Skip to Main Content

A new analysis of the toll of the Covid-19 pandemic suggests 6.9 million people worldwide have died from the disease, more than twice as many people as has been officially reported.

In the United States, the analysis estimates, 905,000 people have died of Covid since the start of the pandemic. That is about 61% higher than the current death estimate from the Centers for Disease Control and Prevention, 561,594. The new figure also surpasses the estimated number of U.S. deaths in the 1918 flu pandemic, which was estimated to have killed approximately 675,000 Americans.


The analysis was conducted by scientists at the University of Washington’s Institute for Health Metrics and Evaluation.

“We’re probably not yet at the global toll of Spanish flu and certainly not at the death rate from Spanish flu. But given what’s unfolding in India right now, given our expectation of continued deaths, Covid is going to rival Spanish flu at the global level in terms of the count, likely, before we see the end of this epidemic,” the institute’s director, Christopher Murray, told reporters in a briefing.

Most countries have underreported deaths, Murray said. In many cases, the under-estimates seem to be a result of health systems being overwhelmed, and of insufficient testing.



But in several cases it appears something else would have to account for the scale of the differences between the deaths countries are reporting and the excess mortality they have experienced during the pandemic, he said, pointing to countries like the Russian Federation and Egypt, where the estimated deaths, 170,000, are 13 times higher than the country’s official death toll, 13,529.

Murray’s group estimates that by September, the global death toll from Covid-19 will reach 9.4 million, with India at that point surpassing the United States as the country with the highest toll. By September, the group’s models suggest that India’s Covid death toll will be 1.4 million people, and the U.S. toll will be 949,000.

The estimates are of deaths directly related to Covid, and do not include deaths that resulted from the pandemic’s disruption of health care — for example, people who did not seek care for heart attacks because they were afraid to go to Covid-swamped hospitals.

Correction: An earlier version of this story misstated the percent change between the current death estimate from the CDC and the estimate in the new analysis.

  • This piece is full of Rubbish – lacks proof / references – and contains errors within its own presentation.
    Please refrain from writing (and reading) this junk.

  • Truthfully, as a layman, I have no idea how many people died of Covid -and apparently, experts do not agree.
    But if actual direct deaths from Covid were 60% higher than CDC saysm that means, out of 16 people who died from Covid, the doctors got the cause of death right 10 times, and wrong the other 6.
    It is hard to imagine how they could screw that badly but if it did happen, it would seem to me the treatment of those 6 must have been inadequate – I mean, presumably they were not being treated for Covid, as their doctors never realized they had it – and it follows from that – they did not get antivirals, or antibody treatments = why would you give those to people who did not have Covid? Dexamethasone ? Tocilizumab? Maybe, but I would think not, as those were suggested as Covid treatments, though they may be generic anti inflammatories.

    Most likely they would only get generic treatment, if any, such as oxygen for example. Very inadequate if the antibody treatments, which seemed to work miracles for some patients, were available.
    Before the epidemic was recognized, it is easy to imagine many old folks were recorded as dying of flu or bacterial infection, but reading about this issue around a year back, it did not sound like that would account for 60% over the official numbers.

  • It would have been useful if this report had contained more information on the methodology used by the University of Washington. This lack of information together with the Liberal use of the word “estimated” tends to devalue the report in my opinion. The figures in the table for the UK seem to have been plucked out of thin air. Nowhere can I find a reported death toll anywhere close to 150,000. In any event, the UK Office for National Statistics are continually updating the excess death figures and these simply do not lie!

    • Then I don’t think you’ve been reading the ONS figures very carefully. Their latest count of deaths registered with Covid comes to 152,491. Up till this year, the proportion where Covid is marked as the main cause of death has been over 90%, though recently this has decreased and the ONS count is not up to date.

  • Who paid for this study and what ties does the U of Washington have to big pharma? Pure attention grabbing/click bait. Why aren’t there links to references to backup these claims?

    • Correction: One link to their own data. Still doesn’t provide where the stats, (if they are stats) came from. KarlPK is correct in sighting the quantity and use of various terms like “Estimate”.

  • Helen, Helen, Helen. Did you actually get your arms around the methodologies in this study? The phrase “building sand castles” doesn’t even begin to describe what these IHME guys are doing. Here’s a great assertion.

    “The main potential increases in excess mortality due to deferred care and increases in drug overdose and depression are hard to quantify at this point or are of a much smaller magnitude. Given that there is insufficient evidence to estimate these contributions to excess mortality, for now we assume that total COVID-19 deaths equal excess mortality.”

    Say what? “Assume??”

    Here’s another “assume.”

    “In addition, we also include in the ensemble a Poisson model with fixed effects on week and year, as well as a model that assumes that the expected mortality rate for 2020 and 2021 is the same as the weekly mortality rate observed in 2019.”

    The sand is starting to get very VERY wet. You’re using ONE year as your benchmark? ONE? Not a multi-year, population adjusted metric? Really?

    And then there’s this.

    “In addition, in many high-income countries, deaths from COVID-19 in older individuals, especially in long-term care facilities, went unrecorded in the first few months of the pandemic.”

    Really? How do you know that? Did you do another “assume” and not even tell us?
    You know, it’s always fun to do a key word search in these sorts of analysis. The piece linked has approximately 2500 words, including figure text. Ready?

    Estimate – appears 26 times.
    Estimating – 6 times
    Model — 21 times
    May and May be – 17 times (excludes the May publication date at the top)
    Likely – 8 times
    Suggest – 6 times

    Behind all the fancy statistics discussion, what we have here the classic problem with epidemiology and why it is NOT a science. It’s weakly predictive and post hoc model building, with all the inherent flaws thereof. There is no falsifiable hypothesis testing. That data is very noisy, if not downright unreliable.

    But the one thing these guys at IHME ARE good at is generating publicity for themselves. I expect this story to be trumpeted far and wide by journos everywhere, because it’s just the sort of thing that will get people all in a tizzy.

  • The US government data on “death from all causes” in 2020 is about 400K higher than it was in 2019, which is constent with the Covid statistics that were widely reported at that time, and not at all consistent with this study. Also, IHME Covid projections have generally been terrible since the start of the pandemic, so take these new numbers with a grain of salt.

Comments are closed.