Background: About 26,000 chemotherapeutic batches were produced in 2003 in the Institute Gustave Roussy and 83% were qualitatively and quantitatively assessed in post-production controls via an analytical platform. The rate of non-conformity (outside the specification limits of the target concentration +/-10%) decreased from 8.9% to 2.2% between years 2001 and 2003. A cost- and time-saving acceptance sampling plan was applied to assay fewer batches whilst maintaining an accurate estimate of the quality level.
Methods: The opportunity to apply a single sampling plan by attributes with an acceptance quality level of 2.2% was evaluated. A prognostic study using a logistic regression model was performed for some drugs to identify risk factors associated with the non-conformity rate of preparations.
Results: Out of 26 drugs, 17 have not been sampled, since they were prepared less than 400 times per year. For six drugs, a reduction of about 50% in the number of assays was estimated. Three drugs were "at risk" of being non-conform: for these drugs, all batches were analysed.
Conclusions: The sampling plan allowed a reduction of almost 8000 analyses with respect to the number of batches analysed for 6 drugs. For the 3 drugs with the higher risk to be non-conform, associated risk factors were identified to set up corrective actions.