Empirical comparison of DRG variants using cardiovascular surgery data: initial results of a project at 18 German hospitals.

In 2000, the responsibility for selecting a DRG variant for use in Germany was assigned to a body comprising representatives of hospitals and insurers called the Self-Administration Board (or Board in this paper). To help the Board, we applied cardiac surgery data from 18 German hospitals to eight different DRG variants. The error caused by bad coding quality could be minimized this way, since all diagnoses and procedures in cardiac surgery must be recorded for quality assurance purposes. To match the German code to the appropriate code required by the DRG variant, we created mapping tables whenever needed. As far as cardiac surgery is concerned, the Australian AR-DRG and the French GHM variants provided the best medical relevance, while the AR-DRG variant considered the level of severity better. Other variants would have to be updated to better reflect the level of medical complexity. Three main causes for wrong grouping could be identified for all systems: incomplete mapping, not enough reference to multidisciplinary treatments, and system construction problems.