• The proposed model utilized DSM-IV categories, age, illness severity, daily activity deficits, dangerousness, and ECT use.

Conclusions:

• A 16-group regression model using clinically valid and administratively feasible variables offers more efficient and equitable Medicare payments.
• The developed casemix classification system is superior to noninteractive, diagnosis-related group-based systems.
• This approach supports improved financial incentives and care delivery in psychiatric facilities.