Tuesday, April 21, 2020

Covid-19 Deaths, Compared

From The New Atlantis, updated on April 21st by Ari Schulman, Brendan Foht, Samuel Matlack:


William M. Connolley said...

It's not a great comparison. It illustrates one thing - that the Covid is increasing strongly - but fails to compare year-long-total deaths, unsurprisingly since things aren't finished. Perhaps it would be better re-drawn as cumulative deaths-to-date starting in January.

Also, I'm pretty sure they've faked their straight-line heart and cancer deaths, because there's no way they are that constant.

Ned said...

Doing it another way could also be dinged as "not a great comparison" because it failed to do as good a job as this one at conveying the rapidity of increase.

For most subjects there is not one optimal way to do data visualization. How you design a visualization depends on what information you're trying to convey.

If you want to be able to simultaneously illustrate a variety of different points, it's probably better to do that via multiple graphics than trying to figure out the one best visualization method that handles them all perfectly.

Ned said...

Promotion on Fox News of hydrochloroquine as a treatment for Covid-19 has dropped 77% in the past ten days:

Trump and Fox News want to send their hydroxychloroquine hype down the memory hole

No explicit mea culpa yet, but at least they've cut back significantly on this one form of disinformation.

Entropic man said...

Did I see Donald Trump and his medical experts arguing on stage last night?

David in Cal said...

I haven't checked this, but the blogger is generally reliable

The total number of deaths in the U.S. for weeks 1 through 14 this year–the year of the coronavirus–is 772,085.

Now do the same for the first 14 weeks of 2019. According to the CDC, as I read the spread sheet, there were 809,704 deaths in the U.S. over the same time period last year. That’s right: through the first 14 weeks of the year, through April 3 or April 10, however the CDC counts the weeks, there were 37,619 fewer deaths this year than last...

There is no way the overall mortality statistics can make sense unless a great many alleged coronavirus deaths are actually people who would have died at the same time, regardless, from other causes.


This demands explanation (if it's correct).


David Appell said...

I've never trusted a thing powerline has ever written. They block you if you question them, always a sure sign of a frightened denier.

Until you give a link to this "spreadsheet," it's impossible to analyze your comment.

David Appell said...

So David, you're saying it's OK if people die of the coronavirus as long as less people die than last year.

That is completely and utterly disgusting.

David Appell said...

One obvious reason why there might be fewer deaths this year than last year is temperature.

For USA48, the average temperature of Jan & Feb 2019 was 32.2 F
For Jan & Feb 2020 it was 35.8 F.

3.6 F higher. A lot.

No critical thinking skills on display at Powerlineblog.

David Appell said...

David in Cal: Sorry. I just saw your like to Powerline.


David Appell said...

uh.... "your *link* to Powerline*..."

I still think the large temperature differential could easily have something to do with fewer deaths this Jan&Feb than last year....

Layzej said...

Auto accidents are probably down. I know I haven't gassed up in over a month...

David Appell said...

"Experimental studies in guinea pigs demonstrated that influenza virus transmission is strongly modulated by temperature and humidity."

Journal of Virology 2014

David Appell said...

Our results support the idea of an important role of climate on the spread of influenza.

BMC Public Health 2016

David in Cal said...

David - Your 7:51 comment is out of line. I didn't make any recommendations or moral judgments. I am just trying to understand a surprising statistic.


David Appell said...

Yes, but auto accident deaths are only ~40,000/yr, IIRC. And driving in Jan20 and Feb20 was fairly normal....

David Appell said...

Only data so far is for Jan2020 traffic, up 2.1% compared to Jan2019.


Might not be big changes in Feb2020, but definitely will in Mar2020.

Ned said...

This thread is a massive embarrassment, first to David in Cal, and anyone taking his comment seriously.

Hinderaker doesn't know any more about epidemiology than your average dairy cow. The CDC statistics he links to are not complete. Data for the past year are still provisional and 2020 data are still being compiled by CDC; the numbers on the linked page will change as death records are assimilated. He is summing data over January and February, when 90% of all US Covid-19 deaths occurred in April ... and the April statistics in the CDC's all-deaths count are woefully incomplete.

If you want to compare 2020 mortality rates to previous years, you either have to wait a year for the data to be finalized by CDC, or you have to look at local jurisdictions that already have compiled their own mortality counts.

Hinderaker's claim that "a great many alleged coronavirus deaths are actually people who would have died at the same time, regardless, from other causes" -- qpromoted here by David in Cal -- is wildly wrong and it's appalling to see this kind of bullshit being uncritically repeated.

In the areas where most coronavirus deaths have occurred, total deaths are WAY up from previous years, with the difference exceeding the number of reported coronavirus cases.

This means that coronavirus is responsible for more than the reported number of deaths. In part this is due to deaths from coronavirus that are occurring outside hospitals and are not being labeled as CV. In part it's due to deaths from other causes that are occurring because hospitals are overwhelmed and patients with other problems are not receiving care, or are afraid to go to the hospital.

For example:

Deaths in New York City Are More Than Double the Usual Total

For those who haven't signed up for the NY Times free coverage of Covid-19, here's the graph from that article:


The Economist did an analysis of mortality rates for countries with large Covid-19 deaths, and found consistently large rates of excess mortality in addition to the official reported deaths directly attributed to Covid-19:

Tracking covid-19 excess deaths across countries
Official covid-19 death tolls still under-count the true number of fatalities

Here's an example for the UK -- which reports statistics more rapidly than the US CDC -- showing the interannual range of variation in deaths per week, for the past decade, compared to 2020:


Virtually all the mortality caused by Covid-19 -- both directly and indirectly -- in the UK and the USA has occurred in the last 4 weeks. But Hinderaker's calculations, promoted here by David in Cal, swamp that by including 10 weeks of data pre-Covid-19, and using incomplete death numbers for the recent weeks to falsely claim that mortality has decreased.

David in Cal's posts here are shameful. First it was promises in early March that tests would "soon" be in the millions. Then it was aping Trump in enthusiastically promoting hydrochloroquine as treatment for CV. This, however, is a much darker turn, spreading disinformation to falsely suggest that the numbers of deaths from Covid-19 are lower than reported when they are actually much higher than reported.

David in Cal said...

Ned - I didn't mean to fool anyone. Your idea is part of the answer to the question I asked. I thank you for your thoughtful explanation.

Regarding tests, I correctly said the administration was taking a number of steps that would lead to big increase. We're now beginning to see that increase. Yesterday, there were 193,000 tests. Today, there are already 225,000 tests as of noon PDT.


Ned said...

6+ weeks ago (March 10) the Administration claimed that "more than 4 million more tests" would be ready by the end of the week. But by the following Sunday night there had actually been less than 30,000 tests.

On March 19, DHHS assistant secretary Brett Giroir told the nation's governors that "10 million" tests were "available"; but the total number of tests completed + pending at the time was 107,000 or about 1% of what Giroir claimed.

Two days later, on March 21, Giroir said that by March 28 there would be "27.6 million" tests ready for use. In reality, only 800,000 tests were completed or pending by the 28th, or 3% of what the Administration had promised.

White House Promised 27 Million Coronavirus Tests By End of March, But U.S. Just Hit 1 Million

April 19: Forbes reports that "Governors From Both Parties Slam Trump's Claims About Testing: 'Delusional' And 'Not Accurate'"

On April 20th, the total accumulated number of tests finally hit the 4 million mark that VP Pence had claimed -- six weeks earlier -- would be available by March 15.

As a reference point, Trump's own former FDA chief (Scott Gottlieb) has suggested that before the country can re-open its economy, we should be doing about 4 million tests per week or 500,000 to 600,000 per day, on a sustained, ongoing basis. Over the past week, the average has been 178,000 per day.

We are forced to strangle the economy with "social distancing" because the Trump Administration has been so inexplicably lethargic and unmotivated about testing. We are finally now starting to have numbers of tests that were needed - and promised! - six weeks ago.

With no way to tell who is contagious, everyone is forced to isolate themselves.