Back to the Top
Dear All,
I have a question regarding the definition of bioequivalence.
According to generally accepted criteria, the 90% CI's for determination of
bioequivalence are 80-125%. We have a situation where the product has been
demonstrated to be bioequivalent, although the lower CI for Cmax is sitting
on 80% when using ANOVA (Ln transforming prior to performing the test).
When using the method of Hauschke et al., this CI is shifted down to 77%.
This is also the case for AUCI and AUCT. The power of the tests is well
above 80%. T-testing using parametric and non-parametric tests has not
shown any differences of statistical significance.
Is this product bioinequivalent?
Regards,
Jennifer Norman
University of Cape Town
Back to the Top
The following message was posted to: PharmPK
The parametric procedure determining bioequivalence should be used for
AUCT, AUCI and Cmax. A nonparametric method could be applied, when
desired, for discrete measures such as Tmax. t-tests evaluating the
significance of differences are not relevant in determinations of
bioequivalence.
Laszlo Endrenyi
University of Toronto
Back to the Top
The following message was posted to: PharmPK
Dear Jennifer,
/one/ possibility for wider CIs after nonparametric testing compared to
their parametric counterparts is the underlying discrete probability function.
Alpha set to 0.05 in a parametric sense is only asymptotically 0.05 (0.05
or smaller), whereas nonparametric alphas are exact - and dependent on
sample size (n).
Below you will find some exact alphas (and resulting CIs) for Hauschke's
method (hoping the table renders in your mail client):
..n....alpha.....CI..
--------------------
..6...0.0465...0.9070
12...0.0444...0.9112
16...0.0469...0.9062
20...0.0482...0.9036
24...0.0486...0.9028
28...0.0499...0.9002
32...0.0487...0.9026
The nonparametric method is a conservative one (alpha is always <0.05, CIs
are always >0.90).
You can get an idea whether this is the reason for a wider CI, if you
calculate the /parametric/ CI with alpha set to the alpha of the
nonparametric method (e.g. for 12 subjects calculate the the t-tests with
alpha 0.0444 instead of 0.05).
> Is this product bioinequivalent?
I don't think so.
If your statistical protocol stated parametric analyses (which I asume...),
you already have shown BE.
You are not supposed to do sequential testing; your alpha is already
"spent", therefore the outcome of the "second stage" is statistically
speaking of "undefined alpha".
best regards,
Helmut
--
Helmut Sch=FCtz
BEBAC
Consultancy Services for Bioequivalence and Bioavailability Studies
Neubaugasse 36/11
1070 Vienna/Austria
tel/fax +43 1 2311746
http://BEBAC.at
Bioequivalence/Bioavailability Forum at http://forum.bebac.at
http://www.goldmark.org/netrants/no-word/attach.html
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)