Bioequivalence and Bioavailability Forum

Hi Sree,

A good place to start is the book by Chow and Liu "Design and analysis of bioavailability and bioequivalence studies".
It describes a way of doing the BE-ANOVA without model fitting, and it also -and this is where I think you easily could gets started with Excel- tells how to derive the statistics without ANOVA, but using t-test in stead.
As it happens, the two methods should ideally give the same results ( in terms of p-values), but they do not in case of imbalance, where the t-test approach gives you p-values that correspond to ANOVA based on model fits with type III sums of squares and they are generally accepted.
All in all, it is a jungle out there. In order to calculate your BE-statistics for a standard 2,2,2-BE study you can thus use a linear model* and play with factor inclusion to obtain the ANOVA (with for example type III SS), or you can do the ANOVA without a model fit by way of the reference above, or you can do it all without ANOVA and just do a series of t-tests (and get p-values corresponding to type III SS).

Best regards
EM.


*: You can even use a (linear) mixed model with subjects as random factor.