## TSD statistical model - with multiple sites [Two-Stage / GS Designs]

Hi d_stat,

» » […] Originally we wanted to go with a variant of ‘Method C’ cause it’s slightly more powerful (esp. when you expect to stop in the IA with BE) and it is preferred by the FDA.
»
» Thus, we are sticking to Potvin.

I wouldn’t. It might be that the site-model mentioned above is not stable. When I tried to add ‘sites’ to Potvin’s ‘Example 2’ (subjects 1–4: 1, 5–8: 2, 9–12: 3), Phoenix/WinNonlin showed me the finger.

» In order to do the validations for multiple-site nature of the study we are thinking of amending the code in the Power2Stage, since it's not part of package yet. Am I wrong to assume the amendment will be connected to decreasing of df for the error term with the number planned sites, …

That’s correct.

» … or is there any other part that I am missing?

You would also have to modify the degrees of freedom in the function to re-estimate the sample size (sampsiz2.R).

» Of course, the alternative is always to encourage (pray to) the holy trinity to update the package

Maurer’s method as implemented in Power2Stage allows already to provide the degrees of freedom and the standard error of the means of a more complex model than the conventional 2×2×2 crossover.
• Modifying Potvin’s method(s) would not be a trivial job and you have to demonstrate control of the Type I Error.
• Maurer’s controls the TIE in the strict sense and is accepted by the FDA as well.
Hence, why the efforts?

Dif-tor heh smusma 🖖
Helmut Schütz

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes