[Value of German Hospital Diagnosis Statistics in epidemiology-- an analysis with subarachnoid hemorrhage as an example].

: Due to the Hospital Statistics Regulations of the 10th April 1990, the German Hospital Diagnosis Statistics were introduced in 1993 with the intention to serve as a database of health care decisions. This purpose requires a high-quality collection of epidemiologically relevant data. From 1993 to 1996 we analysed the datasets obtainable for subarachnoid haemorrhage which is coded with ICD 430. A subset concerning the data of 1996 and the Saarland region was compared to the data of the Saarland medical school at Homburg/Saar. Cases treated in the neurosurgical department were critically reviewed. About 20% of the cases coded with ICD 430 showed no subarachnoid haemorrhage. On the other hand, again about 20% of subarachnoid haemorrhage cases were not coded with ICD 430. The statistics comprise duplicates due to transfers between hospitals. The calculation of incidence is not possible because new bleeding cases cannot be outlined. In the present form the German Hospital Diagnosis Statistics are not suitable as a reliable base of health care decisions. This is partly caused by the inadequacy of the ICD-classification but, also, by the criteria for collecting data. We propose several modifications which can improve data quality in order to meet the intended requirements.