Analysis of coding quality in Germany using health insurance billing data – utilization of qualifying characters in the documentation of diagnoses

Background: As a part of the health care reform 2007 the German risk structure compensation scheme was extended as to connect the financial cash flow towards the payers to morbidity information from outpatient care. The aim of this study was to evaluate coding quality via qualifying characters as well as to identify future challenges. Methods: The study sample included about 350 million diagnoses which were qualified as “assured” or “post-treatment” of about 11k practitioners' treatment of 2.7 million AOK PLUS insurants in Saxony and Thuringia (Germany) during the years 2007 to 2010. We focused on the following four ICD-10-GM Codes: I21, I25, I63 and I64. Results: The number of “post-treatment” diagnoses generally rose from year to year. In the first quarter 2007, 78,097 insurants got a post-treatment diagnosis and in the fourth quarter 2010 91,796. According to the insurants' view it has indicated that this caused not because of short but long coding traditions. 13,947 insurants (I21: 4,953, I25: 1,358, I63: 2,944, I64: 4,692) were identified with a total coding tradition (16 quarters). In particular, general practitioners tend to use this diagnosis codes qualified as “post-treatment” for conditions after certain events. The two sample t-test to the change of the mean coding quality suggest at p<0.01-level an improvement from 2008 to 2010. Conclusion: We identified a lack of diagnosis coding precision regarding the condition after certain medical events. Consequently, we expect adverse effects evaluating the cost of diseases as only “assured” diagnoses are considered within the risk transfer compensation scheme.