Background Drug dispensing is traditionally carried out manually, with a significant risk of errors. While medication preparation and administration accounts for 16% of nurses’ activity, more than a quarter of interruptions occur at these moments. Any distraction during these activities may increase the risk of errors.

Purpose Compare the rates of dispensing errors with and without an automated dispensing cabinet, and evaluate the influence of interruptions on the reliability of this activity.

Material and methods In a simulation environment, volunteer nurses had to prepare 12 pillboxes from a conventional pharmacy (CP, ScanCell^®) and an automated dispensing cabinet (ADC, Pyxis MedStation^®). Six standardised interruptions (INT) were generated: noise, discussion (x2), oral prescription, telephone call and physical intrusion.

The management of these distracting events were categorised (multitasking, task-switching, break of attention, suspending task, sub-optimal performance, no interruption). Errors were also classified (omission, wrong drug, dosage, patient, time).

The contribution of interfacing the ADC with the prescription was estimated.

Results A total of 2808 doses were prepared by 18 volunteer nurses.

With CP, the error rate was 4.13% (2.07% without INT, 2.07% INT), compared to 3.28% with ADC not connected to the prescription (1.28% without INT, 1.99% INT) (p=0. 112). With a connexion to the prescription, the error rate oscillated between 0.71% and 1.85% (p<0. 05). Wrong doses (CP:46%, ADS:34%) and wrong pharmaceutical forms (PC:42%, ADS:43%) were the most frequent errors.

The interruptions’ management were similar with the two systems in case of noise (no INT), oral prescription/telephone call (change of task to answer) and discussion (multi-tasks). During physical intrusion, 50% of the volunteers on ADC refused to be interrupted (8% on CP). The incidence of errors increased by 61% when interrupting tasks on ADC.

Conclusion With an average rate of 4% on a CP, errors are mainly related to dose confusion and lack of knowledge of pharmaceutical forms. This rate can be reduced with an ADC connected to the electronic prescription.

Task interruptions tend to increase the risk of error with ADC, but this effect can potentially be reduced once nurses become accustomed with this tool.

No conflict of interest