Short Info: EfP (France)

The french institute PERNNS ("Pôle d'expertise et de référence national des nomenclatures de santé"), that also developed the GHM system, has proposed a model which could improve the GHM representation of the spectrum of patients.1 It suggested abolishing the CC levels (CMA, CMAS, age) and in turn assigning one or several base GHMs ("GHM élémentaires") to each case. The cost weights would then not be calculated as arithmetic averages, but rather using multivariate statistical procedures.

The grouping is to be done similar to peeling an onion. Initially, a first base GHM is defined using the decision tree of the GHM classification. Then all diagnoses and procedures that belong to this GHM are eliminated. The remaining items are hierachised, and the grouping process is repeated to determine the second GHM. And so forth. GHMs that are coming up repeatedly have to be bundled, and mutually excluding GHMs have to be defined. The number of GHMs per treatment case could be limited, if needed. Cost weights could be calculated based on the pre-existing data of 1996 and 1997, with the aim to establish a coefficient matrix that allows for easy calculation of the cost weight of each case.2 – This method was called "Progressive Defoliation" ("Effeuillage Progressif" = EfP).

First tests revealed that the variance reduction related to costs improved by approx. 10 % up to about 49 % (excluding sameday cases), while only approx. 370 GHMs (instead of 506) were used.

The system never went beyond its test stage.