Bioequivalence and Bioavailability Forum

Dear Eduardo!

❝ Does anybody know how to use SPSS to calculate everything from AUC to TOST?

I don't have any expericence with SPSS, but I guess you easily will get 'lost in cyberspace' (which was my impression from STATISTICA).

❝ or do you recommend another software for this purpose?

A cheaper option is NCSS, which comes for USD 500.

If you already have point estimates and CVs, StudyResult from CreoStat for USD 80 is an option.

Specialized software for BE:
EquivTest from Statistical Solutions for about USD 800. EquivTest is easy to use, but I am not sure whether anything except 2×2 designs is covered. At least the graphical capabilities of my version 2.0 are limited, on the other hand it includes a nonparametric method (which is mandatory for the assessment of t_max in many countries).
A very comfortable system is WinNonlin in the current version 5.0.1 from Pharsight. Almost everything you might need is covered, from noncompartmental PK to the analysis of replicate designs.
Drawbacks: no nonparametric analysis, the price (USD 4000+), and an annual license fee of USD 1000+. If you have to deal with Pharsight's support, you should prepare steady nerves.
Another sytem is Kinetica (v.4.1.1) from Thermo Electron. Similar to WinNonlin in its capabilities and coming for USD 3500 (+annual licence USD 800). Population PK also included (which is only available as a separate product 'WinNonMix' from Pharsight). I think that the user-interface is somewhow 'odd', but you may download a trial version from Thermo's site to find out for yourself. No nonparametrical methods (like in WinNonlin), and acceptance ranges can only be chosen from 'US [0.80/1.25]' and 'Europe [0.70/1.43]', which is very strange (to be polite).

If you start from scratch and have a person around who is interested in some true programming, S+ from Insightful is a very powerful (yet expensive) tool.
A cheaper - or better, free of charge - implementation of the language is the community based freeware R. Remember S/S+/R are programing languages; although developed for statistical and graphical applications, there is no 'one-button-solution' for BE.
Unfortunately the 'learning curve' for both S+ and R is also quite steep.
In R you may use published S/S+ code with only minor (and often no) modifications.
On the other hand the user group of R is just phantastic, beating every commercial software support I know - often you get an answer within minutes!
As a starter I would recommend
Millard, S.P. and A. Krause (eds)
Applied Statistics in the Pharmaceutical Industry
Springer, New York (2001); ISBN 0-387-98814-9
As a teaser you may even download Chapter 7, which deals with the common 2×2 cross-over design.

Personally I would not recommend SAS, according to the license policy of the SAS Institute. Even if you can afford the annual license, you must add the personnel costs of an experienced SAS-programmer. Although SAS has improved over the years, the graphical capabilities are still poor - at least if compared to other software.