All the good stuff, caveats, code, data sources and explanations are linked, some in Update III, and the most important in Update II, Update IV, Update V, Update VI, Update VII, Update VIII, Update IX, Update X, Update XI, Update XII, Bayes Theorem & Coronavirus, and the Sanity Check Perspective, so go to them first before asking what-about-this-and-that. Skip to the bottom for the latest model. Thanks to everybody emailing me sources, including Ted Poppke, Jeff Jorgensen, Jim Fedako, Joe Bastardi, Philip Pilkington, John Buckner, Harry Goff, John Goetz, Warren McGee, Robert Kinney III. https://wmbriggs.com/post/30606/. Sorry I’m slow answering emails.
During a task force meeting Wednesday, a heated discussion broke out between Deborah Birx, the physician who oversees the administration’s coronavirus response, and Robert Redfield, the director of the Centers for Disease Control and Prevention. Birx and others were frustrated with the CDC’s antiquated system for tracking virus data, which they worried was inflating some statistics — such as mortality rate and case count — by as much as 25 percent, according to four people present for the discussion or later briefed on it.
Trump had a press conference and mentioned bad models. CSPAN gave the transcript (they wrote in all caps):
THE MODELS HAVE NOT BEEN EXACTLY ACCURATE. THESE ARE MODELS DONE BY THINK TANKS, UNIVERSITIES. LOOK AT SOME OF THE MODELS. THEY HAVE BEEN WAY OFF. A FEW OF THE MODELS HAVE BEEN ACCURATE. AS FAR AS THE MODELS ARE CONCERNED, IF YOU GO BY THE MODEL, IT YOU COULD LOSE 2.2 MILLION PEOPLE. BECAUSE WE MITIGATED, WE DID THINGS THAT WERE VERY TOUGH FOR OUR COUNTRY TO DO, FRANKLY — THEY TURNED UP A MAGNIFICENT ECONOMY. WE HAD TO TURN DOWN THE WHOLE COUNTRY. WE ARE AT THE LOWEST OF ALL OF THE MODELS. IF YOU LOOK AT 120, 100, 100 20,000 WOULD BE AT THE LOW SIDE. THERE IS NOTHING LOW. LOOK, NOTHING LOW. ONE IS TOO MANY PEOPLE. ONE PERSON TO LOSE WITH THIS IS TOO MANY PEOPLE. IT IS A DISGRACE, WHAT HAPPENED. IF YOU LOOK AT THE MODELS, WE ARE AT THE LOWEST OF THOSE PROJECTIONS. THE MODELS ARE NOT MODELS DONE NECESSARILY BY THE WHITE HOUSE. IF YOU TAKE THE MOST RESPECTED OF THOSE PEOPLE — MANY OF THOSE MODELS ARE WRONG. THEY HAVE BEEN WRONG AS FAR AS THE U.S. IS CONCERNED. THE NUMBERS ARE NOWHERE NEAR WHAT WAS PROJECTED, POTENTIALLY.
Make of this what you will. Because some people are still believing the models. “Citing forecasts that COVID-19 deaths could rise to 3,000 per day in June, they say that reopening without better defenses against infections is reckless.” It’s clear they’re not looking at the data, but at models, to get a number like that.
How about this model? Coronavirus Latest: New University Of Penn Model Predicts 350,000 Deaths By End Of June If All States Fully Reopen. Idiots (a professional opinion). South Dakota never locked down and had one of the lowest (if not the lowest) per capita death rates. Yet these model lovers say up to 350,000 deaths by the end of June? Good grief.
Speaking of models, somebody did a code review of Imperial College’s model. They discovered it stinks. We knew this was true before the code review, because that model predicted the coronavirus would be worse than the Spanish Flu! It had something like 50,000 Americans dying a day at the peak. That model should have been tossed out and never listened to.
Better no model at all than an obviously absurd one.
Propaganda works. Large majority of Americans worried more about reopening too fast than too slow. 71%, they say. I find this most depressing. Maybe Ann Coulter was right.
Below we’ll see evidence lockdowns cause harm. Here’s, of all places, NBC agreeing: “A ‘shocking’ two-thirds of patients recently hospitalized in NY had been staying home“. Who could have guessed that clapping up everybody, healthy and sick together would cause the infection to spread?
Answer: just about anybody.
Weekly All Cause Death
Our source, the COVID Tracking Project, reports 75,107 deaths in the US. The CDC reports 49,876. But they also report 107,825 “Deaths with Pneumonia, Influenza, or COVID-19”, which is quite high. The bulk is coming from pneumonia, which the CDC says is 74,756.
I expressed doubt, but this year’s peak did beat the 2017-2018 flu year! Congratulations to those who were cheering for high death counts. We’ll have to wait on the totals to firm up before we see if this whole flu season beats 2018, though. It’s close, but the flu still has the edge.
Anyway, coronavirus has no chance of catching years from the 1960s and before, as we saw before. This was our first panicdemic, though.
You can see it was a small flu count year. Let’s look at the blow up of the end:
The green is from the CDC’s COVID page, and the black the traditional weekly count, all linked above. There are discrepancies. The dashed green is the all-cause minus the COVID reports. We’re accepting the numbers as accurate.
As you can see from previous years, we’re normally on the way down. The coronavirus killed some, but then that dashed line means there are still more deaths than we normally see, caused by something other than coronavirus. The good news is that, even with late-reporting, this is clearly on the way out.
The cumulative difference between the solid and dashed green line is 49,876, the CDC’s reported death total. Obviously, the late-reporting is causing the drop off. We’ll have to wait, as usual, a few weeks for this to catch up (about two to three to get close enough to the total, though the CDC says up to eight to get full total).
It’s just a guess, but it’s looking like there’s an additional 10 thousand deaths or so caused by, perhaps, the response to the virus. Even NPR is worried about that! Eerie Emptiness Of ERs Worries Doctors: Where Are The Heart Attacks And Strokes?
The patient described it as the “worst headache of her life.”
She didn’t go to the hospital though. Instead, the Washington state resident waited almost a week.
When Dr. Abhineet Chowdhary finally saw her, he discovered she had a brain bleed that had gone untreated.
The neurosurgeon did his best, but it was too late.
“As a result, she had multiple other strokes and ended up passing away,” says Chowdhary, director of the Overlake Neuroscience Institute in Bellevue, Wash. “This is something that most of the time we’re able to prevent.”
Chowdhary says the patient, a stroke survivor in her mid-50s, had told him she was frightened of the hospital.
She was afraid of the coronavirus.
This is proof the panic caused that death. The figures suggest it’s many more. In the US “ER volumes are down about 40% to 50%”. So much for overwhelming the system.
We saw the signal much clearer in England, which had a much more aggressive lockdown. Let’s see.
England & Wales per capita all cause deaths (latest numbers, which only go to 24 April, which is about three weeks behind):
The corona peak doesn’t look like the winner here, sadly, though the ONS does adjust the numbers, too, in time. So there’s still a possibility.
Here’s the blow up (we don’t need per capita here):
Same as before: the dashed line is the all cause minus the official COVID figures. The difference is the official reported COVID total so far, which is 22,351.
There are thus many, many more deaths than normally seen in England and Wales at this time of the year. These are, officially anyway, non-COVID deaths. About 20 thousand. (I’ll firm all these up once the data is final final.)
I’d do Europe as a whole, but they’re only showing the past couple of years, which makes coronavirus really pop. They used to go back a number of years, but lately changed the site so you can’t see the old data. This kind of maneuver always makes me suspicious. And they’re emphasizing z-scores, which is ridiculous.
Don’t trust any data source comparing coronavirus unless that source goes at least back in the 1990s. We see above that this is far, far from the worst thing that every happened. Here’s more perspective.
I’d plot up Taiwan (with 2 and half times the population of Sweden), which had 6 reported deaths, and no lockdown, but a plot of 6 is rather boring.
The evidence is still that of the ancients: panic kills.
Once more, these are models of reports and not actual numbers. Nobody knows the actual numbers. No body. As in nobody.
It’s easy to see where increased testing is driving reported cases up. Anyway, new totals: 4.5 million cases, 300 thousand deaths. Last week: “3.9 million cases, 275 thousand deaths.” Our under-forecast is decelerating, meaning we’re likely looking at Swine Fly totals. Median estimate for that is 375 thousand, with up to 525 thousand.
No lockdown, no panic, no social distancing, no mass insanity with Swine Flu.
Reported naive case-fatality rate:
This is the global reported deaths divided by global reported cases. We’ve looked at this every week, and knew we would be past peak when this plot leveled off and went down.
Again, there is no way this bug is killing 6.8% of the people it infects. Therefore we know that (1) cases are under-counted, (2) deaths are over-ascribed, or (3) both. In what proportion , we don’t know for sure.
Assuming death reports are accurate, then we know where are 58 to 96 million actual cases, if the bug in real life kills 0.5% or 0.3%, numbers which seem more likely.
Tired of writing this, but because this is true, they can use increased testing to reveal already present cases, and make the panic last longer, by implying cases are increasing. When it’s just testing revealing already present infections.
Don’t believe it?
Daily reported cases:
There’s the increase in testing right there. Sure looks like it’s getting worse! Watch for this in the media. Look for stories of cases without deaths, or words about “doubling times” or whatever.
Daily reported deaths:
The naive model can be asymmetric on the right side, but it isn’t doing a great job, hence the under-reporting. Still, Swine Flu territory is a good bet. There is no indication whatsoever it’s a bad bet.
USA! USA! USA!
The US is roughly 4% of the world’s population. Yet the US has 32% of the world’s reported cases, and 26% of the reported deaths. In what most consider the country with the best medical system.
Now either those numbers are correct or the way the US is counting is different from most other places. I say the latter.
Vietnam, for instance, 95.5 million souls: 288 reported cases, and 0 reported deaths. Could be government there keeps things quiet. Could be accurate, too. Good weather in Vietnam; anti-bug weather, that is. Whereas NYC, Detroit, Chicago all had lousy weather; no spring at all.
Or Vietnam could be taking it easy, calmly reporting numbers. Whereas we’re probably juicing them, as we saw above.
Anyway, here the reported totals:
New totals: 1.39 million cases, 79 thousand deaths. Last week: “1.23 million reported cases, 67 thousand reported deaths.” Same kind of under-forecasting as the global model—-which is not surprising, given the USA numbers are a third of it!
Reported naive case-fatality rate:
Reported deaths divided by reported cases. Again, this bug is not killing 5.6% of those it infects. That means there are about 15 to 25 million actual cases in the US, if the bug is killing 0.5% to 0.3%.
Again, watch testing increase cases, and watch the media and dictatorial governors to use the coming increases to extend their power grabs.
Again, don’t believe me?
Reported daily cases:
Testing is slowing down, but not as much as deaths.
Reported daily deaths:
Like I said last week, we picked a bad day to do updates, because there’s always a minimum of reports over the weekend, and our data is always from Monday night. It’s still very much worth looking at the wide variability, though.
Given the under-forecasting of these reports (who knows about the actual deaths), it’s looking past 1 June.
To support this site and its wholly independent host using credit card or PayPal (in any amount) click here