Carl Hennigan

because in 100 people you can see some significant effects and particularly if you're taking people with mortality of 15 20 you can start to see some really important data where you can immediately go right this is the sort of trial we should take forward and what you want to do is to be able to stop the stuff that's not looking so good and go to the next level and the next level and that's the

the point we want to get to, as opposed to still in this point going, oh my gosh, is anything yet working?

We're not quite sure.

No, but I'm going to write it up because I do think you could have this concept of a, because one of the things I found quite difficult to understand is we spend a lot of time trying to come up with sample sizes.

And when we've published it all, they're always underpowered.

Anyhow, so actually we'd be much better if we had a sort of... My thinking is to have a standard sample size.

At the early trials, just do 100 patients.

They have the core outcomes, but some of their smaller baseline problems might be 1,000.

You could actually model that based on the control event rate.

Given the fact here the mortality is so high in hospitalized patients, you'd probably get away with 100.

In the preventive treatments, you'd probably need about 1,000.

And we have looked at that because...

often we get too many trials that even go, we had 250 patients and it was underpowered, and that's why we need systematic reviews.

So we could start to think like this, about here's a minimum sort of sample size that would allow us to have some useful information based on the control event rates.

I'm sure we'll get some statisticians now ringing in going, what does he think he's on about there?

This is ridiculous.

Hi, I'm Carl Hennigan.

I am Professor of Evidence-Based Medicine at the University of Oxford and Editor-in-Chief of BMJ EBM.

Well, there's lots of evidence about seasonal viruses, particularly the ones that have a lipid envelope.

ones like RSV and normal coronaviruses, that three things really have an impact.