Covid-19 Fatalities, Excess Fatalities, Forecasts

From CDC, Atlantic/Covid Tracking Project, and IHME:

Figure 1: Weekly fatalities due to Covid-19 as reported to CDC for weeks ending on indicated dates (black), excess fatalities calculated as actual minus expected (teal), fatalities as tabulated by The Covid Tracking Project/Atlantic (dark red), IHME forecast (light red), all on log scale. Source: CDC 7/29/2020 vintage, Covid Tracking Project/Atlantic accessed 7/29/2020, IHME forecast of 7/22/2020, and author’s calculations.

Two observations: (1) the unofficial count is rising; (2) recent weeks’ (about a month’s worth) CDC data are subject to severe undercounting, so inferring recent trends on the basis of CDC data is not advisable.

From week ending February 29th through the week ending June 27th, the cumulative CDC fatality tally is 127.3 thousand. Cumulative excess fatalities is 164.8 thousand, implying an additional 37.5 thousand Covid-19 fatalities above the official tally over this period.

23 thoughts on “Covid-19 Fatalities, Excess Fatalities, Forecasts”

  1. pgl

    “Two observations: (1) the unofficial count is rising; (2) recent weeks’ (about a month’s worth) CDC data are subject to severe undercounting, so inferring recent trends on the basis of CDC data is not advisable.”

    Of course you are referring to a flow concept here as in the increase in total deaths during the week. As opposed to this:

    “From week ending February 29th through the week ending June 27th, the cumulative CDC fatality tally is 127.3 thousand. Cumulative excess fatalities is 164.8 thousand, implying an additional 37.5 thousand Covid-19 fatalities above the official tally over this period.”

    Most of us get the distinction but it seems it took Sammy a while to figure it out. Of course he is not blatantly lying to us by switching between the two concepts. That would be Bruce Hall.

  2. pgl

    Bruce Hall at first told us that the daily death count would not go up but he had to change this false claim that any increase was a “slight uptick”. Huh – FEMA has a memo that says otherwise:

    https://abc11.com/covid-cases-deaths-coronavirus-masks/6340085/

    ‘An internal FEMA memo obtained by ABC News reports that new cases are going down but deaths are surging. In the seven days ending Monday, new cases nationwide have decreased 0.6% from the previous week but that time period also saw a 30.1% increase in deaths from the disease and deaths have been increasing steadily in recent days.’

    Let’s see if we can find the actual memo. Of course expect Bruce Hall to go off screaming “fake news”.

    1. Bruce Hall

      pgl, still putting words into my comments that were never written.

      Menzie, I presume you are taking the average of the low/high estimates rather than looking at the lower/upper range estimates; e.g., Arizona’s lower/upper variance is about 100/week. And that’s okay since these are estimates and we know that

      https://data.cdc.gov/api/views/xkkf-xrst/rows.csv?accessType=DOWNLOAD&bom=true&format=true%20target=

      Just wondering if all the “excess” should be attributed to Covid-19 rather than some portion being attributed to deaths of people who failed to seek timely treatment for their serious illnesses for fear of becoming infected with Covid-19. Perhaps that has been incorporated into your calculations.
      • https://globalnews.ca/news/6828509/coronavirus-other-illnesses-deaths/
      • https://www.freep.com/story/news/local/michigan/2020/05/05/er-visits-plummet-amid-pandemic-we-know-more-people-dying-home/3067993001/
      • https://www.nejm.org/doi/full/10.1056/NEJMms2009984

      And yes, the CDC’s official numbers are understated for about six weeks or so as death certificates are categorized as noted in my chart using CDC numbers.

      From the chart, it would appear that most of the “excess deaths” that are being attributed to Covid-19 were in April (I’m having a bit of trouble relating specific calendar dates to the months from the start of tracking) and that there has been a “slight” uptick (for pgl’s benefit) when using this methodology.

      This is sort of like estimates of GDP which get adjusted for months. It gives an idea of relative period-to-period changes and scope. Let’s say recent numbers are 95% accurate, maybe on the outside as high as 98%, in the absence of better definitive data. This doesn’t change my earlier observations that the number of cases being reported, the number of hospitalizations being reported, and the number of deaths being reported have much different relationships than in April, primarily because the number of cases reported in April is vastly undercounted using today’s protocols and methodologies. And that’s good news because it takes far more reported cases today to translate into future deaths since so many cases have no or few symptoms and require no hospitalization. But case counts make for great political ads. https://youtu.be/bkMwvmJLnc0

      1. Menzie Chinn Post author

        Bruce Hall: In the CDC file, there is a column “expected”. To calculate excess fatalities, I subtract “expected” from “observed”, exactly as indicated in thenotes to the figure.

      2. pgl

        First you deny you said “slight uptick” but then you massage the data to argue it is only a “slight uptick”. You remind me of someone with multiple personalities – all of them insane.

        1. Bruce Hall

          As can be seen by the latest update to the CDC’s “Provisional” Death Counts” for C-19, there has been the upward count as expected in footnote 2 of the chart I have been providing since the end of June with weekly updates. Relative to the April peak, the uptick is still “slightly” upward. https://www.dropbox.com/s/dee1n3gxqzamdai/Covid-19%20Deaths%2C%20Cases%2C%20and%20Hospitalizations%20-%208-4-20.pdf?dl=0
          As can be seen in the other charts in the file, new cases are already beginning to trend downward. The third chart is an update in the hospitalizations by age which are also showing a slight decline and well below the April peak for the older age groups which have the highest death risk.

          The CDC uses a different tally when looking at cases and deaths by state. In case you missed it, here are the top four states for cumulative cases, deaths, and calculated mortality rate. https://www.dropbox.com/s/jjk9zc0ttpeevbg/Covid-19%20Cases%20-%20Top%20Four%20States.pdf?dl=0

          Contrary to the “spin” that pgl attempts to place on my words, I did not say the deaths counts would not rise.

      3. Ooe

        try +157000 deaths and counting..there is no cure; no control, and no vaccine. Do tell us Bruce how many deaths are acceptable to you?

    1. Baffling

      Let him take it. Send him some bleach to wash it down. We are not a nanny state. Stoooopid republicans are going to have to pee on the electronic fence themselves, before they decide to do the right thing

      1. pgl

        Louie has every right to kill himself but what this jerk expected from his staff was criminal.

        I watched the funeral of John Lewis. His staff loved this man. Louie’s staff may be about to sue their boss.

        1. Moses Herzog

          Is Gohmert going to take it on camera to prove he took it??

          We know donald trump is an exhibitionist (in the worst meaning of the term). We know it would have been the very media circus donald trump loves to take “HCQ” during a press conference, but, no, he sheepishly tells people he took HCQ. Yet we have a lot of “naive” (I’m trying to be kind here) people who keep saying and repeating trump took “HCQ”. We still don’t know that as fact . If Gohmert Pyle wants to prove his MAGA credentials let him stand in front of the cameras, with his doctor standing next to him verifying it is indeed “HCQ” as Gohmert takes it for the cameras.

        1. Bruce Hall

          Menzie, your link refers to a study of critically ill patients. HFHS stated that the effectiveness of HCQ (with zinc and azithromycin is realized when given before symptoms become severe. That’s because HCQ provides a pathway for zinc into cells where the zinc stops the virus replication. Go back to the HFHS link and read that. I have also highlighted that specific factor of early treatment several times.

          Patients treated with hydroxychloroquine at Henry Ford met specific protocol criteria as outlined by the hospital system’s Division of Infectious Diseases. The vast majority received the drug soon after admission; 82% within 24 hours and 91% within 48 hours of admission. All patients in the study were 18 or over with a median age of 64 years; 51% were men and 56% African American.

          “The findings have been highly analyzed and peer-reviewed,” said Dr. Marcus Zervos, division head of Infectious Disease for Henry Ford Health System, who co-authored the study with Henry Ford epidemiologist Samia Arshad. “We attribute our findings that differ from other studies to early treatment, and part of a combination of interventions that were done in supportive care of patients, including careful cardiac monitoring. Our dosing also differed from other studies not showing a benefit of the drug. And other studies are either not peer reviewed, have limited numbers of patients, different patient populations or other differences from our patients.”

          The doctors gave the usual caveats: Dr. Zervos also pointed out, as does the paper, that the study results should be interpreted with some caution, should not be applied to patients treated outside of hospital settings and require further confirmation in prospective, randomized controlled trials that rigorously evaluate the safety and efficacy of hydroxychloroquine therapy for COVID-19 [which apparently suddenly now causes serious side effects in patients who have commonly used it for other pathologies for extended periods with no ill-effects].

          “Currently, the drug should be used only in hospitalized patients with appropriate monitoring, and as part of study protocols, in accordance with all relevant federal regulations,” Dr. Zervos said.

  6. Bruce Hall

    There is something amiss in the Covid-19 reporting. This is a file comparing the top states in “cases”. I’ve also taken the death counts and calculated a death rate based on the number of reported cases. In April, New York had almost one-third of the nation’s deaths attributed to Covid-19. It’s cumulative death rate is 7.8%. The other three states, Texas, Florida, and California, all have higher number of cumulative cases than New York, but none have a death rate over 2%. https://www.dropbox.com/s/jjk9zc0ttpeevbg/Covid-19%20Cases%20-%20Top%20Four%20States.pdf?dl=0

    Even if we say that there are more “active” cases in those other three states, the number of deaths would have to be 4-5 times higher in a few weeks to get close to the death rate in New York.

    So, what are the options for drawing conclusions?
    * New York over-reported deaths
    * New York’s policies (requiring nursing homes to take C-19 infected patients and refusing to use the hospital ship for patients) contributed to significantly more excessive deaths
    * Cases now and cases back in April have a different basis
    * Treatments are vastly better so lower mortality rate
    * The database is garbage
    https://www.cdc.gov/coronavirus/2019-ncov/cases-updates/cases-in-us.html (source)

    Whatever conclusions we draw, it is obvious that the mortality rates are far lower now than at the peak of the epidemic.

    2. Baffling

      “ Whatever conclusions we draw, it is obvious that the mortality rates are far lower now than at the peak of the epidemic.”
      Lets put this in some perspective, bruce. The mortality rates are still an order of magnitude greater than a flu pandemic. And two orders of magnitude greater than a typical flu season. Orders of magnitude, bruce. The trump virus is deadly any way you look at it.

      2. Bruce Hall

        baffling,

        It all depends the denominator. I think everyone will now agree that through most of May, only symptomatic people were tested.
        https://www.nbcnews.com/health/health-news/cdc-says-covid-19-cases-u-s-may-be-10-n1232134 [June 25, 2020 article]

        If this is correct, then except for New York, the mortality rate is in line with the flu. The difference is that the flu affects more younger children more severely while C-19 and the flu affect the elderly severely.

        1. baffling

          “f this is correct, then except for New York, the mortality rate is in line with the flu. ”
          incorrect bruce. try again. ORDERS of MAGNITUDE bruce. idiot.

        2. 2slugbaits

          Bruce Hall So wrong. Just making stuff up. Mean flu death estimates for the 2018-2019 season were 34,157. Flu infections were estimated at 35,520,883 (symptomatic illness). That’s a mortality rate of 0.096%.

          https://www.cdc.gov/flu/about/burden/past-seasons.html

          Worldometer shows the US with 162,728 deaths and 5,028,791 confirmed infections. That’s a 3.2% mortality rate. And even if you assume the number of confirmed cases is actually ten times higher, that still puts the mortality rate at 0.32%. Of course, if you want to use the unconfirmed cases for COVID-19, then you also need to use the unconfirmed cases for the flu, which CDC estimates at 42.9 million.

          he flu affects more younger children more severely

          Wrong again. Children typically die at a lower rate. For example, with the notorious and deadly H1N1 the NIH found: In age-stratified analyses, risk estimates rose monotonically with age, from approximately one death per 100,000 symptomatic cases in children to approximately 1,000 deaths per 100,000

          https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3809029/

          As to hydroxychloroquine, I know you’re unfamiliar with statistical analyses, but statnews.com has a nice takedown on the statistical flaws associated with your favorite Henry Ford hospital group study.
          https://www.statnews.com/2020/07/08/a-flawed-covid-19-study-gets-the-white-houses-attention-and-the-fda-may-pay-the-price/

Comments are closed.