Video-recognition of Erbitux®, Cisplatine and Navelbine® vials: Validation of the vial identification process under real conditions of use
5 October 2012B. Dalifard, M. Dermu, B. Le Franc, E. Griesemann, J-P. Cante Groupe Hospitalier La Rochelle Ré Aunis
As the result of a partnership with the L3I laboratory, the Drugcam® project is now available at the Hospital of La Rochelle. Drugcam® is the first prototype of a system for monitoring the preparation of injectable anticancer drugs in real time, offering the option to review recorded data later on. Drugcam® currently makes it possible to specifically identify 500mg Erbitux®, 100mg Cisplatine and 50mg Navelbine® vials, and to differentiate between all other injectable speciality products available on the market.
The purpose of this study is to validate the identification of the vials by Drugcam® under real use conditions.
Preparation processes using three different specialties identified by Drugcam® were carried out in the isolator, under the control of a pre-recorded computer program. In order to determine the specificity, sensitivity and accuracy of the method being used the number of vials identified as false positives and false negatives were recorded by the traceability module.
In all, 747 vials (full and sealed or opened and partially used and/or fitted with a needle-free draining device were analysed by the system (7 processes carried out, 107 vials on average by process, 22 minutes by process). No false negative or false positive were detected for Erbitux®. No false positive and 5.6% false negatives were noted for Cisplatine. No false positive but 8% false negatives were noted for Navelbine®.
The system shows excellent (100%) specificity. Sensitivity needs improving in order to address the vials detected as false negatives. In any case, false negatives are specifically recognised by the video system at the third camera detection. Average detection and analysis time is 0.5 second, which has a very minor impact on drug processing time and no effect on the quality of the preparation. A review of false negatives with the software design team shows that the issue is related to the detection threshold, which can be adjusted to correct the behaviour of the system.
If no single identification marking is available on the original product containers, video analysis makes it possible to identify the drugs and alert the manipulator, should a mistake occur during the preparation process. A new Drugcam® functionality for detecting syringe volumes is currently under evaluation.