GMS Medizinische Informatik, Biometrie und Epidemiologie (Oct 2008)

Correctness of different prescription forms for narcotics and other drugs in German medication claims data in 2006

  • Pfannkuche, Matthias S.,
  • Glaeske, Gerd,
  • Hoffmann, Falk

Journal volume & issue
Vol. 4, no. 2
p. Doc07

Abstract

Read online

Background: Information on the developing process and on data validity is crucial when using health insurance claims data (on medications) for research purposes. In Germany, different prescription forms are available for narcotics (so-called 'Betäubungsmittelrezepte') and other medications (so-called 'Muster 16'). The aim of this study was to compare the correctness of the recorded research relevant information between them. Methods: We used data of the Gmünder ErsatzKasse (GEK) for the year 2006. Out of 7.2 million prescriptions we drew a simple random sample of 600 prescriptions for narcotics and 600 for other medications. Data on the scanned original prescription, which was used as a gold standard, were validated against that information in claims data. Results: Compared to prescriptions for other medications, dates of prescription (92.5% vs. 68.1%; p<0.0001) and dispensing (85.5% vs. 61.8%; p<0.0001) were recorded less correctly on prescriptions for narcotics. Significant differences were found between pharmacy data processing centres (Apothekenrechenzentren). Compared to all other pharmacy data processing centres, the 'Verrechnungsstelle der Süddeutschen Apotheken GmbH' (VSA), as the biggest institution, showed lower accuracy in recording dates of prescription (56.1% vs. 73.3%; p<0.0001) and dispensing (41.2% vs. 70.8%; p<0.0001) on prescriptions for narcotics. Errors in dates of dispensing at the VSA were associated with the variables prescription form (narcotics vs. other medications; Odds Ratio [OR]: 3.12; 95% confidence interval [CI]: 1.85–5.27), dispensing in claims data on date of prescribing (yes vs. no; OR: 9.33; 95% CI: 1.78–48.87) and date of prescribing on 5., 15. or 25. (yes vs. no; OR: 6.70; 95% CI: 2.85–15.76). Conclusion: Methodological research and validation studies of German statutory health insurance claims data, especially on diagnoses, are urgently needed in the near future.

Keywords