Case selection for a Medicaid chronic care management program.

Area of Science:

• Health Services Research
• Health Economics
• Public Health Policy

Background:

• Medicaid programs increasingly adopt care management strategies to control expenditures and enhance healthcare quality.
• A key challenge lies in accurately identifying individuals at high risk for costly, preventable healthcare service utilization.

Purpose of the Study:

• To compare the predictive performance of three established health risk assessment models for identifying high-cost Medicaid members with chronic conditions.
• To evaluate the efficacy of the Chronic Illness and Disability Payment System (CDPS), Diagnostic Cost Groups (DCG), and Adjusted Clinical Groups Predictive Model (ACG-PM) in Vermont's Medicaid population.

Main Methods:

• Utilized claims data from the State of Vermont's Medicaid program.
• Compared the ability of CDPS, DCG, and ACG-PM to predict the top 10% of members with chronic conditions.
• Assessed model performance for predicting the highest cost percentiles (e.g., 99th percentile).

Main Results:

• The Adjusted Clinical Groups Predictive Model (ACG-PM) demonstrated superior performance in predicting the top 10% of high-cost members with chronic conditions.
• The Diagnostic Cost Groups (DCG) model was found to be more effective in identifying the very highest-cost individuals, specifically those at the 99th percentile.

Conclusions:

• ACG-PM is recommended for general identification of high-cost Medicaid members with chronic conditions.
• DCG models may be preferable when the specific goal is to target the absolute highest-cost individuals for intensive interventions.