Instantaneous detection of nonadherence: quality, strength, and weakness of an electronic prescription database

The "personal electronic medicine profile" (PEM) is a Web-based tool for electronic prescription and monitoring of purchased medicine. It is based on the National Prescription Database and contains data on all prescriptions in Denmark. It includes information on time of drug purchase, number of tablets, and prescribed daily dosage. This allows calculation of the expected time for new purchases.

To study the accuracy of the PEM as a tool for monitoring drug nonadherence as compared with pill counts (PCs).

Five hundred eighty-three randomly selected elderly Danish citizens older than 65 years taking more than four drugs were studied. They were visited three times by a nurse who counted their medicine supply. Contingency table analysis was used to compare drug nonadherence calculated from PC with that revealed by PEM. For PC and PEM, an adherence level of at least 80% was defined as acceptable.

PEM could not accurately process (non)adherence in 44% of all drugs. The probability of identifying drug nonadherence with PEM was low (negative predictive value 23%). Incomplete prescription information (34%) and inaccurate dosage registration (10%) were the major sources of error.

PEM is inferior to PC for accurate monitoring of drug nonadherence. The inaccuracy is due to erroneous prescription information. PEM could be a powerful tool for electronic monitoring of drug nonadherence if prescription information was recorded uniformly and correctly. To increase the accuracy, we recommend informal free-text dosing instruction to be translated into a formal one by use of appropriate software such as library of phrases.