Vinay Prasad

We assume in, we go to London and we assume 5% of the population has been infected.

So we take a thousand people.

Out of a thousand people, 50 people will have the disease.

Yeah, happy with that?

950 will not.

So the sensitivity is 70%.

So sensitivity means that of the people with the disease, 70% will test positive.

So of the 50 people with the disease, 35 will test positive and have the disease and 15 will test positive and not have the disease.

So that's your first problem of the 50.

Then you've got your other group.

Remember, you've then got 950 people who do not have the disease.

And what we said is the specificity is 98%.

And specificity means the proportion of people without the disease who test negative.

what i'm interested in is the proportion of people without the disease who test positive and that's two percent okay why are you interested in that because we're interested really here in the false positives though those who think they've got an antibody but actually they haven't and they're going around saying i've got antibodies but you haven't so of the 950 people without the disease yeah two percent of them will test positive

So 19 people will be false positives in every 1,000 you test.

When you're at low prevalences, you need tests that are up in the 99% sensitivity and 99% specificity.

In fact, in this case, if you want to be certain, you need to have perfect sensitivity and specificity.

That's very difficult to get.

That's the best case scenario of all the tests in effect.

And that's what you try and do here.