Comparison of average and population bioequivalence approach

Aims: We investigated the comparison of average bioequivalence approach and population approach using bioequivalence study data which have been reported. Materials: On MEDLINE, "bioequivalence" was entered as a key word to search in the 3 journals which were published between 1980 and 1989. Consequently, a total of 17 data sets on AUC and 12 data sets on C-max were obtained and analyzed in this review. Method: Assessment of average bioequivalence, assessment of population bioequivalence and assessment of inequality of variance (F-test) were conducted after all data were subjected to logarithmic conversion. Results: Of the data sets which were analyzed in this review, 11 data sets on AUC and 3 data sets on C-max passed the average bioequivalence criterion, and 13 data sets on AUC and 8 data sets on Cmax passed the population bioequivalence criterion. Two data sets on AUC and 1 data set on C-max passed the average bioequivalence criterion, but not the population bioequivalence criterion. Four data sets on AUC and 6 data sets on C-max passed the population bioequivalence criterion, but not the average bioequivalence criterion. The correlation coefficient (r) for the population bioequivalence value and difference in the average bioavailability was 0.412, while the correlation coefficient for the population bioequivalence value and the difference in bioavailability variances was 0.708. Conclusions: In this review using bioequivalence study papers which have been reported in references, the episodes to judge that the test formulation is bioequivalent to the reference formulation occurred more predominantly in the population bioequivalence approach than in the average bioequivalence approach, and population bioequivalence approach might be affected more extensively by the bioavailability variance rather than by the average bioavailability.