Carry-over effects in bioequivalence studies

Carry-over effects are often considered to be one of the main problems of the cross-over design: should we adjust for carry-over or not? We attempt to answer this question by examining the observed frequency of carry-over effects in actual bioequivalence studies. A total of 96 six-sequence, three-period, three-treatment fed-fasted studies are analyzed for carry-over effects and 324 two-sequence, two-period, two-treatment fasted studies are analyzed indirectly for carry-over effects via sequence effects. Two log-transformed pharmacokinetic variables, Cmax and AUC0-t, are modeled in an analysis of variance. The impact of statistically significant carry-over effects on bioequivalence results is examined and the rationale behind not adjusting for carry-over in bioequivalence studies is discussed.