Couple of answers [Power / Sample Size]
❝ 1. I wanna do BE study in european population but all PK data from literature have been obtained from asian volunteers. Is it applicable/correct to use such data for my study sample size calculation?
Lots of studies are performed in India and all are accepted in European applications. If you think about crossovers, no worries.
I would be a little bit cautious if you plan a parallel design and the drug is subjected to polymorphic metabolism. The fraction of extensive/poor metabolizers might differ between populations and the total CV from the Asian population might be misleading. In such a case I recommend a pilot study.
❝ 2. We know that the higher number of subjects the higher chances to pass BE. But on the other hand higher subjects will need higher budget which is not good. Are there any statstical methods to find the optimal decision in such situations?
Well, the producer’s risk (i.e., the type II error) is 1 – power. Study costs are mainly driven by bioanalytics, i.e., the number of samples. Then you have to take into account the clinical overhead (subject remuneration [duration of hospitalisation, number of blood samples, etc], costs of pre-/post study exams, ).
If you have to perform the study in more than one group (limited clinical capacity) or in a Two-Stage-Design, it will be more costly. I’m sure you have a spreadsheet for that.
Then for various sample sizes plot costs vs. power and see what happens.
An example of a simple 2×2×2 crossover of my old CRO. Most costs (clinics, bioanalytics, statistics, etc.) were internal. Only the IEC and pre-/post study lab exams were external. Some costs increase linear with sample size, same are constant (it took me the same time to evaluate 24 or 32 subjects). The Excel file had 110 rows.
Dif-tor heh smusma 🖖🏼 Довге життя Україна!
The quality of responses received is directly proportional to the quality of the question asked. 🚮