Covid 19

[andrewmcleod]

Those questions again:

What was the false positive percentage of the test in the Healy et al study? (2 marks)

Why are so many of the positive tests in the Healy et al study false positives? (5 marks)

Why will this currently be less of an issue (in some ways)? (2 marks)

(I've changed 'false positive rate' to 'false positive percentage of the test' in Q1 for accuracy)

 

[aardgoose]

Interesting use of loaded language

By their own admission they have only sequenced 10% of positive tests

.  10% is a massive number of fully sequenced samples were 10% = 100,000 samples. The likelihood of that 10% being unrepresentative is extremely small. Its not an "admission", with all the implications of guilt that word carries, it is just presenting facts, which actually, for once, represent a significant effort and maybe 'world beating'.

Another thing to take away from the single vs multiple gene results is that the S gene test dropouts (the gene for the spike protein) is a very strong signal of the increase in the new and very likely more infectious variety, where mutations in that gene cause this effect. Ie the number of tests where the S gene is not detected are increasing over time, which in turn will increase the number of single gene only responses as the new variant becomes predominant, where a two gene result would have previously been obtained.

Also confirmed by the full sequencing (as noted above, 10% is a good enough sample to verify this).

The false positive myth has been repeatedly demolished, the only people persisting with this are contrarian journalists with poor numeracy. Again as noted above, a false positive rate should be constant, but yet we are expected to believe it is doubling practically weekly, without any mechanism.

All i can see is motivated reasoning, and an unwillingness understand. You are determined not to believe overwhelming evidence while claiming to be open minded. Even without the testing data, you would have to believe that major hospitals across the country are producing false statistics, and that hundreds of doctors are committing a criminal offence in completing death certificates. Why don't you want to accept the overwhelming evidence? You are implying a massive conspiracy within the medical profession world wide.

 

[maxf]

Those questions again:

What was the false positive percentage of the test in the Healy et al study? (2 marks)

The document states a total of 5110 tests were used in the study, going by the fact that only 19 of the original 31 were retested and could be proven to not test positive again (its best of three with rock paper scissors ...?) I would say that 19/5110 were false positives. 0.37%

Why are so many of the positive tests in the Healy et al study false positives? (5 marks)

I'm guessing because you want me to say that most of them in the false positive bracket were also in the asymptomatic box.

Why will this currently be less of an issue (in some ways)? (2 marks)

I guess you want me to say because we are testing more symptomatic people now than asymptomatic people,

(I've changed 'false positive rate' to 'false positive percentage of the test' in Q1 for accuracy)

A total of 26 people could/ would have been;

i) Told to isolate (increased work pressure on others, trigger more asymptomatic testing in their work environment)
ii) Refused necessary surgery
iii) Been unnecessarily put in a Covid ward where the chances of catching it if you don't already have it are much higher

Based on the results of their test....

There was no need to do this for 19 of them....


How many NHS staff, care workers, teachers, (don't worry we are all equally important I cant be bothered to list every profession) etc etc are not at work right now for the same reason ?

 

[andrewmcleod]

i) Correct; I would have accepted any reasonably justified answer around 0.5% (as there are several ways to interpret 'false positive'). 2 marks, good answer.
ii) Incorrect: 0 marks.
The reason a high number of the positive test results were false positives was because, as stated in the paper, the study was completed when the prevalence was very low (June-July). Consequently, the number of true positive tests will be very low, while the fraction of false positive tests remains high. Thus when the prevalence is very low, a significant fraction of positive tests will actually be false positives. This is well-known and understood.
iii) This question could be fairly disputed as it's arguably a _little_ bit unfair. As the prevalence of the virus has grown dramatically, the fraction of false positives will have remained the same but the fraction of true positives will have increased dramatically. Therefore the fraction of positive tests that are false positives will go down significantly - instead of most positive results being 'false', most positive results will be 'true'. However, the fraction of false positive tests will remain the same, so the same fraction (<0.5%) of tests will result in the wrong advice (to isolate, to miss work) or the wrong treatment (moving into a Covid ward, for example). So arguably while the tests become 'better' at higher prevalence (more likely to be correct), the problem hasn't really gone away.

So, in your infinite wisdom: what do you suggest doing about this very small fraction of tests that results in poor treatment? Doctors will (or should) know about the risk of false positives; I suspect most positive tests will, where possible, we followed up by a second test to verify.

The false positive rate for the ONS study was less than 0.1%, because the false positive rate cannot exceed the minimum positive rate (once accounting for a bit of statistical uncertainty).

Fair play for answering, though...

 

[andrewmcleod]

A total of 26 people could/ would have been;

i) Told to isolate (increased work pressure on others, trigger more asymptomatic testing in their work environment)
ii) Refused necessary surgery
iii) Been unnecessarily put in a Covid ward where the chances of catching it if you don't already have it are much higher

Based on the results of their test....

There was no need to do this for 19 of them....

So what would you have done? Why do you think you know better than the doctors who were treating these patients?

How many NHS staff, care workers, teachers, (don't worry we are all equally important I cant be bothered to list every profession) etc etc are not at work right now for the same reason ?

Well, we can estimate that if we assume a 0.1% false positive rate...
Let's use 1% for the current positivity rate (Fig 1 https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/bulletins/coronaviruscovid19infectionsurveypilot/18december2020 )
About 420,000 tests were carried out on 30th December 2020.

So around 4,200 positive tests, of which 420 would be false positives and about 3780 would be true positives.

Yes, that's a lot of people getting a false positive result. If they get retested, probably none of them will get a second false positive. But do we not tell all 4,200 to isolate just because 10% are false positives? We can't afford to do that. Ideally, we get all those people a second test. But 90% of the time, the positive test will be correct (at 1% positivity and 0.1% false positive, which is actually a very conservative estimate for the ONS data - it's probably lower).

 

[mikem]

Basically health care professionals & government advisors are well aware of the discrepancies and will be mitigating against it. Just the government doesn't wish to introduce another fudge factor to their sales pitch.

 

[maxf]

Test twice before being counted as positive or use two genes minimum, the second for verification. Publish all information about what gene was tested for and at what Ct with test result.

My experience of the testing has not involved contact from any doctors...

 

[maxf]

How do you think we will exit the situation ? We had a nice drop in cases last summer, vaccination by then isn't going to prevent the amount we had around last summer still testing positive given the current methods. Either testing or test methods will have to be relaxed or a certain amount of deaths made 'OK' for us to move on....

 

[Speleofish]

I'm trying to be polite but I can't quite manage it. Please will you stop talking bollocks?

 

[andrewmcleod]

Test twice before being counted as positive

Great if you have the capacity. Probably a bit impractical at the moment. Taking a second test probably means taking a second sample at a different time, not just running the same analysis again on the same sample.

or use two genes minimum, the second for verification.

And dramatically increase the numbers of false negatives, thereby causing people with the virus to carry on spreading it. Making people isolate when they don't need to is mostly just inconvenient. False negatives can help spread the virus and thus kill.

Publish all information about what gene was tested for and at what Ct with test result.

And this is useful information for people because? People don't want to know 'your test returned gene X and Y at CT value of 33'. They want 'best guess yet' or 'best guess no', based on thresholds set by people to maximise the accuracy of the test. You can set your thresholds so that you either minimise the number of false negatives at the cost of false positives, or minimise false positives at the cost of false negatives, or go for some optimal accuracy. Surely false negatives are worse than false positives anyway?

 

[pwhole]

Making people isolate when they don't need to is mostly just inconvenient.

To be fair, that may not be the case - I was exposed to a 'positive' case recently (long story) and I knew for certain I hadn't caught anything but the people who inflicted this ghastly nightmare on me suggested I should self-isolate - three days before Christmas with no warning and no fault of mine. This would have been much more than just inconvenient to me as I live on my own and don't really have 'shoppers' I can call on. I'm sure I could have bribed someone, but then that's more than inconvenient for them too.

If I get offered work at any time in the next month I will have to take it as I need to work, and it probably won't be at home eitther. My life's monkish enough already and I'm fed up of swerving around idiots and assholes who can't even walk straight in a supermarket, never mind navigate a government edict and act appropriately. I've lived like a saint for nine months and hardly done any work and I'm still getting screwed over by people with way more money than me but with far less common sense. I don't qualify for any government help at all as I'm not registered self-employed so it's galling to still be stuck here doing nothing when I was on a wonderful well-paid job when lockdown started in March. I wouldn't have given a shit now if we'd worked that one out to the end and would be eating Toblerone for breakfast every day.

Anyway, the positive case showed no symptoms and the friends of mine who were forced to isolate with the person also showed no symptoms, despite being not that fit and mid-50s. I also showed no symptoms and am much fitter, though still mid-50s. The testee had taken a previous test a week before and was negative, enabling them to leave their university halls and come home for Christmas. A 'just to be sure' test before travelling further afield was then positive. None of us have any idea whether any of the tests were really accurate as no-one suffered any symptoms at all, and in my case I was probably 4m away upwind! I'm not daft. But given the three of them were then living together and two of them are middle-aged, recently ex-smokers and not that fit, how statistically likely is it that all three of them would be asymptomatic?

 

[RobinGriffiths]

So, it's exercise once a day, but not out of your local area. Is Caving exercise or a recreation? Discuss...  I'm off...

 

[Duck ditch]

Can anyone explain the logic of sending schools back for one day.  Let the kids mix and then shut them all down. 
Caving is exercise.  Natural aerobics. 

 

[mikem]

Government on recess...

 

[andrewmcleod]

Anyway, the positive case showed no symptoms and the friends of mine who were forced to isolate with the person also showed no symptoms, despite being not that fit and mid-50s. I also showed no symptoms and am much fitter, though still mid-50s. The testee had taken a previous test a week before and was negative, enabling them to leave their university halls and come home for Christmas. A 'just to be sure' test before travelling further afield was then positive. None of us have any idea whether any of the tests were really accurate as no-one suffered any symptoms at all, and in my case I was probably 4m away upwind! I'm not daft. But given the three of them were then living together and two of them are middle-aged, recently ex-smokers and not that fit, how statistically likely is it that all three of them would be asymptomatic?

Those are, I suspect, hard questions. Firstly Covid's distribution is very clustered, so it's possible the person did actually have Covid but was asymptomatic and didn't pass it on to anyone else (I think a lot of people don't pass it on and a few people pass it on to lots?) - but whether anyone would actually want to estimate a probably I have no idea.

But a positive test is much more likely to be accurate (in the strict statement of 'reveals the presence of SARS-COV-2 genes in the body' not 'diagnoses Covid-19') than not.

It's all a bit of a bugger; it's going to suck for a lot of people in a lot of different ways. I suspect people would be a lot more willing to play by the rules if, as you allude, the government were to support them in doing so.

 

[Pitlamp]

I suspect people would be a lot more willing to play by the rules if, as you allude, the government were to support them in doing so.

It might help if Dominic Cummings was to be punished finally for breaking the law last year.

 

[AR]

At least the SNP MP has been called to account over her behaviour - https://www.bbc.co.uk/news/uk-scotland-scotland-politics-55537974

 

[Pitlamp]

Yes - but - going for a drive, with the family, to see if your eyes are working . . . . ?

Boris Johnson's answer was we should put it all behind us. Try pleading that when you perform a motoring offence.

 

[Fjell]

It?s the difference between rules and common sense.

1. Going for a drive and walk.
2. Getting on a train when you know you have covid.

Not that I agree with blatant political prosecution. I think Scotland should think long and hard before leaving the Dear Leader in complete charge. Makes Hungary look liberal.

 

[Duck ditch]

Why aren?t we locking down properly.  Leaving the airports open and passengers free to roam is like trying to empty a bath while leaving the taps running.
It?s a real shame Sweden?s plan didn?t work.  God I wished it had.  It would be the simplest solution. I?m sure we would have adopted that policy if that had worked.  However what has success is countries like South Korea and Vietnam who have locked down properly with success.  Why oh Why don?t we adopt there ideas?  I think we haven?t because they have slanty eyes.