Australian Refined Diagnosis Related Groups. Formal and inherent problems of grouping with the example of stroke care

With the Health Reform 2000, the Australian Refined Diagnosis Related Groups (AR-DRG), Version 4.1 have been chosen as the basis for the future German costing system for hospitals. With regard to Stroke Severity (Barthel Index [BI]) we investigated to what extent the grouping according to AR-DRGs can reproduce healthcare expenditures for such patients. Options to adapt and optimize the system are discussed. 632 patients who had suffered a cerebrovascular accident and were discharged from conservative acute care in 1999, were classified according to the AR-DRGs. For the grouping we alternatively used data from the current hospital information system and a stroke database for quality assurance. The results were also compared with the clinical profiles for the public hospital sector of the corresponding DRGs in Australia (1997-98). On average 0.99 additional diagnoses per case were documented in the hospital information system, compared to 3.65 in the stroke database. In the stroke database 177 cases (36.8%) were assigned to the DRG with the highest cost weight. 53.7% of these patients suffered a serious stroke (BI < 30). Grouping on the basis of hospital information system data led only to 14 cases (2.8%) assigned to the DRG with the highest cost weight. Type and extent of additional diagnoses are crucial for the grouping process. From a clinical and economic point of view, measures of disability and impairment should be assigned to the grouping process to improve homogeneity under both aspects. Scores can also serve for determining reliable outcome parameters. For the development of an outcome related reimbursement system, procedures must be included in the definition of medical DRGs. In future, DRGs, which cover overlapping healthcare sectors, should be developed for patients with poststroke rehabilitation.