Scrutable

jdc wrote: Fri Oct 23, 2020 7:08 pm So winning all round then.

I think the phrase you are looking for is "World-beating".

Oven-ready.

Countries' case numbers are undoubtedly useful for monitoring the trajectory within each country (unless there are drastic changes to testing rates).

But it's difficult to compare between countries with different testing policies. Portugal, for instance, does seem to have unlimited tests for whoever wants them, whereas the UK has an explicit policy of restricting tests to people with symptoms, along with several other choices (self administration, postal delivery) known to increase the false positive rate. So if both countries were showing the same per capita case numbers it would be wise to assume the UK probably has more - but I would have no idea what the likely ratio would be.

That's why I was asking if anyone is modelling it, as there are loads of statistical methods for estimating detection rates from observational data.

An example of someone I know recently having problems with access to tests in the UK: she is a household member of someone who'd had a positive test, and had sore throat and mild fatigue herself. But this wasn't enough to qualify for a test. When she lied and said she'd lost her sense of smell, the system then allowed her to book one. Her test came back positive too. It's like we've designed the system to minimise the chances of identifying mild/asymptomatic cases, despite the fact we know they can spread the virus too.

There's the ONS survey, which gives a realistic estimate of the true prevalence of people infected with the virus by sampling people at random (though I think there a few caveats). Latest figures are 1 in 130 for England, a bit better in W and S, worse in NI. I'm sure other countries will have similar surveys.

(I've gotta run and haven't been able to check any of the assertions in this post, which are based on my incomplete understanding.)

Everyone...

Yes, I agree, other measures of Covid infections would be better for a cross-national comparison. I just used infections over the past two weeks because at the time I happened to be looking at a map showing that data.

I think that the best comparison would be excess deaths as that will avoid problems with different national testing systems, and secondary deaths due to overstretched healthcare systems are also relevant. However I’m looking after the kids now so can’t go looking for the data.

Sciolus wrote: Sat Oct 24, 2020 8:04 am There's the ONS survey, which gives a realistic estimate of the true prevalence of people infected with the virus by sampling people at random (though I think there a few caveats). Latest figures are 1 in 130 for England, a bit better in W and S, worse in NI. I'm sure other countries will have similar surveys.

Not sure about the ONS survey - but knowing someone who was invited to take part in the similar Imperial College (REACT) survey - my hunch is the false negative rate is going to be higher for these surveys than Tier 2 testing. Couple of reasons...

- A proportion (not all) of Tier 2 tests are carried out by people trained to carry out the swabbing procedure. The ONS and Imperial (REACT) survey swabs are all self-administered.

- Not sure about the ONS, but the Imperial survey is using different swab instructions to some Tier 2 testing. Tier 2 testing instructions normally ask for something close to a nasal mid-turbinate swab (the Randox instructions ask for the swab to be inserted about 2.5cm up the nostril (until resistance is felt) and rotated 5 times against the nasal walls. However, the Imperial (REACT) instructions simply ask for the swab to inserted horizontally and pushed "gently back" rather than "up" and the instruction video just says to roll it between the finger and thumbs.

Presumably less aggressive nasal swabbing is appropriate for volunteer participants who aren't taking part for their own clinical benefit (don't want anyone leaking CSF at home for a research study).

But while these might seem like trivial differences, if you're looking for something with perhaps 1% prevalence in a sample of 30,000 a few false negatives can make a difference.

AMS wrote: Sat Oct 24, 2020 7:25 am An example of someone I know recently having problems with access to tests in the UK: she is a household member of someone who'd had a positive test, and had sore throat and mild fatigue herself. But this wasn't enough to qualify for a test. When she lied and said she'd lost her sense of smell, the system then allowed her to book one. Her test came back positive too. It's like we've designed the system to minimise the chances of identifying mild/asymptomatic cases, despite the fact we know they can spread the virus too.

That's really concerning.

My neighbour - who is in the highly vulnerable category due to ongoing chemotherapy - has been having building works done to adapt the house for her husband who's been in a care home for the last few months after requiring an operation on his knee. Last Tuesday one of the builders went into isolation as their partner has coronavirus. No-one told my neighbour until the end of the week. He can't get a test without symptoms so she has no idea if she's been exposed. She's self-isolating as a precaution. She was due to have chemo next week but is going to have to miss it as a result, and rather than go to the hospital to have her PIC line changed they have to send someone out to do it.

When you consider that Johnson was talking about creating a test for everyone to take every day, to not even allow people to take a test if they've been exposed is a really pathetic state of affairs.

Fishnut wrote: Sat Oct 24, 2020 11:18 am

AMS wrote: Sat Oct 24, 2020 7:25 am An example of someone I know recently having problems with access to tests in the UK: she is a household member of someone who'd had a positive test, and had sore throat and mild fatigue herself. But this wasn't enough to qualify for a test. When she lied and said she'd lost her sense of smell, the system then allowed her to book one. Her test came back positive too. It's like we've designed the system to minimise the chances of identifying mild/asymptomatic cases, despite the fact we know they can spread the virus too.

That's really concerning.

My neighbour - who is in the highly vulnerable category due to ongoing chemotherapy - has been having building works done to adapt the house for her husband who's been in a care home for the last few months after requiring an operation on his knee. Last Tuesday one of the builders went into isolation as their partner has coronavirus. No-one told my neighbour until the end of the week. He can't get a test without symptoms so she has no idea if she's been exposed. She's self-isolating as a precaution. She was due to have chemo next week but is going to have to miss it as a result, and rather than go to the hospital to have her PIC line changed they have to send someone out to do it.

When you consider that Johnson was talking about creating a test for everyone to take every day, to not even allow people to take a test if they've been exposed is a really pathetic state of affairs.

That's so much worse, as at least in my example it's a household with no health issues that increase vulnerability.

They did actually raise it with their (Labour) MP - perhaps your neighbour might too?

AMS wrote: Sat Oct 24, 2020 11:22 am

Fishnut wrote: Sat Oct 24, 2020 11:18 am

AMS wrote: Sat Oct 24, 2020 7:25 am An example of someone I know recently having problems with access to tests in the UK: she is a household member of someone who'd had a positive test, and had sore throat and mild fatigue herself. But this wasn't enough to qualify for a test. When she lied and said she'd lost her sense of smell, the system then allowed her to book one. Her test came back positive too. It's like we've designed the system to minimise the chances of identifying mild/asymptomatic cases, despite the fact we know they can spread the virus too.

That's really concerning.

My neighbour - who is in the highly vulnerable category due to ongoing chemotherapy - has been having building works done to adapt the house for her husband who's been in a care home for the last few months after requiring an operation on his knee. Last Tuesday one of the builders went into isolation as their partner has coronavirus. No-one told my neighbour until the end of the week. He can't get a test without symptoms so she has no idea if she's been exposed. She's self-isolating as a precaution. She was due to have chemo next week but is going to have to miss it as a result, and rather than go to the hospital to have her PIC line changed they have to send someone out to do it.

When you consider that Johnson was talking about creating a test for everyone to take every day, to not even allow people to take a test if they've been exposed is a really pathetic state of affairs.

That's so much worse, as at least in my example it's a household with no health issues that increase vulnerability.

They did actually raise it with their (Labour) MP - perhaps your neighbour might too?

Liam Fox is our MP so I can't imagine it'll do anything. My mum's a councillor though so I'm going to see if she at least find out why she wasn't notified straight away. It's the council paying for the works so they should have said something.

Fishnut wrote: Sat Oct 24, 2020 11:18 am

AMS wrote: Sat Oct 24, 2020 7:25 am An example of someone I know recently having problems with access to tests in the UK: she is a household member of someone who'd had a positive test, and had sore throat and mild fatigue herself. But this wasn't enough to qualify for a test. When she lied and said she'd lost her sense of smell, the system then allowed her to book one. Her test came back positive too. It's like we've designed the system to minimise the chances of identifying mild/asymptomatic cases, despite the fact we know they can spread the virus too.

That's really concerning.

My neighbour - who is in the highly vulnerable category due to ongoing chemotherapy - has been having building works done to adapt the house for her husband who's been in a care home for the last few months after requiring an operation on his knee. Last Tuesday one of the builders went into isolation as their partner has coronavirus. No-one told my neighbour until the end of the week. He can't get a test without symptoms so she has no idea if she's been exposed. She's self-isolating as a precaution. She was due to have chemo next week but is going to have to miss it as a result, and rather than go to the hospital to have her PIC line changed they have to send someone out to do it.

When you consider that Johnson was talking about creating a test for everyone to take every day, to not even allow people to take a test if they've been exposed is a really pathetic state of affairs.

These are probably the kinds of people Matt Hancock accused of lying about symptoms to get a test and blaming them for the shortage. Your poor neighbour, I hope she is OK.

discovolante wrote: Sat Oct 24, 2020 11:45 am

Fishnut wrote: Sat Oct 24, 2020 11:18 am

AMS wrote: Sat Oct 24, 2020 7:25 am An example of someone I know recently having problems with access to tests in the UK: she is a household member of someone who'd had a positive test, and had sore throat and mild fatigue herself. But this wasn't enough to qualify for a test. When she lied and said she'd lost her sense of smell, the system then allowed her to book one. Her test came back positive too. It's like we've designed the system to minimise the chances of identifying mild/asymptomatic cases, despite the fact we know they can spread the virus too.

That's really concerning.

My neighbour - who is in the highly vulnerable category due to ongoing chemotherapy - has been having building works done to adapt the house for her husband who's been in a care home for the last few months after requiring an operation on his knee. Last Tuesday one of the builders went into isolation as their partner has coronavirus. No-one told my neighbour until the end of the week. He can't get a test without symptoms so she has no idea if she's been exposed. She's self-isolating as a precaution. She was due to have chemo next week but is going to have to miss it as a result, and rather than go to the hospital to have her PIC line changed they have to send someone out to do it.

When you consider that Johnson was talking about creating a test for everyone to take every day, to not even allow people to take a test if they've been exposed is a really pathetic state of affairs.

These are probably the kinds of people Matt Hancock accused of lying about symptoms to get a test and blaming them for the shortage. Your poor neighbour, I hope she is OK.

She's fine so far. She's had a pretty tough time with covid as her brother-in-law who lived with her died of coronavirus at the beginning of the pandemic. That she didn't catch it (she did a LOT of disinfecting) despite being in the same house as him gives me hope that she'll be ok this time but it's still a massive worry.

And given the number of asymptomatic cases, you'd think they'd want to test everyone who thought they needed it. Unless there's some fetish for sticking things really far up your nose I'm not aware of*, I can't see people getting tests just for shits and giggles.

* if there is, don't tell me. I'm enjoying my blissful ignorance on this subject!

Fishnut wrote: Sat Oct 24, 2020 11:51 am

discovolante wrote: Sat Oct 24, 2020 11:45 am

Fishnut wrote: Sat Oct 24, 2020 11:18 am

That's really concerning.

My neighbour - who is in the highly vulnerable category due to ongoing chemotherapy - has been having building works done to adapt the house for her husband who's been in a care home for the last few months after requiring an operation on his knee. Last Tuesday one of the builders went into isolation as their partner has coronavirus. No-one told my neighbour until the end of the week. He can't get a test without symptoms so she has no idea if she's been exposed. She's self-isolating as a precaution. She was due to have chemo next week but is going to have to miss it as a result, and rather than go to the hospital to have her PIC line changed they have to send someone out to do it.

When you consider that Johnson was talking about creating a test for everyone to take every day, to not even allow people to take a test if they've been exposed is a really pathetic state of affairs.

These are probably the kinds of people Matt Hancock accused of lying about symptoms to get a test and blaming them for the shortage. Your poor neighbour, I hope she is OK.

She's fine so far. She's had a pretty tough time with covid as her brother-in-law who lived with her died of coronavirus at the beginning of the pandemic. That she didn't catch it (she did a LOT of disinfecting) despite being in the same house as him gives me hope that she'll be ok this time but it's still a massive worry.

And given the number of asymptomatic cases, you'd think they'd want to test everyone who thought they needed it. Unless there's some fetish for sticking things really far up your nose I'm not aware of*, I can't see people getting tests just for shits and giggles.

* if there is, don't tell me. I'm enjoying my blissful ignorance on this subject!

Well exactly. Sorry to sound like a stuck record but it sort of reminds me of when e.g. councils are trying to kick people out of their homes, or refuse accommodation - 'we have a shortage of housing/long list of people in need of housing' - yes and the person you are kicking out is one of them!

Bird on a Fire wrote: Fri Oct 23, 2020 11:02 pm Countries' case numbers are undoubtedly useful for monitoring the trajectory within each country (unless there are drastic changes to testing rates).

But it's difficult to compare between countries with different testing policies. Portugal, for instance, does seem to have unlimited tests for whoever wants them, whereas the UK has an explicit policy of restricting tests to people with symptoms, along with several other choices (self administration, postal delivery) known to increase the false positive rate. So if both countries were showing the same per capita case numbers it would be wise to assume the UK probably has more - but I would have no idea what the likely ratio would be.

That's why I was asking if anyone is modelling it, as there are loads of statistical methods for estimating detection rates from observational data.

Well in the UK there is the ONS serology survey and the covidzoe symptom study that also uses serology to estimate the proportion of symptoms reported that are associated with actual covid.

Or here, there's the datianalisicoronavirus instagram account.
That's new positives per million in each age group by week.

shpalman wrote: Sat Oct 24, 2020 2:03 pm Or here, there's the datianalisicoronavirus instagram account.

photo_2020-10-24_15-17-30.jpg

That's new positives per million in each age group by week.

Why do they do the ages upside down? I know it's only a silly thing and doesn't affect the interpretation... but it's more intuitively easy to read with age increasing with the y axis

jimbob wrote: Sat Oct 24, 2020 2:10 pm

shpalman wrote: Sat Oct 24, 2020 2:03 pm Or here, there's the datianalisicoronavirus instagram account.

photo_2020-10-24_15-17-30.jpg

That's new positives per million in each age group by week.

Why do they do the ages upside down? I know it's only a silly thing and doesn't affect the interpretation... but it's more intuitively easy to read with age increasing with the y axis

Does a virus burn upwards or trickle down??

Age generally goes upwards. At least, that's been my experience in my life so far.

Yeah but Instagram is for teh kidz. Rule 1 of data visualisation: adapt to your audience.

jimbob wrote: Sat Oct 24, 2020 1:45 pm

Bird on a Fire wrote: Fri Oct 23, 2020 11:02 pm Countries' case numbers are undoubtedly useful for monitoring the trajectory within each country (unless there are drastic changes to testing rates).

But it's difficult to compare between countries with different testing policies. Portugal, for instance, does seem to have unlimited tests for whoever wants them, whereas the UK has an explicit policy of restricting tests to people with symptoms, along with several other choices (self administration, postal delivery) known to increase the false positive rate. So if both countries were showing the same per capita case numbers it would be wise to assume the UK probably has more - but I would have no idea what the likely ratio would be.

That's why I was asking if anyone is modelling it, as there are loads of statistical methods for estimating detection rates from observational data.

Well in the UK there is the ONS serology survey and the covidzoe symptom study that also uses serology to estimate the proportion of symptoms reported that are associated with actual covid.

Serology would be another good way of comparing between countries, as long as the sampling strategies were comparable (e.g. random).

I'm not sure how useful the ONS/Zoe approaches are for estimating detection rates per se, as presumably folk who get tested via that route don't also need to try to get tested via the usual government pathways.

I'm really talking about statistical techniques to analyse the case, testing and death data, but estimating an additional parameter (detection). It's simple enough that I'm sure lots of people have already done it, but I haven't seen it mentioned anywhere in reports.

Bird on a Fire wrote: Sat Oct 24, 2020 6:41 pm Yeah but Instagram is for teh kidz. Rule 1 of data visualisation: adapt to your audience.

teh kidz is all on snaptok innit

But as you can see from the data, there are a lot of cases amongst my fellow youths.

bob sterman wrote: Sat Oct 24, 2020 9:13 am Not sure about the ONS survey - but knowing someone who was invited to take part in the similar Imperial College (REACT) survey - my hunch is the false negative rate is going to be higher for these surveys than Tier 2 testing. Couple of reasons...

My parents have taken part in those - mum did a self-swab a couple of weeks ago, dad doing a fingerprick this week for REACT. Pulled at random from GP listings, the letters said, but that they both got selected within a couple of weeks seemed a bit odd to me. Also, dad has been extreme cautious throughout.

I wondered if they asked about behaviour/exposure at all, but apparently not. I suppose that's because it's just a straight survey of incidence.

Fishnut wrote: Thu Oct 22, 2020 3:55 pm I've not really dived deep into the coronavirus news because it's all so f.cking depressing I just can't face it. But I was wondering if there's been any comparison between now and the 1918 flu - not biologically, but socially, politically, etc? Did it just go away on its own eventually or did something happen to actively end the pandemic? How was the economy affected? How did countries deal with it? I realised that I've heard of the Wall Street Crash in 1929 but I don't remember any significant recessions as a result of the pandemic. Were economies significantly different then so they weren't as badly affected, or was there an economic hit that just gets ignored in the write-ups? I've seen a few photos of people wearing masks at the time but what was done to slow the spread? Is there anything we can learn from the past?

I ask because I read this article yesterday that made the point that the focus on finding tech solutions is blinding us to the analogue solutions that already exist. That it's not just the UK that's struggling does suggest that the problems aren't unique to us but I do find it interesting that it's so many "western" countries that are finding it hardest to combat. Are we so wedded to the idea that we can tech our way out of problems that we fail to see the simple solutions?

Well, from the couple of articles I've read they did many of the same things we have now - closed public buildings, put curfews on social venues, staggered commutes/shop opening hours, limited numbers in shops, encouraged mask wearing. (They had anti-mask protestors in San Fransico too...) Healthworkers quarantined themselves separately from families to avoid taking it home, all that kind of thing.

Plus ca change & all that.

There were 3 or 4 major waves, starting in the spring of 1918, and of those the autumn 1918 one was the worst iirc. In the US, individual states responded differently, and those that locked down early and hardest suffered the least deaths and I think also the least economic impact.

Some links:
https://mashable.com/article/-coronavir ... urope=true
https://virus.stanford.edu/uda/fluresponse.html
https://www.theguardian.com/world/2020/ ... -outbreak
https://papers.ssrn.com/sol3/papers.cfm ... id=3561560 -- this one's about the economic impact
https://www.nationalgeographic.com/news ... ce-health/

In terms of cases missed by government testing, I've just been reading on the Uni of East Anglia's covid page:

UEA is actively encouraging staff and students to undertake asymptomatic testing to identify cases and make campus safer for everyone and is working in partnership with the Earlham Institute (EI) on the Norwich Research Park to do this as part of the Norwich Testing Initiative (NTI). Our own testing initiative scheme, working in partnership with the EI, has helped to identify more than three quarters of the students who have tested positive on campus, which demonstrates the importance of testing and making campus safer for everyone.

Tests undertaken via the NTI are collected Monday to Friday, with tests collected on a Friday processed into Saturday. There are then no more NTI tests processed until the Monday collection, which will mean lower results reported for Sunday and Monday.

So at least under those circumstances (young people living together - cases are amongst students rather than staff) the government testing scheme seems to be missing about 75% of cases. Seeing as the current wave appears to be more amongst those age cohorts, this does rather suggest that the UK's official figures are a mahoosive underestimate. Do many other countries have a policy of no-asymptomatic-testing?

(Also sounds like they're able to get results to people within ~24 hours, showing the value of a decentralised approach that doesn't waste precious days on postal services)

https://www.uea.ac.uk/about/news/statem ... -19-advice

Bird on a Fire wrote: Sun Oct 25, 2020 1:01 pm In terms of cases missed by government testing, I've just been reading on the Uni of East Anglia's covid page:

UEA is actively encouraging staff and students to undertake asymptomatic testing to identify cases and make campus safer for everyone and is working in partnership with the Earlham Institute (EI) on the Norwich Research Park to do this as part of the Norwich Testing Initiative (NTI). Our own testing initiative scheme, working in partnership with the EI, has helped to identify more than three quarters of the students who have tested positive on campus, which demonstrates the importance of testing and making campus safer for everyone.

Tests undertaken via the NTI are collected Monday to Friday, with tests collected on a Friday processed into Saturday. There are then no more NTI tests processed until the Monday collection, which will mean lower results reported for Sunday and Monday.

So at least under those circumstances (young people living together - cases are amongst students rather than staff) the government testing scheme seems to be missing about 75% of cases. Seeing as the current wave appears to be more amongst those age cohorts, this does rather suggest that the UK's official figures are a mahoosive underestimate. Do many other countries have a policy of no-asymptomatic-testing?
...

Italy does as of about now, and more and more countries either already do or will have soon, as they have to re-prioritise who to test as the number of symptomatic cases rises.

As spread becomes more "community" and everything but the essentials gets locked down, knowing whether your mild symptoms are covid or not becomes less and less useful since the advice to self-isolate is the same and there isn't much hope or use in tracing everyone's contacts.