Accuracy of adverse-drug-event reports collected using an automated dispensing system

The accuracy of adverse-drug-event (ADE) reports collected using an automated dispensing system was evaluated.

ADE reports were collected by requiring nurses on five units in a tertiary care facility to select a reason for removing two tracer drugs (dextrose injection 50% [D50] and naloxone) from an automated dispensing system (Medstation 2000, Pyxis, San Diego, CA). The accuracy of the ADE reports during a period of 4.5 months was evaluated through retrospective chart review. The sensitivity, specificity, positive predictive value, and negative predictive value of the reports were calculated.

A review of 61 D50 transactions found that the appropriate reason for removal was selected by nursing staff 62% of the time. Twenty-seven transactions were recorded as occurring due to an ADE, and 70% of these were confirmed in the medical record. The sensitivity and specificity of the ADE reports for D50 were 55.9% (95% confidence interval [CI], 39.2-72.6%) and 70.4% (95% CI, 53.2-87.6%), respectively. A review of 32 naloxone transactions found that nurses correctly selected the reason for removal 88% of the time. Twenty-three transactions were recorded as occurring due to an ADE, and 87% of these were confirmed in the medical record. The sensitivity and specificity of the ADE reports for naloxone were 95.2% (95% CI, 86.1-104.4%) and 72.7% (95% CI, 46.4-99.1%), respectively.

A Pyxis ADE reporting mechanism using the tracer drugs D50 and naloxone increased the overall reporting of ADEs.