Seven months experience with a robot for the compounding of sterile chemotherapy preparations: activity analysis, performance and limitations

Background
Automating chemotherapy compounding aims to improve safety, traceability and staff organization in centralized compounding units. We report seven months of experience with the integration of the Kiro Isolator^® robot (Grifols) into the compounding unit of a cancer center.

Materials and methods
Retrospective analysis covering July 2025 to January 2026. Data were extracted from the Kiro^® software, from Chimio^®, and from statistics supplied by Grifols, and covered activity, robot-compounded preparations, the drugs involved, non-conformities, training, and organizational impact.

Results
Of 19,797 preparations, 27.5% (n = 5,456) were compounded by the robot, with a median of 42 preparations produced per day (Q1–Q3: 30–50). Fifteen drugs were handled, mainly paclitaxel (n = 1,424), cyclophosphamide (n = 791), 5-fluorouracil in elastomeric pumps (n = 599) and doxorubicin (n = 528). Selection favored viscous products, large volumes, drugs requiring lengthy reconstitution, and drugs compounded in advance. The rejection rate was 0.86% (47/5,456), more than half of these being due to a gravimetric deviation exceeding ±5%. Four drugs (pembrolizumab, bevacizumab, cytarabine, Etopophos^®) were subsequently excluded, their cycles being inefficient and monopolizing the carousel. Three technicians were trained, each over one week.
Compounding time averaged 7.4 min per preparation, compared with 4.1 min using DrugCam^® with an experienced technician. The mean turnaround time for a released preparation, from start of pre-processing to end of post-processing, was approximately 60 min, reflecting operation in cycles of 6.8 preparations on average (median 7), each preparation being released only once the full cycle was complete. One episode of environmental microbiological contamination required a half-day sterilization of the isolator. Five days of breakdown occurred, two requiring technical intervention.

Discussion / Conclusion
This first experience confirms the feasibility of integrating the Kiro Isolator® into routine practice, with 27.5% of activity robot-compounded and drug selection guided by the expected compounding benefit. That benefit is primarily organizational: relief from the most demanding preparations, greater technician autonomy and the ability to compound in advance, with no gain in throughput, since the robot cycle time remains longer than that of manual compounding. Technical dependence and the contamination episode underline the need for a business continuity plan and close microbiological monitoring. Further optimization will rely on rigorous selection of the drugs that are genuinely efficient in cycle mode, and on continuous performance evaluation.

See the poster