Background Chemotherapy preparation units have to face increasing activity with constant staff. Safety is therefore threatened.

Purpose The purpose of our experiment was to measure the effect of work overload on preparation accuracy and error.

Material and methods A simulation study using tracers (lidocaine and phenylephrine) was conducted in an operational context. 12 operators had to produce 1, 2 or 3 sets of 8 preparations in a fixed time of 1 h. For each series of 8 preparations, 4 syringes at different dosages and volumes, starting from 2 concentrations of stock solutions, were compounded for each tracer. Results were analysed according to qualitative (visual observation, choice of stock solution, diluents and label) and quantitative (validated CE methods; accurate: <5% deviation from the target concentration; weakly accurate: 5–10%; inaccurate: 10–30%; error: >30%) criteria.

Results A gradual reduction in preparation time, inversely correlated with workload, was obtained (4 min 11 s, 3 min 07 s and 2 min 35 s for sessions with 8, 16 and 24 syringes, respectivelyp <0.0001).

No difference in the accuracy of the doses was observed between the 3 levels of workload (p = 0.23, Cox model regression). The distribution of quantitative analysis for the production of 8, 16 and 24 syringes was as follows:

• accurate: 57%, 51% and 49%;

• weakly accurate: 26%, 25% and 32%;

• inaccurate: 16%, 23% and 17%; and

• error: 1%, 1% and 2%.

The observed error rate (qualitative and quantitative analysis) for the preparation of 8, 16 and 24 syringes was 1.1%, 2.1% and 4.5%, respectively. The difference in errors rates between the 3 levels was not statistically significant (mixed effects logistic regression, p = 0.15), possibly due to a lack of power.

Conclusion Our pilot study showed that operators are able to increase their working speed without impacting on dose accuracy. However, a large proportion of inaccurate preparations were observed and inclusion of robust control methods in the process is recommended. Acceleration of the manual production rate appears to be possibly associated with a greater probability of making a mistake, but this trend has to be confirmed in a larger sample size study.

No conflict of interest.