Bioequivalence and Bioavailability Forum

Hi arjunroy,

❝ 1. R code is working fine,

So I guess you have n_t=n_r otherwise the code will give wrong result. I noticed it some time ago so I'll modify the code when I have some time.

❝ though SAS IML v9.1.3 did not, got problem in calculating Cov matrix. Execution error noted in log despite it can be done manually.

Do you add the COV function before the run? As I mentioned in the post, for SAS 9.2 and older version there's no COV function so you should add it manually. The example of COV I gave only work for no missing value which in most of the case it should be. But if not, well, you have to adapt the code.

❝ 2. Any idea how dm-max is calculated in a 2 parameter weibull function, I am unable to verify in DDSolver output. For all kind of data it sets dm_max=11.82

❝ Any further insights will be appreciated.

If I remember it correctly, that's a certain default value in DDSolver. You should use you own criteria instead of 11.82.

I'm not aware any guideline that clearly define how to set similarity limit but in my opinion the criteria can be obtained by, e.g., fitting suitable model at least 2 different reference batches (more batches would be better) and determine the MSD between them (let's call it MSD_R). If you have more batches, pair-wise compare the MSD_R (A vs B, B vs C, A vs C, ...) and pick the biggest value.

The rational is that the difference between test and reference (upper limit of MSD) should not be greater than the maximum difference observed between reference batches themselves. So the maximum value in your pair-wise comparison of difference between reference can be used as dm_max.