[Model project on coding accuracy].

Abstract

First experiences with the external evaluation of coding accuracy in view of the German DRG-System are reported. 387 randomised inpatient cases of three departments of a municipal hospital were evaluated. 1.648 diagnosis codes and 946 procedure codes were evaluated with complete clinical data. Before and after correction by the reviewers the cases were grouped (AR-DRG 4.1) and the casemix index of the sample was calculated. 45.9% to 56.7% of primary diagnosis in the department samples were rated as correct. 25.2% to 37.5% of secondary diagnosis were rated as correct, 8.3% to 14.2% were corrected and 49.2% to 60.5% were rated as not relevant with regard to the German coding standards. 7.2% to 22.7% of secondary diagnosis had to be completed in the data. Evaluation of procedure codes resulted in 54.2% to 65.7% accepted codes, 5.9% to 12.1% corrected codes and 23.1% to 39.9% not accepted with regard to the German coding standards. 30.8% to 37.0% of procedure codes had to be completed in the data. After review, remarkable shift in DRGs was seen and casemix index increased 6.9% in average (0.25-12.1%). General and department-specific implications for improvement of coding accuracy could be evaluated. Consequences of potential coding errors in a prospective payment system on DRG basis were seen under conditions of daily hospital practice. External evaluation of coding accuracy used in this study could be the methodological basis for further investigations on this topic.