Bioequivalence and Bioavailability Forum

Dear D. Labes,

I agree with our Capt'n. Since the study is intended for European submission, I wouldn't bother too much. The NfG states (Section 3.6.1):

The statistical analysis (e.g. ANOVA) should take into account sources of variation that can be reasonably assumed to have an effect on the response variable.

(my emphasis)

In many cases subjects are split into two (or more) groups due to logistic reasons, but only in a minority of cases 'group' is included in the model. I have seen just a few problems in the EU - but a lot of questions from the Gulf States... I would say, the same principle applies to more than one center. Since you use the same protocol, my simple mind doesn't see a big deal. I would have an eye not only on the clinical performance, but on identical handling of samples in the centers (e.g., time interval between sample draw and centrifugation, cooling, g-force and duration of centrifugation, storage).
I wouldn't include center as a factor. From my personal experience (talking about the analogous 'group' factor), a significant effect is rarely seen (as the Capt'n said: between-subject factor has low power). But, what if? We can expect to see one in ~alpha of cases...
To be prepared for the worst (the pooled analysis is not accepted), I would suggest to split subjects not in roughly equal group sizes amongst centers, but to try to allocate as many subjects as possible to the 'largest' center. An example: Clinical capacity 24 beds, T/R 93%, CV 20%, power 90% for n=32.
If we allocate the 32 subjects to two centers of 16 subjects each, a significant effect is seen and pooling is not accepted by authorities, power drops to 64.6% (n=16). If we allocate 24/8, power in the larger center (n=24) would be still 81.7%.

On the other hand it's interesting that FDA states:

If the study is carried out in two or more groups and those groups are studied at different clinical sites, or at the same site but greatly separated in time (months apart, for example), questions may arise as to whether the results from the several groups should be combined in a single analysis. Such cases should be discussed with the appropriate CDER review division.

(my emphases)

Personal experience: nil.