Bioequivalence and Bioavailability Forum

Dear all,

with the IR GL operational for almost three years now and the MR GL approaching (see also this post) I’m curious what is your practice. I’m talking only about formulations where t_max is clinical relevant (mostly rapid onset, rarely AE related).
We all know that “A statistical evaluation of t_max is not required.” Does that mean that it is not accept­able as well? I read that “A non-parametric analysis is not acceptable”, but

<nitpicking>
this paragraph is related to ANOVA and log-transformation.
</nitpicking>

“[…] there should be no apparent* difference in median t_max and its X between test and reference product.”
where X = “variability” (IR GL) and “range” (MR draft).

My practice (as it was since the mid 1980s):

• Descriptive statistics of t_max of the formulations, i.e.,
  x̃, 25/75 percentiles, minimum, maximum; box plots.
  
• Ignore the “not acceptable” statement and perform a nonparametric test (generally I state an acceptance range of ±something in the protocol).
  Calculate PE and ~90% CI.
  
• Report no relevant difference in t_max, CI including zero, and similar

Experience so far: no problems.

Questions:
1. What do you do now? _____________
2. Do you consider changing your practice in the near future? [ ] yes [ ] no
3. If yes, what are you planning to do? _____________

---

• Would somebody be so nice and explain to me what “apparent” in this context possibly might mean? Any non-zero number is apparent to me. I guess it should read “relevant”.