Back to the Top
Dear
We are conducting a pilot study with high variability. What is the to choose the sample size for the same.
Thanks & regards,
Mr. Manoj Kumar
Biostatistician
Back to the Top
Dear Mr. Manoj Kumar
First of all you have to decide what design suits the study, then see whether you are going to use controls and cases only then you can decide which formula to use and most likely will be under the continous response variables.
regards
Rafid Dr.Rafid S. Jabir, MD, M.Sc(Clinical Biochemistry) PhD Pharmacology Student
Pharmacotherapeutics Unit,
Department of Medicine
Faculty of Medicine and Health Sciences
Universiti Putra Malaysia, 43400 UPM
Serdang, Selangor
Malaysia
Back to the Top
The following message was posted to: PharmPK
Hi Manoj,
The sample size can be chosen based on intra subject CV of the drug, which can be obtained from literature. If you don't have that information and you are planning to do a pilot study then you can start with 12 volunteers to get an idea for the sample size required for the pivotal study.
Thanks
Tausif Ahmed, Ph.D.
Principal Scientist, DMPK, Sai Advantium Pharma Ltd., Pune
Back to the Top
Dear Manoj,
Sample size is depends upon the variability, so nobody can deside with ds information
regards,
venkaetsan
Back to the Top
The following message was posted to: PharmPK
Dear Tausif Ahmed,
you wrote:
> The sample size can be chosen based on intra subject CV of the drug, > which can be obtained from literature. If you don't have that > information and you are planning to do a pilot study then you can start > with 12 volunteers to get an idea for the sample size required for the > pivotal study.
> No, you can't estimate the sample size of a parallel design study based on the intra-subject CV. Sample size estimations for such a design must be based on the total variance (inter- plus intra-subject). Since for most drugs CVinter > CVintra a parallel design study based on CVintra will by grossly underpowered. BTW, don't mix up the CV of a drug with the CV of a formulation. Example: rectal diclofenac formulations show a CVintra of <15%, whereas enteric coated formulations are HVDPs...
Literature data are always a mixed blessing. Essentially you have no idea about covariates which may influence variability (e.g., posture & physical activity in the clinical phase, bioanalytics). If have seen studies failing where variability was higher due to a 'better' analytical method - the 'old' one was an nice HPLC/UV, whereas the 'better' one was struck by a matrix effect in LC/MS-MS. Of course this doesn't happen any more - everybody should have learned the lesson by now. ;-)
I agree with your statement about a pilot study; but what do you mean by "12 subjects"? Are you talking about twelve in total, i.e., six per group? That's by far too few to get a reliable estimate of both the GMR and the variability. The smaller the sample size in any pilot study is, the higher the degree of uncertainity of both estimates. Even for a cross-over twelve subjects I would call quite small, but for a parallel design...
Best regards,
Helmut
-
Ing. Helmut Schuetz
BEBAC - Consultancy Services for
Bioequivalence and Bioavailability Studies
Neubaugasse 36/11
1070 Vienna, Austria
Back to the Top
Dear Helmut,
I would like to add to your comment regarding inter and intrasubject variability of drug. Intersubject variability can be overcome by crossover randomised study whereas intrasubject variability is difficult to overcome. There are different ways to overcome intrasubject variability, i.e. one can use t1/2 or dosing interval or combination of both but it can't be understood as the best method, for this you have have to understant the PK behaviour of drug in the body.
Kind Regards
Md. Tarique Imam
|
Want to post a follow-up message on this topic?
If this link does not work with your browser send a follow-up message to PharmPK@boomer.org with "Sample size for a parallel study" as the subject | Support PharmPK by using the |
Copyright 1995-2011 David W. A. Bourne (david@boomer.org)