Abstract
The case-mix classification consists in identifying a series of clinical conditions, so that they can be univocally clustered in groups which are mutually exclusive. Homogeneous conditions are assigned to the same "group" or "class". The criteria for homogeneity can be much different, depending on the goal of the classification itself. Many grouping systems are acknowledged internationally. They may follow the criterion of homogeneity with respect to aetiology (ICD system), resource consumption during an inpatient stay in acute care facilties (DRG system), in a rehabilitation unit (FIM-FRG systems), or in a nursing home (RUG system). Each of these systems strives to keep some distinction across groups, based on clinical criteria, even when the dominant grouping criterion (e.g. cost homogeneity) would rather suggest group merging. The main statistical technique adopted for case-mix grouping is the classification and regression tree (CART). CART allows to find the best compromise between the need to maximise variance explanation of the dependent variable (i.e. the homogeneity criterion: clinical severity, cost, length of stay and the like), and the need to retain clinically identifiable distinctions (e.g. based on the organ impaired). Using a classification system as a payment system entails other constraints: easiness of accountability, disincentive against upcoding, disincentive against adverse selection of more demanding patients. In rehabilitation medicine a FIM-FRG system is already adopted as a payment system for post-acute rehabilitation in USA. Some trials on FIM-FRG classification have been successfully completed in Italy, too.
Original language | English |
---|---|
Pages (from-to) | 201-204 |
Number of pages | 4 |
Journal | Europa Medicophysica |
Volume | 39 |
Issue number | 4 |
Publication status | Published - Dec 2003 |
Keywords
- CART
- Case-mix classification
- FIM-FRG
- Rehabilitation medicine
ASJC Scopus subject areas
- Rehabilitation