The national raw numbers report is very misleading. As pointed out, it does not show the per capita rate, nor does it show the differences in localities. And they do not indicate the number who have recovered.
For instance, in the US, much of the country has little to no infections. Most of the infections are in metropolitan areas, though there are hot spots in rural areas too as indicated in the discussions centered around meatpacking plants. (My bet is based on this we will have to significantly modify how we produce meat products in the US, BTW)
Even in my state. Washington, we have been bumping along for the past two weeks statewide at 200 infections per seven-day average. Yet, there are some counties that have not reported any infections since the beginning of the stay at home order, and there are other counties that have not had any new reports for two weeks. In my county, we have not had a new case for a week. And of those, only four remain in strict quarantine. While the governor has extended the stay at home rule until 31 May, he will be allowing some of the counties to begin opening sooner provided they meet certain criteria).
We are really hoping for a small family reunion in the middle of June so we can have a memorial service for my brother. The problem is, we live in different jurisdictions that have different timelines for lifting their stay home/stay healthy orders.
IMHO we really don't have enough evidence even to say that, begging your pardon, Gramps49, although perhaps your state actually does---because my state, and most of them, simply aren't testing. Like, hardly at all. And there's no likelihood of that changing any time soon. In my heart of hearts, I'm not sure some of our leaders even WANT testing, because they're afraid of what it would show. And so we slog on with no real idea of just how much of the population is infected/infectious, and how much might (please God) be immune, and who is still very much at risk.
Oh, our stats look pretty fine, but the studies I've seen suggest there may be between 40 and 80 extra real-but-unofficial cases out there in the population for every officially counted one. And if that's the case, my small suburb has something like 600 fucking cases right now. But we won't know for years, until some grad student does a dissertation and analyzes the excess mortality rate. Probably about 2025, then.
IMHO we really don't have enough evidence even to say that, begging your pardon, Gramps49, although perhaps your state actually does---because my state, and most of them, simply aren't testing. Like, hardly at all. And there's no likelihood of that changing any time soon. In my heart of hearts, I'm not sure some of our leaders even WANT testing, because they're afraid of what it would show. And so we slog on with no real idea of just how much of the population is infected/infectious, and how much might (please God) be immune, and who is still very much at risk.
Oh, our stats look pretty fine, but the studies I've seen suggest there may be between 40 and 80 extra real-but-unofficial cases out there in the population for every officially counted one. And if that's the case, my small suburb has something like 600 fucking cases right now. But we won't know for years, until some grad student does a dissertation and analyzes the excess mortality rate. Probably about 2025, then.
You do have a point. No, we are not testing very well now, but Inslee says the feds have promised him enough test kits in a couple of weeks to do it. They will be key to easing up on our restrictions statewide.
I had asthma symptoms and dutifully contacted my doctor who spent most of the visit telling me I couldn't be tested without a fever of 101 Fahrenheit, because allegedly 90% of COVID suffers get one (what HAS this man been reading?) though I hadn't even asked about testing for COVID, and I know damn well that's not CDC guidance. I conclude that's local guidance, designed to put off all but the most desperate cases, because we simply have no tests and no near prospect of getting any.
I should say our municipal testing site, such as it was, is now closing because it is not getting enough referrals.
There could be a feedback loop there. There have been instances where, because of the paucity of tests, the standards for referrals have been set very high (e.g. must have high fever, persistent dry cough, and fatigue. Only two of the three means no referral.) This leads to very few referrals, leading to a shutdown of testing because of so few referrals. One wonders what the results would be if the bureaucratic hurdles were eliminated and testing was done without a referral, or with much lower standards for a referral.
^ UK testing got choked off for a while last week (with capacity exceeding tests done) before referral procedures were eased off to un-choke it - so that Hancock's 100,000 a day pledge could be met.
(I don't have much of a view on this - I'm sure meeting it was way more political than clinical, but given how govt and the press works, his pledge has been useful in prioritising a major increase in testing, and it seems we're going to need that to work out how to safely ease lock down.)
My county (in coordination with other SF Bay counties) is easing slightly on Monday; outdoor gardening services (and plant nurseries) will now be allowed as long as social distancing is maintained and other rules followed. The county has had 113 known deaths and over 2,000 cases, 132 currently hospitalized with 48 of those in ICU. We mostly dodged the first bullet and I think many of us know it.
We have only had 7 cases in our county. 6 recovered and one doing well at home. All can trace to people outside of the county. However very little testing, was done here and traces of disease were found in four of the counties sewage treatment plants. Last week sewage again tested and all came back no traces. So now they are about to open up just about everything, including our lake which is a big draw for fisherman from out of the area. We will see what happens. The good news is starting on Monday they now have enough test kits to start drive by testing with an appointment. This is a poor county and many people depend on local business. They do not plan on opening restaurants. Fingers Crossed.
^ UK testing got choked off for a while last week (with capacity exceeding tests done) before referral procedures were eased off to un-choke it - so that Hancock's 100,000 a day pledge could be met.
(I don't have much of a view on this - I'm sure meeting it was way more political than clinical, but given how govt and the press works, his pledge has been useful in prioritising a major increase in testing, and it seems we're going to need that to work out how to safely ease lock down.)
Given that the "we've met the target" message that Thursday had 122,000 tests was only possible by include 40,000 tests sent out in the post on Thursday, which won't be measured for another week, the evidence is that it was political rather than clinical. Clinically lead testing would concentrate on people showing symptoms to guide treatment (and, associated precautions for those doing the treating). Policy lead testing would concentrate on identifying those who need to be at work (if capacity available, even without symptoms). Epidemiology lead testing would be conducting random testing to determine the extent of actual infection.
Setting a target, especially when you need to fiddle the books to make a claim to having met it, is simply political. It's to be seen to be doing something, even if it's not a particularly useful something.
This is probably going to be a long post. Some of this I've said before but it was a few pages back and with a thread this size, I think repetition is gonna be necessary. Some of what I am going to say is in total agreement with above posters.
So; Excess mortality is the only figure that really means anything - for lots of reasons that I will explain shortly - and it's difficult to get an instant read on it. However, we won't have to wait 5 years in developed countries. In the US, the CDC will have good data in 6-12 months and the same goes for the ONS in the UK. I know many of you know this but just to be clear in case it's a new term to some, it has a simple definition: that is a mortality rate that's higher than expected. How you specify expected is a little more tricky but only a little. A 5 year average is typical and seasonally adjusted. I.e. you compare April 2020 with April 2015, 16, 17, 18 and 19. If you compare June with January, you'll get spurious results because more people die in winter than summer.
There are three categories of deaths that are relevant here. Or potentially 6, because in each of the three boxes there will be people who have been tested and those who haven't.
1. People who died from Covid-19 who would have died anyway from something else.
2. People who died from Covid-19 who otherwise had a normal or near-normal life expectancy.
3. People who die from something else who would had survived but for the burden on healthcare or other knock-on effects of the pandemic.
Taking those in turn: 1. We know that the majority of Covid-19 deaths are in the elderly. People over 80 have a 10% mortality in 12 months anyway. Thus if someone was dying from something else, even if Covid-19 infection was the terminal event it is not necessarily accurate to count that as a Covid-19 death. Here time becomes a factor: we are all dying people; if you set your time window at 100 years, then there's no such thing as excess mortality because essentially we'll all be dead in 100 years or so. Conversely, if someone is dying from cancer, it's not unusual that bacterial pneumonia will be the terminal event and thus a Covid-19 infection is not really the cause of death it's just supplanted the bacterial pneumonia and the cancer is the cause of death. The only way to unpick this on a population level is to look at how many people you expect to die in a given period (based on the 5 year seasonal average) and compare that to the actual deaths. The thing that is very alarming is the preliminary data from the ONS suggests this could be really high. Two things to note here: firstly some of those deaths belong in box 3 but are still due to the pandemic. Another key point here is that if you only look at those who tested positive, these box three people will get missed. The second point is that if you have a bad winter with flu, you see a spike in deaths followed by a lower than average mortality in the following weeks as the most vulnerable is now a smaller population because of the ones who have died off. The lower mortality is unlikely to be enough to offset the higher rate but it is still there. So it's a question of timeframe. I would suggest 3 months is probably about right. For the individual and the family, this is ridiculous and callous. I've been close family to someone terminally ill and often those extra days and weeks are very precious. I am not saying that it's ok if people died a couple of months early as they were old anyway. I am not saying that. However, to evaluate how badly afflicted a country is by the pandemic and the effectiveness of a country's response, some leeway here is needed on a population level. I would suggest that a 3 month window makes sense on a population analysis level, whilst not for one moment meaning to minimise personal tragedies.
2. These are easier to count. These will include the majority of key-workers who have died. To a large extent, with proper PPE and a proper, coordinated approach to a pandemic, these deaths are avoidable.
3. This is a very mixed bag and, as I said, if your mortality count is just a number of people who died and tested positive, this number will be wrongly given a value of zero. Some examples: I have no doubt that there will be a spike in cardiovascular deaths and stroke deaths because we know that people are not seeking medical treatment when they need to. More indirectly if the economic cost of Covid-19 and lockdown causes homelessness (as it inevitably will) that will result in increased mortality as there is a measurable mortality from homelessness. We have had a child die of sepsis locally who's parents brought her in very late because they were worried about Covid. She probably would have survived if she got to hospital earlier. Conversely there will be a negative number in this box too. Less journeys means less traffic accidents and deaths. The net result will still be an up spike in mortality overall.
All of this takes some serious analysis to unpick fully but the headline excess mortality rate is very informative and the detailed analysis will show some information on what proportion of these deaths might have been avoided.
The other reason why the excess mortality number is so informative is because, for reasons that are hopefully now evident, it's completely independent of testing. Given how the UK and US are totally failing at testing, this is particularly vital.
So if it turns our that permanent immunity is impossible, does that mean we’ll have to keep social distancing forever?
I would have thought more likely it would mean regular booster injections of a vaccine, when one becomes available.
@Croesos, thanks for those articles on meat plants. I bet the management of those places are all working from home. How do they sleep?
Booster injections, or a cure, or some kind of meds like the AIDS cocktail to prevent the infection from having its usual effects. There are other ways to fight disease besides vaccines.
There are some important limitations to these data, including the fact that they are England only and that these are only the in-hospital deaths.
If you follow the link to the daily-chart in the excel file the downward trend in deaths is very clear. The tail is going to be quite long - it's clear that the rise was steeper than the fall, but it also appears to be linear and thus single-figure death rates are likely to be reached in a week or two.
As has been discussed at length on this thread, there are significant problems with the UK-testing program - or rather the lack of one. The fact that the testing criteria has changed more than once and there is no random sampling means that the number of confirmed cases is not very helpful for tracking the epidemic. Hence, I've only been looking at the mortality numbers.
As @Barnabas62 has noted, it is an assumption that community (i.e. care-home) deaths are following the same trend. I think they probably will but he's right, we don't currently know.
Anyway, I've been thinking - dangerous, I know. If you're managing an epidemic, the new infection rate is the key data. However, for the reasons I've mentioned, we don't have a reliable read on this number. The mortality number is reliable despite the caveats above and shows a downward trend that I think I would describe as 'encouraging.' Certainly, it is a major relief looking at the trends, if 'encouraging' is too strong a word. However, there is a variable time delay between being exposed to the virus and dying. Some people are dying very quickly, many after several days, some a few weeks. This is important because the peak of mortality was around 17th April in England. Most likely, the peak of infection was more than a week before that.
In terms of the care-home deaths, it's a good bet that the time-to-death (sorry, correct technical term) is shorter. Thus if the death rate in care homes is parallel or even behind the hospital one that suggests the actual spread of infection in care homes is later than in the community at large. This does have implications for managing the epidemic.
All just idle wondering from me - thank you for tolerating my waffle...
Just a quick note on my five year mention--I figure it'll be five years before my LOCAL stats on excess mortality are known, not the national or even the state ones.
Thanks AFZ - I've found your posts very helpful, informative and clear. I've also found (to my surprise) that thinking about actuarial science is more interesting than I had anticipated - though I have some track record here, remembering what happened when I gave up my career and was thinking about how long I expected to maunder along in a future retirement
I did a little googling for '% mortality vs age' stats, but didn't get very far - any suggestions? It would do me good to put some odds on a Difficult Relative, not from the point of view of looking forward to their passing, more from the perspective of helping me appreciate their continued company!
Thanks AFZ - I've found your posts very helpful, informative and clear. I've also found (to my surprise) that thinking about actuarial science is more interesting than I had anticipated - though I have some track record here, remembering what happened when I gave up my career and was thinking about how long I expected to maunder along in a future retirement
I did a little googling for '% mortality vs age' stats, but didn't get very far - any suggestions? It would do me good to put some odds on a Difficult Relative, not from the point of view of looking forward to their passing, more from the perspective of helping me appreciate their continued company!
For those who are interested in American excess death statistics the CDC has a nice visualization dashboard here. It allows you to filter by states. New York's graph looks particularly harrowing. And yes, the last two or three weeks are probably a little low because of the time it takes data to reach the CDC, but the shape is already there.
I know many of you know this but just to be clear in case it's a new term to some, it has a simple definition: that is a mortality rate that's higher than expected. How you specify expected is a little more tricky but only a little. A 5 year average is typical and seasonally adjusted. I.e. you compare April 2020 with April 2015, 16, 17, 18 and 19. If you compare June with January, you'll get spurious results because more people die in winter than summer.
It may be a difference in methodology or AFZ not wanting to get too bogged down in statistical details, but the CDC considers "excess deaths" to be anything in excess of the upper bound of the 95% confidence interval, not in excess of the average. As such the CDC's numbers are a bit on the conservative side. You can see that the weekly death count in that chart is almost always lower than the threshold for excess deaths. If excess deaths were based on an average you'd expect the threshold to be exceeded about half the time just from random fluctuations. Using the upper bound of the confidence interval you'd expect random fluctuations to exceed the threshold only 2.5% of the time.
The 3rd group also contains those that we need to offset against the 1st group. There are a ton of issues doing it simply. But when e.g. people express concern for the "hairdresser's" against "grannies", that's sort of where we need to look, except:
1) We need to compare across alternative universes, that is the "starving hairdressers we've lost" against the "grannies we've saved" (not the ones we still lost).
2) That set of numbers is a horribly mixed bag (it contains +'s and -'s from all sides of the equation)
3) Although the young are at much less risk, it's still a risk. It's not just "hairdresser" v "granny" but "starving hairdresser" v "pneumatic hairdresser"
4) We can, (and actually have to some extent) done something. It's a false dilemma (by parties that care about neither) that this is the contest, we can save (as many as possible of) both.
[ETA 5) It's all hindsight]
Domestic violence is another case where the alternatives are much weaker, so in this case I am very pleased that the UK government has taken steps to protect the men, women and children at risk.
For those who are interested in American excess death statistics the CDC has a nice visualization dashboard here. It allows you to filter by states. New York's graph looks particularly harrowing. And yes, the last two or three weeks are probably a little low because of the time it takes data to reach the CDC, but the shape is already there.
Hawaii's is nice, and to a lesser extent the western and 'A' rural states (and indeed California)
Mississipi and Virginia look very worrying (also what on earth happened to Connecticut!)
IMHO we really don't have enough evidence even to say that, begging your pardon, Gramps49, although perhaps your state actually does---because my state, and most of them, simply aren't testing.
The limited number of random tests, tests of municipal sewerage facilities etc. that have been done suggest that the actual rate of Covid-19 infection might be higher than the published case numbers by a factor of between 10 and 20 (probably depending on where you are).
Certainly in these parts, when they doubled the number of tests they performed per day, they also doubled the number of positive cases, which is consistent with this.
They have been showing an increase in the number of deaths attributed to alzheimers / dementia but last week there were slightly fewer deaths from cancer than the five year average.
Brave new world uses as some statement about women (I'm guessing big busted).
I was looking for a short phrase to express "a hairdresser with (Cov-19 induced) pneumonia", I may have got it wrong.
The point was it's not just a them v us situation (although there are partially elements that way).
United Arab Emirates - 13,599 (10,816 / 2,664 / 119)
Poland - 13,375 (8,949 / 3,762 / 664)
Romania - 12,732 (7,414 / 4,547 / 771)
Ukraine - 11,411 (9,634 / 1,498 / 279)
Indonesia - 10,843 (8,347 / 1,665 / 831)
South Korea - 10,793 (1,360 / 9,183 / 250) 2.7%
The listings are in the format:
X. Country - [# of known cases] ([active] / [recovered] / [dead]) [%fatality rate]
Fatality rates are only listed for countries where the number of resolved cases (recovered + dead) exceeds the number of known active cases by a ratio of at least 2:1. Interestingly none of the eight countries where the resolved cases ratio meets this criteria has a fatality rate close to the 1% figure we've been told is the most likely for COVID-19. The closest is South Korea, which is nearly three times that rate. This implies that either the virus is more deadly than we're being led to believe, or undetected cases are at least double (or in some cases as much as six-fold) the official number of detected cases.
Italics indicate authoritarian countries whose official statistics are suspect. Other country's statistics are suspect if their testing regimes are substandard.
If American states were treated as individual countries twenty of them would be on that list. New York would be ranked at #2, between "everywhere in the U.S. except New York" (#1) and Spain (#3). New Jersey would be between Turkey and Russia.
No countries have joined the 10,000 case club since the last compilation.
Mississipi and Virginia look very worrying (also what on earth happened to Connecticut!)
Connecticut is right next to New York City, with a lot of people who commute back and forth.
If there is no check on the virus it only takes one case and it’ll eventually spread state wide. (or world wide, as we know). No state should be complacent at this stage of the pandemic.
but, without an effective, persistent and globally administered vaccine much of what we currently think of as "normal" will need to change.
What, permanently? Not an option.
Not changing isn't an option either. "Normal" is something that, even in the absence of a pandemic, is in constant flux. We'll come out the other side of this pandemic different, the pandemic itself will have changed us. The question is, what of the old normal will be retained, what will we give up and what new things will we introduce.
If the government wants any chance of meeting carbon targets, it will need to support an ongoing shift to home working now so many workers and businesses have set up for it. I expect that a larger proportion of people will work from home after this, for a larger proportion of the time than used to be the case.
I think greeting rituals may change semi permenantly, shaking hands was already reducing a lot anyway.
It is possible what we consider acceptable personal space will also change.
I think the shift to online ordering and delivery / click n collect will accelerate.
For those who are interested in American excess death statistics the CDC has a nice visualization dashboard here. It allows you to filter by states. New York's graph looks particularly harrowing. And yes, the last two or three weeks are probably a little low because of the time it takes data to reach the CDC, but the shape is already there.
I know many of you know this but just to be clear in case it's a new term to some, it has a simple definition: that is a mortality rate that's higher than expected. How you specify expected is a little more tricky but only a little. A 5 year average is typical and seasonally adjusted. I.e. you compare April 2020 with April 2015, 16, 17, 18 and 19. If you compare June with January, you'll get spurious results because more people die in winter than summer.
It may be a difference in methodology or AFZ not wanting to get too bogged down in statistical details, but the CDC considers "excess deaths" to be anything in excess of the upper bound of the 95% confidence interval, not in excess of the average. As such the CDC's numbers are a bit on the conservative side. You can see that the weekly death count in that chart is almost always lower than the threshold for excess deaths. If excess deaths were based on an average you'd expect the threshold to be exceeded about half the time just from random fluctuations. Using the upper bound of the confidence interval you'd expect random fluctuations to exceed the threshold only 2.5% of the time.
Thanks for that. I keep wondering how detailed to make these posts. Torn between precision and clarity...
Peter Hitchens is still burbling away, saying that the lockdown is excessive, but I noticed one thing he says, which is common among skeptics, that people have been frightened by Boris, or by the govt, and unnecessarily. I don't agree.
It's rational to be afraid of dying. And the virus has spread quickly. In my household, the news that the football manager Arteta, had caught it, was a shock, and made it real. And made the closing down of sport inevitable. And of course, Boris himself was seriously ill.
Skeptics also often dismiss projections of possible deaths, but here they are dismissing mathematics itself. That doesn't make sense.
(You remind me of an obstetric surgeon I know, who was describing a night shift with a load of mothers in labour lining up, some in difficulty, some wanting pool births etc etc. 'So, we're runnin' out of time, an' you know if you go in that way, you could take its 'and off, but if you go this way you might put an 'ole in 'er bladder - but we can fix that, so this way it is'. Precise it wasn't, but the job got done, bish bash bosh. He doesn't speak like that, but it makes the joke better. ETA I couldn't take the pressure - I'd have to go and read the internet for a couple of weeks while I thought about the best way to do it.)
In case it wasn't obvious, it was the likes of him, Simon Jenkin's of the guardian and the liberate movement, I was thinking of.
I don't think there case works out, either morally using their numbers, or with the 'right' numbers using their (alleged) morality.
While forcing the old normal pre-vaccine is just suicidal, the new (mid-term) normality has to be one that works (including new mum's, who almost go through this anyway and new dad's who were expecting different pressures).
Re Connecticut, it's missing Mar/Apr data, but has a clear anomaly that peaks Feb 1. I think it must be a recording issue or something independent (if it were Covid, to get that it would have to predate Wuhan).
Hopefully there will be a big push to improve cycling networks in cities so that less public transport is needed. Fewer cars and planes will also mean better lung health for everyone.
My source of information on Virginia is https://www.vdh.virginia.gov/coronavirus, combined with newspaper articles. The counties in northern Virginia which are hotspots are suburbs of Washington, DC, which has serious problems. I have seen reports of the COVID sweeping through nursing homes in Henrico and Chesterfield counties, which are also hotspots. I suspect that the city of Harrisonburg, which has a much higher rate than the surrounding county, also has a nursing home problem.
Montgomery county, where I live has 56 cases, seven hospitalizations, and 1 death. Unfortunately, in the past week there have been two new cases with one hospitalization. The county population is 98,000.
but, without an effective, persistent and globally administered vaccine much of what we currently think of as "normal" will need to change.
What, permanently? Not an option.
Not changing isn't an option either. "Normal" is something that, even in the absence of a pandemic, is in constant flux. We'll come out the other side of this pandemic different, the pandemic itself will have changed us. The question is, what of the old normal will be retained, what will we give up and what new things will we introduce.
Foremost in my mind are sports and socialising (which includes seeing family and going to church). If social distancing becomes a permanent part of our lives then I don’t see how either of those is going to be possible, and if either of those are allowed then I don’t see how social distancing is going to be possible.
How many people here are willing, however theoretically, to never see friends and family or be able to go to church again?
I was going to make a quip about our great grandchildren living a solitary, socially isolated life as if that’s just what people do. But of course if social distancing becomes permanent then none of us will ever have great grandchildren, because how the hell are people who aren’t already in relationships going to meet prospective partners if they’re not allowed to socialise with people they don’t know?
In case it wasn't obvious, it was the likes of him, Simon Jenkin's of the guardian and the liberate movement, I was thinking of.
I don't think there case works out, either morally using their numbers, or with the 'right' numbers using their (alleged) morality.
While forcing the old normal pre-vaccine is just suicidal, the new (mid-term) normality has to be one that works (including new mum's, who almost go through this anyway and new dad's who were expecting different pressures).
Re Connecticut, it's missing Mar/Apr data, but has a clear anomaly that peaks Feb 1. I think it must be a recording issue or something independent (if it were Covid, to get that it would have to predate Wuhan).
The right wing journalists, (Jenkins is rather different), tend to find a left field epidemiologist, who argues that many people have already caught the virus, or that it has burned out. Plus, of course, the argument that the lockdown is itself killing people, and is wrecking the economy.. Plus, Sweden is wonderful.
My impression is that they mostly start with their conclusion, and hunt around for stuff to support it. This seems grotesque in the case of Toby Young, who notoriously argued that a few hundred thousand dead people is a price worth paying. He also says "We are a nation of bedwetters". Wow.
But I thought maybe Boris was sympathetic to this view, but the projections from Imperial College scared the living bejasus out of him, and presumably he has learned a practical lesson from nearly dying. Of course, IC is mocked by these writers as project fear Mk 2, and they usually ignore the fact that they are projections of deaths without a lockdown.
I was going to make a quip about our great grandchildren living a solitary, socially isolated life as if that’s just what people do. But of course if social distancing becomes permanent then none of us will ever have great grandchildren, because how the hell are people who aren’t already in relationships going to meet prospective partners if they’re not allowed to socialise with people they don’t know?
Foremost in my mind are sports and socialising (which includes seeing family and going to church). If social distancing becomes a permanent part of our lives then I don’t see how either of those is going to be possible, and if either of those are allowed then I don’t see how social distancing is going to be possible.
How many people here are willing, however theoretically, to never see friends and family or be able to go to church again?
I was going to make a quip about our great grandchildren living a solitary, socially isolated life as if that’s just what people do. But of course if social distancing becomes permanent then none of us will ever have great grandchildren, because how the hell are people who aren’t already in relationships going to meet prospective partners if they’re not allowed to socialise with people they don’t know?
Like this it is unsustainable for long (well I suppose there is Tinder, so if it did become worse and mutative).
My guess is that the big (100,000) sports/socials will be thinned out/refimented and that we'll have to move back to more local things.
But when it's down to the levels it now is in China, there's no reason why you can't frequently go to the recreation ground, your family and your pub.
If you know there's a case in Solihul, then you can take action in Solihul (and warn the inflicted's immediate relatives) and be 90% sure that you've now told everyone who might now be contagious.
Comments
For instance, in the US, much of the country has little to no infections. Most of the infections are in metropolitan areas, though there are hot spots in rural areas too as indicated in the discussions centered around meatpacking plants. (My bet is based on this we will have to significantly modify how we produce meat products in the US, BTW)
Even in my state. Washington, we have been bumping along for the past two weeks statewide at 200 infections per seven-day average. Yet, there are some counties that have not reported any infections since the beginning of the stay at home order, and there are other counties that have not had any new reports for two weeks. In my county, we have not had a new case for a week. And of those, only four remain in strict quarantine. While the governor has extended the stay at home rule until 31 May, he will be allowing some of the counties to begin opening sooner provided they meet certain criteria).
We are really hoping for a small family reunion in the middle of June so we can have a memorial service for my brother. The problem is, we live in different jurisdictions that have different timelines for lifting their stay home/stay healthy orders.
Oh, our stats look pretty fine, but the studies I've seen suggest there may be between 40 and 80 extra real-but-unofficial cases out there in the population for every officially counted one. And if that's the case, my small suburb has something like 600 fucking cases right now. But we won't know for years, until some grad student does a dissertation and analyzes the excess mortality rate. Probably about 2025, then.
You do have a point. No, we are not testing very well now, but Inslee says the feds have promised him enough test kits in a couple of weeks to do it. They will be key to easing up on our restrictions statewide.
There could be a feedback loop there. There have been instances where, because of the paucity of tests, the standards for referrals have been set very high (e.g. must have high fever, persistent dry cough, and fatigue. Only two of the three means no referral.) This leads to very few referrals, leading to a shutdown of testing because of so few referrals. One wonders what the results would be if the bureaucratic hurdles were eliminated and testing was done without a referral, or with much lower standards for a referral.
(I don't have much of a view on this - I'm sure meeting it was way more political than clinical, but given how govt and the press works, his pledge has been useful in prioritising a major increase in testing, and it seems we're going to need that to work out how to safely ease lock down.)
Setting a target, especially when you need to fiddle the books to make a claim to having met it, is simply political. It's to be seen to be doing something, even if it's not a particularly useful something.
So; Excess mortality is the only figure that really means anything - for lots of reasons that I will explain shortly - and it's difficult to get an instant read on it. However, we won't have to wait 5 years in developed countries. In the US, the CDC will have good data in 6-12 months and the same goes for the ONS in the UK. I know many of you know this but just to be clear in case it's a new term to some, it has a simple definition: that is a mortality rate that's higher than expected. How you specify expected is a little more tricky but only a little. A 5 year average is typical and seasonally adjusted. I.e. you compare April 2020 with April 2015, 16, 17, 18 and 19. If you compare June with January, you'll get spurious results because more people die in winter than summer.
There are three categories of deaths that are relevant here. Or potentially 6, because in each of the three boxes there will be people who have been tested and those who haven't.
1. People who died from Covid-19 who would have died anyway from something else.
2. People who died from Covid-19 who otherwise had a normal or near-normal life expectancy.
3. People who die from something else who would had survived but for the burden on healthcare or other knock-on effects of the pandemic.
Taking those in turn: 1. We know that the majority of Covid-19 deaths are in the elderly. People over 80 have a 10% mortality in 12 months anyway. Thus if someone was dying from something else, even if Covid-19 infection was the terminal event it is not necessarily accurate to count that as a Covid-19 death. Here time becomes a factor: we are all dying people; if you set your time window at 100 years, then there's no such thing as excess mortality because essentially we'll all be dead in 100 years or so. Conversely, if someone is dying from cancer, it's not unusual that bacterial pneumonia will be the terminal event and thus a Covid-19 infection is not really the cause of death it's just supplanted the bacterial pneumonia and the cancer is the cause of death. The only way to unpick this on a population level is to look at how many people you expect to die in a given period (based on the 5 year seasonal average) and compare that to the actual deaths. The thing that is very alarming is the preliminary data from the ONS suggests this could be really high. Two things to note here: firstly some of those deaths belong in box 3 but are still due to the pandemic. Another key point here is that if you only look at those who tested positive, these box three people will get missed. The second point is that if you have a bad winter with flu, you see a spike in deaths followed by a lower than average mortality in the following weeks as the most vulnerable is now a smaller population because of the ones who have died off. The lower mortality is unlikely to be enough to offset the higher rate but it is still there. So it's a question of timeframe. I would suggest 3 months is probably about right. For the individual and the family, this is ridiculous and callous. I've been close family to someone terminally ill and often those extra days and weeks are very precious. I am not saying that it's ok if people died a couple of months early as they were old anyway. I am not saying that. However, to evaluate how badly afflicted a country is by the pandemic and the effectiveness of a country's response, some leeway here is needed on a population level. I would suggest that a 3 month window makes sense on a population analysis level, whilst not for one moment meaning to minimise personal tragedies.
2. These are easier to count. These will include the majority of key-workers who have died. To a large extent, with proper PPE and a proper, coordinated approach to a pandemic, these deaths are avoidable.
3. This is a very mixed bag and, as I said, if your mortality count is just a number of people who died and tested positive, this number will be wrongly given a value of zero. Some examples: I have no doubt that there will be a spike in cardiovascular deaths and stroke deaths because we know that people are not seeking medical treatment when they need to. More indirectly if the economic cost of Covid-19 and lockdown causes homelessness (as it inevitably will) that will result in increased mortality as there is a measurable mortality from homelessness. We have had a child die of sepsis locally who's parents brought her in very late because they were worried about Covid. She probably would have survived if she got to hospital earlier. Conversely there will be a negative number in this box too. Less journeys means less traffic accidents and deaths. The net result will still be an up spike in mortality overall.
All of this takes some serious analysis to unpick fully but the headline excess mortality rate is very informative and the detailed analysis will show some information on what proportion of these deaths might have been avoided.
The other reason why the excess mortality number is so informative is because, for reasons that are hopefully now evident, it's completely independent of testing. Given how the UK and US are totally failing at testing, this is particularly vital.
AFZ
Booster injections, or a cure, or some kind of meds like the AIDS cocktail to prevent the infection from having its usual effects. There are other ways to fight disease besides vaccines.
I have been following the English NHS death statistics here:
https://www.england.nhs.uk/statistics/statistical-work-areas/covid-19-daily-deaths/
There are some important limitations to these data, including the fact that they are England only and that these are only the in-hospital deaths.
If you follow the link to the daily-chart in the excel file the downward trend in deaths is very clear. The tail is going to be quite long - it's clear that the rise was steeper than the fall, but it also appears to be linear and thus single-figure death rates are likely to be reached in a week or two.
As has been discussed at length on this thread, there are significant problems with the UK-testing program - or rather the lack of one. The fact that the testing criteria has changed more than once and there is no random sampling means that the number of confirmed cases is not very helpful for tracking the epidemic. Hence, I've only been looking at the mortality numbers.
As @Barnabas62 has noted, it is an assumption that community (i.e. care-home) deaths are following the same trend. I think they probably will but he's right, we don't currently know.
There's some really good charts on this point (and addressing my previous post) here: https://fullfact.org/health/covid-deaths/
Anyway, I've been thinking - dangerous, I know. If you're managing an epidemic, the new infection rate is the key data. However, for the reasons I've mentioned, we don't have a reliable read on this number. The mortality number is reliable despite the caveats above and shows a downward trend that I think I would describe as 'encouraging.' Certainly, it is a major relief looking at the trends, if 'encouraging' is too strong a word. However, there is a variable time delay between being exposed to the virus and dying. Some people are dying very quickly, many after several days, some a few weeks. This is important because the peak of mortality was around 17th April in England. Most likely, the peak of infection was more than a week before that.
In terms of the care-home deaths, it's a good bet that the time-to-death (sorry, correct technical term) is shorter. Thus if the death rate in care homes is parallel or even behind the hospital one that suggests the actual spread of infection in care homes is later than in the community at large. This does have implications for managing the epidemic.
All just idle wondering from me - thank you for tolerating my waffle...
AFZ
I did a little googling for '% mortality vs age' stats, but didn't get very far - any suggestions? It would do me good to put some odds on a Difficult Relative, not from the point of view of looking forward to their passing, more from the perspective of helping me appreciate their continued company!
Thank you.
ONS data on life expectancy is pretty good:
https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/lifeexpectancies/datasets/nationallifetablesunitedkingdomreferencetables
It may be a difference in methodology or AFZ not wanting to get too bogged down in statistical details, but the CDC considers "excess deaths" to be anything in excess of the upper bound of the 95% confidence interval, not in excess of the average. As such the CDC's numbers are a bit on the conservative side. You can see that the weekly death count in that chart is almost always lower than the threshold for excess deaths. If excess deaths were based on an average you'd expect the threshold to be exceeded about half the time just from random fluctuations. Using the upper bound of the confidence interval you'd expect random fluctuations to exceed the threshold only 2.5% of the time.
1) We need to compare across alternative universes, that is the "starving hairdressers we've lost" against the "grannies we've saved" (not the ones we still lost).
2) That set of numbers is a horribly mixed bag (it contains +'s and -'s from all sides of the equation)
3) Although the young are at much less risk, it's still a risk. It's not just "hairdresser" v "granny" but "starving hairdresser" v "pneumatic hairdresser"
4) We can, (and actually have to some extent) done something. It's a false dilemma (by parties that care about neither) that this is the contest, we can save (as many as possible of) both.
[ETA 5) It's all hindsight]
Domestic violence is another case where the alternatives are much weaker, so in this case I am very pleased that the UK government has taken steps to protect the men, women and children at risk.
Hawaii's is nice, and to a lesser extent the western and 'A' rural states (and indeed California)
Mississipi and Virginia look very worrying (also what on earth happened to Connecticut!)
The limited number of random tests, tests of municipal sewerage facilities etc. that have been done suggest that the actual rate of Covid-19 infection might be higher than the published case numbers by a factor of between 10 and 20 (probably depending on where you are).
Certainly in these parts, when they doubled the number of tests they performed per day, they also doubled the number of positive cases, which is consistent with this.
https://www.nrscotland.gov.uk/files//statistics/covid19/covid-deaths-infographic-week-17.pdf
They have been showing an increase in the number of deaths attributed to alzheimers / dementia but last week there were slightly fewer deaths from cancer than the five year average.
I was looking for a short phrase to express "a hairdresser with (Cov-19 induced) pneumonia", I may have got it wrong.
The point was it's not just a them v us situation (although there are partially elements that way).
What, permanently? Not an option.
Connecticut is right next to New York City, with a lot of people who commute back and forth.
The listings are in the format:
X. Country - [# of known cases] ([active] / [recovered] / [dead]) [%fatality rate]
Fatality rates are only listed for countries where the number of resolved cases (recovered + dead) exceeds the number of known active cases by a ratio of at least 2:1. Interestingly none of the eight countries where the resolved cases ratio meets this criteria has a fatality rate close to the 1% figure we've been told is the most likely for COVID-19. The closest is South Korea, which is nearly three times that rate. This implies that either the virus is more deadly than we're being led to believe, or undetected cases are at least double (or in some cases as much as six-fold) the official number of detected cases.
Italics indicate authoritarian countries whose official statistics are suspect. Other country's statistics are suspect if their testing regimes are substandard.
If American states were treated as individual countries twenty of them would be on that list. New York would be ranked at #2, between "everywhere in the U.S. except New York" (#1) and Spain (#3). New Jersey would be between Turkey and Russia.
No countries have joined the 10,000 case club since the last compilation.
If there is no check on the virus it only takes one case and it’ll eventually spread state wide. (or world wide, as we know). No state should be complacent at this stage of the pandemic.
If that's what floats your boat...
I think greeting rituals may change semi permenantly, shaking hands was already reducing a lot anyway.
It is possible what we consider acceptable personal space will also change.
I think the shift to online ordering and delivery / click n collect will accelerate.
Thanks for that. I keep wondering how detailed to make these posts. Torn between precision and clarity...
AFZ
It's rational to be afraid of dying. And the virus has spread quickly. In my household, the news that the football manager Arteta, had caught it, was a shock, and made it real. And made the closing down of sport inevitable. And of course, Boris himself was seriously ill.
Skeptics also often dismiss projections of possible deaths, but here they are dismissing mathematics itself. That doesn't make sense.
Hitchens is currently on Mail on Sunday, online.
(You remind me of an obstetric surgeon I know, who was describing a night shift with a load of mothers in labour lining up, some in difficulty, some wanting pool births etc etc. 'So, we're runnin' out of time, an' you know if you go in that way, you could take its 'and off, but if you go this way you might put an 'ole in 'er bladder - but we can fix that, so this way it is'. Precise it wasn't, but the job got done, bish bash bosh. He doesn't speak like that, but it makes the joke better. ETA I couldn't take the pressure - I'd have to go and read the internet for a couple of weeks while I thought about the best way to do it.)
I don't think there case works out, either morally using their numbers, or with the 'right' numbers using their (alleged) morality.
While forcing the old normal pre-vaccine is just suicidal, the new (mid-term) normality has to be one that works (including new mum's, who almost go through this anyway and new dad's who were expecting different pressures).
Re Connecticut, it's missing Mar/Apr data, but has a clear anomaly that peaks Feb 1. I think it must be a recording issue or something independent (if it were Covid, to get that it would have to predate Wuhan).
My source of information on Virginia is https://www.vdh.virginia.gov/coronavirus, combined with newspaper articles. The counties in northern Virginia which are hotspots are suburbs of Washington, DC, which has serious problems. I have seen reports of the COVID sweeping through nursing homes in Henrico and Chesterfield counties, which are also hotspots. I suspect that the city of Harrisonburg, which has a much higher rate than the surrounding county, also has a nursing home problem.
Montgomery county, where I live has 56 cases, seven hospitalizations, and 1 death. Unfortunately, in the past week there have been two new cases with one hospitalization. The county population is 98,000.
I don't think we have a serious problem.
Re "dismissing the math itself":
I suspect the math is way beyond most people's understanding.
Foremost in my mind are sports and socialising (which includes seeing family and going to church). If social distancing becomes a permanent part of our lives then I don’t see how either of those is going to be possible, and if either of those are allowed then I don’t see how social distancing is going to be possible.
How many people here are willing, however theoretically, to never see friends and family or be able to go to church again?
I was going to make a quip about our great grandchildren living a solitary, socially isolated life as if that’s just what people do. But of course if social distancing becomes permanent then none of us will ever have great grandchildren, because how the hell are people who aren’t already in relationships going to meet prospective partners if they’re not allowed to socialise with people they don’t know?
The right wing journalists, (Jenkins is rather different), tend to find a left field epidemiologist, who argues that many people have already caught the virus, or that it has burned out. Plus, of course, the argument that the lockdown is itself killing people, and is wrecking the economy.. Plus, Sweden is wonderful.
My impression is that they mostly start with their conclusion, and hunt around for stuff to support it. This seems grotesque in the case of Toby Young, who notoriously argued that a few hundred thousand dead people is a price worth paying. He also says "We are a nation of bedwetters". Wow.
But I thought maybe Boris was sympathetic to this view, but the projections from Imperial College scared the living bejasus out of him, and presumably he has learned a practical lesson from nearly dying. Of course, IC is mocked by these writers as project fear Mk 2, and they usually ignore the fact that they are projections of deaths without a lockdown.
Yente's back in business via Zoom.
Like this it is unsustainable for long (well I suppose there is Tinder, so if it did become worse and mutative).
My guess is that the big (100,000) sports/socials will be thinned out/refimented and that we'll have to move back to more local things.
But when it's down to the levels it now is in China, there's no reason why you can't frequently go to the recreation ground, your family and your pub.
If you know there's a case in Solihul, then you can take action in Solihul (and warn the inflicted's immediate relatives) and be 90% sure that you've now told everyone who might now be contagious.
Gotcha. I haven’t seen Fiddler yet. Maybe I never will now.
But even online matchmaking isn’t much good if the people so matched can never physically meet. You can’t conceive a child over Zoom.