The Comparison Of Statistical Property Of Three 2×4 Crossover Designs In Bioequivalence Test

Object: Bioequivalence study is a key component of evaluating the consistency between quality and efficacy of generic drugs,and which bioequivalence trial design is recommended to carry out the equivalence tests is an urgent problem to be solved.For the 2×4 crossover design,although the FDA gave the recommended sequence of medication(TRTR-RTRT),some researchers displayed that TRRT-RTTR was the optimal 2×4 crossover design.Therefore,the purpose of this study is to compare the statistical properties of TRTR-RTRT,TRRT-RTTR,and TTRR-RRTT,and to provide references for bioequivalence clinical trials.Methods: Based on the two bioequivalence evaluation methods recommended by EMA and FDA,considering whether carryover effect exists,comparing three 2×4 crossover designs under different conditions(including formulation effect,carryover effect,sample size,and within-subject variability).Type I error,power,and coverage of confidence interval are used as indicators to investigate statistical properties of three bioequivalence trial design methods.Results: The results of three crossover designs have very little difference when the carryover effect does not exist.No matter which methods in ABEL and RSABE is used,the results are similar when the carryover effect exists.In terms of the type I error and power,if the carryover effect is contrastive to the formulation effect of the previous period( < 0),the type I error of TTRR-RRTT gradually inflates with the increase of ||,at the same time,the power increases.While the type I errors of TRTR-RTRT and TRRT-RTTR tend to be conservative with the increase of ||,and the power decreases;If the carryover effect is assimilative to the formulation effect of the previous period( > 0),the type I error of TTRR-RRTT tend to be conservative with the increase of ||,at the same time,the power decreases.While the type I error rates of TRTR-RTRT and TRRT-RTTR inflated with the increase of||,and the powers increase.In terms of confidence interval coverage,the coverage rate will not be affected by the carryover effect,but ||.The confidence interval coverage of the three crossover designs decrease with the increase of ||,and TRTR-RTRT has the largest decrease.When || = 0.1,the coverage rates of TRRT-RTTR and TTRR-RRTT are still maintained at around 90%,and TRRT-RTTR is slightly higher than TTRR-RRTT when the sample size is relatively large.Conclusions: When there is no carryover effect,the results of the three 2×4 crossover designs are the same.When carryover effect is exists,TRRT-RTTR is recommended.With the influence of the carryover effect,TRTR-RTRT and TTRR-RRTT is easy to draw conclusions that are paradoxical to the facts.It is worth noting that TRRT-RTTR is only applicable when the carryover effect is small(such as || = 0.1).If the carryover effect is large,none of the three crossover designs can reveal the true average bioequivalence.