Evaluation of statistical power function for various diclofenac bioequivalence trials with different subject numbers

• Published in: European journal of drug metabolism and pharmacokinetics Vol. 34; no. 2; pp. 85 - 91
• Main Authors: Popović, Jovan, Mikov, Momir, Sabo, Ana, Jakovljević, Vida
• Format: Journal Article
• Language: English
• Published: Paris Springer-Verlag 01.04.2009
• Subjects: Administration, Oral; Adult; Analysis of Variance; Anti-Inflammatory Agents, Non-Steroidal - administration & dosage; Anti-Inflammatory Agents, Non-Steroidal - pharmacokinetics; Biological Availability; Biomedical and Life Sciences; Biomedicine; Cross-Over Studies; Data Interpretation, Statistical; Delayed-Action Preparations; Diclofenac - administration & dosage; Diclofenac - pharmacokinetics; Female; Human Physiology; Humans; Male; Medical Biochemistry; Models, Statistical; Pharmaceutical Sciences/Technology; Pharmacology/Toxicology; Pharmacy; Sample Size; Therapeutic Equivalency; analysis of variance; diclofenac; Power; bioequivalence
• ISSN: 0378-7966; 2107-0180
• DOI: 10.1007/BF03191156

Summary This study presents application of statistical power function for the t -test and ANOVA F -test on the evaluation of diclofenac bioequivalence in trials with the wide variations in sample sizes (N=12, 18 and 24). The power function, together with appropriate equations tables and figures, is used to calculate the power of the ANOVA for crossover design, the number of subjects for a given value of power and the minimum detectable difference in treatment means for different pharmacokinetic parameters of the formulations. The power of the trial with a small, sample size (N=12) to detect 20% differences between diclofenac formulations is shown to be more than 0.9 and almost the same as the power of the trial with a large sample size (N=24). In all trials for all pharmacokinetic parameters the power to detect 20% difference is shown to be more than 0.8. For the power of 0.8, the needed subject number to detect 20% difference in treatment means is the same or smaller than used and the minimum detectable difference is smaller than 20% in all our trials. This investigation shows that bioequivalence studies with small number of subjects (N=12) may be quite adequate for valid conclusions.