Contribute Try STAT+ Today

When STAT first compared projected U.S. deaths from Covid-19, in early April, there seemed to be a glimmer of good news: A prominent model had just lowered its estimate for total deaths through Aug. 4 from about 100,000 to 60,000, reflecting the apparent success of three weeks of social distancing across much of the country.

On Wednesday, April 29, the country blew past 60,000, more than three months before the Institute for Health Metrics and Evaluation projected. (The 60,000 almost certainly undercounts Covid-19 deaths, by about 9,000.) IHME, whose model has been criticized by many epidemiologists, now says the most likely death toll on Aug. 4 will be 72,433, though it could be as low as about 60,000 (obviously impossible) and as high as 115,000.

20.04.29 Epidemic Comparison Area
Hyacinth Empinado/STAT

As some models stumble, and many no longer even try to project more than a few weeks, the Centers for Disease Control and Prevention has added models to those it highlights on its website. One stands out for the fact that it is not associated with a big-name institution. Produced by independent data scientist Youyang Gu, it has been tracking actual U.S. cases and deaths better than many.


Gu’s model projects that U.S. Covid-19 deaths by early August could be as low as 88,217 and as high as 293,381, with a most likely toll of 150,760.

As with all the numbers swirling around the pandemic — more than 3.2 million cases and 230,000 deaths worldwide — these are challenging to grasp. Gu’s lower number is roughly the capacity of the Florida Gators’ stadium in Gainesville; imagine a homecoming game crowd all dying. It is the number of passengers in 360 full long-haul 767s. It is nearly double the number of U.S. combat deaths in the Vietnam War. And 293,381, of course, is nearly triple any of the above.


With U.S. deaths from Covid-19 passing 61,000 this week, the question is as difficult as ever: Are these large numbers or small numbers?

At the beginning of the pandemic’s spread in the U.S., President Trump dismissed early projections of thousands, even tens of thousands, of U.S. deaths as no  worse than the lives lost in an average influenza year. So far this season, flu deaths total 24,000 to 62,000 (data from the Centers for Disease Control and Prevention are estimates with wide uncertainties). Whether you think a tally as much as 12 times higher is a tragedy or merely unfortunate is a personal call.

For comparison, about 8,000 Americans die every day from, almost overwhelmingly, diseases and other natural causes. Those, of course, haven’t stopped; Covid-19 deaths are in addition to those (with a caveat noted below).

NEW 20.04.29 Disease Graph Still (1)

To make these numbers easier to understand, we show how 88,000 to 293,000 compares to some leading causes of death and to previous pandemics.

After our last story, many readers complained that our numbers of annual deaths from other causes were wrong, so one note about the methodology we used then and now: The models project that Covid-19 cases and deaths will decline in the next few weeks, as they have already done in Seattle, San Francisco, and other places. If stay-at-home orders and other countermeasures keep working, there should be few deaths after July. We therefore treated the total projected deaths as occurring over five months, from March to the end of July, and therefore calculated five months worth of cancer, heart disease, and other deaths.

Support STAT: If you value our coronavirus coverage, please consider making a one-time contribution to support our journalism.

Because Covid-19 deaths are “truncated into weeks,” as Anthony Fauci, a member of the White House Covid-19 task force and director of the National Institute of Allergy and Infectious Diseases, has said, they both capture our attention and overwhelm hospitals.

Covid-19 is particularly severe in — and more likely to kill — the elderly and people with existing illnesses, including heart disease. Some people taken by Covid-19 would likely have died from these diseases even in the five-month time frame. We do not try to calculate how many of the Covid-19 deaths “substitute” for other deaths; that is an important calculation that researchers will be eager to do once the crisis passes.

You may believe a different methodology paints a truer picture of how Covid-19 deaths compare to others. As we said, how to think about deaths is deeply personal.

  • Like many, I am struggling to put this new virus in context. This article helps, but I’d much rather see a comparison of deaths on an annual basis (not the five month peak of the virus impact). Also concerned about the issue of death from the virus vs death with the virus – and other mis-counting issues that have been reported. Would it be valid to compare the 2020 total deaths (at the end of 2020 of course) – to the average/expected annual deaths – to see if a spike occurred in 2020? This might give a more “real” estimate of the mortality impact of the virus. The actual death counts (from all causes) are a much more reliable data point it seems.

  • Great article. Nice to see accurate numbers and information. Some people will believe what they want to believe. Maybe it helps their little psychies cope.

    • Thanks Willie – very interesting data here… been looking for a comparison of actual to expected deaths. Really puts this in perspective!

    • The link you’ve provided for the 37,000 figure is the total for the week ending 4/25. It does not include any Covid 19 deaths from the last 7 days.

  • Cannot understand why Aus – has only 93 cases – approx. 26 Million Population – so many travellers and cruise boats. In NSW can exercise outside, go to hair dressers, supermarket, DYI – furniture shops open, $2 shops – takeaway food and coffee at cafes and restaurants. We are the Lucky Country … hope it will last through winter.

  • p.s. The CDC site states the following:

    “This page will be updated daily at noon. Numbers close out at 4 p.m. the day before reporting”. That means that their numbers are actually for the previous day

  • US is going 2X/3Wks, 6wks(?) after NYC began lockdown when it was 2X/2Days. Compared with France and Spain (see charts links below), their lockdown brought the exponential growth down much better. Now with lockdown lifted in some states, US exponential growth can only go steeper. Your estimate 50k additional deaths by June 1 is a little on the conservative side. Would be useful to have a investigative report of COMPARATIVE STUDY OF LOCKDOWNS BETWEEN DIFFERENT US AND FRANCE/SPAIN to see what leaks the virus may be exploiting in the US.

  • Blew past 60,000 deaths when CDC.Gov has it listed at 37,000 to date?? Explain that one to me? Where is the truth in reporting?

    • Honestly. I’m sick and tired of these reporters running around with numbers like this. 60,000 deaths include anyone with the flu, pneumonia AND “possible” cases. News reports like this need to stop. The death toll is at 37,000 and that’s it.

    • According to CDC it is indeed 65000+. You guys can believe what you want but don’t blame reporters for it.

  • Coronavirus vs Common Flu

    As of 5/2/20

    So Far. Common
    Coronavirus. Flu
    Annual Deaths 244k 300-650k
    Mortality Rate. 7% * 0.1% US

    * Ignores (1) people who died but were not tested for coronavirus and (2) people asymptomatic but never tested

    The WHO projected in mid Feb an ultimate mortality rate for the virus of 3.4%. The mortality rate for SARS was 10% and for MERS 34%. Iceland has the highest testing rate of any country by far, 12.7%, and it’s mortality rate is only 0.56%


    1,798 cases. 12.7% of the pop
    10 deaths
    Mortality Rate 0.56%

    Total Worldwide
    3,456,223 cases. 243,024 deaths
    Mortality Rate 7%

    So far, 46,000 of Iceland’s 360,000 citizens or 12.7% have been tested, a higher percent than any other country. Admittedly this was not a random study but tested individuals were self selected so there is some testing bias.

Comments are closed.