If you didn't see yesterday's White House briefing on the coronavirus -- I didn't until this morning -- the first 10-15 minutes are worth a look. It was like there was a completely different president. Trump looked stunned, like he'd seen a ghost. Grim, no swagger, no arrogance, presidential. In a way I almost felt sorry for him.
After Trump, Dr. Blix shows the projections that must have hit Trump hard.
PS: With the recent changes to Blogger, now videos look stupid too.
11 comments:
1000+ new deaths today (USA) per Worldometers. Two days later than I had expected, this time last week. So the milestone is bad, but the slight delay is good.
A couple of days ago, I linked to these projections (which accompany a preprint research article). They seem to be automatically updating the projections daily.
The forecast for the US is now slightly worse - 94,000 deaths, with a peak in hospital usage on April 16, and shortages of 85,000 hospital beds and 19,000 ICU beds.
They also have forecasts for each individual state. E.g., Oregon is supposed to peak quite late (May 6), and to have no shortfall of hospital/ICU beds. The state's death toll is forecast to reach 550 by late summer.
http://covid19.healthdata.org/projections
Again, note the caveat - the error bars around every date and estimate are quite large. So ... it's not clear how much this is really helping.
The figures at Worldometers are discouraging. 26,398 new cases today (as of 5:24 PDT), and that's not a final number. Every day brings a greater number of new cases.
The new Abbott test is supposed to make possible a far greater number of tests, but there's no sign so far that it's helping. Daily number of tests has been stuck at just over 100,000 for several days.
days.https://www.abbott.com/corpnewsroom/product-and-innovation/detect-covid-19-in-as-little-as-5-minutes.html
Thank you for the briefing. That was much more informative than anything I've seen from the US government so far and pulled very few punches. It even mentioned the 2.1 million deaths worst case that has been circulating.
I hope it was not symbolic that one of the two flags dropped to half mast 20 minutes into the briefing.
BTW today figures showed that 26% of the tests turned out positive. The figure had been running around 17% to 18%. I wonder why it jumped like that.
Cheers
Good question, David. Anyone know? I haven't followed testing stats much.
Yeah, that's odd. It had been even lower - ca. 15%. As I've said before, you really want the % positive to be way below 10% or it means you're not testing enough. But the fact that it increased when the number of tests went from 50k/day to 100k/day is very strange.
I wonder if it's possible that some of the independent labs that are doing testing now are not reporting all the negatives, mostly just the positives? That would explain why the total # of tests stopped increasing (not all are being reported) and why the % positive went up. If true, that would be slightly reassuring, though it would make it hard to interpret what's happening.
That is pure speculation on my part ... been too busy with work to follow this closely for the past few days ...
Aha, yes, this does seem to be happening:
Some States Are Reporting Incomplete COVID-19 Results, Blurring The Full Picture
"Several states are reporting only positive COVID-19 test results from private labs, a practice that paints a misleading picture of how fast the disease is spreading.
Maryland, Ohio and others are posting the numbers of new positive tests and deaths, for instance, but don’t report the negative results, which would help show how many people were tested overall....."
Also, there are signs the backlog of "pending" tests is ballooning, with private labs unable to handle the demand:
Private Labs Are Fueling a New Coronavirus Testing Crisis
Within the clinical-testing world, it is an open secret that Quest Diagnostics—one of the industry's two big players, along with Labcorp—has struggled to scale up its operations in California. And yet, Quest has continued to accept specimens from across the country, leading to a huge backlog of tests at the company's facility in San Juan Capistrano.
This failure accounts for at least some of the tens of thousands of pending tests reflected in the state's reported numbers....
Hopefully all these wrinkles will get sorted out.
Bottom line - right now, the "percent positive" data for the US may not be reliable, because some labs aren't reporting negative tests, only positive. This also means the total testing counts are not reliable.
Is there any way to estimate the ultimate US death rate from the projections, i.e.
total_deaths/total_confirmed_cases
even if no one really knows the number of true cases? It'd be interesting to compare to death rates at the present. But I haven't seen projections for the total number of confirmed cases in the US.
Death rates so far (using WHO data):
World 5.1%
China 4.0%
US 2.4%
Canada 1.1%
Italy 11.9%
My bio-statistician wife tells me that normal requirements for a reportable disease would require only the positives. Negative test results are not normally required.
In yesterday's White House presser, Pence reminded labs to report test results. He may have been referring to those labs that don't report negative tests.
Post a Comment