Coronavirus (Non-Politics)

[Member]

"Official data shows the test positivity rate among secondary pupils is around one in 2,000. Just 1,805 of the 3.8million lateral flow tests taken last week were positive.

Health bosses say around one in 1,000 positive results from rapid kits are false. This would be the equivalent of around 3,800 positives from 3.8million tests."

Reading an article on the Daily Mail website (please don't judge). I inteprept the second paragraph differently - if 1 in a 1000 positive results are a false positive, and there were 1,805 positive test results, doesn't it mean just 2 are incorrect? Daily Mail are suggesting that most of the 1,805 positives are false.

Full article here

'Vast majority' of positive rapid Covid tests are 'likely to be false'

Data from March 11 to 18 shows 1,805 positive results from 3.8million tests taken. Given the current best-guess false-positive rate, only 645 of those are likely to be genuine infections.

www.dailymail.co.uk

In your second paragraph this is the sensitivity of the test. So this is saying that if you give the test to 1,000 people who have the virus, then it'll be correctly positive 999 times, and falsely negative 1 time. So this is giving us the 'true positive' (sensitivity) statistic. So if you only give the test to people who definitely have the condition then it works extremely well. But in reality we give the test to a mixture of people who are true positive and true negative.

What you need to remember is that when cases are low (low prevalence of the virus), you get more false positives because you're testing more people who are truly negative. If the prevalence is high then there are fewer people without the virus to falsely test positive. So the prevalence impacts on false positives.

So in those numbers, there might be 10,000 true positive cases, but you've only managed to test a fraction of those and correctly identify them, but you've also tested 1,000,000 negative people and that's resulted in 1,000 false positives.

 

[Benfleet A1]

Bit of help and advise from Pubey & Tinks please. I work with a guy who's daughter has been diagnosed with breast cancer. Hopefully its been caught in time but because she will be spending quite a lot of time in hospital having treatment, she's been given the Astra Zenica vaccine just over a week ago. She started her first course of chimo last monday and everything seemed fine but tonight she has been taken in suffering from a blotchy rash feeling extremely poorly.

I'm asking if there is any connection between the jab and chemo treatment or have either of you heard of any such cases. The guy is pulling his hair out with worry as you can imagine and I just want to try and reassure him a little if possible.

 

[Holy Joe]

In your second paragraph this is the sensitivity of the test. So this is saying that if you give the test to 1,000 people who have the virus, then it'll be correctly positive 999 times, and falsely negative 1 time. So this is giving us the 'true positive' (sensitivity) statistic. So if you only give the test to people who definitely have the condition then it works extremely well. But in reality we give the test to a mixture of people who are true positive and true negative.

What you need to remember is that when cases are low (low prevalence of the virus), you get more false positives because you're testing more people who are truly negative. If the prevalence is high then there are fewer people without the virus to falsely test positive. So the prevalence impacts on false positives.

So in those numbers, there might be 10,000 true positive cases, but you've only managed to test a fraction of those and correctly identify them, but you've also tested 1,000,000 negative people and that's resulted in 1,000 false positives.

Can I just run through that to double check I've understood please.

We have 3.8m tests. If we knew in advance everyone was positive we'd expect to get 3,796,200 correct positive results and 3,800 incorrect negative results. If we knew everyone was negative, we'd expect to get 3,800 incorrect positive results with 3,796,200 correct negative results.

If you don't know what you're testing, but you know prevelance is roughly 0.05%, then out of the 3.8m tests you'd expect 1,888 positive results (0.05% of 3.8m is 1,900 but 1 in thousand of those will give false negative). But you'd also expect false positives, so you have 3,798,100 people who are negative, so you'd expect 3,798 positives from them. So in total you'd expect 5,686 positive tests, of which most, 3,798 will be false?

 

[Special Blue]

Can I just run through that to double check I've understood please.

We have 3.8m tests. If we knew in advance everyone was positive we'd expect to get 3,796,200 correct positive results and 3,800 incorrect negative results. If we knew everyone was negative, we'd expect to get 3,800 incorrect positive results with 3,796,200 correct negative results.

If you don't know what you're testing, but you know prevelance is roughly 0.05%, then out of the 3.8m tests you'd expect 1,888 positive results (0.05% of 3.8m is 1,900 but 1 in thousand of those will give false negative). But you'd also expect false positives, so you have 3,798,100 people who are negative, so you'd expect 3,798 positives from them. So in total you'd expect 5,686 positive tests, of which most, 3,798 will be false?

Can I just run through this to make sure I'm thick!?
Just joking there are some very clever people on this site!

 

[Massimo Giovanni]

Hancock needs to make it clearer, the press, the TV, all need to remind folk that the covid dance isn't over.
This morning, walking past Aldi, Priory Crescent, the place seemed to be family and chums day trip out!
What happened to the shop alone, and AGAIN, some of the masks looked about as useful as wearing a string vest over one's visage.
And it appeared the traffic lights for red no entry was being observed, not a lot.

 

[Tinks]

Bit of help and advise from Pubey & Tinks please. I work with a guy who's daughter has been diagnosed with breast cancer. Hopefully its been caught in time but because she will be spending quite a lot of time in hospital having treatment, she's been given the Astra Zenica vaccine just over a week ago. She started her first course of chimo last monday and everything seemed fine but tonight she has been taken in suffering from a blotchy rash feeling extremely poorly.

I'm asking if there is any connection between the jab and chemo treatment or have either of you heard of any such cases. The guy is pulling his hair out with worry as you can imagine and I just want to try and reassure him a little if possible.

Oh my poor thing.
I have not heard of any direct link to be honest and in fact the AZ was recommended for people with underlying health conditions including chemotherapy. There is also advice for people suffering with auto-immune to have the vaccinations.
hopefully it is nothing more sinister than a mild reaction
Please let me know how she is going

 

[Benfleet A1]

Oh my poor thing.
I have not heard of any direct link to be honest and in fact the AZ was recommended for people with underlying health conditions including chemotherapy. There is also advice for people suffering with auto-immune to have the vaccinations.
hopefully it is nothing more sinister than a mild reaction
Please let me know how she is going

Thanks Tinks, I knew you would step up. I'll let him know.

 

[Ricey]

Hancock needs to make it clearer, the press, the TV, all need to remind folk that the covid dance isn't over.
This morning, walking past Aldi, Priory Crescent, the place seemed to be family and chums day trip out!
What happened to the shop alone, and AGAIN, some of the masks looked about as useful as wearing a string vest over one's visage.
And it appeared the traffic lights for red no entry was being observed, not a lot.

Give people an inch and they’ll take a mile. It happened the first time and it’s happened everytime since. Difference this time is we have a vaccination being rolled out and hopefully it will make the difference and those of us who aren’t selfish ****s get to return to normality.

 

[jackj13]

Give people an inch and they’ll take a mile. It happened the first time and it’s happened everytime since. Difference this time is we have a vaccination being rolled out and hopefully it will make the difference and those of us who aren’t selfish ****s get to return to normality.

Last time hospitalisations and deaths were at this level we were pretty much open and this time we have over 30million people vaccinated. I can understand why. Can’t really see how queuing for a shop is selfish either. You are also allowed to meet people outside.

 

[dcblue]

Vaccine passport for sporting events. Looks like I won't be going to the hall again.

 

[Massimo Giovanni]

Last time hospitalisations and deaths were at this level we were pretty much open and this time we have over 30million people vaccinated. I can understand why. Can’t really see how queuing for a shop is selfish either. You are also allowed to meet people outside.

Queuing wasn't the problem I saw. When the green light came on a whole herd went in, not the 1 in 1 out.

 

[Massimo Giovanni]

Took an 86 yr old neighbour to SUH yesterday for blood test, well organised, efficient and ran like clockwise. The NHS is making changes.

 

[jackj13]

Queuing wasn't the problem I saw. When the green light came on a whole herd went in, not the 1 in 1 out.

Have to be honest it hasn’t been 1 in 1 out at any supermarket I’ve been to for a long time now.
In other news very low death rates last few days. Down as low as 10. Putting that into perspective we average 1400 deaths a day from all causes.

 

[RobM]

Vaccine passport for sporting events. Looks like I won't be going to the hall again.

There's a positive then!

 

[Sandbach Shrimper]

Hancock needs to make it clearer, the press, the TV, all need to remind folk that the covid dance isn't over.
This morning, walking past Aldi, Priory Crescent, the place seemed to be family and chums day trip out!
What happened to the shop alone, and AGAIN, some of the masks looked about as useful as wearing a string vest over one's visage.
And it appeared the traffic lights for red no entry was being observed, not a lot.

It pretty much is over. 10 deaths reported yesterday and a couple of thousand 'cases'. If people weren't wearing masks and you turned off your TV, would you know there's supposedly a pandemic going on?

Not everyone can shop alone, and what does it matter anyway? If you've been vaccinated, why do you care what others are doing?

 

[Massimo Giovanni]

It pretty much is over. 10 deaths reported yesterday and a couple of thousand 'cases'. If people weren't wearing masks and you turned off your TV, would you know there's supposedly a pandemic going on?

Not everyone can shop alone, and what does it matter anyway? If you've been vaccinated, why do you care what others are doing?

So you think lockdown was a good thing and it is okay to repeat it when numbers go up? New strains appear, that risks to those who medically can not be jabbed are not worth care?
By using "supposedly" does that indicate a mindset towards the pandemic?
I do not believe all I read, or hear, or see AND in years ahead facts about the origin and dealings with covid will, for sure, be interesting.

 

[Member]

Can I just run through that to double check I've understood please.

We have 3.8m tests. If we knew in advance everyone was positive we'd expect to get 3,796,200 correct positive results and 3,800 incorrect negative results. If we knew everyone was negative, we'd expect to get 3,800 incorrect positive results with 3,796,200 correct negative results.

If you don't know what you're testing, but you know prevelance is roughly 0.05%, then out of the 3.8m tests you'd expect 1,888 positive results (0.05% of 3.8m is 1,900 but 1 in thousand of those will give false negative). But you'd also expect false positives, so you have 3,798,100 people who are negative, so you'd expect 3,798 positives from them. So in total you'd expect 5,686 positive tests, of which most, 3,798 will be false?

Yes this is basically correct. I the first paragraph there is a difference in how good a test is at spotting true positive (sensitivity) and how good it is at spotting true negatives (specificity). For something like COVID it's more important to spot true positives at the expense of spotting true negatives, because false negatives are more harmful than false positives.

The lateral flow test has about 75% sensitivity (it'll test positive in true positive cases 75% of the time). This is pretty good, because often people using them are asymptomatic and so may not have high viral load, but obviously it'll miss out on 1/4 of positive cases. It has a specificity of 99%, so it'll be negative 99% of times in people without COVID. Obviously you expect specificity to be high, because there is nothing there in terms of virus to actually register a positive result.

So if you test 1,000,000 people but the prevalence is 1% (so 10,000 cases), you'll get:
7,500 positive results (10,000 x 75% sensitivity) which are true positives
2,500 negative results (10,000 x 25% [1-sensitivity]) which are false negatives
980,100 negative results (990,000 x 99% specificity) which are true negatives
9,900 positive results (990,000 x 1% (1-specificity]) which are false positives

Therefore when prevalence is low a positive result has a high chance of being false positive (~25% of a positive test being false)

If you have the same sensitivity (75%) and specificity (99%) but 50% prevalence (500,000 cases), you'll get:
375,000 true positives
125,000 false negatives
495,000 true negatives
5,000 false positives

Therefore with a high (symptomatic) prevalence then there's a 1% chance of a positive test being a false positive.

 

[Holy Joe]

Yes this is basically correct. I the first paragraph there is a difference in how good a test is at spotting true positive (sensitivity) and how good it is at spotting true negatives (specificity). For something like COVID it's more important to spot true positives at the expense of spotting true negatives, because false negatives are more harmful than false positives.

The lateral flow test has about 75% sensitivity (it'll test positive in true positive cases 75% of the time). This is pretty good, because often people using them are asymptomatic and so may not have high viral load, but obviously it'll miss out on 1/4 of positive cases. It has a specificity of 99%, so it'll be negative 99% of times in people without COVID. Obviously you expect specificity to be high, because there is nothing there in terms of virus to actually register a positive result.

So if you test 1,000,000 people but the prevalence is 1% (so 10,000 cases), you'll get:
7,500 positive results (10,000 x 75% sensitivity) which are true positives
2,500 negative results (10,000 x 25% [1-sensitivity]) which are false negatives
980,100 negative results (990,000 x 99% specificity) which are true negatives
9,900 positive results (990,000 x 1% (1-specificity]) which are false positives

Therefore when prevalence is low a positive result has a high chance of being false positive (~25% of a positive test being false)

If you have the same sensitivity (75%) and specificity (99%) but 50% prevalence (500,000 cases), you'll get:
375,000 true positives
125,000 false negatives
495,000 true negatives
5,000 false positives

Therefore with a high (symptomatic) prevalence then there's a 1% chance of a positive test being a true positive.

Thanks very much - increased my understanding greatly. Just checking last sentance should be a false postive?

 

[Member]

Thanks very much - increased my understanding greatly. Just checking last sentance should be a false postive?

Yes. Corrected.

 

[Holy Joe]

Yes. Corrected.

phew - I did understand! Thanks very much for taking the time to explain