Representativity of dispensations of non-steroidal anti-inflammatory drugs (NSAIDs) in the German Pharmacoepidemiological Research Database

Background: Claims databases warrant investigation of their validity and representativity for research purposes. NSAIDs belong to the most frequently dispensed drugs in Germany. Due to the withdrawal of the Cox-2 selective NSAID rofecoxib in 2004, major changes in NSAID drug prescription habits were to be expected. Objectives: To compare representativity of drug dispensations in the German Pharmacoepidemiological Research Database (GePaRD) with data from all statutory health insurances (SHI) in Germany before and after withdrawal of rofecoxib. Methods: GePaRD consists of claims data from four German SHIs covering about 14 million insurants throughout Germany (approximately 17% of the German population. For the years 2004 to 2006, NSAID dispensations in the database were compared to data from the German Drug Prescription Report (Arzneiverordnungs-Report) which is based on reimbursement of drug dispensations by all German SHIs. The number of defined daily doses (DDDs) was compared on the level of chemical substance as well as for coxibs as a group vs. traditional NSAIDs (tNSAIDs). Results: In both data sources, the number of coxib dispensations declined markedly and amounted 2006 to one third of the dispensations in 2004. The concomitant increase in dispensations of tNSAIDs was more pronounced in the database. On the level of chemical substance, time trends were concordant for some substances (e.g. ibuprofen), but showed different patterns for others e.g. meloxicam). The proportion of DDDs for each substance within class of NSAIDs and the resulting order were in good concordance between both data sources. In 2006, diclofenac was with 59% (German Drug Prescription Report) and 57% (database) by far the most frequently prescribed drug, followed by ibuprofen. Conclusions: Overall, data from GePaRD.were sufficiently representative for Germany and enabled identification of changes in prescription patterns to a similar extent as the overall data from Germany. Some differences, especially regarding rarely prescribed drugs, need further examination.