to explain why testing lots of asymptomatic people everyday is a bad idea

[chomalungma]

They want to do lots of testing on asymptomatic people everyday.
Aim is to see if people are positive, and if so, then those people have to self isolate.

Issue is: There will be many people who will have to self isolate who don't need to because of false positives

All to do with how specific a test is, how sensitive it is, and the prevalence of disease in the population you are screening.

Let's say the levels of disease are 1 in 10,000

So in 1 million people, 100 people will have the disease

And 999,900 people won't have the disease.

They get a really good test; It detects disease in 99% of the people who have it.

So in the 'ill' population, we have 99 positives and 1 negative

The test is also very specific. Say 99.5% specific. That means in the 999,900 people who don't have the disease, we will get 994,900 people who test negative and 5000 who test positive (False positives)

So we have 5099 positive tests.
And 99 of them are actual positives

So 98% of the positive results are false positves

(this is based on a level of 1 in 10,000 and those figures for specificity and sensitivity) Obviously a test needs to have really good specificity and sensitivity and it needs to be done on a population with a relatively high chance of having the disease.

This also explains it

www.forbes.com/sites/startswithabang/2020/05/07/this-is-how-false-positives-and-false-negatives-can-bias-covid-19-testing/#7f4715c5174

Can you please state what qualifications/experience you have that make you qualified to posset this opinion

M.Sc Clinical Biochemistry

10 years working in path labs

Plus

It's maths.
Not opinion
Maths

It would be helpful to have a solid understanding of how infectious asymptomatic people actually are.

The tests do seem very sensitive.

www.hdruk.ac.uk/projects/false-positives/
Here's another explanation. False positives and negatives aren't a new corona thing.

Also

www.dailymail.co.uk/news/article-8717747/Experts-banging-heads-wall-Boriss-Moonshot-Covid-testing-plan.html

@chomalungma

Can you please state what qualifications/experience you have that make you qualified to posset this opinion

M.Sc Clinical Biochemistry

10 years working in path labs

mic drop

Here's another explanation. False positives and negatives aren't a new corona thing

This

100% OP. I've been banging on about this for ages. You are more qualified than me though

The test is also very specific. Say 99.5% specific.

I thought latest data was showing a specificity of 99.9%?

@chomalungma

Can you please state what qualifications/experience you have that make you qualified to posset this opinion

M.Sc Clinical Biochemistry

10 years working in path labs

Boom! Haha

I was explaining the same thing to my husband yesterday. I dont have your Qualifications, just a degree which involved statistics and a decent understanding of maths.

When the actual prevalence of the virus is very low, false positives are a really big problem.

This link www.mathsisfun.com/data/probability-false-negatives-positives.html explains it nicely. Not covid specific but principle is the same.

I'm interested in how contagious asymptomatic and people with very mild symptoms are.

I thought latest data was showing a specificity of 99.9

I was just talking in general about tests - not quoting specific data

They are talking about developing new tests similar to pregnancy tests.

That's really interesting, thanks for posting. I had thought false positives weren't a but g deal. That yes it's an inconvenience for those people but probably worth it for the greater good, as it's a very small percentage. But when you scale up the numbers you realise the impact and if people don't trust the results, some of the people with genuine positive tests are likely to think theirs is a false positive too and not take it seriously.

Brilliant @chomalungma 😀😀😀 I agree with you but don’t have any qualifications to back it up like you so will ride along on your coat tails 😂

@PinkFondantFancy

100% OP. I've been banging on about this for ages. You are more qualified than me though

It's just maths though.

This is why screening needs to be done on the right populations.
Otherwise a positive test isn't that helpful.

It can certainly reassure you if you are negative, but if you get a positive result, there is a very good chance you are negative*

*I am talking in general about any screening test

I heard Naga trying to understand it on BBC this morning.

It really needs a table to understand.

However, they have been saying that testing needs to be more reliable before mass testing happens.

Plus they'd be testing the population, not asymptomatic people. They won't find out someone is asymptomatic until they're tested.

Plus they'd be testing the population, not asymptomatic people. They won't find out someone is asymptomatic until they're tested

All depends on the prevalence of disease in the population and of course, how good the tests are.

Can you please state what qualifications/experience you have that make you qualified to posset this opinion. M.Sc Clinical Biochemistry. 10 years working in path labs

Touché

I understand the maths here, but have no idea what specificity of a test means and why it would result in false positives. I am off to ask Google.

I am off to ask Google

Let's say you have 100 people who do NOT have the disease.

If the test was 100% specific, then all of them would test negative.
99% specific would mean that 99 would be negative and 1 would be positive.

@chomalungma
Nothing else to add and totally agree with you. But also wanted to say..... you’re awesome

Please can you replace Chris Twitty asap?

@chomalungma

Can you please state what qualifications/experience you have that make you qualified to posset this opinion

M.Sc Clinical Biochemistry

10 years working in path labs

That was impressive!

I agree. This is well known basic maths and applies to pretty much all medical tests.

(Cambridge maths degree, qualified actuary, 15 years post qualification experience, 3 years marking statistics practice papers for actuarial education.)