Abstract Two automation methods for aseptic preparation in hospital pharmacy, robot and peristaltic pump, were compared to manual process both for risk analysis using Failure Modes Effects and Criticality Analysis (FMECA) method and for productivity using time analysis grids built for each process. The results obtained with the different workflow organizations showed that the worst-case conditions for productivity was production “on demand” of tailor-made preparations. in that case, the manual process was not significantly different from the robotic process (p-value=0.72). For the standardized preparations, the semi-automatic process preparing a batch from bulk solution from “to be reconstituted” drugs was significantly superior to the robotic process preparing repetitive series of doses (p-value<0.01). Productivity of the robot was dramatically increased when the robot performed standardized preparations either from ready to use solutions or mixed cycles due to the robot design. When different processes were FMECA analyzed for risk analysis the robotic process was found as the safer process in comparison to others with a total of Criticality Indexes of 1060, 719, 656 for manual, semi-automatic and robot, respectively. Except for the robotic, semi-automatic and manual processes needed additional IT control systems to limit the risk of failures.

The aim of this study was to develop a specific simulation program for the validation of a cytotoxic compounding robot, KIRO® Oncology, for the preparation of sterile monoclonal antibodies and anti-infectious drugs. The impact of excipient formulations was clearly measured using simulation accuracy tests with worst case excipient (i.e. viscous, foaming) and allowed to correct the robotic settings prior to real production. Corrections brought accuracies within the acceptable range of ±5%. KIRO® Oncology robot has also the capacity of self-cleaning and a simulation combining media fill test, and environmental monitoring was able to validate the aseptic process including simulation of worst case conditions and highlighting the areas not accessible to self-cleaning to be corrected by additional manual cleaning measures. The risk of chemical contamination was simulated by using fluorescent dye of the process with high-risk excipient formulation and overpressure vials. Quality control reliability was simulated by using a model drug, and final concentration was determined by high-performance liquid chromatography-ultraviolet detection. Finally, productivity was simulated using different models of production showing the impact of the type of drug, the number of vials and the poor standardization of the process.