Sources of outlier data in the bioanalytical and clinical part of a piroxicam bioequivalence study

• Published in: International journal of clinical pharmacology and therapeutics Vol. 58; no. 11; pp. 652 - 663
• Main Authors: Sandulovici, Roxana, Mircioiu, Ion, Aboul-Enein, Hassan Y., Manolache, Mihai, Mircioiu, Constantin, Voicu, Victor, Anuta, Valentina
• Format: Journal Article
• Language: English
• Published: Munich Dustri - Verlag Dr. Karl Feistle GmbH & Co. KG 01.11.2020
• Subjects: Bioequivalence; Clinical trials; Hypotheses; Metabolism; Metabolites; Pharmaceutical sciences; Pharmacokinetics; Pharmacy; Physiology; Statistical analysis
• ISSN: 0946-1965
• DOI: 10.5414/CP203794

This paper analyzes the potential outliers in the bioanalytical and clinical part of a bioequivalence study, the effect on bioequivalence decisions whether or not it is appropriate to eliminate them from the statistical evaluation of bioequivalence.OBJECTIVEThis paper analyzes the potential outliers in the bioanalytical and clinical part of a bioequivalence study, the effect on bioequivalence decisions whether or not it is appropriate to eliminate them from the statistical evaluation of bioequivalence.The clinical part was a cross-over, two periods, two sequences bioequivalence study concerning two piroxicam formulations, on healthy subjects. A simulation study evaluated the influence of 10% errors on the percent bias of calculated concentrations from nominal ones.MATERIALS AND METHODSThe clinical part was a cross-over, two periods, two sequences bioequivalence study concerning two piroxicam formulations, on healthy subjects. A simulation study evaluated the influence of 10% errors on the percent bias of calculated concentrations from nominal ones.In bioequivalence studies, it is not possible to distinguish between relevant types of outliers based only on statistical criteria. The "problem" is particularly acute when the omission of outliers leads to a bias in the decision concerning bioequivalence from rejection to acceptance. In such cases, there is the suspicion of subjective analysis and torture of data. The effect of analytical errors at high plasma levels was criticized for the calculated concentrations in the neighborhood of lower limit of quantification. Errors at low concentrations have a less significant effect. In the pharmacokinetic analysis, several types of outliers were shown: single points, curves, pairs of curves corresponding to the same subject, intrasubject ratios of areas under curves and maximum concentrations. These pharmacokinetic outliers could have had, at the same time, bioanalytical, physiological and physicochemical causes.RESULTSIn bioequivalence studies, it is not possible to distinguish between relevant types of outliers based only on statistical criteria. The "problem" is particularly acute when the omission of outliers leads to a bias in the decision concerning bioequivalence from rejection to acceptance. In such cases, there is the suspicion of subjective analysis and torture of data. The effect of analytical errors at high plasma levels was criticized for the calculated concentrations in the neighborhood of lower limit of quantification. Errors at low concentrations have a less significant effect. In the pharmacokinetic analysis, several types of outliers were shown: single points, curves, pairs of curves corresponding to the same subject, intrasubject ratios of areas under curves and maximum concentrations. These pharmacokinetic outliers could have had, at the same time, bioanalytical, physiological and physicochemical causes.Considering the results, it was proposed the following algorithm in the analysis of outlier data and outlier subjects in bioequivalence studies: evaluation of the implications of the decision concerning elimination of outliers on the decision concerning bioequivalence; application of the statistic tests for detection of outliers data; evaluations from the point of view of physiological pharmacokinetics, final decision concerning elimination of outliers.CONCLUSIONConsidering the results, it was proposed the following algorithm in the analysis of outlier data and outlier subjects in bioequivalence studies: evaluation of the implications of the decision concerning elimination of outliers on the decision concerning bioequivalence; application of the statistic tests for detection of outliers data; evaluations from the point of view of physiological pharmacokinetics, final decision concerning elimination of outliers.