Bioequivalence and Bioavailability Forum

To whom it may concern!

Not only in the forum but in my daily work as a consultant I’m increasingly often confronted with “cre­ative” study designs where the underlying statistical methodology is either dubious or currently non­existent at all. Just to name a few:

• “Two-stage replicate designs” intended for reference-scaling. Method? Patient’s risk?
  
• Replicate designs with more than two treatments. Method?
  
• Futility rules in two-stage designs. Power?
  
• Two-stage designs for parallel groups. Heteroscedasticity? Patient’s risk? Power?
  
• Planning a pivotal study based on a single published CV…

Rather scaring in many cases doubts arise when the study is already underway. How is that possible? Were IECs/IRBs and regulators sleeping as well? In many cases it turned out that people designing those studies had limited no knowledge about existing methods, their underlying assumptions/limitations and/or pur­poses. “Classical” examples are blindly following FDA’s partial replicate design or believing that the 94.12% confidence interval is applicable to all two-stage designs (aka Pocock’ natural constant α 0.0294). When asked a common answer is “We planned to evaluate the study with SAS…”. Bingo! It also often unfolds that their main sources of information are guidelines, presentations found on the inter­net, or forum posts / blogs.
Only a minority might have read original publications beyond their abstracts.

To call the statistician after the experiment is done
may be no more than asking him to perform a postmortem examination:
he may be able to say what the experiment died of. RA Fisher

Amazingly these (Internet-Generation?) attitudes are not limited to people working in the industry but to regulators as well. Recently I received a question from an ASEAN-assessor which boiled down to:

“A generic product has demonstrated bioequivalence in a well-powered in vivo study. However, the formulation failed to show f₂ ≥50 in all p_H-values tested. Can I still accept the application?”

—
SCNR,
Helmut (the grumpy old man)
Profile «accessible to registered users only»
The quality of responses received is directly proportional to the quality of the question asked. ☼
→ Science Quotes