Optimizing information on drug exposure by collection of package code information in questionnaire surveys

PURPOSE: The thorough analysis of special drug characteristics requires information on the specific brand of a drug. This information is often not sought in pharmacoepidemiologic surveys although in many countries packages are labelled with an unequivocal code (in Germany called Pharmazentralnummer (PZN)). We aimed to assess the benefit and quality of PZN information collected in self-completed questionnaires. METHODS: We performed a survey in 905 ambulatory patients who were asked to list brand name, strength, and the PZN of all drugs they were taking. RESULTS: The medication list was completed by 97.5% (n = 882) of the responding patients (mean age 67.3 years). Altogether 5543 drugs (100%) were mentioned in the questionnaires and for 4230 (76.3%) the exact drug package could be allocated on the basis of the PZN. When PZN was considered in addition to the drug name the quality of drug coding was significantly improved (p < 0.001) with regard to the allocation of drug package (74% versus 2%), brand (90% versus 70%), and strength (96% versus 86%). The time needed for drug coding was three times shorter. CONCLUSIONS: The high response rate and high fraction of correct PZN indicate that the collection of package code information is a valuable method to achieve more accurate drug data in questionnaire surveys and to facilitate the drug coding procedure.