Bioequivalence and Bioavailability Forum

Since the last meeting of ASEAN BA/BE taskforce in Brunei, 29 Jul 2008. They mention for BE study reporting format in section 9.4 Statistical analyses that

"Summary of statistical significance for AUC and Cmax (based on log-transformed data calculated as 90 % CI of test/reference Geometric Means) and for tmax (based on non-transformed data calculated as p value)"

I'm in Thailand, under ASEAN, I'm clear about AUC and Cmax but for tmax I'm not. As everyone knows statistical inference for tmax, we use nonparametric method with non-transformed data, I have some questions following:

How to calculate p-value? Because we use crossover design we should concern about sequence effect. By Wilcoxon method, we treat data as parallel and the Wilcoxon hypothesis is testing for equality (two-sided test) not for equivalence (two one-sided tests). If I misunderstand, please correct me else please suggest other method or some reference books. (I have the 2nd ed. Design and Analysis of BA/BE by SC. Chow and JP. Liu, they suggest three nonparametric methods but there are no method which present result in exact p-value.)