Back to the Top
Dear Members,
greetings.
I have some question regarding outliers in bioequivalence studies.
We have conducted a two-way crossover bioequivalence study in 40
healthy human subjects. In that we found one subject response was
very different in period-2 as compared to period-1. (i.e. his Cmax in
period-2 is 10 times lower than period-1). And the ratio for this
subject is very low.
By observing the data we applied outliers test and found that subject
as an outlier.
1. When we remove that subject data from the statistical calculation
then the 90% confidence interval is with in the acceptance limit for
the drug as well the metabolite.
2. When we are not removing the subject data from the statistical
calculation then the 90% confidence interval is not with in the
acceptance limit for the drug but it is in the limit for metabolite.
Below are my questions:
What exactly should I have to do?
1. Should I have to remove that subject data and carryout the
analysis and submit the results?
Or
1. Should I have to present the results with and with out the outliers?
2. What is the regulatory requirement when we find an outlier in the
bioequivalence study data?
I am grateful to here your suggestions and thoughts.
Regards,
kanimozhi.A
Back to the Top
The following message was posted to: PharmPK
Dear
kanimozhi.A
The following information have been taken from Bolton S and Bon C
(Pharmaceutical Statistics * practical and clinical applications,
Marcel Dekker, INC 2004)
An outlier is a value that does not appear to fit the pharmacokinetic
profile. If repeat analyses verify these values, one has little
choice but to retain these values in the analysis. If such values
appear rarely, they will usually not affect the overall conclusions
since the individual results are a small part of the overall average
results, such as in the calculation of AUC. An exception may occur if
the aberrant value occurs at the time of the estimated Cmax, where
the outlier could be more influential.
The current FDA position is to disallow the exclusion of an outlier
from the analysis solely on a statistical basis. However, if a
clinical reason can be determined as a potential cause for the
outlier and when the outlier appears to be due to the reference
product, an outlier may be omitted from the analysis at the
discretion of the FDA. The FDA also suggests that the outlier be
retested in a sample of 6-10 subjects from the original study to
support the anomalous nature of the suspected outlier. Part of the
reasoning for not excluding outliers is that one or two individual
outliers suggests the possibility of a sub-population that shows a
difference between the products.
Although theoretically possible, this is a highly unlikely event
without definitive documentation. Also, using this reasoning, an
outlying observation due to the reference product would suggest that
the reference did not act uniformly among patients, suggesting a
deficiency in the reference product. Another possible occasion for
discarding an individual subject's result is the case where very
little or no drug is absorbed. Explanations for this effect could be
product-related or subject-related, but the true cause is unlikely to
be known.
Hoping it will be of some help.
++
Nadeem Irfan Bukhari
Lecturer Pharmaceutical Technology,
International Medical University,
Bukit Jalil 57000, Kuala Lumpur, Malaysia
Web: http://www.imu.edu.my
Tel: +60 3 8656 7228, Ext. 1186; Fax: 86567229
Back to the Top
The following message was posted to: PharmPK
Dear Ms. Kanimozhi,
Issue regarding the acceptance of Outliers is still a debatable one
around the world. To my knowledge, it all depends on the regulatory
agency to which the submission is made. For eg, ANVISA accepts data
without outliers, whereas, FDA has some reservations on that. In any
case you should have both data, with and without the outliers for
submission. And also, it is always better to hear it from the horse's
mouth (regulatory agency).
Hope this is useful.
Regards,
Dr. M. Joseph MD
Back to the Top
The following message was posted to: PharmPK
What I have done in my earlier studies with such instances, better to
provide in final report both with and without outliers, so regulatory
body
reviewer will get exact picture and which will minimise the future
queries.
Also specify the statistical test performed to detect the ouliers.
Better to write in protocol, how the outliers will be handled.
Thanks,
Dr.Manoj Bose, MBBS, MD
Principal Investigator
Back to the Top
Dear Group,
According to Brazilian Regulatory Agency (Anvisa), the data with and
without outliers have to be submitted, but we rarely accept data
excluding outliers.
Regards,
Paula Macedo Cerqueira, PhD
Pharmacist- Technical Consultant
National Health Surveillance Agency - ANVISA
Bioequivalence Unity - Sao Paulo -SP- Brazil
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)