REFINEMENTS TO THE DIAGNOSTIC COST GROUP (DCG) MODEL

The Diagnostic Cost Group (DCG) model, originally developed by Ash et al. (1986, 1989), has been proposed as an alternative to the existing payment system for reimbursing Medicare health maintenance organizatio ns, the Adjusted Average Per Capita Cost (AAPCC). The DCG model is a l inear regression model that uses both demographic and diagnostic infor mation to predict total plan payments for health care. This paper exte nds previous work by estimating the model using 1984-85 data and by de veloping a more thorough method for classifying hospitalizations by de grees of discretion. It also explores the loss of predictive power res ulting from not using diagnoses for the most discretionary hospitaliza tions for calculating payments. The paper examines a number of extensi ons and refinements to the basic DCG model.