Bioequivalence and Bioavailability Forum

Hi John!

❝ Can you point me to some of those publications?

One example is “Method A” of Potvin’s orginal paper. Although the algo contains an intermediate power calculation (which could lead to stopping in the first stage) α is not adjusted (i.e., kept at 0.05 in all steps). In other words, it is a – little bit more conservative – add-on design. However, they faced inflation. Quote:

The type I error for Method A is seen to be inflated in certain scenarios. For example, with a first stage of 12 subjects, the inflation in type I error rate is up to about 16% (to 0.058). […] Because Method A clearly did not satisfy our requirement for at most a neg­li­gible in­crease in type I error rate, we […] did not consider Method A further.

Hauck et al.¹ reported a type I error of up to 0.0614 for medium variability (CV 24%). Cui et al.² showed an example, where the actual type I error was 0.037 (at a nominal α of 0.025). That’s ~50% inflation. Coffey and Muller³ wrote about internal pilot studies:

Unfortunately, for some values […] the design may nearly double α_t.

Note that Ref’s 2&3 dealt not with BE in a cross-over design but with superiority testing / parallel groups.
For the Japanese Add-On I would expect ~50% inflation since the second group is ½ of the first and no adjustment is done.

❝ ❝ The only regulation where Add-on designs are still acceptable is the Japanese.

❝ This is surprising since the Japanese are so strict on quality.

Two years ago at the BA/BE-conference in Kobe I tried to discuss this topic. By my standards I was very polite and started by praising their new guidelines. Next I talked about patient’s risk, which is compromised in their Add-On approach. I tried to explain it in simple words, like “The entire α is already ‘spent’ in the failed study. Nothing is ‘left’ for the add-on part. Any additional evaluation will inflate the patient’s risk.” The head of the Japanese authority replied: “Thank You!” I waited a good while for something more to come, but she gave me nothing else than a Siamese smile.

❝ When I was visiting Japan, I noticed that their groceries (packaged items) have very short expiry dates (like 2 weeks)...

What about?

---

1. Hauck WW, Preston PE, Bois FY. A group sequential approach to crossover trials for average bioequivalence. J Biopharm Stat. 1997;7(1):87–96. doi:10.1080/10543409708835171.
   
2. Cui L, Hung HMJ, Wang S-J. Modification of Sample Size in Group Sequential Clinical Trials. Biometrics. 1999;55(3):853–7. doi:10.1111/j.0006-341X.1999.00853.x.
   
3. Coffey CS, Muller KE. Controlling Test Size While Gaining the Benefits of an Internal Pilot Design. Biometrics. 2001;57(2):625–31. doi:10.1111/j.0006-341X.2001.00625.x.