Study on the residual contents of empty drug vials from intelligent dispensing robots and manual compounding of cytotoxic drugs in PIVAS
Hui Liu, Linke Zou, Yujie Song, Junfeng YanIntroduction
The residual drug content in empty vials post-compounding is a crucial indicator for assessing the accuracy of intelligent dispensing robots. Excessive residue not only reduces therapeutic efficacy but also increases occupational exposure risks during handling. This study aims to investigate the differences in residual drug levels between intelligent dispensing robotic and manual preparation methods for cytotoxic drugs.
Method
This experimental study was conducted in the Pharmacy Intravenous Admixture Service (PIVAS) of a provincial People's Hospital. Four cytotoxic drugs were prepared using both an intelligent dispensing robotic and manual preparation methods under a designed protocol. High-performance liquid chromatography (HPLC) was used to measure residual drug contents in empty vials. The residual proportions were analyzed and compared under various conditions.
Results
A total of 270 vials were analyzed, including 150 prepared by the intelligent dispensing robot and 120 from manual preparation. The residual drug content in vials prepared by the robot ranged from 1.050 to 29.450 mg, with a residual proportion of 0.73% to 4.90%. For manually prepared vials, the residual drug content ranged from 1.030 to 43.350 mg, with a residual proportion of 0.73% to 8.67%. The residual drug proportion was significantly lower for robotic preparation [2.23% (1.66%, 3.36%)] compared to manual preparation [3.02% (1.52%, 4.40%)] (P < 0.001).
Conclusion
This study identified differences in residual drug proportions between robotic and manual compounding. Under the study conditions, robotic compounding resulted in a lower overall residual proportion; however, the direction and variability of the differences varied among drug–packaging combinations. Differences were also observed among formulations with different reconstitution characteristics, stopper types, and container types. These findings reflect the overall performance of the two workflow and provide valuable evidence to support the clinical application and regulatory development of intelligent dispensing robot systems.