7 Pages

Appendix V: Outlier Tests and Chemical Assays

In a recent landmark decision resulting from a trial involving the Federal Government and Barr Laboratories, Judge Wolin made many judgments based on his constant probing and the testimony of expert witnesses [1]. Remarkably, most of what he had to say was clear, correct and to the point, despite his sparse background in the subject material. Much of the Decision related to testing drug products during their production when failing results (out of specification) were observed. A summary of the Decision is available from the FDA [2]. A previous paper by this author [3] presented some alternatives to retesting when a single out of specification result was observed for which no obvious cause was apparent, a situation which is common in my experience. This paper discusses some issues related to the elimination of an out of specification (OOS) result with no obvious cause, based on an outlier test. The Judge, in his Decision, stated that tests for outliers which can be used to exclude an aberrant observation are not appropriate for chemical tests. His reasoning was that the USP includes tests for outliers, but presents these tests only in the context of biological assays, which tend to be very variable. This, he suggests, is appropriate because of the large variability of these kinds of procedures. Judge Wolin further suggests in his Decision that such outlier analyses should not be used for chemical assays, because if they were appropriate, the USP would have recommended the procedure for chemical assays. Thus, the judgement is that, by default, outlier tests for chemical assays should not be used. All of this raises several questions, including (a) Was it the USP’s intention to exclude outlier tests for chemical assays? (b) Was this an oversight or was it intentional? (c) Does the USP not discuss outlier tests for chemical assays because the issue is complex with many possible alternatives?