Log in
Register|
pash413 ★ India, 2012-04-06 16:56 (5196 d 23:47 ago) Posting: # 8393 Views: 2,275 |
|
|
Dear All In the recent USFDA individual guidance on capacitabine tablet bioequivalence study, it is recommended that – 'Since each patient in the study will be receiving different doses, dose should be included in the statistical model'. But in general in the crossover study even though different doses are used for each subject (as per their BSA), any given subject receives the same dose of Test and Reference. Hence is there any rational to include the term 'dose' in the statistical model? Because each subject will acts as his/her own control providing a Test/Reference comparison which is independent of the dose of drug received. Is it that the regulatory expect the Test/Reference ratio should not significantly differ across the doses in this case? Kindly share your thoughts on above requirement by USFDA and also suggest what would be the Statistical model after including this term for two way crossover study. Edit: Guidance linked. [Helmut] |
|
ElMaestro ★★★ Denmark, 2012-04-07 15:07 (5196 d 01:36 ago) @ pash413 Posting: # 8394 Views: 1,800 |
|
|
Hello Pash, I must admit that I also find it a little weird. The drug is dosed according to the approximate surface area. If you wish to include dose in the stats model it must be absolute dose rather than dose divided by area (expressed as mg/m2 will be the same across all subjects). Absolute dose then comes in as a covariate, I guess. But as you say, for each subject the absolute dose will be the same in both periods within any subject. So I can't imagine inclusion of dose achieves much. I guess (and I literally mean this is a guess) that the model is: y=intercept+Seq+Subj+Trt+Per+Dose, where the first four are fixed factors and the last one is a covariate. Or...? — Pass or fail! ElMaestro |
Ing. Helmut Schütz