Robotic chemotherapy compounding: A multicenter productivity approach

Introduction The aim of this study is to compare productivity of the KIRO Oncology compounding robot in three hospital pharmacy departments and identify the key factors to predict and optimize automatic compounding time. Methods The study was conducted in three hospitals. Each hospital compounding workload and workflow were analyzed. Data from the robotic compounding cycles from August 2017 to July 2018 were retrospectively obtained. Nine cycle specific parameters and five productivity indicators were analysed in each site. One-to-one differences between hospitals were evaluated. Next, a correlation analysis between cycle specific factors and productivity indicators was conducted; the factors presenting a highest correlation to automatic compounding time were used to develop a multiple regression model (afterwards validated) to predict the automatic compounding time. Results A total of 2795 cycles (16367 preparations) were analysed. Automatic compounding time showed a relevant positive correlation (r(s)|>0.40) with the number of preparations, number of vials and total volume per cycle. Therefore, these cycle specific parameters were chosen as independent variables for the mathematical model. Considering cycles lasting 40 minutes or less, predictability of the model was high for all three hospitals (R-2:0.81; 0.79; 0.72). Conclusion Workflow differences have a remarkable incidence in the global productivity of the automated process. Total volume dosed for all preparations in a cycle is one of the variables with greater influence in automatic compounding time. Algorithms to predict automatic compounding time can be useful to help users in order to plan the cycles launched in KIRO Oncology.