Drug Res (Stuttg) 2013; 63(02): 98-103
DOI: 10.1055/s-0032-1333296
Original Article
© Georg Thieme Verlag KG Stuttgart · New York

A Preliminary Model to Avoid the Overestimation of Sample Size in Bioequivalence Studies

E. Ramírez
1   Clinical Pharmacology Service, Hospital Universitario La Paz, IdiPaz, School of Medicine, Universidad Autónoma de Madrid, Madrid, Spain
,
V. Abraira
2   Clinical Bioestatistics Unit, Hospital Universitario Ramón y Cajal, IRYCIS, CIBERESP, Madrid, Spain
,
P. Guerra
1   Clinical Pharmacology Service, Hospital Universitario La Paz, IdiPaz, School of Medicine, Universidad Autónoma de Madrid, Madrid, Spain
,
A. M. Borobia
1   Clinical Pharmacology Service, Hospital Universitario La Paz, IdiPaz, School of Medicine, Universidad Autónoma de Madrid, Madrid, Spain
,
B. Duque
1   Clinical Pharmacology Service, Hospital Universitario La Paz, IdiPaz, School of Medicine, Universidad Autónoma de Madrid, Madrid, Spain
,
J. L. López
1   Clinical Pharmacology Service, Hospital Universitario La Paz, IdiPaz, School of Medicine, Universidad Autónoma de Madrid, Madrid, Spain
,
B. Mosquera
1   Clinical Pharmacology Service, Hospital Universitario La Paz, IdiPaz, School of Medicine, Universidad Autónoma de Madrid, Madrid, Spain
,
R. Lubomirov
1   Clinical Pharmacology Service, Hospital Universitario La Paz, IdiPaz, School of Medicine, Universidad Autónoma de Madrid, Madrid, Spain
,
A. J. Carcas
1   Clinical Pharmacology Service, Hospital Universitario La Paz, IdiPaz, School of Medicine, Universidad Autónoma de Madrid, Madrid, Spain
,
J. Frías
1   Clinical Pharmacology Service, Hospital Universitario La Paz, IdiPaz, School of Medicine, Universidad Autónoma de Madrid, Madrid, Spain
› Author Affiliations
Further Information

Publication History

received 20 September 2012

accepted 18 December 2012

Publication Date:
20 February 2013 (online)

Abstract

Often the only available data in literature for sample size estimations in bioequivalence studies is intersubject variability, which tends to result in overestimation of sample size. In this paper, we proposed a preliminary model of intrasubject variability based on intersubject variability for Cmax and AUC data from randomized, crossovers, bioequivalence (BE) studies. From 93 Cmax and 121 AUC data from test-reference comparisons that fulfilled BE criteria, we calculated intersubject variability for the reference formulation and intrasubject variability from ANOVA. Lineal and exponential models (y=a(1−e-bx)) were fitted weighted by the inverse of the variance, to predict the intrasubject variability based on intersubject variability. To validate the model we calculated the coefficient of cross-validation of data from 30 new BE studies. The models fit very well (R2=0.997 and 0.990 for Cmax and AUC respectively) and the cross-validation correlation were 0.847 for Cmax and 0.572 for AUC. A preliminary model analyses allow us to estimate the intrasubject variability based on intersubject variability for sample size calculation purposes in BE studies. This approximation provides an opportunity for sample size reduction avoiding unnecessary exposure of healthy volunteers. Further modelling studies are desirable to confirm these results especially suggestions of the higher intersubject variability range.