No, it's relevant if your testing strategy, while unchanged domestically, is not the same as the country next door, i.e. Germany. The more people you test, the higher the R-rate will appear, and vice versa.
No, it's relevant if your testing strategy, while unchanged domestically, is not the same as the country next door, i.e. Germany. The more people you test, the higher the R-rate will appear, and vice versa.
Thanks for the link. Interesting work. I don't know the epidemiology literature and especially the mathematical modelling well enough to know whether such a low level of the population exposed is more likely than the models claiming much higher levels of sars-cov-2 in the population. However, as the authors point out, if these data are anywhere near correct then releasing the lock-down without effect testing and tracing in place would virtually guarantee another large wave of infections.
I think there is a common fantasy that you can beat the virus down, and it will stay down. It's quite hard to accept the whack a mole nature of it, and that without testing or a vaccine, the game goes on, and on, and on ...
I don't know the epidemiology literature and especially the mathematical modelling well enough to know whether such a low level of the population exposed is more likely than the models claiming much higher levels of sars-cov-2 in the population.
If it is, then it suggests that lockdown was an effective "flatten the curve" strategy. But that also has the lockdown-lifting implications you mention. And if it isn't, then people will complain the lockdown was a huge economic and mental waste of time. This really is like the Y2K bug.
Here in Arizona, the temperature in the Phoenix area is predicted to hit 100°F this weekend for the first time this year. Will be an interesting test of whether or not hot weather has any effect on the virus.
If you can test and trace and are dealing with penny-money new case numbers then getting out of lockdown is OK. Getting out prematurely runs a high risk of jacking up the R nought number and is therefore a bad idea. Anyway that's how I read it. Alan's right. A fizzle or anothet chain reaction. And for God's sake avoid another chain reaction. That seems to be the current wisdom.
No, it's relevant if your testing strategy, while unchanged domestically, is not the same as the country next door, i.e. Germany. The more people you test, the higher the R-rate will appear, and vice versa.
No. That is completely wrong. It's a day-to-day comparison of French figures against previous French figures.
You really should read what the growth-rate is before making statements like that. It has precisely nothing to do with what percentage of tests are positive.
So far as I can tell, France could have far higher numbers of infected people out there than we know about, and thus a higher than reported R-rate. But we simply don't know, because we're not doing mass testing.
So far as I can tell, France could have far higher numbers of infected people out there than we know about, and thus a higher than reported R-rate. But we simply don't know, because we're not doing mass testing.
Unless you think that the number of hidden cases is fluctuating wildly from day to day, this still has precisely nothing to do with the page I linked to.
Okay? Suppose France is only detecting 20% of cases every day. So long as that is true each and every day, the effect on the page that I linked to is zero. If France detecting 75% of cases every day, the answer would be the same. If France was detecting 5% of cases every day, the answer would still be the same.
The ratio of cases from one day to the next is not affected by the testing efficacy if that efficacy is roughly constant, regardless of whether that efficacy is extremely good or extremely bad.
No. That is completely wrong. It's a day-to-day comparison of French figures against previous French figures.
You really should read what the growth-rate is before making statements like that. It has precisely nothing to do with what percentage of tests are positive.
I didn't say anything about testing positive. I said testing.
So far as I can tell, these figures, in France, only reflect cases for which symptoms have developed seriously enough to warrant a test, and mass testing is what we have emphatically not been doing so far. So the French figures derived from tests tell you basically nothing about mild or asymptomatic cases which nevertheless can, perhaps, be contagious. Unless you can show me that Germany (or any other country) is testing a similar category of person, I don't think the comparison of that ratio between countries is very meaningful.
There are a few scientists with left field theories, e.g., that many people have already caught the virus
The nature of this virus is such that some people who catch it show only mild - or even no - symptoms, so it's 100% certain that more people - possibly a lot more people - have had it than the official stats indicate. That's not really that left-field a theory.
Given that all the people who have had it without anyone (even themselves) knowing have - obviously - survived without needing treatment, that means the currently-published hospitalisation and mortality rates (the ones informing public policy) are higher - possibly considerably higher - than the actual rates*.
We need more testing, especially antibody testing, so that we can find out what the real numbers are. The ideal would be universal testing regardless of symptoms (or the lack thereof), but that's never going to happen for several perfectly good reasons. Either way, the current amount of testing being done is woeful.
.
*= of course, a low percentage of a high number can still be larger overall than a high percentage of a low number. It depends on what the numbers actually are. Which is why we need to find that out.
No. That is completely wrong. It's a day-to-day comparison of French figures against previous French figures.
You really should read what the growth-rate is before making statements like that. It has precisely nothing to do with what percentage of tests are positive.
I didn't say anything about testing positive. I said testing.
So far as I can tell, these figures, in France, only reflect cases for which symptoms have developed seriously enough to warrant a test, and mass testing is what we have emphatically not been doing so far. So the French figures derived from tests tell you basically nothing about mild or asymptomatic cases which nevertheless can, perhaps, be contagious. Unless you can show me that Germany (or any other country) is testing a similar category of person, I don't think the comparison of that ratio between countries is very meaningful.
I'm sorry, but you are continuing to completely misunderstand the nature of the data I linked to.
The fact that you openly said you did not read how the figures were calculated, but continue to make posts as if you understand what they mean and can comment on their validity, seems very atypical for you.
The question as to what percentage of cases each country is successfully detecting is completely different. What I linked to says nothing about that at all.
EDIT: Except, of course, that if your detection rate is low you're highly unlikely to contain the spread of the virus, and this will ultimately show up even in a poor testing regime.
Well, educate me then. The growth rate applies consistently to each country's number of cases over time, I can see that (assuming they don't change the extent of their testing, a big assumption). But if non-tested people also pass the virus on to other non-tested people, and the proportion of non-tested people is different in different countries, surely that makes comparing those growth rates across countries difficult?
[ETA maybe what I'm objecting to fundamentally is the title of your linked article "The one number to watch". It seems to me that this is all far more complex than watching ONE number]
Well, educate me then. The growth rate applies consistently to each country's number of cases over time, I can see that (assuming they don't change the extent of their testing, a big assumption). But if non-tested people also pass the virus on to other non-tested people, and the proportion of non-tested people is different in different countries, surely that makes comparing those growth rates across countries difficult?
No. Why would it? The only way it could make a difference is if non-tested people somehow have a capacity to cause less sickness in new cases compared to the capacity of tested people to cause sickness in new cases.
If 20% of people that you infect will get sick enough to be detected, then whether you yourself got sick enough to be detected is completely irrelevant.
I've already said to you, it doesn't matter if the detection rate is 20% or 90%, so long as it's constant the answer doesn't change.
Let's set up some figures for the sake of argument.
TRUE NUMBER of new infections on consecutive days: let's say 5,000 new cases yesterday, 5,500 new cases today (figures only God knows).
Country with really excellent and thorough testing regime that detects 90% of these cases says: we had 4,500 new cases yesterday and 4,950 new cases today. They're a bit off. The growth rate figure they get is 4,950 divided by 4,500 = 1.1
Country with rather poor testing regime that only checks people who are pretty ill and only detects 20% of cases says: we had 1,000 new cases yesterday and 1,100 today. They're massively wrong. The growth rate figure they get is 1,100 divided by 1000 = 1.1.
Okay? Your detection rate does not affect this figure in and of itself. What will affect the growth rate is the practical result of poor testing: you won't catch cases, you won't prevent new ones, and the growth rate will spiral out of control. But your inability to stop the spread will show up in this figure so long as you're still detecting that same percentage of cases.
The only way it won't show up is if your detecting rate goes down, for example if your testing criteria is based on only doing say, the sickest 1,000 people regardless of how many people are really sick. But if your testing is based on saying "this is how sick you have to be to get tested", and 20% of infected people consistently get that sick, then you will keep getting the right growth rate even if all your case figures are 5x lower than they ought to be.
Given certain assumptions I don't think differences in testing regime between countries makes a difference to the growth rates. If the number of people who are seriously ill is a fixed proportion of the number of people infected then the growth rate of people who get seriously ill should reflect the growth rate of people who get infected even if you don't know the proportion of people infected who get seriously ill.
The problem is that the assumption may not hold if for example people who are liable to serious illness are not evenly distributed through the population.
Given certain assumptions I don't think differences in testing regime between countries makes a difference to the growth rates. If the number of people who are seriously ill is a fixed proportion of the number of people infected then the growth rate of people who get seriously ill should reflect the growth rate of people who get infected even if you don't know the proportion of people infected who get seriously ill.
The problem is that the assumption may not hold if for example people who are liable to serious illness are not evenly distributed through the population.
I'm not certain that particular example would affect it, unless what you're saying is that there'll be a change as the infection reaches a different group of people, in which case yes.
One thing that definitely would affect the figure is if you don't have enough testing kits available to maintain the same criteria for testing as the number of ill people soars. Which is actually why it was a good reason for countries to keep an eye on testing supplies. Certainly here in Canberra, people were actively refused testing if they didn't fulfil the criteria that had been set out, and the local Department of Health's website said this.
Whereas I suspect the early US figures were a total mess (and indeed they still might be poor) because testing was partly based on people's capacity to access (and pay for?) a test. If you were rich and had the sniffles, you got checked. Poorer people died without ever being diagnosed.
I know that Australian testing is beginning to widen / have less strict criteria, which we can do because the system is not overwhelmed. That could cause the growth rate to go up as our detection ability improves, though I understand there was an international study a week or two ago that already put our detection ability as the best in the world.
To date there have been 1616 deaths in Scotland, of which 910 were in hospitals, 537 in care homes, 168 at home and 1 in another institution (prison AIUI).
This includes all deaths in which COVID 19 was included on the death certificate as a cause, even if it was not the primary cause.
74% of deaths have been of those aged 75 or over. 35.3% were aged 85 or over.
55% were male.
I know that Australian testing is beginning to widen / have less strict criteria, which we can do because the system is not overwhelmed. That could cause the growth rate to go up as our detection ability improves, though I understand there was an international study a week or two ago that already put our detection ability as the best in the world.
That can be corrected for providing that everyone who would have been tested previously is still tested, and the test is flagged as being under the previous criteria. You can then use the expanded testing regime to help control the disease spread while still monitor the progress of those controls by only comparing those test results flagged as being taken under the same criteria. If you don't have that additional data for the reason why someone was tested then things become trickier (and, if you move from testing because a physician says the patient meets particular criteria to mass testing of anyone who turns up at the drive-through station you may lose that train of testing criteria).
The news reports I've heard simply announce the number of deaths in people that have tested positive that have been announced for the 24 hour period. As I have previously noted that number will include some people for whom the positive test is coincidental and not related to their cause of death. The corollary is also true that there will be deaths caused by Covid-19 that are not in these data. (Such as care home deaths).
Anyway, what I hadn't appreciated until I looked at the graph (click on one of the links to open up the excel file) is the distorting effect of the reporting date vs the date of death. Over the past week or so, the numbers for the UK have been roughly in the region of 600-900. That would suggest a plateau. However, if you look at the graph there is a marked downward trend. The caveat is that these figures remain provisional for a few days as they get updated by later reporting (often because the test result is not available at the time of death). However the graph clearly shows that the national peak was 7th April with an encouraging downward trend.
This doesn't solve the problem of unwinding the lock-down safely but it is very good evidence that the lock-down is working.
To get to ICU you have to be already dying. So to have a 50-50 chance of surviving is actually a miracle.
I thought ICU was where you went if they intubated you for invasive ventilation. And normally isn't ICU just where you go if you've potentially life-threatening symptoms, not necessarily because you're dying?
Yes, the lockdown is working, because it's reducing the R0 to below 1. But without accurate statistics, by date and by cause of death, then any decision made on when to relax or end the restrictions will inevitably be made on false assumptions.
If the FT is right, and we've had over 41,000 deaths, then the numbers might be falling, but from an unacceptably high level that precludes any loosening of the ties.
Yes, the lockdown is working, because it's reducing the R0 to below 1. But without accurate statistics, by date and by cause of death, then any decision made on when to relax or end the restrictions will inevitably be made on false assumptions.
If the FT is right, and we've had over 41,000 deaths, then the numbers might be falling, but from an unacceptably high level that precludes any loosening of the ties.
The lockdown is working because fewer people are getting infected, but that does not change the R₀.
The R₀ is how infectious a disease potentially is, not how many people are actually getting infected at a given time.
In epidemiology, the basic reproduction number (sometimes called basic reproductive ratio, or incorrectly basic reproductive rate, and denoted R₀, pronounced R nought or R zero[17]) of an infection can be thought of as the expected number of cases directly generated by one case in a population where all individuals are susceptible to infection.
The only reason the R₀ seems to vary of this virus, besides incorrect usage of the term, is that there is not enough data at the moment.
Yes, the lockdown is working, because it's reducing the R0 to below 1. But without accurate statistics, by date and by cause of death, then any decision made on when to relax or end the restrictions will inevitably be made on false assumptions.
If the FT is right, and we've had over 41,000 deaths, then the numbers might be falling, but from an unacceptably high level that precludes any loosening of the ties.
The lockdown is working because fewer people are getting infected, but that does not change the R₀.
The R₀ is how infectious a disease potentially is, not how many people are actually getting infected at a given time.
You just copied and pasted that from wikipedia without ever understanding it. Well done.
R0 has two modes, basic and effective. The effective R0 is literally what every layperson is talking about when they talk about R0 - the actual rate of transmission.
The only reason the R₀ seems to vary of this virus, besides incorrect usage of the term, is that there is not enough data at the moment.
Much like the fatality rate, as I noted earlier, R₀ is not a fixed constant like gravity but depends a lot on human factors, such as frequency and type of interactions between people. R₀ is going to be a lot lower for a virus released into a population that greets each other with a polite nod than it will in a population that greets each other with big, sloppy kisses (for example). So R₀ can be varying for reasons other than lack of data.
R0 has two modes, basic and effective. The effective R0 is literally what every layperson is talking about when they talk about R0 - the actual rate of transmission.
I'm not even sure how one would measure the basic R0 in any given population, it would assume no changes in behaviour as a result of the presence of the illness - not even someone sick going to bed. And, then there would be the population to population variations that could reflect different cultural practices (how often people shake hands, for example), different environmental conditions (does a disease spread at the same rate in a hot humid climate compared to a cold one?) or even genetic variations between populations. Basic R0 at best is a way of categorising different diseases in some semi-quantitative measure, with little bearing on how a disease will actually spread in different populations.
Review on CBC radio today indicated that up to 30% of anti-body tests are in accurate, and that the American FDA has fast tracked approvals of things without the normal requirements for evidence. That there may be places where vaccinations are having human trials, but this is mostly about the safety of unproven vaccines: does the vaccine cause side effects. The proof of a vaccine is that it prevents disease, and this requires extensive testing. Which is probably mid-2021 before one is actually available.
Last night our governor of Washington State, Gov. Inslee, gave what I thought was a well-reasoned description of what to expect when the state begins to open up. He said it will be more like dialing back the restrictions gradually instead of flipping the switch. I was encouraged by his remarks. How about you @mousethief?
But the director of the CDC did warn if we are not diligent, the pandemic will likely come back next year and be even worse because it will be coupled with the influenza epidemic. I know I get my influenza shot every year--not that it will help against coronavirus, but it will protect me from the one disease in any case.
Here in Arizona, the temperature in the Phoenix area is predicted to hit 100°F this weekend for the first time this year. Will be an interesting test of whether or not hot weather has any effect on the virus.
Well as the virus is appearing in tropical areas of the world I doubt it.
But the director of the CDC did warn if we are not diligent, the pandemic will likely come back next year and be even worse because it will be coupled with the influenza epidemic. I know I get my influenza shot every year--not that it will help against coronavirus, but it will protect me from the one disease in any case.
Hopefully the 'flu shot provides protection. The people who produce each years 'flu shot do a certain amount of gazing into a crystal ball to try and predict what the next serious strains of 'flu will be in time to get the vaccine produced and distributed. They don't always succeed in getting it perfect and we have a strain that the vaccine doesn't protect against. The worst case scenario is for that to happen at the same time as this coronavirus makes a resurgence.
But the director of the CDC did warn if we are not diligent, the pandemic will likely come back next year and be even worse because it will be coupled with the influenza epidemic. I know I get my influenza shot every year--not that it will help against coronavirus, but it will protect me from the one disease in any case.
Hopefully the 'flu shot provides protection. The people who produce each years 'flu shot do a certain amount of gazing into a crystal ball to try and predict what the next serious strains of 'flu will be in time to get the vaccine produced and distributed. They don't always succeed in getting it perfect and we have a strain that the vaccine doesn't protect against. The worst case scenario is for that to happen at the same time as this coronavirus makes a resurgence.
This is true. I almost died twenty years ago from complications of influenza. I learned from that to take a flu shot even if it is off, it has always provided some protection for me.
And now they say the coronavirus now has at least 30 different strains--though that might not be all that bad since previous vaccines from other diseases used strains of that particular virus to fight the particular disease.
The news reports I've heard simply announce the number of deaths in people that have tested positive that have been announced for the 24 hour period. As I have previously noted that number will include some people for whom the positive test is coincidental and not related to their cause of death. The corollary is also true that there will be deaths caused by Covid-19 that are not in these data. (Such as care home deaths).
Anyway, what I hadn't appreciated until I looked at the graph (click on one of the links to open up the excel file) is the distorting effect of the reporting date vs the date of death. Over the past week or so, the numbers for the UK have been roughly in the region of 600-900. That would suggest a plateau. However, if you look at the graph there is a marked downward trend. The caveat is that these figures remain provisional for a few days as they get updated by later reporting (often because the test result is not available at the time of death). However the graph clearly shows that the national peak was 7th April with an encouraging downward trend.
This doesn't solve the problem of unwinding the lock-down safely but it is very good evidence that the lock-down is working.
AFZ
I'd looked at those graphs before. Because there is a spread of actual death dates included in the announced death total. the more recent columns of deaths are provisional. But it does look as though there is a downward trend in hospital deaths - which is a relief.
However I think the "hidden death" factor i.e COVID-19 deaths in care homes or hospices or at home is a challenge in interpreting the trends.
I found this interesting article which may explain why our ICU death rates are so high. Apparently people are getting to the hospital way later in the course of the disease than one would expect because they simply don't realize how terrible their oxygen levels are.
Yes, the lockdown is working, because it's reducing the R0 to below 1. But without accurate statistics, by date and by cause of death, then any decision made on when to relax or end the restrictions will inevitably be made on false assumptions.
If the FT is right, and we've had over 41,000 deaths, then the numbers might be falling, but from an unacceptably high level that precludes any loosening of the ties.
The lockdown is working because fewer people are getting infected, but that does not change the R₀.
The R₀ is how infectious a disease potentially is, not how many people are actually getting infected at a given time.
You just copied and pasted that from wikipedia without ever understanding it. Well done.
Indeed, lb could have pasted the next paragraph from the wikipedia article:
R0 is not a biological constant for a pathogen as it is also affected by other factors such as environmental conditions and the behaviour of the infected population.
Going back to the question of how useful 12% immunity in your population is (if indeed that's what 12% exposure means, and I don't think we're certain of that), the math in the models isn't especially difficult. Indeed, further down that same wiki article, it says "For simple models, the proportion of the population that needs to be effectively immunized (meaning not susceptible to infection) to prevent sustained spread of the infection has to be larger than 1 − 1/R0".
What does that mean in practice? Well, we know that our lockdown behaviour has reduced R0 to something significantly under 1. The people who are fitting models to the data will have a fit value for R0 in the current lockdown conditions. With no immune people, R0 >1 means you get a runaway epidemic. With 12% immunity, your epidemic doesn't run away until R0>1/(1-12%), or R0>1.14. That's just describing herd immunity with math.
So more or less, it means that you could have social contact with 14% more individuals than you could at the start of the outbreak without infection rates exploding.
Think back to your life in March. If I changed the number of people in your grocery store by 14%, would you notice? If you had lunch with a table containing 7 people rather than 6, would you notice? I don't think you would. A 14% change in social contacts is hardly noticeable. So basically, that means that the current level of herd immunity isn't useful for un-lockdown. (Having a cadre of immune individuals is useful, but that's something entirely different.)
A more reasonable rationale for relaxing restrictions is to say that we currently have R0 well below 1, and that once the rate of infections has subsided a lot, we can relax the restrictions a bit, and still keep R0 below 1 (or 1.14). But without rigorous testing, knowing how much to relax the restrictions is basically slightly educated guesswork. We know March was bad - we can't just go back to doing that, or we'll have a second blowup. The question is how much of life can we turn back on again without going there.
This is sort of modelable, and models can help inform your guesses, but the data is still rather poor. And the thing that we need for better data is widespread testing.
The news reports I've heard simply announce the number of deaths in people that have tested positive that have been announced for the 24 hour period. As I have previously noted that number will include some people for whom the positive test is coincidental and not related to their cause of death. The corollary is also true that there will be deaths caused by Covid-19 that are not in these data. (Such as care home deaths).
Anyway, what I hadn't appreciated until I looked at the graph (click on one of the links to open up the excel file) is the distorting effect of the reporting date vs the date of death. Over the past week or so, the numbers for the UK have been roughly in the region of 600-900. That would suggest a plateau. However, if you look at the graph there is a marked downward trend. The caveat is that these figures remain provisional for a few days as they get updated by later reporting (often because the test result is not available at the time of death). However the graph clearly shows that the national peak was 7th April with an encouraging downward trend.
This doesn't solve the problem of unwinding the lock-down safely but it is very good evidence that the lock-down is working.
AFZ
I'd looked at those graphs before. Because there is a spread of actual death dates included in the announced death total. the more recent columns of deaths are provisional. But it does look as though there is a downward trend in hospital deaths - which is a relief.
However I think the "hidden death" factor i.e COVID-19 deaths in care homes or hospices or at home is a challenge in interpreting the trends.
Possibly. My best guess is that the non-hospital deaths are following the same pattern. Simply because there's no good reason for the two to diverge.
It's the cluster issue AFZ. Care homes concentrate vulnerable people and to date the staff have not been protected all that brilliantly. The stat to watch now is the new cases stat. It still looks pretty high to me. Certainly high enough to preclude effective tracing, even if we have testing capability and staff to do it promptly.
Given the underlying infectiousness and the fact that we can be infectious before symptoms emerge, tracing needs to be orompt.
Chris Whitty, chief medical officer, in an obviously prepared reply, said that no vaccine or anti-viral would be produced this year. Hence, social measures must continue.
I think they got Whitty to say this, as no politician wants to say it.
On the Coronavirus newscast they were talking about the date of death data & peak, saying with a lot of caveats, that the peak started to drop slightly - given the lag between infection & death - with the introduction of of pre-lockdown measures, so it is likely people are to be going back over that data with a fine tooth comb trying to work out which partial measures had the most effect. (In order to have strategies for the partial unlock.)
So if, for example, they found the melee distance rule had a noticeable impact but shutting the schools made little noticeable difference - than reopening schools might be the first unlock they go for.
It's the cluster issue AFZ. Care homes concentrate vulnerable people and to date the staff have not been protected all that brilliantly. The stat to watch now is the new cases stat. It still looks pretty high to me. Certainly high enough to preclude effective tracing, even if we have testing capability and staff to do it promptly.
Given the underlying infectiousness and the fact that we can be infectious before symptoms emerge, tracing needs to be orompt.
Indeed. But the ONS data on non-hospital deaths (which also has important caveats) is about a week out of date when we get it... and to my eye (squinting slightly) I think the peaks will line up.
Of course, there really is no excuse for not protecting staff. And, as I've said more than once, effective testing and tracing is crucial for any relaxation of restrictions. However, none of that changes my view that the lockdown is working...
So if, for example, they found the melee distance rule had a noticeable impact but shutting the schools made little noticeable difference - than reopening schools might be the first unlock they go for.
All the schools round here have already said they're not opening until fall, because opening the schools again for 2 or 3 weeks before closing for the summer is unlikely to produce any actual education (many school buildings aren't air conditioned, so extending the school year into the summer isn't practical even if there weren't a whole raft of legal and contractual issues involved.) Our schools stop for the summer at the end of May.
For the UK, where school typically runs through much of July, going back to school before the summer makes more sense (assuming it's sensible vis-a-vis the virus)
I'm sure that the UK lockdown has flattened the curve. One thing that is becoming increasingly clear is that the journey down the curve is going to be slower than the journey up the curve. So I guess, like others, I'm pretty nervous about the effects of progressive relaxation.
Spain has extended their lockdown to 10 May and we're running a couple of weeks behind them I think. Both Spain and ourselves are still reporting 4K new cases a day. End May begins to look possible for some limited relaxation of lockdown constraints.
Not much help to me I'm afraid. I'm in the need to shield category so Mrs B and I are expecting to be pretty much housebound for some time to come. But I'll be delighted for others if they are able to enjoy the summer. Autumn already looks dodgy with a second wave quite likely.
I'm sure that the UK lockdown has flattened the curve. One thing that is becoming increasingly clear is that the journey down the curve is going to be slower than the journey up the curve.
The main problem with flattening the curve is that it becomes much longer.
It's not a flattened sine curve, it's a flattened skewed sine curve. Evidence from Italy and Spain suggests strongly that the decline from the peak is very slow and a similar pattern seems to be emerging re the UK.
Meanwhile in the USA the death toll for the USA for 22 April (using the worldometer GMT clock) again exceeds 2,000 and the new cases total exceeds 27,000. And the new cases spread is showing some 70% of cases outside New York and New Jersey, spread over lots of States.
And some States move ahead with relaxing restrictions and other say "not on your life". The virus has a massive and widespread hold in the majority of States, most are miles away from having new case numbers down to feasible tracing levels, testing levels are still way below where they need to be, and there is no remedial treatment.
Politics has overtaken common sense to a potentially lethal degree. Not everywhere. But in enough States to threaten all the others. It's enough to make you weep with frustration.
You make it sound as if an outbreak going on for months or years is a good thing. Lengthening the curve may be the lesser of two evils, but unless we can make it so flat that it's no more than a minor nuisance, it's still far from appealing. Ideally we need to find a way to reduce the area under the curve (flattening it does of course give us more time in which to do that).
Comments
Thanks for the link. Interesting work. I don't know the epidemiology literature and especially the mathematical modelling well enough to know whether such a low level of the population exposed is more likely than the models claiming much higher levels of sars-cov-2 in the population. However, as the authors point out, if these data are anywhere near correct then releasing the lock-down without effect testing and tracing in place would virtually guarantee another large wave of infections.
AFZ
No. That is completely wrong. It's a day-to-day comparison of French figures against previous French figures.
You really should read what the growth-rate is before making statements like that. It has precisely nothing to do with what percentage of tests are positive.
Unless you think that the number of hidden cases is fluctuating wildly from day to day, this still has precisely nothing to do with the page I linked to.
Okay? Suppose France is only detecting 20% of cases every day. So long as that is true each and every day, the effect on the page that I linked to is zero. If France detecting 75% of cases every day, the answer would be the same. If France was detecting 5% of cases every day, the answer would still be the same.
The ratio of cases from one day to the next is not affected by the testing efficacy if that efficacy is roughly constant, regardless of whether that efficacy is extremely good or extremely bad.
I didn't say anything about testing positive. I said testing.
So far as I can tell, these figures, in France, only reflect cases for which symptoms have developed seriously enough to warrant a test, and mass testing is what we have emphatically not been doing so far. So the French figures derived from tests tell you basically nothing about mild or asymptomatic cases which nevertheless can, perhaps, be contagious. Unless you can show me that Germany (or any other country) is testing a similar category of person, I don't think the comparison of that ratio between countries is very meaningful.
The nature of this virus is such that some people who catch it show only mild - or even no - symptoms, so it's 100% certain that more people - possibly a lot more people - have had it than the official stats indicate. That's not really that left-field a theory.
Given that all the people who have had it without anyone (even themselves) knowing have - obviously - survived without needing treatment, that means the currently-published hospitalisation and mortality rates (the ones informing public policy) are higher - possibly considerably higher - than the actual rates*.
We need more testing, especially antibody testing, so that we can find out what the real numbers are. The ideal would be universal testing regardless of symptoms (or the lack thereof), but that's never going to happen for several perfectly good reasons. Either way, the current amount of testing being done is woeful.
.
*= of course, a low percentage of a high number can still be larger overall than a high percentage of a low number. It depends on what the numbers actually are. Which is why we need to find that out.
I'm sorry, but you are continuing to completely misunderstand the nature of the data I linked to.
The fact that you openly said you did not read how the figures were calculated, but continue to make posts as if you understand what they mean and can comment on their validity, seems very atypical for you.
The question as to what percentage of cases each country is successfully detecting is completely different. What I linked to says nothing about that at all.
EDIT: Except, of course, that if your detection rate is low you're highly unlikely to contain the spread of the virus, and this will ultimately show up even in a poor testing regime.
[ETA maybe what I'm objecting to fundamentally is the title of your linked article "The one number to watch". It seems to me that this is all far more complex than watching ONE number]
No. Why would it? The only way it could make a difference is if non-tested people somehow have a capacity to cause less sickness in new cases compared to the capacity of tested people to cause sickness in new cases.
If 20% of people that you infect will get sick enough to be detected, then whether you yourself got sick enough to be detected is completely irrelevant.
I've already said to you, it doesn't matter if the detection rate is 20% or 90%, so long as it's constant the answer doesn't change.
Let's set up some figures for the sake of argument.
TRUE NUMBER of new infections on consecutive days: let's say 5,000 new cases yesterday, 5,500 new cases today (figures only God knows).
Country with really excellent and thorough testing regime that detects 90% of these cases says: we had 4,500 new cases yesterday and 4,950 new cases today. They're a bit off. The growth rate figure they get is 4,950 divided by 4,500 = 1.1
Country with rather poor testing regime that only checks people who are pretty ill and only detects 20% of cases says: we had 1,000 new cases yesterday and 1,100 today. They're massively wrong. The growth rate figure they get is 1,100 divided by 1000 = 1.1.
Okay? Your detection rate does not affect this figure in and of itself. What will affect the growth rate is the practical result of poor testing: you won't catch cases, you won't prevent new ones, and the growth rate will spiral out of control. But your inability to stop the spread will show up in this figure so long as you're still detecting that same percentage of cases.
The only way it won't show up is if your detecting rate goes down, for example if your testing criteria is based on only doing say, the sickest 1,000 people regardless of how many people are really sick. But if your testing is based on saying "this is how sick you have to be to get tested", and 20% of infected people consistently get that sick, then you will keep getting the right growth rate even if all your case figures are 5x lower than they ought to be.
The problem is that the assumption may not hold if for example people who are liable to serious illness are not evenly distributed through the population.
I'm not certain that particular example would affect it, unless what you're saying is that there'll be a change as the infection reaches a different group of people, in which case yes.
One thing that definitely would affect the figure is if you don't have enough testing kits available to maintain the same criteria for testing as the number of ill people soars. Which is actually why it was a good reason for countries to keep an eye on testing supplies. Certainly here in Canberra, people were actively refused testing if they didn't fulfil the criteria that had been set out, and the local Department of Health's website said this.
Whereas I suspect the early US figures were a total mess (and indeed they still might be poor) because testing was partly based on people's capacity to access (and pay for?) a test. If you were rich and had the sniffles, you got checked. Poorer people died without ever being diagnosed.
I know that Australian testing is beginning to widen / have less strict criteria, which we can do because the system is not overwhelmed. That could cause the growth rate to go up as our detection ability improves, though I understand there was an international study a week or two ago that already put our detection ability as the best in the world.
The report is here:
https://www.nrscotland.gov.uk/files//statistics/covid19/covid-deaths-report-week-16.pdf
To date there have been 1616 deaths in Scotland, of which 910 were in hospitals, 537 in care homes, 168 at home and 1 in another institution (prison AIUI).
This includes all deaths in which COVID 19 was included on the death certificate as a cause, even if it was not the primary cause.
74% of deaths have been of those aged 75 or over. 35.3% were aged 85 or over.
55% were male.
The official releases are here: https://www.england.nhs.uk/statistics/statistical-work-areas/covid-19-daily-deaths/
The news reports I've heard simply announce the number of deaths in people that have tested positive that have been announced for the 24 hour period. As I have previously noted that number will include some people for whom the positive test is coincidental and not related to their cause of death. The corollary is also true that there will be deaths caused by Covid-19 that are not in these data. (Such as care home deaths).
Anyway, what I hadn't appreciated until I looked at the graph (click on one of the links to open up the excel file) is the distorting effect of the reporting date vs the date of death. Over the past week or so, the numbers for the UK have been roughly in the region of 600-900. That would suggest a plateau. However, if you look at the graph there is a marked downward trend. The caveat is that these figures remain provisional for a few days as they get updated by later reporting (often because the test result is not available at the time of death). However the graph clearly shows that the national peak was 7th April with an encouraging downward trend.
This doesn't solve the problem of unwinding the lock-down safely but it is very good evidence that the lock-down is working.
AFZ
I thought ICU was where you went if they intubated you for invasive ventilation. And normally isn't ICU just where you go if you've potentially life-threatening symptoms, not necessarily because you're dying?
If the FT is right, and we've had over 41,000 deaths, then the numbers might be falling, but from an unacceptably high level that precludes any loosening of the ties.
The R₀ is how infectious a disease potentially is, not how many people are actually getting infected at a given time. The only reason the R₀ seems to vary of this virus, besides incorrect usage of the term, is that there is not enough data at the moment.
You just copied and pasted that from wikipedia without ever understanding it. Well done.
R0 has two modes, basic and effective. The effective R0 is literally what every layperson is talking about when they talk about R0 - the actual rate of transmission.
Much like the fatality rate, as I noted earlier, R₀ is not a fixed constant like gravity but depends a lot on human factors, such as frequency and type of interactions between people. R₀ is going to be a lot lower for a virus released into a population that greets each other with a polite nod than it will in a population that greets each other with big, sloppy kisses (for example). So R₀ can be varying for reasons other than lack of data.
Not heard of the series -- what network? I pretty much watch either CBC or BBC World.
But the director of the CDC did warn if we are not diligent, the pandemic will likely come back next year and be even worse because it will be coupled with the influenza epidemic. I know I get my influenza shot every year--not that it will help against coronavirus, but it will protect me from the one disease in any case.
This CBS Eye on Earth Report was quite interesting this morning.
Well as the virus is appearing in tropical areas of the world I doubt it.
This is true. I almost died twenty years ago from complications of influenza. I learned from that to take a flu shot even if it is off, it has always provided some protection for me.
And now they say the coronavirus now has at least 30 different strains--though that might not be all that bad since previous vaccines from other diseases used strains of that particular virus to fight the particular disease.
I'd looked at those graphs before. Because there is a spread of actual death dates included in the announced death total. the more recent columns of deaths are provisional. But it does look as though there is a downward trend in hospital deaths - which is a relief.
However I think the "hidden death" factor i.e COVID-19 deaths in care homes or hospices or at home is a challenge in interpreting the trends.
Indeed, lb could have pasted the next paragraph from the wikipedia article:
Going back to the question of how useful 12% immunity in your population is (if indeed that's what 12% exposure means, and I don't think we're certain of that), the math in the models isn't especially difficult. Indeed, further down that same wiki article, it says "For simple models, the proportion of the population that needs to be effectively immunized (meaning not susceptible to infection) to prevent sustained spread of the infection has to be larger than 1 − 1/R0".
What does that mean in practice? Well, we know that our lockdown behaviour has reduced R0 to something significantly under 1. The people who are fitting models to the data will have a fit value for R0 in the current lockdown conditions. With no immune people, R0 >1 means you get a runaway epidemic. With 12% immunity, your epidemic doesn't run away until R0>1/(1-12%), or R0>1.14. That's just describing herd immunity with math.
So more or less, it means that you could have social contact with 14% more individuals than you could at the start of the outbreak without infection rates exploding.
Think back to your life in March. If I changed the number of people in your grocery store by 14%, would you notice? If you had lunch with a table containing 7 people rather than 6, would you notice? I don't think you would. A 14% change in social contacts is hardly noticeable. So basically, that means that the current level of herd immunity isn't useful for un-lockdown. (Having a cadre of immune individuals is useful, but that's something entirely different.)
A more reasonable rationale for relaxing restrictions is to say that we currently have R0 well below 1, and that once the rate of infections has subsided a lot, we can relax the restrictions a bit, and still keep R0 below 1 (or 1.14). But without rigorous testing, knowing how much to relax the restrictions is basically slightly educated guesswork. We know March was bad - we can't just go back to doing that, or we'll have a second blowup. The question is how much of life can we turn back on again without going there.
This is sort of modelable, and models can help inform your guesses, but the data is still rather poor. And the thing that we need for better data is widespread testing.
Possibly. My best guess is that the non-hospital deaths are following the same pattern. Simply because there's no good reason for the two to diverge.
I could very easily be wrong.
Given the underlying infectiousness and the fact that we can be infectious before symptoms emerge, tracing needs to be orompt.
I think they got Whitty to say this, as no politician wants to say it.
So if, for example, they found the melee distance rule had a noticeable impact but shutting the schools made little noticeable difference - than reopening schools might be the first unlock they go for.
Indeed. But the ONS data on non-hospital deaths (which also has important caveats) is about a week out of date when we get it... and to my eye (squinting slightly) I think the peaks will line up.
Of course, there really is no excuse for not protecting staff. And, as I've said more than once, effective testing and tracing is crucial for any relaxation of restrictions. However, none of that changes my view that the lockdown is working...
AFZ
P.s. I may be wrong.
All the schools round here have already said they're not opening until fall, because opening the schools again for 2 or 3 weeks before closing for the summer is unlikely to produce any actual education (many school buildings aren't air conditioned, so extending the school year into the summer isn't practical even if there weren't a whole raft of legal and contractual issues involved.) Our schools stop for the summer at the end of May.
For the UK, where school typically runs through much of July, going back to school before the summer makes more sense (assuming it's sensible vis-a-vis the virus)
Spain has extended their lockdown to 10 May and we're running a couple of weeks behind them I think. Both Spain and ourselves are still reporting 4K new cases a day. End May begins to look possible for some limited relaxation of lockdown constraints.
Not much help to me I'm afraid. I'm in the need to shield category so Mrs B and I are expecting to be pretty much housebound for some time to come. But I'll be delighted for others if they are able to enjoy the summer. Autumn already looks dodgy with a second wave quite likely.
It’s definitely working. The question now is when it will have worked well enough that we can say it’s done it’s job and is no longer needed.
The main problem with flattening the curve is that it becomes much longer.
Meanwhile in the USA the death toll for the USA for 22 April (using the worldometer GMT clock) again exceeds 2,000 and the new cases total exceeds 27,000. And the new cases spread is showing some 70% of cases outside New York and New Jersey, spread over lots of States.
And some States move ahead with relaxing restrictions and other say "not on your life". The virus has a massive and widespread hold in the majority of States, most are miles away from having new case numbers down to feasible tracing levels, testing levels are still way below where they need to be, and there is no remedial treatment.
Politics has overtaken common sense to a potentially lethal degree. Not everywhere. But in enough States to threaten all the others. It's enough to make you weep with frustration.
You make it sound as if an outbreak going on for months or years is a good thing. Lengthening the curve may be the lesser of two evils, but unless we can make it so flat that it's no more than a minor nuisance, it's still far from appealing. Ideally we need to find a way to reduce the area under the curve (flattening it does of course give us more time in which to do that).