Facts about Covid-19

[BeElBeeBub]

Matt Hancock was indeed wrong but so was some of the further explanations on this.

The 0.8% estimate (and it's an estimate not fully proven) means that if you test 1000 people who definitely do NOT have Covid, you would get 80 positive results.

Agreed up to here.

The big issue from a statistics point of view is we are not only testing people who are known not to have covid, nor are we even testing a random sample - we are actually testing people who think that they have it already.

Not really. We are testing people who are being forced into a test, whether they want one or not, because their school, workplace, or nursery won't allow them back without a negative test. They have colds, highly prevalent at this time of year when the new term starts. Happens every year.

So first, the sample is not random, and second, the probability calculations depend very much on the prevalence of the disease in society.

The test results and accuracy thereof are not influenced by the prevalence of the disease, but the rate of false positives is.

You cannot conclude from the daily testing stats that most of the results are false positives - this is nonsense.  The statistician guy was on yesterday explaining how a lot of journalists are getting these nuances wrong because they are not trained in statistics.

As Thomas often says. Please quote me, using the quote tags where I have concluded from the daily testing stats that most of the results are false positives. Don't misrepresent my position, as this thread is about facts, I've done my best to stick to facts.

Apart from anything else, the positive rate as a % of tests is over 4% right now and rising, so even if you could conclude (you can't) that 0.8% of those are false positives, there is still a massive rise in relative terms.

Either you have different figures or your maths is defective, but 6634 positives from 416,363 tests is 1.59%

Further, some of the ONS data on some weeks where there was very low rates in August showed positive test rates of only about 0.2%, so this seems to throw into question the idea that 0.8% of results are false positives, even if you assume that every single one of  those ONS tests was a false positive.

Yes, that is a good and very reasonable argument. However, the testing numbers have tripled since then, and the greater the pressure on labs, the greater the likelihood of error.

A couple of points.

The number of test carried out per day has been around the 200-250k mark (from 10th Sept to 20th Sept).  That includes multiple test per person i.e. the actual number of people tested is lower than the number of tests by an unknown number.

The average number of new cases over the same time period was around 3.5k (again round numbers).

That gives a positive rate of at least  1.4% (3.5k/250k).

This is above the false positive rate for the testing which is thought to be less than 1%.

But we can dig down in weekly figures for an idea of what the number of people tested as opposed to tests conducted.

If you look here https://www.gov.uk/government/publications/nhs-test-and-trace-england-and-coronavirus-testing-uk-statistics-10-september-to-16-september-2020 and look at the data tables you can see for the week 10-16 Sept 1,700k tests were recorded but only 587k people were tested (roughly 3 tests for every person) and 19k positives were recorded (a bit over 3.25%).  By looking back over the weeks to June the lowest positive rate of 1.12% was in 9-15 July where 360k people were tested and 4k were positive.

July and August saw +ve rates of between 1 and 1.5% for 350-450k tests.

If you made the argument that there was no CV in the UK over July/August then the false +ve rate of the test would be 1-1.5%  and we are currently seeing double that.

[Borchester]

No more dogging for you Borchester.
https://uk.yahoo.com/news/matt-hancock-awkwardly-bans-casual-101408452.html

Damn. And I was so looking forward to a traditional Christmas.

[srb7677]

No more dogging for you Borchester.

And no more sheep worrying for you, Sheepy

[Sheepy]

Matt Hancock apparently showed his ignorance when he answered a question on false positives. He didn't understand the maths.

This is how it works:
NHS now testing 200,000 people a day.
Suppose 2000 (1%) are actually positive.
Say the false positive rate is 0.8% (It is known to be less than 1% but depends on lab procedure)
198,000 x 0.8%  = 1584 people are added to the positive tests, making 3,584. The result is now inflated by 79%.

Matt Hancock thought that 16 of the 2000 would be incorrect. Go to the bottom of the class.
How's my maths, Mr Borchester?

The 0.8% estimate (and it's an estimate not fully proven) means that if you test 1000 people who definitely do NOT have Covid, you would get 80 positive results.

Actually Javert, I think that you will find that 0.8% of a 1000 is 8, not 80. However, as long as this nonsense is over by Christmas and we can all have a bit of a bop I am not that bothered.

No more dogging for you Borchester.
https://uk.yahoo.com/news/matt-hancock-awkwardly-bans-casual-101408452.html

[Borchester]

Matt Hancock apparently showed his ignorance when he answered a question on false positives. He didn't understand the maths.

This is how it works:
NHS now testing 200,000 people a day.
Suppose 2000 (1%) are actually positive.
Say the false positive rate is 0.8% (It is known to be less than 1% but depends on lab procedure)
198,000 x 0.8%  = 1584 people are added to the positive tests, making 3,584. The result is now inflated by 79%.

Matt Hancock thought that 16 of the 2000 would be incorrect. Go to the bottom of the class.
How's my maths, Mr Borchester?

The 0.8% estimate (and it's an estimate not fully proven) means that if you test 1000 people who definitely do NOT have Covid, you would get 80 positive results.

Actually Javert, I think that you will find that 0.8% of a 1000 is 8, not 80. However, as long as this nonsense is over by Christmas and we can all have a bit of a bop I am not that bothered.

[papasmurf]

Here are some facts about Covid-19:-

More at link:-

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/coronaviruscovid19relateddeathsbydisabilitystatusenglandandwales/2marchto14july2020#breakdown-of-deaths-involving-covid-19-by-age-sex-and-disability-status

The number of deaths involving the coronavirus (COVID-19) and their percentage distribution across disability categories among the study population to July 2020 is shown in Table 1. Disabled people (those limited a little or limited a lot) made up 6 in 10 (59.5%) of all deaths involving COVID-19 for the period to July 2020 (27,534 of 46,314 deaths). Disabled people made up around 16% of the study population followed from 2 March.

[Barry]

Important notes and usage information
Main points from latest release
The number of deaths registered in England and Wales in the week ending 11 September 2020 (Week 37) was 9,811; this was 2,072 deaths higher than in Week 36.

In Week 37, the number of deaths registered was 5.4% above the five-year average (505 deaths higher).

The number of death registrations may have been affected by the August Bank holiday (31 August). This can cause delays in deaths being registered in Week 36, resulting in an increase in deaths being registered in Week 37.

Of the deaths registered in Week 37, 99 mentioned "novel coronavirus (COVID-19)", accounting for 1.0% of all deaths in England and Wales.

This is the second lowest number of deaths involving COVID-19 since Week 12; an increase of 21 deaths compared with Week 36 (26.9% increase).

The number of deaths in hospitals was below the five-year average in Week 37 (371 fewer deaths), while the numbers of deaths in private homes and care homes were above the five-year average (830 and 57 more deaths respectively).

The number of deaths involving COVID-19 increased across five of the nine English regions; the North West was the only English region to have lower overall deaths than the five-year average.

In Wales, the number of deaths involving COVID-19 decreased from four deaths (Week 36) to one death (Week 37), while the total number of deaths in Week 37 was similar to the five-year average (one death higher).

The number of deaths registered in the UK in the week ending 11 September 2020 (Week 37) was 11,145, which was 576 deaths higher than the five-year average and 2,149 deaths higher than Week 36; of the deaths registered in the UK in Week 37, 110 deaths involved COVID-19, 27 deaths higher than Week 36.

Source: ONS
99 of 9811 mentioned COVID-19. 1% of deaths.

[Barry]

Matt Hancock was indeed wrong but so was some of the further explanations on this.

The 0.8% estimate (and it's an estimate not fully proven) means that if you test 1000 people who definitely do NOT have Covid, you would get 80 positive results.

Agreed up to here.

The big issue from a statistics point of view is we are not only testing people who are known not to have covid, nor are we even testing a random sample - we are actually testing people who think that they have it already.

Not really. We are testing people who are being forced into a test, whether they want one or not, because their school, workplace, or nursery won't allow them back without a negative test. They have colds, highly prevalent at this time of year when the new term starts. Happens every year.

So first, the sample is not random, and second, the probability calculations depend very much on the prevalence of the disease in society.

The test results and accuracy thereof are not influenced by the prevalence of the disease, but the rate of false positives is.

You cannot conclude from the daily testing stats that most of the results are false positives - this is nonsense.  The statistician guy was on yesterday explaining how a lot of journalists are getting these nuances wrong because they are not trained in statistics.

As Thomas often says. Please quote me, using the quote tags where I have concluded from the daily testing stats that most of the results are false positives. Don't misrepresent my position, as this thread is about facts, I've done my best to stick to facts.

Apart from anything else, the positive rate as a % of tests is over 4% right now and rising, so even if you could conclude (you can't) that 0.8% of those are false positives, there is still a massive rise in relative terms.

Either you have different figures or your maths is defective, but 6634 positives from 416,363 tests is 1.59%

Further, some of the ONS data on some weeks where there was very low rates in August showed positive test rates of only about 0.2%, so this seems to throw into question the idea that 0.8% of results are false positives, even if you assume that every single one of  those ONS tests was a false positive.

Yes, that is a good and very reasonable argument. However, the testing numbers have tripled since then, and the greater the pressure on labs, the greater the likelihood of error.

[Javert]

Matt Hancock apparently showed his ignorance when he answered a question on false positives. He didn't understand the maths.

This is how it works:
NHS now testing 200,000 people a day.
Suppose 2000 (1%) are actually positive.
Say the false positive rate is 0.8% (It is known to be less than 1% but depends on lab procedure)
198,000 x 0.8%  = 1584 people are added to the positive tests, making 3,584. The result is now inflated by 79%.

Matt Hancock thought that 16 of the 2000 would be incorrect. Go to the bottom of the class.
How's my maths, Mr Borchester?

Matt Hancock was indeed wrong but so was some of the further explanations on this.

The 0.8% estimate (and it's an estimate not fully proven) means that if you test 1000 people who definitely do NOT have Covid, you would get 80 positive results.

The big issue from a statistics point of view is we are not only testing people who are known not to have covid, nor are we even testing a random sample - we are actually testing people who think that they have it already.

So first, the sample is not random, and second, the probability calculations depend very much on the prevalence of the disease in society.

You cannot conclude from the daily testing stats that most of the results are false positives - this is nonsense.  The statistician guy was on yesterday explaining how a lot of journalists are getting these nuances wrong because they are not trained in statistics.

Apart from anything else, the positive rate as a % of tests is over 4% right now and rising, so even if you could conclude (you can't) that 0.8% of those are false positives, there is still a massive rise in relative terms.

Further, some of the ONS data on some weeks where there was very low rates in August showed positive test rates of only about 0.2%, so this seems to throw into question the idea that 0.8% of results are false positives, even if you assume that every single one of  those ONS tests was a false positive.

[Sheepy]

There are rumours Barry, you ain't scared enough and have been an instigator of freedom.
https://uk.yahoo.com/news/coronavirus-mutation-could-evolving-around-022142707.html
So the Torygraph needs a word.

[Barry]

[Nick]

Matt Hancock apparently showed his ignorance when he answered a question on false positives. He didn't understand the maths.

This is how it works:
NHS now testing 200,000 people a day.
Suppose 2000 (1%) are actually positive.
Say the false positive rate is 0.8% (It is known to be less than 1% but depends on lab procedure)
198,000 x 0.8%  = 1584 people are added to the positive tests, making 3,584. The result is now inflated by 79%.

Matt Hancock thought that 16 of the 2000 would be incorrect. Go to the bottom of the class.
How's my maths, Mr Borchester?

Also the fact that a CV-19 death is anyone who had it in the 28 days up to their death.

You had it mildly, got tested positive and on your way out to the pub you get knocked down and killed. Another CV-19 death.

[Borg Refinery]

Hopefully 'events' will 'disrupt' things.

https://pol-tics.com/index.php/topic,1864.0.html

Hard headed pragmatism worked well for Genghis Khan too! 

[Thomas]

All true Tommy. Sometimes I think that you and I are the only folk with the sort of hard headed cynicism to run the country 

[Borchester]

I don't see the point in testing 200,000 a day. At such a rate it would take nearly a year to test the entire country, by which time most everybody would either be dead or recovered.

Its like the placebo effect borkie.

The governemnt know fine well this flu is going to go wround every human in these islands ,and the world , so they have to be seen to be doing something to calm the peoples fears , when in reality they know they can't do anything about it.

Thats the hard feckin truth no one wants to talk about.

All true Tommy. Sometimes I think that you and I are the only folk with the sort of hard headed cynicism to run the country