Coding accuracy of administrative drug claims in the Ontario Drug Benefit database.
Journal Articles
Overview
Research
Identity
Additional Document Info
View All
Overview
abstract
BACKGROUND: Every year in Ontario, the records of over 42 million prescriptions dispensed to persons eligible for Ontario Drug Benefit (ODB) benefits are transmitted to a central database. The ODB database is the second largest database of medications in Canada, containing records on almost half of all medications dispensed in Ontario. There is no information about the reliability of the coding on the ODB drug claims database and, therefore, the objective of this study was to estimate the reliability of coding of the Drug Identification Number, and the date, quantity and duration of the dispensation on claims sent to the ODB. METHODS: To meet this objective, approximately 100 randomly selected prescriptions dispensed from each of 50 pharmacies in southern Ontario between July 1, 1998 and December 31, 1999 were audited. For each claim, the written information on the prescription was compared with the electronic information submitted to the ODB database. Logistic regression was used to test the association between coding errors and the location, owner affiliation, and productivity of each pharmacy (defined as the annual volume of dispensations divided by the annual number of hours worked by all pharmacists and pharmacy assistants). RESULTS: Of the 183 pharmacies owners invited to participate, consent to abstract information was obtained in 50, yielding a participation rate of 27%. Of the 5155 dispensed prescriptions, 37 errors were found, yielding an overall error rate of 0.7% (95% CI 0.5% to 0.9%). None of the characteristics of pharmacies that were examined (location, owner affiliation, productivity) was associated with coding errors. CONCLUSIONS: Pharmacists almost always dispense the medication that is prescribed and this information is reliably transmitted to the ODB drug claims database. This means that any conclusions drawn by researchers using these data are not likely to be compromised by low coding reliability.
