In a separate (published) study, a sorbitol oral solution (60 mgsorbitol) was bioinequivalent to risperidone tablet, with AUC and Cmax confidence intervals falling outside lower the bound (90% CIs 75.0–92.6 for Cmax and 76.2–92.9 for AUC) [20]. Garcia-Arieta reports that the bioequivalence failure in each of these studies was due to a few subjects who appeared to be outliers in terms of the test:reference ratio, and pulled down the lower bound of the mean confidence intervals; if these subjects were excluded, the comparisons passed the bioequivalence criteria. This is attributed to greater sensitivity to the effects of sorbitol on intestinal transit for these subjects.