Medication errors detected in non-traditional databases: Types of errors in methotrexate dosing as listed in four different Danish registers

© 2015 Springer-Verlag Berlin Heidelberg. Aims: We have looked for medication errors involving the use of low-dose methotrexate, by extracting information from Danish sources other than traditional pharmacovigilance databases. We used the data to establish the relative frequencies of different types of errors. Methods: We searched four databases for cases involving low-dose methotrexate between 1999 and 2011: the Danish Patient Safety Databases (DPSD), controlled by the Danish National Agency for Patients' Rights and Complaints, the Patient Compensation Association (PCA), the Danish Poison and Information Centre (DPIC), and the online database of the Department for Patient Complaints (DPC). We categorized the place where the error occurred, the processes and types of error involved, the person responsible, and the clinical outcome. Results: We identified 173 errors. In 109 (63 %), either harm resulted or could not be excluded; of these, 26 (15 %) resulted in serious harm, including nine deaths (5 %); 53 (31 %) involved incorrect daily administration; and 107 (62 %) involved a dosing error. Sixteen events (9.2 %) concerned insufficient or faulty monitoring, with four serious outcomes and two deaths. Prescription errors involving daily rather than weekly administration, by hospital physicians, were most likely to result in serious outcomes, including deaths. The error mechanism was evaluated in 129 events. Action-based errors comprised 50 % and knowledge-based errors 34 %. Action-based errors were more likely to result in completed errors, whereas knowledge-based errors more often resulted in near misses. Conclusions: The medication errors in this survey were most often action-based (50 %) and knowledge-based (34 %), suggesting that greater attention should be paid to education and surveillance of medical personnel who prescribe and monitor methotrexate, particularly physicians, who accounted for 40 % of the errors.