Bayer's OOS oops
The FDA cited Bayer HealthCare for an analytical anomaly at a facility in Bergkamen, Germany, in an August 5, 2009, warning letter (WL: 320- 09- 09). Bayer had established procedures allowing for the averaging of out-of-specification (OOS) and within-spec analytical test results from separate samples, resulting in the release of API batches based on passing averaged assay results.
The FDA concluded that the Bayer method fails to ensure that manufactured APIs meet established specifications. It listed six batches that the company had released based on reportable assay results obtained from the average of two independent sample results. One of the sample results was out of spec while the other was within. The averaged passing reportable assay result was compared against the established specifications, and the batches were released to the marketplace.
The FDA disagreed with this use of averaging. "An assay test is used to determine potency, not method variability. The validation of your analytical method should address robustness or variability, while system suitability is designed to address instrument variation performance, which was met in each of these instances. We believe that these results were true OOS values and that these batches should not have been released for distribution."
Bayer pushed back in a response letter, stating that its averaging technique is appropriate and in line with the regulator's OOS guidance document. The FDA again disagreed. "Your firm prepares two to three separate samples, which are assayed individually. We expect you to treat each of these results independently, and not to average an OOS result with a passing individual result. The hiding of an OOS result in the average is an unacceptable practice."