Sharewise Presentation 22/7/24, page-84

Mango

I find it really interesting, and a little bit hard to explain, but here goes. The idea of trialling is that you are taking a sample of a large population, and then try to extrapolate how your sample is representative of the whole population - then do that for a placebo and a treatment group to show that if you treat the whole population with the drug or placebo, the drug would give a better outcome on average.

Say you have 2 dartboards, one with pain scores for placebo and one with pain scores for treatment. Your placebo dartboard may go from say 6 to 10. Say your treatment dartboard goes from 5 to 9. How many darts do you need to throw to show that your dartboard has a lower average than the placebo. Maybe quite a few because there is a big overlap of scores

But if your PPS dartboard painscore outcomes are 3 to 7. You could throw far fewer darts and very quickly you will see there is a significant difference in the average of the scores between that and placebo.

So @Mozzarc is right that its about the treatment effect - the greater the differentiation between treatement and placebo, the fewer the samples required to show statistically that one is meaningfully different from the other, and will continue to be that way if you keep throwing millions of darts