Action limit outlier test: a novel approach for the identification of outliers in bioassay dose-response curves.

Guidelines like United States Pharmacopeia 1032 [1] and pharm.Eur. [2] acknowledge that cell-based bioassays are complex methods and thus prone to outliers. However, investigations into root causes of outliers are often inconclusive. We have established a procedure (including quality control and documentation) implemented in a freely available software application which includes not only the experience of the analyst but also information of historical data. This action limit outlier test is unique to our knowledge. Action limit outlier test allows the determination of outliers efficiently which lead to a significant reduction of false positives in comparison with the traditional outlier test ROUT [3] or Rosner [4] alone as shown by our simulated data (58 and 44% reduction of false positives for ROUT and Rosner, respectively).