Background Our centralised cytotoxic preparation unit generates over 50,000 preparations per year, of which about 4% are standardised rounded dose preparations performed manually. In order to optimise the production and to increase volume and quality of these preparations, the unit acquired a semi-automatic Repeater pump.

Purpose To test the analytical performance of the pump by establishing a qualification protocol in order to integrate it in the current functioning of the unit.

Materials and methods The parameters assessed were linearity, precision (repeatability and reproducibility) and accuracy. To this end, transfers of solvent (5% glucose and 0.9% sodium chloride) were performed for different volumes: 1, 10, 20, 50, 100, 150, 250, 500 and 900 mL (which is the maximum volume used). Each volume was repeated six times and the reproducibility was studied on three different days. These volumes were determined by weighing. The linearity was evaluated by testing the slope, the coefficient of determination r² and the coefficient of variation (CV) of response factor. The acceptance levels of fidelity and accuracy were determined on the basis of our practice: CV less than 2% for fidelity and less than 5% for accuracy.

Results No difference was found between the two solvents: linearity was demonstrated throughout the range of volume. Repeatability was proved from 10 to 900 ml. For 1 ml volumes, although the CV was higher than the threshold set, it remained very low (2.6% maximum). Reproducibility was satisfied on the whole range of volume. We could however notice higher values of CV(but still within the threshold of 2%) for extreme volumes: 2% for 1 mL, 1,6% for 500 mL and 1.8% for 900 mL (versus 0.2–0.8% from 10 to 250 mL). Accuracy was validated from 20 ml to 900 ml (maximum relative bias = 3.8%). Results were not satisfactory for volumes of 1 ml (CV about 15%) and 10 mL (CV about 8%) but this did not affect the quality of our procedure which involved no volumes lower than 25 mL. These results should be considered in units performing transfers of volumes less than 20 mL. It should be noted that the performance of the pump has not been evaluated with liquid of higher viscosity than 5% glucose. We believe that it was not required since we distribute accurately only dilute solutions (0.04% to 0.8% [w/v] of active ingredient).

Conclusions To the standards set, the distribution of volumes with the Repeater pump was shown to be linear, repeatable, reproducible and accurate from 20 ml to 900 ml whereas it was insufficient for 1 and 5 mL volumes. This study guarantees the quality of our preparations with this new production process. It also opens up new organisational possibilities: currently, 15 different lots (20 to 45 units/lot) are prepared to achieve 10% of the total unit production activity.

No conflict of interest.