Lots of people are talking about Sweden. I’ve been avoiding it because the daily data has a patter that SteveF noticed in the Florida data: It looks like there’s a whoppin’ oscillation riding on top of the main trend. I told SteveF that until I see two successive oscillations where the “peak” and the “trough” on the daily death rates both decline, I’m not going to buy anyone’s claim that they know Sweden’s death rate is being reigned in, or flattened, or whatever.
We’ve certainly not seeing two successive oscillations with the peak in daily deaths declining. Nor the trough declining. Both rose on the most recent oscillation in deaths. These graphs are from Worldmeter. (I prefer semi-log graphs for data that are generally expected to be exponential, but these do show the oscillation which is quite large.
Data source: https://www.worldometers.info/coronavirus/country/sweden/ (I merged the graphs.)
Update
Mike M found a better graph at Wikipedia and I think I’ve tracked down the underlying data. I’ll try to plot that up tomorrow. Meanwhile, this is the latest Illinois data. (I noticed I entered an incorrect shift for the date of the lockdown. So this graph will look a bit shifted compared to the previous. Also, some people complained 15 days is too short for death lag. So I added 18, which is what the value in a paper James originally sited. It’s just a trace to help you eyeball anyway. )
The ‘troughs’ seem to line up with weekends. Perhaps death certificate issuance is a weekday occupation. The three day decline around Easter weekend might be related.
ace,
I think that’s SteveF’s theory on a similar feature in Florida. I’m sure we have some of that in Illinois.
It’s a big oscillation in Sweden.
The chart at Wikipedia for cases looks like the one above. But the one for deaths looks extremely different:
https://en.wikipedia.org/api/rest_v1/page/graph/png/2020_coronavirus_pandemic_in_Sweden/0/fb22dd01b572c08d7d45314573bdc703fa695cb2.png
I suspect that the Wikipedia graph uses data classified by date of death while the Worldmeter is using reporting date.
Lucia,
I don’t know for sure what is going on in Sweden, but it does look like more diligent reporting on week days; after all, they are at restaurants and sidewalk cafe’s on the weekends. 😉
.
The healthdata.org projections for Sweden keep pushing the peak death rate to later dates (now May 19) and lowering the total number of projected deaths; they are now projecting 9,191 total by August (1,581 as of two days ago). I will be shocked the total reaches 9,000+.
MikeM,
Definitely smoother! Now to find the underlying data. (Wikipedia says they are updating daily…. but of course link to a whole graphics intensive site. And in Swedish. )
There is a link to an excel sheet, and that includes a column for “Antal avlidna per dag”
For above comment, spreadsheet at
https://www.folkhalsomyndigheten.se/smittskydd-beredskap/utbrott/aktuella-utbrott/covid-19/bekraftade-fall-i-sverige
The web page with the spread sheet has some notes (which I put through google translate)
“The number of cases reported in the statistics is based on laboratory-confirmed cases reported in accordance with the Infection Protection Act according to the reporting date. Data on the number of intensive care providers is obtained from the Swedish Intensive Care Register’s special reporting module SIRI. The number of reported cases is constantly changing as people seek care and are examined for the virus. There is some delay in reporting and supplementing data on new cases and deceased cases, so the number of days in recent days (especially during holidays) should be interpreted with caution. The statistics are compiled daily with notifications received until 11.30 the same day. The current and previous day’s number is incomplete and will be complete only the following day.
The statistics of deceased persons are based on data reported so far to the Public Health Agency, differences with the regional reporting. The statistics show the number of people with confirmed covid-19 who have died, regardless of the cause of death. The time series with the number of deceased per day contains only those cases where the date of the death is known, therefore the total number of deceased can differ from the number reported in the time series.
Laboratory-confirmed cases are an indicator for monitoring developments. Initially, people who were living in areas with known spread of covid-19 were sampled. But since mid-March, people with symptoms consistent with covid-19 infection who are in need of inpatient hospital care, healthcare personnel and elderly care, and people who are being tested in sentinel testing are primarily being tested. This means that people with mild symptoms are not asked to contact the health care and therefore do not end up in the statistics of reported cases. When using data, specify the source Public Health Authority.”
Italics mine.
I’m guessing this is the better data than at world meter, but be cautious about the past few days. I think I’ll look at it for a day or two to see which days of data change.
(I wonder if I can find similar cleaned up data for Illinois? I know Annan says similar thing about different data sources in UK.)
lucia,
In case you hadn’t figured it out from context, that’s number of deaths per day.
Unless the Swedes are really relaxed about reporting avlidnas, it sure looks like their antal avlidna per dag is already falling, even if you discount the last 4 to 5 dags on the graph. Wait… no, that can’t be right, they are supposed to be suffering for the sin of adopting a different policy.
Dewitt,
I guess it, then put it into google translate. 🙂
SteveF,
Yep. Looks like falling too. The Swedes do say lots of people are staying home. I think they are also the country with the largest single person households in the world and already has lots of telecommuting. So what worked for them migth not work for others.
Neverthelss, it’s important to have a look at them. It’s just all the data I was seeing processed has those YUGE wiggles. I was like…. You want me to conclude something from t h h h aaat?!
Lucia,
“So what worked for them migth not work for others.”
.
Sure. Probably would not work for the NYC metropolitan area, and probably not for most large cities. This is a major pandemic of the densely populated regions, not of the countryside; I suspect the Swedish model would work fine in many places. But more importantly, more places should be free to try that model. One wag (can’t remember who) asked if 52% of the covid deaths were in North and South Carolina, and the NYC metropolitan area had but a handful of deaths, would NYC shut itself down? To even ask the question is to answer it.
SteveF,
Sure. I agree population density is probably a huge driver. Most our deaths are in Chicago. Then Cook County outside Chicago. Then the collar counties.
We’ll see what gets implemented for coming out of the lock down going forward.
I plan to plot up Sweden with some sort of line to show “last week” on log paper. I’ll just plot and post. Then I’ll try to see if there were any dates of important advisories. I know there was one. It’s just not the shutdown other places have.
Other than NYC, what evidence do we have that density is a major factor? Real question.
McKinley County, New Mexico has a population of 73,000 and has had 519 confirmed cases. That is 718/100K, higher than any state other than New York or New Jersey, on which they are gaining rapidly. The population density is 5.5 people per square mile.
Mike M.,
McKinley County has had only 7 deaths giving a low death/case ratio. New Mexico has a very high test to case ratio at 19:1. More tests are going to find more cases. Also, can you say ‘anecdotal’? You would need lots more than one example of low population density and high case number to call the high population density increases transmission rate hypothesis into question.
I think contact rates between different people is the most important parameter. One can easily see how dense urban environments produce a higher number but there can be outliers such as small communities that meet for church every morning for example.
Mike M,
There is quite strong correlation between deaths per million population and population density. http://imgur.com/a/1N50na1
The correlation with cases per million population is even stronger. not perfect, of course, but pretty strong.
Not sure who was asking about ventilator mortality …
.
In New York’s largest hospital system, 88 percent of coronavirus patients on ventilators didn’t make it
https://www.washingtonpost.com/health/2020/04/22/coronavirus-ventilators-survival/
“Eighty-eight percent of the 320 covid-19 patients on ventilators who were tracked in the study died. That compares with the roughly 80 percent of patients who died on ventilators before the pandemic, according to previous studies “
Tom Scharf
Sure. Then they kill each other, but it ends there.
DeWitt Payne (Comment #183504): “You would need lots more than one example of low population density and high case number to call the high population density increases transmission rate hypothesis into question.”
.
No, I don’t need *any* examples in order to question an assumption and ask for evidence.
Yes, a dependence on population density is plausible. But it seems there is also a correlation with poverty. And with race. And density, poverty, and race are correlated with each other. So it is hard to separate which is responsible.
So we have a county that is 86% minority, 33% below the poverty line, with a very low population density and a lot of cases. That suggests that maybe density is not the key factor.
New Mexico isn’t Iowa or Rural Illinois. People are clustered. So where people live the density isn’t all that low.
The Zuni’s in McKinley aren’t spread out all over the desert. The city of Zuni Pueblo, New Mexico in McKinley New Mexico evidently has a population density of 720 people/ square mile. That’s not bit, but it’s also not 5 people/square mile.
https://en.wikipedia.org/wiki/Zuni_Pueblo,_New_Mexico
The city of black Rock in NM has a population density of 206.3 per square mile. That’s a more spread out.
What matters is contact. All other things being equal, population density results in more contact. But evidently tribal elders are trying to persuade people not to engage in large religious dances.
My wife’s best friend married a Swede and they live in a suburb of Stockholm. She reports that weekends are sacrosanct in Sweden and they don’t even try and report numbers then. So analysts are optimistic about their approach on Monday morning and pessimistic starting around Wednesday.
Dang… the file I found still has the wiggles. I guess I’ll find the one Wikipedia draws from later… sigh…
>There is quite strong correlation between deaths per million population and population density.
I’m surprised they haven’t put out a press release yet – ‘COVID is caused by global warming’.
MikeN,
“I’m surprised they haven’t put out a press release yet – ‘COVID is caused by global warming’.â€
.
No, but lots of global warming activists have already said the draconian restrictions on private actions, command, control, and economic contraction the pandemic has brought on are EXACTLY what they want….. permanently.
MikeN,
There’s a lot of tweeting about how Covid shows we do need to respond to collective threats and so on. And so, of course, Covid means we should all now ‘see’ that we need to act on climate change.
This article discussed “Lived” population density. That is, measuring population density based on where people acutually live rather than just dividing the number of people in a country by the area in the footprint of the whole country.
https://www.citylab.com/life/2018/02/theres-a-better-way-to-measure-population-density/552815/
I found the article by way of a tweet that included a figure showing the correlation between lived density and COVID-19
Lucia,
Here is a site with all the Swedish data: https://experience.arcgis.com/experience/09f821667ce64bf7be6f9f87457ed9aa
The interesting comparison graphics for me are the cases by age group and the deaths by age group. There are almost no cases for people younger than 30. There are lots of cases for people older than thirty. There are almost no deaths below age 60, and total deaths are dominated by people over 70, and especially over 80. Looking at the Swedish data, it becomes clear this is a serious pandemic, but only for the elderly.
SteveF,
I’ve seen that site. But where to I click to get the underlying data?
That’s what lots of sites look like right now– graphics they chose. But hard to find a link to the underlying data. (It would help if I spoke Swedish. But based on the ones in English, it wouldn’t help much!!)
Lucia,
I had Chrome translate all the pages at that site to English (Chrome automatically enlists the help of Google Translate). I can find no link to a spreadsheet or similar form for the same data. They have 82 days of daily historical data, broken down by cases, intensive care cases, and deaths, in total and by region, but you would have to transcribe to use it… nothing downloadable.
lucia (Comment #183512): “The Zuni’s in McKinley aren’t spread out all over the desert.”
.
I have not heard anything about an outbreak in Zuni Pueblo, although some central New Mexico pueblos have been hit hard.
To a large degree the Navajo *are* spread out all over the desert, although these days there are also towns. The Navajo Nation has been hit really hard by the Wuhan virus. But I don’t know if that is confined to the towns.
.
lucia: “New Mexico isn’t Iowa or Rural Illinois. People are clustered. So where people live the density isn’t all that low.”
.
But that is pretty much the case in any rural area.
lucia (Comment #183529): “I found the article by way of a tweet that included a figure showing the correlation between lived density and COVID-19”
.
Now that is starting to look like evidence. But although it suggests that population density is a factor (unsurprising), it is not *the* factor.
MikeM
Not the same everywhere. In Iowa, people on farms really live on their farm. The state gets rainfall uniformly everywhere. The farms are spread out . There are no big inaccessible mountains. The people and towns are spread out.
No one said it’s “the” factor. It seems to be a big one.
Tom Scharf,
That was me. It’s likely that at least half of those ‘survivors’ will die in the next year. And hospital administrators were pushing doctors to put people on ventilators to try to limit infections.
That’s a lot like heart attack victims that need CPR and/or defibrillators. They usually don’t live very long even in the unlikely event they are revived and discharged from a hospital.
MikeM
If you could find specific link to the data (not a figure) for the daily death figure you found at Wikipedia, that would be great. The one I found was the one that oscillates. . .
Sadly, Wikipedia gave the “source” to the top of the site, and try as I might, when I find data it’s only the oscillating stuff. (Why people don’t post links to the actual data . . . sigh. . .)
If only 12% survive after being on a ventilator, then it is possible a fraction of those would have survived anyway without being on a ventilator, and a lesser chance that the ventilator actually killed a few people. The ventilator may only be saving a very few number of lives.
lucia (Comment #183545): “Not the same everywhere. In Iowa, people on farms really live on their farm.”
.
No, the farmers live on their farms. But these days, there are not so many farmers, even in Iowa.
When I think of rural Iowa, the first place name that springs to mind is Ottumwa, I suppose because it was the home town of Radar O’Reilly. So I decided to compare its population to the county it is in. Then I wondered how representative that might be, so I checked the surrounding counties and their county seats. Here are the results:
Wapello County 35,625
Ottumwa 25,023
Jefferson County 16,843
Fairfield 9,464
Monroe County 7,970
Albia 3,766
Davis County 8,753
Bloomfield 2,640
Keokuk County 10,511
Sigourney 2,059
Mahaska County 22,335
Oskaloosa 11,463
Over half the population in that six county area lives in the county seats. And there might be other towns.
————
p.s. – I am starting to fear for my mental health. 🙂
lucia (Comment #183550): “If you could find specific link to the data (not a figure) for the daily death figure you found at Wikipedia, that would be great.”
.
Indeed. I have searched for the actual numbers for a number of such Wikipedia articles, but always in vain.
The worldometers.info page for Sweden list this page as its source.
It’s in Swedish, but it’s fairly obvious what’s what. The case distribution by age is fairly flat except for the two youngest decades, 0-9 years and 10-19 years with very few cases and no deaths. It drops off some for the 90+ age group. The deaths, though, are concentrated at the high end. Nearly 90% of the deaths are age 70+ with the peak at 80-89, 807 deaths, 40% of the current total of 2,021 deaths. 70-79 and 90+ have nearly equal numbers of deaths at 476 and 474, slightly less than 24% each. 60-69 has 160 deaths, or 7.9% of the total.
According to this site, ~2 million people, 20.4% of the Swedish population is 65 and over.
Tom Scharf,
I posted links to articles about ventilators earlier. I could probably find them again, but the gist was that ventilators were being used when blood oxygen levels dropped below some critical level, but the ER doctors were reporting that many of the patients were not, in fact, in severe distress, so it seems at least as likely that some of the patients on ventilators that died would have lived if they had not been put on ventilators. That is, the ventilators killed them.
DeWitt,
It is clear that your personal Hell would involve a mechanical ventilator. I hope you have made your preferences plain to your family.
SteveF,
Since I would be in an induced coma if I were on a ventilator, it’s not so much Hell, but a waste of resources.
DeWitt,
Fair enough. Trust your family members.
Temperature correlates to population density, because of urban heat islands. Population density correlates to COVID cases.
Trump cancelled Paris Accord.
Therefore Trump killed 100,000 Americans and a million globally from COVID-19.
https://wattsupwiththat.com/daily-coronavirus-covid-19-data-graph-page/
.
Per Willis, Sweden is still running parallel to the US in deaths per million.
.
Also per Willis, “ in the US, new cases is a function of new tests. For every one hundred additional tests that we do, we find an additional nineteen confirmed cases of coronavirus. â€
.
“.. Conclusion? Don’t use confirmed cases as a metric of the spread of the virus—the number of cases is indeed a function of the numbers of tests. ..â€
.
https://wattsupwiththat.com/2020/04/22/of-tests-and-confirmed-cases/
.
Or maybe tests requested are a function of how many people are sick.
Ed Forbes,
Maybe that used to be true. We now have testing on demand in my area. The results are that only about 1% of the tests are positive. That’s single digit positive results as the number of infected in my area seems to be small.
lucia,
Here are the daily deaths from the Wikipedia figure. I just gave in and transcribed them from the figure.
Mar/18 6
Mar/19 7
Mar/20 9
Mar/21 8
Mar/22 12
Mar/23 11
Mar/24 20
Mar/25 23
Mar/26 31
Mar/27 32
Mar/28 35
Mar/29 39
Mar/30 43
Mar/31 47
Apr/01 53
Apr/02 69
Apr/03 78
Apr/04 71
Apr/05 86
Apr/06 91
Apr/07 83
Apr/08 110
Apr/09 84
Apr/10 89
Apr/11 98
Apr/12 96
Apr/13 84
Apr/14 91
Apr/15 108
Apr/16 107
Apr/17 72
Apr/18 72
Apr/19 73
Apr/20 70
Apr/21 42
Apr/22 40
Apr/23 31 incomplete data
They clearly leveled off starting two weeks ago and have been declining for the last week. Illinois looks like it might be still increasing.
Sweden had 2129 deaths as of 4/22 while Illinois has 1795 as of today. But deaths started rising in Sweden about one week ahead of Illinois and Illinois has about as many deaths as Sweden had a week ago. Illinois does have about 20% more people than Sweden.
It seems to me that if the lock down is working in Illinois, it is working even better in Sweden.
Mike M,
I want the find source of the data. Not a list of the numbers. A list of the numbers is useless if I don’t know what the source is!!
.
The analysis you have afterwards isn’t taking the correct factors into consideration to decide if something is “working” in Sweden. I would like to find the source of the data. I’m guessing you didn’t find it.
lucia,
I believe the original data comes from here.
You can select from four graphs, cases/day, cases cumulative, intensive care cases/day and deaths/day. It looks like Mike M.’s data is off by a day. I also don’t believe that deaths/day has dropped off that fast. I suspect the data aren’t complete. Note that this looks very different from worldometers.info for Sweden although worldometers lists this page as its source.
Dewitt,
When I click to that page, I see graphics. How do I get the underlying numbers for those graphs? There must be something to click.
Also, which is death’s per day? I see “Sjukdomsfall per ” which is cases. I want deaths.
DeWitt
This is an image of the site I see when I click the you gave
The Sjukdomsfall per dag is cases. I don’t see death’s per day. I don’t see a link to data (though it may be there.)
Yes, Dewitt, that looks like it. And yes, it looks like I was off by a day in reading the numbers from the Wikipedia graph; my 4/22 is their 4/21. I guess we can check in a few days to see if the numbers change.
Ok. I get it. I need to click that little arrow icon. (I still want a link to the source!!) It does look very different from the worldmeter data and from the other source of data I found!
Lucia,
It took a while to find, but if you click around, you will find a list of 82 pages. Each page contains data for a date: the number of cases by region, total for that day, deaths for that day, and more. I believe all the graphs come from that data. The Stockholm cases and deaths dominate, of course.
(One of the reasons I want to see the source is to see if it’s something that is being updated. One of the sources explicitly says not to trust the recent data, but doesn’t say how far back we need to wait to consider it trustworthy.
lucia,
If you want to go exploring, click on the ‘Information om datakällor’ link under the map.
DeWitt,
Yes…. clicking that reveals this
I tried clicking all of four of those. None of those goes to a list of data.
Not trying to be difficult here. Has anyone found a series of clicks or a url that actually has the source of the death data? If so, what series of clicks worked. SteveF, what exactly did you click to find the deaths per day?
lucia,
I don’t know what you are trying to find. You found the graph. If you hover over a bar on the graph, you get the date and number of deaths. Do you want more than that?
Mike M
I’ve said I want the actual source for the data. I think I found it:
For example, if I go here:
https://www.folkhalsomyndigheten.se/smittskydd-beredskap/utbrott/aktuella-utbrott/covid-19/bekraftade-fall-i-sverige
I find a link to
“Data som statistiken ovan bygger pÃ¥ kan laddas ner här (Excel)”
That gets me an excel spread sheet with
““Antal avlidna per dag—
This one looks like your data. Now I can read that page for information. (Because other pages warn recent data are not necessarily updated.)
I can also watch over a few days to see if the recent data changes.
Maybe this is part of why things have been so bad in New York City:
https://nypost.com/2020/04/18/gov-cuomo-urged-to-shut-down-nyc-subways-to-stop-coronavirus-spread/
People are told to stay home, but there are a lot of essential workers who still have to get to their jobs. On public transit, of course, because New York City. So the city made big transit cuts and the buses and subway cars are packed.
But don’t worry. They are minorities and deplorables, so it is OK.
———-
The people responsible for that decision belong in prison. I say that because I am opposed to capital punishment.
———
p.s. – Have I finally found an exception to Poe’s Law?
Ok.. This page says not to trust the recent data especially the past two days. It’s not more data than Mike M has (meaning it has 4/23, 4/24 supposedly.) Then it has a final row that says “incomplete data” with a 17.
So this seems to be the source which is what I want. (Not just the numbers.)
I guess I wait two days, go back and see if the final number changed.
Sweden updated the data at their underlying source for the graph Mike M posted. I had downloaded last night and checked this morning.
.
No numbers before 4/12 changed. Those for 4/12 increased by 1. It looks like you can’t trust the numbers in the final 10 days on the list. They go up. The didn’t go up a huge amount — but they weng up. So for example overnight, the value posted for 4/23/20 went from 31 to 41. The value posted for 4/24 went up from 6 to 19.
.
My guess is Sweden’s daily death rate is going down. But we won’t really know true daily death rate data for a while.
.
I’d guess the “world meter” data is somehow what was reported up to that day. Then people go back and attribute it to the correct day of death. Similar things may happen in all states and countries. The process takes the “wiggles” out, but it also may mean the smooth data is not up to date.
.
Mike M: Being able to compare these is why I wanted the source, not just to pick off current numbers from the chart. Now it’s fairly easy to so how they get updated so we can have a clue how much to trust the shape of that graph. Basically: the appearance of a drop since the max death rate in Sweden is going to lessen. (It might not go away. But it’s going to lessen. I’ll be able to say about how much in a few days. Or a week.
The number for 4/21 (4/22 on the list I posted above) has jumped from 31 to 41. For the prior week, it looks like a couple per day have been added.And now there are 19 listed for 4/22.
It seems pretty obvious what is happening. The Worldmeter data are plotted according to the date reported, so that each day is complete before going on to the next day. But the FOHM site is using date of death, so that the reports being filed today are being added to the appropriate prior day, as recorded on the death certificates. So they are still waiting for the paperwork for all the people who died the last two days, many of those who died on Wednesday, and some who died on previous days.
———-
Addition: The FOHM data is probably just as up to date as the Worldmeter data. It is just more obvious that it is not as up to date as one might imagine.
As of just now, both sites say 2192 total deaths.
If I were worldometers.info, I would record the total number of deaths reported each day by FOHM at a fixed time, I believe they use midnight Zulu time. Then subtract the previous day’s number to get deaths/day.
Mike M,
Yes. That’s what I think both are doing. But it means we can’t just eyeball the FOHM data and determine the current trend in deaths. I’m looking at it over the course of a few days to get an idea of the number of as-yet-unreported deaths for some number of days back.
The worldmeter data is “uptodate” for what it is which is number of reported deaths to date with no attempt to actually claim those deaths happened on that day.
The FOHM is not up to date for what it is which appears to be an attempt to actually find the number of deaths on a particular date. That takes more time.
Earlier on, I said I would need to wait to be certain deaths are dropping based on eyeballing worldmeter data because it’s got that oscillation. Then we found the FOHM, which looks like it’s dropping. But now we know that the numbers for at least the final 7 days aren’t necessarily on the correct dates. In the case of both counts, some of the deaths that already happened have probably not been reported. So they aren’t in the 2192 tally. For worldmeter, they will appear to happen in the future. For FOHM, they’ll added to the correct date.
lucia,
It’s the weekend. We know from the worldometers.info bar graph that during the week the numbers can go up a lot. I don’t think you can safely draw any inferences or try to model changes in the death rate from the FOHM data based on data from the last ten days or so.
Dewitt,
That’s what I think. It looks like we can’t safely conclude what’s happening in the past week to 10 days from either source. In the case of the worldmeter data, if you are dispassionate and unemotional ( 🙂 ) you can’t help but notice the oscillation and so know you sort of need to see something over a couple of oscillations. If you see the FOHM data and you take the trouble to track down the source, then you should see the deaths in recent days aren’t fully incorporated.
.
My guess is it will turn out death’s are dropping. But right now, I can’t know for sure. So I’m at the same place I was when I wrote this post. BUT, I might have a better idea by … oh… Tuesday. At that point, I’ll have a better idea how far behind the FOHM data is. (And we may get part way through another oscillaation on the worldmeter data!)
More on filtered UV light used internally. Actually makes sense, but unclear how effective it will be in practice.
.
It seems that Trump was right again on something that needs to be looked at.
.
https://www.sgtreport.com/2020/04/youtube-takes-down-bio-tech-firms-video-explaining-potential-uv-light-treatment-for-coronavirus-patients/
Ed Forbes,
He did manages to sound befuddled. . . And then later he claimed it was a joke!!
lucia,
The Lysol and bleach were clearly inventions of the media since in the next sentence he clearly referred to UV light being the disinfectant. But putting words in the BOM’s mouth seems to be required in modern journalism, much like the ‘pandemic is a hoax’ meme.
Very alarming data from Pennsylvania, according to the COVID tracking project. The daily deaths for the last 10 days:
16-Apr 60
17-Apr 49
18-Apr 80
19-Apr 276
20-Apr 92
21-Apr 360
22-Apr 58
23-Apr -201
24-Apr 71
25-Apr 45
April 19 and 21 are pretty scary, but April 23 is downright terrifying. It appears that the zombie apocalypse has begun. 🙂
.
Oh. They retroactively classified several hundred deaths as “probable” Wuhan, then rescinded:
https://www.dailywire.com/news/pennsylvania-forced-to-remove-hundreds-of-deaths-from-coronavirus-death-count-after-coroners-raise-red-flags
.
Whew!
New York’s not looking good. I’m seeing the infection rate in the next layer of counties north of New York going up. The number of new cases outside the city was 1273 larger than the number in the city, As I said above somewhere, it looks like diffusion. Also, at best, Rt is about 1 as the new case/day plot has been more or less flat for the month of April. If any state should have tried to put restrictions on travel, it was New York.
Nationally, new cases on 4/24 was the highest ever at 38,958. The previous high was 34,517 on 4/3. Worldometers says that they should have state level pages soon.
I don’t think we’re going to see extraordinarily high death numbers from Sweden.
There’s a whole lot of hype about Sweden not going into lock down… but the reality is that they are still imposing measures to slow the spread which are not really being reported by the media. They have banned gatherings of more than 50 people. They are doing a lot of other things like the rest of the world, but they are asking their citizens to do these things voluntarily rather than forcing them to do it by law… and the people are complying. People are working from home where they can and they are avoiding unnecessary travel. The government did closed their high schools and all students are now distance learning. They also banned people visiting nursing homes (but they only did that recently after a whole bunch of nursing homes became corona hot-spots). The only real difference is that restaurants remain open but they have been ‘advised’ by their health authority on table spacing. The borders remain open, but the government has issued advisories against travel.
The Swedish concept is to ask their people to take some responsibility for their own health and the health of others. The lack of a formal lock down doesn’t mean that people can just do whatever they like.
They are going to have more deaths per head of population than other places, as their policy is basically trying to achieve a slow burn into a herd immunity… but their death count is not going to spiral out of control as some people would expect.
I have run the latest daily death data for the UK with the Annan Bayesian model and linked the results below. There are 3 previous model runs by Annan and my run on the most current daily death data.
3/29/2020: Ro= NA ; Rt=1.02+/-0.94
4/5/2020: Ro=3.06 +/-0.25; Rt=0.87+/-0.47
4/14/2020: Ro=2.98+/-0.32; Rt=0.49+/-0.27
4/26/2020: Ro=3.1+/-0.5; Rt=0.78+/-0.10
The Bayesian model Rt value appears to change with the most current trends in daily deaths and thus does not appear to be wedded to the Bayesian priors.
The tabled results appear valid to me, but the data and labels in the end of the report plot can be a bit different or unexplained as noted by posters here and those that I have observed. I think I will have time today to go through the model code line by line and look for explanations.
I plan to run the most current US daily death data with the Annan model today.
https://www.dropbox.com/s/12kgeee6me9o3a8/UK_Daily_Deaths_Bayesian_02_19_to_04_26.pdf?dl=0
skeptical,
Yes. Interviews of Swedes themselves say they are social distancing. They just haven’t closed everything by fiat.
.
But as I’ve said before: What works in Sweden might not have worked elsewhere. For example Swedes probably aren’t having packed parties in apartments like some are having in the south west side of Chicago (despite the lockdown.)
https://wgntv.com/news/coronavirus/viral-video-shows-large-house-party-on-south-side-amid-pandemic/
The south west side of ChiTown is already hard hit. It’s going to get worse there. The reason our numbers aren’t going down is there are probably pockets of areas where people are not social distancing at all.
Other things different in Sweden (from memory): Most single person homes in the world. Very high on telecommunting pre corona. Scandanavia in general is a country of sort of introverts (relative to, for example, Italians and Greeks.) Despite no stay at home, lots cancelled planned destinations and are doing “stay cations”.
Keeping people from packing beaches, bringing kegs and partying in the US is difficult.
I think Sweden’s curve is looking like New York’s. I’m going to go plot up some charts to show updating of the “smooth” one over time so people can see the danger of interpreting it too quickly. But it does look like it’s going down.
What’s going on in Australia?
–
Unlike Sweden we have had a massive lock down. Australia is America writ small. Same size, but population 25 million to 328 million so 13 times less potential density.
Most people live in the big cities.
6000 cases 85 deaths only 6 per day new cases.
–
Is it seasonal and we will get a big rise as winter approaches?
Is it the warmer climate, healthier population or just the phine addicted society where people did not talk to each other much anyway.
–
I would guess that 75 of the 85 who died are over 80 [note a guess only]. Old cruise ship travelers and Nursing Home victims.
–
Children are being allowed back to school with the messaging that kids do not need social distancing! The logic escapes me though the practicalities do not.
– The Swedish model seems better. A slightly higher initial death rate compared to a death by 100 cuts. Even New York will work out fine, great herd immunity by accident.
Ken, as a UK resident I have been following the numbers with a degree of interest and concern. Both your forecast and Annan’s latest one suggest that the epidemic is turning down but that is not really apparent in the numbers. Essentially the numbers have been flatlining for about 3 weeks so the forecast downturn keeps getting delayed. I suspect the reason is that the figures for recoveries is very low, much lower than countries such as France or even Turkey or Brazil. So far only 780 cases have recovered. Whereas in France the number of active cases is now virtually flat, it is still increasing in the UK. A week ago, Annan’s forecast suggested we were at the peak as does your forecast. My view of the numbers suggest that we still haven’t quite reached the elbow
angech,
Depending which antibody screening studies you believe, there are likely more than 20 times as many asymptomatic cases as confirmed positives…. so more than 120,000 who already got the virus. My guess, based on a few visits to Australia, is that Aussies tend to be less likely than Americans to throw a keg party in defiance of of no-gathering rules. The kids really are at zero risk, and their not-so-old parents at nearly zero as well. Grandparents may want to void contact for a while.
Graeme wrote: “So far only 780 cases have recovered.”
.
I’m going to call BS on that. A ball is being dropped somewhere.
lucia (Comment #183946)
Lucia, good luck with that.
I’ve read a lot of conflicting stories online about how the Swedes report their deaths… but what appears to be most likely is that they count a death on the day that it is ‘reported’ rather than the actual day of death. Apparently there is some government body which sorts out the mess and produces figures based on day of death, but I haven’t been able to locate that data… and even if I could find it, the last two weeks of data would still not be accurate. The numbers on worldometers is the stuff that’s counted on day of reporting. Not many people working on the weekends, so not many reported deaths on those days (as you’ve already observed). The whole thing can be skewed by any longer than normal delays in reporting, so you’re going to have fun trying to produce something which is an accurate representation of what’s actually happening on the ground over there.
skeptikal (Comment #183941): “There’s a whole lot of hype about Sweden not going into lock down… but the reality is that they are still imposing measures to slow the spread which are not really being reported by the media. They have banned gatherings of more than 50 people. They are doing a lot of other things like the rest of the world, but they are asking their citizens to do these things voluntarily rather than forcing them to do it by law… and the people are complying.”
.
NOBODY(*) is suggesting that we do nothing. The argument is between the light touch of the Swedes and the heavy hand of most U.S. jurisdictions. The light touch of the Swedes is working, just like things were working here before a bunch of petty despots got power drunk.
Yes, some people will be irresponsible under the Swedish model. But as lucia points out, people are being irresponsible under harsh lock down rules. As usual, the Left ignores the fact that laws and rules are usually followed by the people who are inclined to follow laws and rules. You can’t control the rule breakers by imposing harsher rules on the rule followers.
———–
* Note: “Nobody” in the sense of “nobody still thinks the Earth is flat”.
skeptikal (Comment #183951): “I’ve read a lot of conflicting stories online about how the Swedes report their deaths … The whole thing can be skewed by any longer than normal delays in reporting, so you’re going to have fun trying to produce something which is an accurate representation of what’s actually happening on the ground over there.”
.
That has been discussed and largely sorted on this site. The reporting in Sweden is more accurate than the reporting here. The total numbers are up to date, but they are also tabulating information on the actual date of death; not available here at all, I think.
lucia (Comment #183946)
Ahh, that’s probably why I’m still detecting a gain in your daily deaths. I thought it was odd that your deaths hadn’t peaked yet, but I just put it down to a high number of cases being imported into your state, rather than transmission by irresponsible people not playing by the rules.
I think a model – and especially a Bayesian model that can be updated with previous posteriors as priors and any other model that can change projections with new data – with capabilities to look at all the data in an objective view of the model (at least within the framework of the model that was fixed from the start) can put bounds on what we as individuals might want to see over short time periods. It is informative of human inclinations and fun to have humans inputs of guesses where the data is leading and even more so when the guesses can eventually be compared in reasonably near time to what actually occurred.
@ skeptikal
“Ahh, that’s probably why I’m still detecting a gain in your daily deaths. I thought it was odd that your deaths hadn’t peaked yet, but I just put it down to a high number of cases being imported into your state, rather than transmission by irresponsible people not playing by the rules.”
It could very well be that the rule-breakers are spreading the virus far beyond what would otherwise occur. Or not. We should be skeptical of allowing our biases to lull us into into believing that we have it figured out. Every data point so far suggests that we don’t.
Oh no! The predicted catastrophe never happened, and the damned Swedes are opening up even more… those evil murders of the elderly. https://www.forbes.com/sites/gabrielleigh/2020/04/27/the-predicted-coronavirus-catastrophe-hasnt-arrived-in-sweden-whats-next/#7ea224963caf
.
People are returning to get haircuts, eat at restaurants, drive to work in Stockholm, and even have barbecues without social distancing. Kids up to 16 never stopped going to school either. What if they gave a catastrophe and nobody came? Maybe Sweden will give us an answer.
skeptical
Well… you probably didn’t try as hard as I did. We discussed the two sets of data right here. In comments.
If you’d been reading here, you could have found the link to the data you tried but failed to find posted in comments. Heck. it’s in comments on this thread!!!! 🙂
I’ve plotted the data were unable to locate. It’s show in today’s post. 🙂
Outdoor transmission to others of COVID-19 is apparently very rare.
https://www.medrxiv.org/content/10.1101/2020.04.04.20053058v1
.
“Conclusions: All identified outbreaks of three or more cases occurred in an indoor environment, which confirms that sharing indoor space is a major SARS-CoV-2 infection risk.”
.
Chinese study said for outbreaks of 3 or more people 80% of them were home outbreaks. 318 outbreaks were studied, only one was traced to an outdoor environment. Note that “transportation” is a separate category which did have outbreaks.
.
This may of course be related to a severe lock down order.
Earle,
The mass of other people staying at home will tend to protect them relative to if they were mixing. Basically, the people at the party (assuming it happened) will contact fewer people that they otherwise would have.
.
Some people suggest the party video is fake news with the party happening sometimes in the past. I think investigations are ongoing.
Tom Scharf,
“Outdoor transmission to others of COVID-19 is apparently very rare.”
.
All the more reason to play golf.
Mike M,
“You can’t control the rule breakers by imposing harsher rules on the rule followers.”
.
Yes, but just like making gun ownership difficult for law abiding citizens, ever more onerous corona virus rules make the nanny state left feel so powerful, righteous, and morally superior. Screw the Constitution.
lucia (Comment #183960)
I was late to join the conversation on this thread so I thought I could get away with just skimming the comments… that didn’t work out very well for me.