Quality of hospital data and DRGs.

Research on and application of DRG is frequently performed with the aid of computerized data. This study has been made to assess the importance of errors in hospital data bases for DRG research, and the financial consequences if DRG is used as a reimbursement system. For the same hospital stays information on the DRG grouping variables has been collected from medical records and a data base. Frequencies of errors in the data base for the DRG grouping variables (principal diagnosis, secondary diagnosis, complications, procedures, age and discharge status of death) and length of stay are given. DRG grouping is performed on the basis of both medical record and data base information. The analysis shows the DRG system to be robust towards errors in the data bases. The procedure used in many countries for testing the US DRG definitions with national data found in data bases is also considered to be robust towards errors. Data quality in the studied data base is sufficient for DRG research purposes. The factors giving the robustness are of a general nature and may also apply to other data bases. Quality of computerized data is a less critical factor for DRG than for other applications of such data. If the present data base had been used in financial connection the errors would have had almost no deteriorating effect on hospital incomes.