Bioequivalence and Bioavailability Forum

Dear Valen!

Unfortunatelly I don't have access to v2.1 (the ANOVA module was replaced by the BE module [Linear Mixed Effects Modeling] in v3.2).

Maybe some remarks are helpful:

• you should ln-transform your data, resulting in
  Bioequivalence Limits: (A.H.)Lower= 0.800 (A.H.)Upper= 1.250
  
• in a 3×3×3 design (as well as in a conventional 2×2) carry-over is confounded with other effects; therefore carry-over generally is not included in the model (lack of carry-over is verified by taking a plasma sample prior to administration in all periods, which should show concentrations <5% C_max).
  
• for drawbacks of a 3×3×3 design have a look at this post and the followings.

The statistical test for carry-over^[1] is considered obsolete.^[2,3,4]

BTW, what data does your variable Carryovr contain?

References

1. JE Grizzle
   The two-period changeover design and its use in clinical trials
   Biometrics 21, 467-480 (1965)
   
2. PR Freeman
   The Performance of the Two-Stage Analysis of Two-Treatment, Two-Period Crossover Trials
   Statistics in Medicine 8, 1421-1432 (1989)
   
3. S Senn
   Cross-over Trials in Clinical Research
   2^nd Edition, Wiley, Chichester, pp 35-88/295-322 (2002)
   
4. Jones B and MG Kenward
   The 2x2 cross-over trial. Consequences of preliminary testing, in:
   Design and Analysis of Cross-Over Trials
   2^nd Edition, Chapman & Hall, Boca Raton, pp 44-50 (2003)