Covid 19 testing

amunro11
Posts: 3
Joined: Thu Jul 30, 2020 8:18 pm

Covid 19 testing

Post by amunro11 »

Hi.
Forgive me if this subject has been raised elsewhere. What are users views on efficacy of C-19 testing?
I work in a care home for the elderly and we are under huge pressure to be tested, either by nurses who visit to test staff or using self testing kits. Test involves taking swab from nostril and throat.
I am sceptical about reliability of test results and have no faith in U.K. Gov handling of whole pandemic issue. I have refused test so far but today I was interviewed by my manager and asked what my reasons were for my refusal to test. He is under enormous pressure from his manager to have all staff tested so was sympathetic to my feelings but also managed to hint that action would be taken if I continued to refuse a test. Views please.
I work on Scotland, I am not a carer but a cleaner employed by local authority. My union is useless. It was hinted that management would ‘risk assess’ my situation and probably remove me from my workplace and place me somewhere I can do less harm!

guy153
Posts: 48
Joined: Wed Apr 22, 2020 5:29 pm

Re: Covid 19 testing

Post by guy153 »

The RT-PCR test works by looking for little bits of viral RNA and then amplifying them.

The number of "cycles" of amplification you do allows you to tune the test depending on whether sensitivity or specificity is more important.

Sensitivity is how good the test is at finding the thing you're looking for. Set the cycle threshold too low and you might miss cases, especially if you didn't get a decent swab.

Specificity is how good the test is at not giving you false positives. Crank the cycle threshold right up and the test will start detecting its own primers and report that it has found SARS2 RNA In a pineapple.

The UK tests will be using a reasonable cycle threshold (or were the last time I checked) giving them a sensitivity of about 70% and a specificity of at least 95% probably a bit better. So this means if you test 100 people who are infected it will miss about 30 if them. But if you test 100 people who are not infected it will falsely tell you about 5 of them are. It can also find smashed up bits of virus in people who have recovered so this will reduce the specificity a bit more (but I don't know by exactly how much, it's fairly academic anyway, see below).

The specificity of the test becomes dramatically more important the lower the actual prevalence of the thing you're looking for. Because levels of Covid in the UK are now so low, if you test positive it doesn't mean there is a 95% you actually have Covid. If 5 out of 100 tests are false positives but only 0.03% of people really have Covid (which is about the current UK prevalence) then there is only about a 0.6% chance that having tested positive you really have Covid (0.03/5).

If you test negative it also means you probably don't have Covid because even though 30% people will test negative anyway if they have Covid, you only had a 0.03% of having Covid anyway.

So in conclusion the test is a perfectly good test with a reasonable sensitivity and specificity. At the height of the epidemic if you'd had one that gave you results quickly it would have been useful. But now that the prevalence in the UK is so low, the test is not able to give a meaningful result to an individual just because of the math. The test result just tells you whether it is exceedingly unlikely you have Covid or extremely unlikely you have Covid.

It only has a use now when doing population level studies with very large samples. You need to test tens of thousands of people at random to try to gauge the real prevalence.

guy153
Posts: 48
Joined: Wed Apr 22, 2020 5:29 pm

Re: Covid 19 testing

Post by guy153 »

I should mention another source of false positives is the test finding not SARS2 but a different related coronavirus. Scientifically speaking it ought to be reasonable to require a match on sequences that are unique to SARS2 (at least out of known coronaviruses) but there have been reports of the WHO playing silly buggers with the official test criteria. I don't know how significant that is.

Ianric
Posts: 10
Joined: Fri Jul 31, 2020 11:46 pm

Re: Covid 19 testing

Post by Ianric »

guy153 wrote: Fri Jul 31, 2020 9:42 am I should mention another source of false positives is the test finding not SARS2 but a different related coronavirus. Scientifically speaking it ought to be reasonable to require a match on sequences that are unique to SARS2 (at least out of known coronaviruses) but there have been reports of the WHO playing silly buggers with the official test criteria. I don't know how significant that is.
I have heard coronavirus are a family of viruses and I was also wondering if people testing positive have another coronavirus.

Kevin_Sceptic
Posts: 30
Joined: Wed Jul 29, 2020 11:04 am

Re: Covid 19 testing

Post by Kevin_Sceptic »

Very informative, @guy153, thank you.

guy153
Posts: 48
Joined: Wed Apr 22, 2020 5:29 pm

Re: Covid 19 testing

Post by guy153 »

Ianric wrote: Fri Jul 31, 2020 11:51 pm I have heard coronavirus are a family of viruses and I was also wondering if people testing positive have another coronavirus.
They might but it's unlikely the test would falsely identify another known coronavirus as it deliberately looks for genes that are found only in SARS2 (although some of them are also found in SARS1, the closest known human-infecting relative, but that seems to have died out).

But if you are infected with some unknown close relative of SARS2 the effect is the same: try not to infect anyone else, get plenty of rest, and seek medical attention if you have difficulty breathing.

There's a good graphical calculator here that uses "Bayes Formula" to work out the probability you have Covid if you have a positive test based on prevalence, test sensitivity and specificity:

https://www.bmj.com/content/369/bmj.m1808

The box on the top left defaults to 80%. This is hopelessly unrealistic as the number of people infected with SARS-2 at any one time will only ever get that high at the peak of a very rapid outbreak in a small crowded environment (a prison or care home perhaps).

The lowest you can set that to is 1%, which means a positive result means you have a 16% chance of actually having Covid. But the real prevalence is much lower than that-- 0.09% in the most recent ONS Covid-19 Infection Survey. But even that is so low that the 95CI (if you account for the specificity of the test, which the ONS don't appear to) starts well below zero.

Ianric
Posts: 10
Joined: Fri Jul 31, 2020 11:46 pm

Re: Covid 19 testing

Post by Ianric »

guy153 wrote: Sat Aug 01, 2020 9:19 am
Ianric wrote: Fri Jul 31, 2020 11:51 pm I have heard coronavirus are a family of viruses and I was also wondering if people testing positive have another coronavirus.
They might but it's unlikely the test would falsely identify another known coronavirus as it deliberately looks for genes that are found only in SARS2 (although some of them are also found in SARS1, the closest known human-infecting relative, but that seems to have died out).

But if you are infected with some unknown close relative of SARS2 the effect is the same: try not to infect anyone else, get plenty of rest, and seek medical attention if you have difficulty breathing.

There's a good graphical calculator here that uses "Bayes Formula" to work out the probability you have Covid if you have a positive test based on prevalence, test sensitivity and specificity:

https://www.bmj.com/content/369/bmj.m1808

The box on the top left defaults to 80%. This is hopelessly unrealistic as the number of people infected with SARS-2 at any one time will only ever get that high at the peak of a very rapid outbreak in a small crowded environment (a prison or care home perhaps).

The lowest you can set that to is 1%, which means a positive result means you have a 16% chance of actually having Covid. But the real prevalence is much lower than that-- 0.09% in the most recent ONS Covid-19 Infection Survey. But even that is so low that the 95CI (if you account for the specificity of the test, which the ONS don't appear to) starts well below zero.
A common skeptic question is that if coronavirus is dangerous and produces severe symptoms, why would they need to test people to see if they have the illness. For instance, the symptoms of illnesses such as bubonic plague and Ebola are so horrific, you wouldn't need to test people to see if they have the illness. Do you feel this is a valid point.

guy153
Posts: 48
Joined: Wed Apr 22, 2020 5:29 pm

Re: Covid 19 testing

Post by guy153 »

Ianric wrote: Sat Aug 01, 2020 11:25 am A common skeptic question is that if coronavirus is dangerous and produces severe symptoms, why would they need to test people to see if they have the illness. For instance, the symptoms of illnesses such as bubonic plague and Ebola are so horrific, you wouldn't need to test people to see if they have the illness. Do you feel this is a valid point.
Yes this is a valid point. For some things you want to do a test in order to decide on a treatment. But at the height of the pandemic, we either didn't have tests (in the UK) or the results took about a week to come back. Either is useless if you need to treat someone who is having trouble breathing. Doctors got pretty good at spotting Covid from lung X-rays and were using that for diagnosis.

So what is the point of testing? It can be useful if you are trying to contain an epidemic before it has spread. If you catch cases early, trace contacts, and quarantine anyone who tests positive, even if asymptomatic, it is possible to keep the infection rate in your country very low without having a lockdown. Iceland achieved this. But it's not clear how much more you achieve with this than people just voluntarily quarantining themselves when they have symptoms of any kind of cold or flu, and using simple devices like IR thermometers. Although asymptomatic people and presymptomatic people can spread the virus in theory, it's likely only to be a small amount, so it makes more sense to concentrate your efforts on the symptomatic. This is far easier to do, requires no technology, and has almost no collateral damage to society or health.

The other thing testing lets you do, which is actually useful, is just to know if the pandemic has reached your country yet at all. If not then it might be worth being very careful at your borders and quarantining people coming in. But SARS2 had got almost the whole way around the world before enough tests were available to make this viable. As soon as most places had the tests it was already too late-- they were finding hundreds of cases and there wasn't much point testing. But New Zealand got the virus late enough relative to the tests for this to be useful.

Even then it has very limited use as a tool to keep SARS2 out of your country. All it does is tell you that you don't have SARS2 there yet and so it's worth having a stab at keeping things that way. But the test is only about 70% sensitive. If a plane landed from NY in mid-March, there could easily have been 100 active infections on board, 30 of whom would have tested negative. So if you want to keep SARS2 out indefinitely your only option really is to ban flights from places where the epidemic isn't over and then require everyone to quarantine anyway just in case. This may sounds extremely strict but it is what you need to keep infections at zero.

There is an enormous difference between the strategies you need to keep Covid at zero (which is practically impossible, and most importantly, you have to know when to give up) and just damping it down a bit (which is quite easy and doesn't have to destroy society).

Ianric
Posts: 10
Joined: Fri Jul 31, 2020 11:46 pm

Re: Covid 19 testing

Post by Ianric »

guy153 wrote: Sat Aug 01, 2020 12:15 pm
Ianric wrote: Sat Aug 01, 2020 11:25 am A common skeptic question is that if coronavirus is dangerous and produces severe symptoms, why would they need to test people to see if they have the illness. For instance, the symptoms of illnesses such as bubonic plague and Ebola are so horrific, you wouldn't need to test people to see if they have the illness. Do you feel this is a valid point.
Yes this is a valid point. For some things you want to do a test in order to decide on a treatment. But at the height of the pandemic, we either didn't have tests (in the UK) or the results took about a week to come back. Either is useless if you need to treat someone who is having trouble breathing. Doctors got pretty good at spotting Covid from lung X-rays and were using that for diagnosis.

So what is the point of testing? It can be useful if you are trying to contain an epidemic before it has spread. If you catch cases early, trace contacts, and quarantine anyone who tests positive, even if asymptomatic, it is possible to keep the infection rate in your country very low without having a lockdown. Iceland achieved this. But it's not clear how much more you achieve with this than people just voluntarily quarantining themselves when they have symptoms of any kind of cold or flu, and using simple devices like IR thermometers. Although asymptomatic people and presymptomatic people can spread the virus in theory, it's likely only to be a small amount, so it makes more sense to concentrate your efforts on the symptomatic. This is far easier to do, requires no technology, and has almost no collateral damage to society or health.

The other thing testing lets you do, which is actually useful, is just to know if the pandemic has reached your country yet at all. If not then it might be worth being very careful at your borders and quarantining people coming in. But SARS2 had got almost the whole way around the world before enough tests were available to make this viable. As soon as most places had the tests it was already too late-- they were finding hundreds of cases and there wasn't much point testing. But New Zealand got the virus late enough relative to the tests for this to be useful.

Even then it has very limited use as a tool to keep SARS2 out of your country. All it does is tell you that you don't have SARS2 there yet and so it's worth having a stab at keeping things that way. But the test is only about 70% sensitive. If a plane landed from NY in mid-March, there could easily have been 100 active infections on board, 30 of whom would have tested negative. So if you want to keep SARS2 out indefinitely your only option really is to ban flights from places where the epidemic isn't over and then require everyone to quarantine anyway just in case. This may sounds extremely strict but it is what you need to keep infections at zero.

There is an enormous difference between the strategies you need to keep Covid at zero (which is practically impossible, and most importantly, you have to know when to give up) and just damping it down a bit (which is quite easy and doesn't have to destroy society).
We hear about the number of cases. How are cases determined. Is it based on test results alone or people reporting symptoms without testing being classed as cases.

guy153
Posts: 48
Joined: Wed Apr 22, 2020 5:29 pm

Re: Covid 19 testing

Post by guy153 »

Traditionally a "case" is someone who's ill and turns up needing treatment of some kind. But in recent weeks the meaning has changed for the purposes of Covid-19 to mean anyone who tests positive.

A clue that most of the positives being found now are false positives is that we keep reading they are "almost all asymptomatic". If they were true positives you would expect the number of asymptomatic infections to be between 30% and 80% as this is the sort of range it was in during the actual epidemic.

Post Reply