Quality assessment of drug sales data: the case of antibacterials in Iceland

Background: Two sets of drug sales data, published by the Icelandic Ministr y of Health, did not match for antibacterials in 1989. The search for cause s turned out to be a project in itself. Objective: To analyze quality problems in the sales data on antibacterials and describe a method for systematic quality assessment of drug sales data. Methods: Documentary analysis based on the following sources: 1) Nordic Sta tistics on Medicines, 1975-95; 2) Drug Use (Notkun lyfja), 1975-93; 3) Icel andic Drug Market, 1975-94; 4) Unpublished data from the Icelandic Ministry of Health. The following framework was developed to evaluate the quality o f drug sales data: 1. Completeness of registration; 2. Accuracy and degree of completeness of data; 3. Size and coverage of the data source; 4. Data format; 5. Data accessibility, availability and cost. Results: Four discrepancies were found, two due to changes in DDD, and two larger ones stemming from errors in calculating DDD, resulting in an overes timation of the contribution of the respective products to the total DDD/10 00 inhabitants/day. Errors were detected in available sales data at least b ack to 1980, resulting in sales being overestimated by up to 13%. The reaso ns for the discrepancies were found mostly under point 2 in the framework. Conclusion: The errors uncovered by this study indicate a possible low qual ity of drug statistics which might lead to wrong conclusions about the leve l and development of sales of drugs. As a tool, the framework might be used for quality assessment of drug sales data.