COVID Mortality Idiocies: Irrelevant Comparisons, Bad Context, and Under-estimating Deaths
by meep
Some people have been asking me about some pretty stupid COVID “memes” going around, one so stupid that Facebook actually broke out fact-checking on it. So stupid that SLATE of all places debunked it.
This is really not the season for this sort of thing, and I’m going to try to keep my ranting to a minimum. One of the following got me so angry at how stupid it was, I started doing actuarial nit-picking as only other actuaries would believe. I will spare you that experience (for now).
WaPo and its White Christmas
The tweet that launched it all:
Every time you listen to Bing Crosby’s “White Christmas,” about five people have died from the coronavirus between the beginning and the end of the song. https://t.co/Rpuc9mDsuz
— The Washington Post (@washingtonpost)
First: I hate the “Every x minutes, y happens” sort of thing, especially as it indicates a steady process… which most of these things aren’t. COVID-19 deaths definitely aren’t a constant thing – they wax and wane, which is a driver of various policy insanities.
Second: I hate using such a cheap hook as referencing “White Christmas”. I don’t particularly like the song, but there are loads of songs that last only 2 and a half minutes.
Here — my favorite song that will get you to just over one COVID death (well, 1.42 deaths, evidently), by this reckoning: Minimum Wage by They Might Be Giants.
Other people had thoughts:
Using this math, "Last Christmas" by Wham! kills about 7 people. Please limit your celebrations to songs that are "Rudolph the Red-Nosed Reindeer" length or shorter. Be safe out there, folks.#Karin4Congress
— Dr. Congresswoman Karin Baker-Thompkins (@ResitsTrump)
Gives new meaning to Whamaggedon.
Think about how many people "Imagine" has killed
— Robert Kroese (@robkroese) December 19, 2020
Yeah, I hate that song, too.
Studies show "White Christmas" is listened to 2.7 million times a day in December. At five people dead per playing, over the span of 18 days so far, that would mean 243 million people have died from the coronavirus.
— Jeff Dobbs (@jeffdobbs) December 19, 2020
This has not happened.
WE RATE THIS CLAIM: Four Pinocchios
This shows one of the biggest flaws in this particular construction.
Well stop playing it!
— Franklin Harris (@FranklinH3000) December 19, 2020
Man, that’s an old joke.
As is WaPo.
Deadliest day in American history?
Here we go: Deadliest days in U.S. history? [posted December 7]
Okay, this is the piece facebook links to for fact-checking, but let me link to Slate as well.
Slate hits the first point:
For one thing, a list of the “deadliest days” in American history would include days with the most deaths, not the most deaths from one discrete event.
And, if we really want to be specific, I would say the “deadliest day” would be when the highest percentage of the U.S. population died.
As noted in the Slate piece, about 675,000 Americans died directly from the Spanish flu pandemic, which is a much bigger impact given that the U.S. population of 1918 was less than a third of what it is now in 2020.
The Slate piece still annoys me because it compares against war deaths, when they really should compare against, oh, the cholera outbreaks of the 19th century, which killed off 5 to 10% of the total population in Cincinnati, St. Louis, and Detroit in 1849-1851. Then there was the yellow fever outbreak of Philadelphia of 1793 which killed 10% of its population.
Those would be appropriate comparisons, even if the transmission wasn’t directly human-to-human.
I said I would keep the ranting to a minimum, so I won’t keep picking for nits, but there are lots in this one meme.
I will link to this prior post to get at one of the other things that bugs me about this meme, and just drop it.
Oh, and in January 2018, 286,744 died in the U.S., which works out to 9,250 people per day [note: January had the highest such average for the year, which is why I picked it — deaths are usually highest in January due to flu/pneumonia]. Just so you can compare against recent stats. If you want to check me, I got it straight from CDC WONDER. Hmm, I need to do more demo videos.
A respite: something to make you smile
I will get back to idiocy in a moment.
But for now, enjoy: Tables Turned: Restaurants Organize To Ban Gov. Cuomo From ‘all New York City establishments’
New York governor Andrew “Killer” Cuomo’s ridiculous lockdown orders have angered New York City’s restaurateurs to the point that they are organizing to ban him from all NYC establishments for life.
….
‘According to bar owner Larry Baird “ He can eat at some shitty roadside diner outside of Albany but he will not be served anywhere in New York City, known universally as the world’s greatest dining destination! If he has to use the restroom he can go pee on my street-corner! That’s what he wants anyway!”’
….
If you’re wondering why a restaurateur would suggest the governor “pee on my street-corner,” this is a response to one of the most bizarre edicts thus far. Cuomo decreed that while socially-distanced dining outdoors in the snow is currently permissible, patrons using the establishment’s restrooms is not.
Mind you, the people quoted aren’t owners of tony Manhattan restaurants, but Coney Island establishments. Still, it’s a good gag.
By the way, Cuomo backed down on the “pee on the street corner” edict.
Oh, and Cuomo evidently thinks Saturday Night Live is essential in a way that restaurants are not.
Undercounting deaths
So we had an over-emphasizing deaths idiocy. Let’s look at the other side:
Despite the virus, US deaths this year are literally no worse than they've been for the last 10, and may even be on track to go DOWN.
— Jon Miller (@MillerStream) December 13, 2020
2010: 2.5M
2011: 2.5M
2012: 2.5M
2013 :2.6M
2014: 2.6M
2015: 2.7M
2016: 2.7M
2017: 2.8M
2018: 2.8M
2019: 2.9M
2020: 2.5M (as of Nov.)
That tweet has 27.3K retweets.
And this is just plain wrong… he claims it’s from the CDC, and the tweet is from December 13.
I know what mistake he made….because he linked all his sources (and because I know what the CDC has been doing all this time and I’ve been really annoyed as it both underemphasizes and overemphasizes the mortality).
Note: the 2010-2019 numbers are fine. The 2020 one is wrong.
I will show you the 2020 link he provided – it’s cached so we have it as of a specific date/time.
Here is the relevant table on that page:
Here is the problem: the table covers deaths from the week ending February 1, 2020 to week ending November 14, 2020. (The other issue is that as of the reporting date, November 27, 2020, the first couple weeks of November has undercounted deaths). But let us pretend that the death count is complete for that period — that excludes January 1 – 25, 2020 and November 15 – December 31. That’s 72 days. Almost 20% of the year — and we don’t have those deaths.
I will be updating my projections soon, but it is clear there will be more than 3 million total deaths in the U.S. for 2020. Dale Hall of the Society of Actuaries wrote about it here and you can look at our earlier conversation.
I want to commend this person for providing all his data sources for the numbers — fabulous! More tweeters should do so!
But, dude, you missed some super-important details. The most important detail being that 2.5 million deaths is for about only 80% of the year. Not 11/12ths (~92%).
CDC: Good data, poor presentation
I somewhat understand why the CDC has done what it did, as it usually does not report total deaths in real-time. COVID did not officially appear in death statistics until after the end of January, which is why they started from that date.
But seriously, they should have the numbers going back to the beginning of the year and provide an estimate of the percentage completion of death reporting so that people understand the uncertainty (such as February deaths are likely 99.9% complete, but more recent weeks are only 50% complete).
There have been people who exaggerate the COVID mortality impact by focusing only on when deaths are reported as opposed to when they actually occurred, and who look at COVID death counts in isolation from total death counts. Note: the mortality impact is really bad as it is without exaggerating it.
There are others who under-estimate the mortality impact by looking at incomplete data and treating it as if it were complete. Or assume COVID deaths are over-counted, and also ignoring the total number of deaths in the population. Note: no, this is not a Spanish flu pandemic situation, but it is worse than all other short-term mortality increases we’ve had since 1918.
One can’t expect the CDC to directly communicate on everything, but dammit, this is not my job, and it IS the job of some CDC Communications person.
I would have liked to be discussing COVID public policy, such as vaccine prioritization, or about pandemic modeling, but I keep finding the most basic fact is under dispute: the total number of people who died (from anything). It is tough to get into those fine details when people cannot agree on death counts.
If you told me a year ago that I would spend this year ranting about people who cannot appropriately count dead people, I would think you were nuts. Well, now I know.
Related Posts
Vaccines Reduce Risk: A Look at the Changing Age-Related Mortality Risk of COVID
Top Causes of Death by Age Group, 2021: Finalized U.S. Stats
Deaths in the Dominican Republic: Bad Booze?