# Revisiting my COVID-19 death estimate

In the comments to “Why do we care about COVID-19 deaths more than driving-related deaths?” (March 26, 2020), a reader asked a great question:

How good of an estimate today can you make of traffic-related deaths in the US in the next year? This estimate is quite uncertain given that miles driven will likely plummet and depend on the duration of various shutdowns. You can still probably guesstimate the total miles driven will be some fraction of the previous year and be within 0.25X to 4X. You won’t be wrong by 100X.

Now predict today the number of coronavirus deaths in the next year in the US. Your estimate could very easily be off by 100X or more.

This was just a few days after New York went into shutdown (March 22) and before the typical U.S. hospital had seen even one Covid-19 patient. Here was my prediction:

why can’t we extrapolate from Lombardy to get a reasonable estimate? Out of 10 million people, COVID-19 has killed 5,000 to date. Assume that becomes 10,000 by the end of the year. That’s 1 in 1,000 people. Applying that to the U.S. we get 330,000. Horrific, of course, but about the same as the driving deaths expected for next 10 years (which didn’t seriously concern the nation). I don’t think this can be off by 100X. 33 million is too much. 3300 is, sadly, likely to be exceeded (1,301 as of right now). Maybe we can do 4X better than Italy due to advances in knowledge and drug therapies that are available. That brings us down to 82,500 deaths, not too different from what the Imperial College folks are predicting per capita for the UK. We’re not as competent as other countries when it comes to health care, so estimate 150,000 COVID-caused deaths through February 2021? Let’s come back to this post on March 1, 2021 and see if the mortality was, in fact, within 0.25X to 4X of 150,000.

Our heroic anonymous reader gave me some wide error bars (0.25-4X), there, but not nearly as wide as what the IHME prophets gave themselves regarding Sweden (“They’re fairly confident that on May 23, Sweden will have between 11 deaths and… 2,789 deaths”).

So… my proposed method back in March was to extrapolate from Lombardy to the entire U.S., reduce for the worldwide effort to develop treatments and add back in for American incompetence at organizing health care. This boiled down to 45 percent of whatever the Lombardy death rate was. And then there was an additional guess that, as of March 26, Lombardy was halfway through its total COVID-19 deaths.

So… let’s put two questions to the readers.

How did my guess that Lombardy was halfway through its total COVID-19 suffering hold up? As of February 27, the Google said that 28,275 people in Lombardy had died with a COVID-19 tag (Italy had a big second wave of deaths tagged to COVID-19 starting in the fall of 2020). I’m having a bit of trouble finding the death rate through December 31, 2020 (maybe a reader can help out). My guess that 10,000 would be the death toll in Lombardy “by the end of the year [2020]” seems to have been a little over 0.5X of the actual.

How did my guess that the U.S. toll would be “45 percent of Lombardy” hold up? As of February 27, 2020, the Google said that 510,000 Americans had suffered “COVID-19 death” (keep in mind that, with a median age of 80-82 for “COVID-19 death”, we are saying that a 92-year-old with cancer, diabetes, and COPD “died of COVID-19” so long as a positive PCR test can be obtained).

Through February 27, 2020, Lombardy has had a COVID-tagged death rate of 0.28 percent (28,275 divided by 10 million). The U.S. has had a death rate of 0.155 percent (510,000 divided by 330 million, but perhaps the divisor should be 350 million?).

The guess was 45 percent and, as of February 27, the statistic was 55 percent (0.155 divided by 0.28).

What about the 150,000 number that I tossed out? That is 0.29X of the 510,000 number that we’re being fed. So, unlike our heroes at IMHE and other epidemiology institutions, the reality was within the error bars that I set up.

(One reason that my estimate came out on the low side, I think, is that I underestimated the extent to which Americans would want to wallow in coronasadness and maximize the count of very old, very frail people who purportedly died “of COVID-19”. This can be seen on Facebook as people claim that the impact of coronavirus has been worse than all of the wars that the U.S. has ever fought, except maybe the War of Northern Aggression. At least some subset of Americans wants to equate a healthy 18-year-old marching off and never returning with an 82-year-old who was expected to die within 1-2 years meeting his/her/zir/their final end within a few weeks of a positive PCR test. Admittedly this method of counting is not unique to the U.S. For example, the Swedes have a computer system automatically tag “COVID-19” to anyone who dies within 30 days of a positive test, even if the person dies in a traffic accident.)

From my Italy photos, a square in Burano, 1996:

#### 8 thoughts on “Revisiting my COVID-19 death estimate”

1. Mememe says:

There is a five hundred year-old man out there who just keeps beating the odds of dying, day after day, year after year.

2. Mememe says:

The dinosaurs did not go extinct because of a single cataclysm. They all, individually, happened to get really unlucky, due to unrelated circumstances, on the exact same day.

3. Michael says:

Just in case people take the “people are dying with COVID-19 but not of COVID-19” seriously, here’s a page with some data from the UK:
https://www.covidfaq.co/Claim-People-are-dying-with-Covid-but-not-of-Covid-e9d5af56419a438bb626bb08271b3b69

Yes, with older rather than younger people dying, often the QALYs (quality-adjusted life year, https://en.wikipedia.org/wiki/Quality-adjusted_life_year ) loss is lower. I don’t think there’s been an accurate assessment of the effects of long-COVID to adjust the quality of future years for those youngsters yet.

• philg says:

Thanks for that link. “A study by academics at the University of Glasgow suggested people who had died of Covid typically had over a decade to live, based on their age and prior conditions.” Here in Maskachusetts, the median age of a “Covid-19 death” was 82 with more than 98 percent having preexisting conditions. Do we really believe that a typical sick 82-year-old could expect to live to 92?

• Jarle says:

@philg, You need only to consult an actuarial life table. The latest SSA version for 2017 (whole US, not just MA) states that on average at 82, males are still expected to live to 89 and females to 91: https://www.ssa.gov/oact/STATS/table4c6.html

• philg says:

Jarle: looking at that table for all older adults would make sense to me if COVID-19 killed older adults at random. Influenza is much better studied. Death among the elderly from flu is not random and those who die did not have average life expectancy for their age. https://www.medscape.com/viewarticle/919668 is a 2019 article:

“The risk for influenza-related mortality is five times higher among people with heart disease, 12 times higher among those with chronic lung disease, and 20 times higher among those with both heart and lung disease.”

We’ll have a better idea if COVID-19 knocked over those who were likely to die soon when we have data regarding deaths for 2020 through 2022. We can compare those to 2017-2019, for example, and adjust for population growth and see if the “excess deaths” number is as high as we’ve been told. (This is complicated by the fact that 2020-2022 will have a lot of extra deaths caused by shutdown, e.g., from medical care being suspended, from alcohol and opioid consumption, from increased obesity due to quarantine-next-to-fridge and gym closure, from depression due to unemployment, from additional murders perpetrated by those who have more time on their hands, etc. Most of the deaths from shutdown, of course, won’t show up for decades, e.g., shorter lives for today’s children who have missed a year of education.)

• George A. says:

And this just in “Covid-19 death rates 10 times higher in countries where most adults are overweight, report finds” [1]

“But in countries where more than 50% of the population was overweight, the Covid-19 death rate was much higher — more than 100 per 100,000.”

4. George A. says:

We know seatbelts saves lives, they are effective for all ages, but they are *most* effective for infants when used with car seats. In effect infants are the most vulnerable group to car crashes without car seats. So if you remove car seats, death rate and injuries from car crashes will be much higher. The same principle applies to coronapanic. Those death rate and infection numbers are coming off people with age and/or with pre-existing condition. I am all in to make sure that every-single-one-of-us is safe and live to our full life expectancy. But putting the whole community/city/country/world into a lock down and panic mode is like saying infants or not, *everyone* must be in a car seat because we must save every-single-life from car accidents.