Bioequivalence and Bioavailability Forum

Hi ammulu,

❝ Hi guys can any one tell me why we should log transform the data for

❝ calculations of BA-BE data?

1. Because regulators want you to.
2. Because the model used for evaluation assumes that the residuals of the model fit follow a normal distribution, and log transformation is "widely" believed to make that assumption justified in "most cases".
2½. Because AUC and Cmax-values cannot be negative, there is a restriction on the residuals if you use a non-transformed evaluation. That restriction is in contradiction with the prinicple of normality.

❝ whether natural log should be used or any other ?

Doesn't matter. Your result will be the same. Try and see for yourself.

EM.

Dear ammulu!

I agree with ElMaestro, but I would rate his #1 as my #3.
In the future please search before starting a new thread; we had this question already a couple of times, e.g. here and there.

Hi ammulu,
I am answering this question in a simple language "Log functions tend to squeeze the larger values in your data and streches out the smaller values", this might be useful when you might be dealing with large data.