Can we use video preparation control system for performance management and training programs?

Introduction
The Drugcam^® (DC) video control system was installed in 2017 on the Pharmacotechnical unit to monitor the stages of chemotherapy preparations.

Objective
Evaluate the DC system as a performance management tool (number of errors and speed of completion) in order to adapt the preparation training program (initial and ongoing).

Method
Descriptive analysis of DC raw data since January 2018. For each prepared molecule, a preparation speed interval target was determined (median time ± 20%) and then each preparation was identified ("normal" (belonging to the speed interval), "slow", "fast"). The number of errors identified per DC (vial, sample, packaging, label) was noted, enabling the identification of each preparation (with or without errors). An additional analysis, integrating manual indexing (MI) as an error was performed. The results were interpreted according to the seniority of the pharmacy technicians (PT) when handling with DC.

Results
40236 preparations were made by 17 PTs. For the PTs whose seniority was from 4 to 21 months, 43±2% of the preparations performed by one PT were performed with a preparation time in the "normal" speed range (26±2% "fast"; 31±2% "slow"). On average, 76±3% of the preparations performed by a PT had no errors identified by DC (77% performed at normal speed, 71% slow, 84% fast). When considering the MI, 60±4% of the preparations were classified without errors (60% normal speed, 50% slow, 72% fast).
To reach this performance target, an average of 3 months is required.

Discussion
Learning DC takes several months to reach a performance plateau. The training path has therefore been adapted accordingly.
Less than half of the preparations were completed within the target time frame. Preparations classified as "slow" as opposed to "fast" are related to a higher proportion of errors and a recent training program. Studying the profile of these errors allowed us to better target and illustrate training points.
Almost a quarter of the preparations were identified by DC as having at least one error. These errors are of different types: poor detection, real error... MIs are a source of error as such, removing the security of automated recognition. Therefore, they should not be neglected while remaining attentive to the corrections of MI sources (e.g. parameterization defect) which can lead to a preparation error by removing vigilance.

Conclusion
DC is a good performance management tool. Indeed, with the implementation of indicators, it allows to follow the production process and to adapt the initial and continuous training plan of thePTs.

See the poster