Abstract

Up to now, little research has been done to analyse how often the date of the prescription and the dispensing of the drugs are correct in German statutory health insurance claims data. We used data of the Gmünder ErsatzKasse (GEK). Out of 6.99 million prescriptions in 2005 we drew a simple random sample of 1,000. Dates of filling and dispensing on the scanned original prescription (SOP) were validated against that information in computerized claims data. Factors associated with nonagreement were also evaluated. In 90.4 % (95 % CI: 88.5-92.1) the SOP date of prescribing was identical to the information in the claims data. Compared to printed dates, handwritten or stamped dates were less frequently recorded correctly (95.4 % vs. 40.4 %; p<0.0001). In 76.6 % (95 % CI: 73.8-79.1) of the SOPs the dispensing date was consistent with the information in the claims data. Compared to good legible prints, moderately legible dispensing dates were less frequently recorded correctly (80.4 % vs. 68.2 %; p<0.0001). Because of weak printing about one third of the dispensing dates were moderately legible. In cases where drug exposure has to be measured accurately, original prescriptions should be viewed. There is further need for clarification on the correctness of variables in claims data.