When it comes to the drastic increases in COVID-19 cases lately, some are desperate to deny that it has anything to do with the fact that states like Florida have loosened the restrictions that were designed to prevent exactly that.
Hence the latest COVIDiot idiocy: the claim that “the increase isn’t because cases are on the rise, it’s only because we’re testing more.” Those damn tests!
The worst offender of course is Donald Trump, who is so addled he might actually believe we can overcome the disease by ignoring it. I’d rather not pay the millions of lives his “plan” will cost.
Worldometer provides data by state for the number of positive and negative tests. Here’s what they list for total number of tests by day in Florida:
The number of tests has certainly increased over time. But it hasn’t increased much recently, if at all. It’s been nearly a month since a sizeable increase in the average daily testing.
Most of those tests came back negative, which is why the graph of negative tests looks a lot like the graph of total tests:
But (and this is a big one) the graph of positive tests does not:
I can hear you thinking, “OK, it’s already obvious, there’s no doubt, the increased caseload is not due to the increase in testing. Any more evidence would just be ‘piling on.’”
Permit me to pile on. Here’s the ratio of positive tests to total tests recently:
People who suggest that maybe cases aren’t really increasing it’s just increased testing, who go on facebook and twitter and blogs and repeat the stupidity without looking into it or even, really, thinking it through, aren’t just falling for one of the stupidest sucker moves of the Trump administration (which is saying a lot), they’re actively promoting that idea. They help Trump, the racist-in-chief rapist-in-chief, advance his murderous agenda on dealing with COVID-19.
Trump’s agenda on COVID-19: save the wealth of his wealthy friends, let the people die.
This blog is made possible by readers like you; join others by donating at My Wee Dragon.
Open Mind 
Wow, cases are rocketing worldwide. Over 30,000 new cases in the US today (so far) and over 49,000 in Brazil (so far), with more than 165,000 cases worldwide, easily the largest increase to date. New deaths are not currently increasing but this may change as the new cases progress through the disease.
Agree – the problem starts when there aren’t any treatment options for the new cases because the beds are all occupado.
That turns the whole thing into a nightmare. We are however (no thanks to the Tyrannosaurus Rump) getting better at treatment. We have a few additional options that may help as we get better at treating this disease but the most important missing link here is regular and frequent testing of every person who goes in and out of a nursing home or aged care facility.
That was the worst thing that Sweden did wrong and it is behind most of the most deadly results. As someone who is 68 going on 69 the prospect of being effectively in lockdown for the rest of my life is a bit annoying but it beats not having the rest of my life to do the things I still can do.
We understand something about this now – we know the masks help – mostly by reducing the spread from us when we are sick and do not know it so thy are worn to help OTHER people. That makes the difference between the ideology on the right and the left a significant thing in this disease. Far be it from me to tell a right wing-nut what to do – but I have to be glad I don’t live anywhere near any of them – here in Wellington, NZ.
[Response: I think a lot of us are envious of those living in NZ.]
Masks also help reduce the innoculum, potentially resulting in a milder disease. Tamino’s envious of us NZers and quite rightly. But don’t think we controlled the virus only through good leadership (though that was important); we also got very lucky and the fiasco of two recent new cases illustrates perfectly that this virus can always take advantage of any loophole.
On Florida, I see it’s just had its largest new case count (3,822) and appears to have now gone to an exponential curve.
It is natural for those of living in a failing state to envy those living in a functioning democracy.
It is not a sharp cut-off, but a positive rate above 10% also means that you are probably missing a lot of cases/infections and that makes fighting the virus with tracking and tracing harder. Florida is back to the situation where the measures to fight the virus will have to assume everyone has it: mandatory masks, no large meetings (especially inside), lock downs, etc.
It looks as if the Trump-idiot governor of Florida is not willing to do such things, but there will be a moment that so many people die that he will be forced to do them. Latest when the morgues flow over and the military has to come in to remove bodies from people’s homes. It would save many lives and make the crises shorter to do this earlier, but learning from what happens in other regions seems to be too much to ask in 2020.
Tamino
As usual: thanks for your good work.
As an European guy, I’m a bit disappointed by the fact that you so heavily concentrate on COVID’s US data. We would enjoy your global view!
Some weeks ago, I posted a graph showing uniform, percentile-based comparisons of 10 countries, with running means over the percentiles.
That was obviously wrong! I should have made a graph of the percentiles of the running means instead (the mistake was best visible when looking at… Sweden, the ‘good’ lockdown-free guy which now shows even more like a bad, bad boy than it did before).
Here is the new variant, comparing Brazil, Germany, Spain, France, Italy, Mexico, Russia, Sweden, UK and US (bold):
https://drive.google.com/file/d/1BdnW6PGtLWmvOySr43P7B7g4Yz_6nj-b/view
I still do not know if this anonymous, percentile-based plotting is correct; but it looks better to me than plots based on data like cases per million, because these imho do not take population density into account.
What is the sense of comparing Australia and Belgium (the country with the highest death toll per capita) ?
One could add lots of other countries like Peru, Chile, Ecuador etc. All are climbing up Worldometers’ hill. But they would tell the name story I guess.
J.-P. D.
Speaking Sweden, they seem to have stopped publishing figures (nothing for the last two days on Worldometers or the Johns Hopkins dashboard). I couldn’t find any information on why; does anyone here know?
https://www.covid19insweden.com/en/
Victor Venema
Here we can see why I don’t like logarithmic representations…
Rgds
J.-P. D. in Germoney
The reason that Sweden has not published to worldometer for some days is due to midsummer national holiday.
Some things worth noting about the worldometer data by the way. The data is not that accurate, at least not the European data but I would guess this goes for all countries. It differs from meteorological data in that it is not at all done coherently across the world.
Firstly it assigns the day of report submission as day of death. This is problematic and misleading as it takes a while for the death reason to be confirmed, up to two weeks or even more in many European countries. This means that the “new deaths” information at worldometer for each given day is off, those deaths usually has happened a while back.
Secondly and perhaps even more misleading but not due to worldometer is that there is no standard how to report. Each country has it’s own way for testing and death confirmation therefore the numbers between countries are not reliable.
So be careful about using the data, do not make too big assumptions based on it. As they say, if you put garbage in you get garbage out. These are reasons why many european media has discarded worldometer as a source for the data.
Complete accuracy is impossible, especially as countries don’t publish reliable data and data that are inconsistent across countries. I don’t think any source does better but the Johns Hopkins app shows roughly the same data but probably reset the days on a US basis (worldometers resets at 000 GMT (roughly)). Johns Hopkins does reference worldometers as a source, though, so it can’t be as bad as you seem to think. Which data source would you recommend as an alternative to the two mentioned above?
20 June and the reported number of daily US Covid-19 cases is again up in the 30,000s, The next couple of days will be worrying times for those watching Covid-19 infections in the US.
I note the malignant narcissism of Trump at Tulsa explained away these signs of a second peak (to a half-empty auditorium) as being the result of too much testing. Strangely (or not in the case of The Donald) he forgets that back in May he was praising himself because the US was testing more than anywhere else in te whole wide world.
He also was on form at Tulsa trying to nail-down the name this desease-of-so-many-names. Of course, Donalsd T Rump is breath-takingly stupid and missed out the name of Covid-19 most significant to the US.
That is of course “Trump’s Plague”.
So with that “couple of days” mentioned above now behind us, what does the damage from Trump’s Plague look like now.
As there is a weekly element to the reported COVID-19 cases, the numbers below are the percentage increases on 7-days-before for the last two weeks. (Data used from Worldometers)The last two days show no sign of those increases in number-of-reported-cases having run their course. Rather, they show an acceleration that is iteself presently increasing. (Over this 14-day’s-worth of data points, acceleration is rising linearly at a rate of +3% per day.) This is not a good place to be.
-14%
+1%
+3%
+8%
+12%
+7%
+10%
+34%
+25%
+19%
+23%
+31%
+29%
+51%
Yeah, just by eyeball, we’re basically at the levels of daily increases that obtained in early April. And as you say, the rational expectation at this point would be that it’s just going to be getting worse for some time.
Another day, another datapoint. 36,015 cases reported.
So the increase on the week = +41%, slap on the projection of the exponential acceleration being boosted by 3% per day.
By such count, projecting the next data-point sees the number of cases centred on 38,300 in a range 34,600 to 42,000.
The one saving grace is that this present level of US cases, while at the same level as the first peak, the incidence is spread further over the US than the carnage in NY/J in the first peak. NY/NJ had 45% the US cases back then but with only a tenth the population.
But saying that, with the present exponential growth and that 3% daily boost, the level of infection experienced in NY/NJ in the first peak will arrive across the whole US shortly after 4th July.
I just hope there are sensible people your side of the pond who recognise that situation and are in a position to stop it.
There are, and omens are looking increasingly good that they will show up at the polls in force in November:
https://thehill.com/homenews/the-memo/504213-the-memo-trumps-law-and-order-bet-falling-flat?fbclid=IwAR12UMXFY9jrbj-HcRnOApGz8fwcDQ294qqE6143WhygaAcd-pCVyr61WBs
Note that Trump has given up all pretense of trying to address the epidemic. All he says about it now is that:
1) he was brilliant to impose travel bans (which everyone can see were totally ineffective in blunting the epidemic) in January, and
2) we test more than anybody–uncharacteristically true in absolute terms, but we’re only #26 in per capita terms–and again, it hasn’t kept us from a really horrendous practical result. Worse, he just publicly admitted, repeatedly, insisting despite aides’ multiple claims to the contrary that he “was not kidding”, to having sought to slow down testing so our stats wouldn’t look so bad.
Yes, really:
https://www.cnn.com/2020/06/22/politics/donald-trump-testing-slow-down-response/index.html
Normal people, even if they are Trump supporters, are not oblivious to the incoherence of touting testing as an accomplishment while admitting you opposed it.
Yet another day, yet another data point. 38,386 cases reported. Not just “slap on” but a bull’s eye. The increase on the week 46%.
So tomorrow 46%+3% = 49% which would suggest a central projection of 41,600 in the range 37,800 to 45,600.
The bad place remains the projected future. But, Doc Snow above, on these numbers the bad place will arrive next month, not in November.
Another day. Another data point. 40,184 reported cases, the highest on record, topping the previous highest back on 24th April.
The projection goes a little strange through the next few days as it is amplifying the weekly cycle caused by weekend reporting. The central values run:-
51,000
51,800
41,400
51,000
59,500
64,700
71,500
within a range of roughly +/ 10%.
From this side of the Atlantic there is no news of any efforts being made to confront this rising level of infection. We do have our own major lockdown reversals to worry about, some of which was very local to me, with our Vikki appearing on national news.
There’s a scientific news summary piece in Nature dated 16 June 2020 which addresses estimates of the “infection fatality rate” (IFR) or, in other terms,
. While it does not explicitly link to the supporting papers, they are not difficult to find using Google Scholar. I’ll link them below.
The IFR is converging to about 0.006 for the general population under age 65 and 0.06 or ten times higher for people 65 and older. There are probability intervals available about these.
(1) T. W. Russell, et al, “Estimating the infection and case fatality ratio for coronavirus disease (COVID-19) using age-adjusted data from the outbreak on the Diamond Princess cruise ship,” February 2020. Euro Surveill. 2020;25(12):pii=2000256. https://doi.org/10.2807/1560-7917.
ES.2020.25.12.2000256.
(2) A. Clark, et al, “Global, regional, and national estimates of the population at increased risk of severe COVID-19 due to underlying health conditions in 2020: a modelling study”, Lancet Glob Health 2020,
15 June 15 2020, https://doi.org/10.1016/S2214-109X(20)30264-3.
And two really interesting papers about estimating latent population sizes:
(3) T. Jombert, et al, “Inferring the number of COVID-19 cases from recently reported deaths”, Wellcome Open Research 2020, 5:78 Last updated: 26 MAY 2020, https://doi.org/10.12688/wellcomeopenres.15786.1.
(4) T. W. Russell, et al, “Using a delay-adjusted case fatality
ratio to estimate under-reporting”, https://fondazionecerm.it/wp-content/uploads/2020/03/Using-a-delay-adjusted-case-fatality-ratio-to-estimate-under-reporting-_-CMMID-Repository.pdf (35 citations).
By the way, “(3)” notes their model can use a Negative Binomial in place of Poisson to model heterogeneity in transmission. From what we now know about superspreaders, that’s critically important. Also, presumably their CFR (= IFR) could be reset at 0.6%, although the proper way is doing a population weighted two endpoint mix,
, where 
is the fraction of older and susceptible individuals. I frankly would like to see results sampled along the unit interval for 
. Don’t know what to do about 
. That seems more to be a function of how well local authorities are clamping down.