RE: Morning all !2 Jun 2023 23:19
Do you know how many patients are needed in a trial to make the results meaningful?
Defender, If I may add a more general, clinical trials, answer to Crumb's precise estimate. If what you regard as "meaningful" is a 5% probability of the results happening by chance, rather than a 1% chance, then you can manage with fewer samples/subjects. If the difference you are 'expecting to prove' is very small (say a 5% increase in survival rate) then you need a bigger sample than if you are expecting to prove a difference of, say, 50%. In other words, there is no general answer because the sample sizes depend upon the hypothesis, the experimental design and the expected variability of the samples etc etc etc. However, statisticians have formal methods to estimate the sample-sizes they will need to support the hypothesis they entertain. The problem is that, when you are dealing with humans rather then lab rats, the sample sizes are almost always smaller than you would like...