Refinements to the Diagnostic Cost Group (DCG) model

Area of Science:

• Health economics
• Medical informatics
• Public health policy

Background:

• The Adjusted Average Per Capita Cost (AAPCC) is the current Medicare reimbursement system for health maintenance organizations.
• The Diagnostic Cost Group (DCG) model offers a data-driven alternative using demographic and diagnostic predictors.
• Previous iterations of the DCG model require updates and refinements for practical application.

Purpose of the Study:

• To estimate the DCG model using recent Medicare data (1984-85).
• To develop a refined method for classifying hospitalizations based on discretion.
• To evaluate the impact of diagnostic information on payment prediction accuracy.

Main Methods:

• Linear regression analysis applied to 1984-85 Medicare data.
• Development of a novel classification system for hospitalizations by degree of discretion.
• Comparative analysis of predictive power with and without discretionary diagnostic data.

Main Results:

• The DCG model was estimated using 1984-85 data, providing updated predictive parameters.
• A more granular classification of hospitalizations by discretion was established.
• Excluding diagnoses for highly discretionary hospitalizations led to a measurable loss in predictive accuracy.

Conclusions:

• The DCG model shows promise as a more accurate reimbursement system for Medicare.
• Refinements in hospitalization classification enhance the model's precision.
• Incorporating diagnostic data, even for discretionary cases, is crucial for robust payment prediction.