Leiter Reports: A Philosophy Blog

News and views about philosophy, the academic profession, academic freedom, intellectual culture, and other topics. The world’s most popular philosophy blog, since 2003.

  1. Fool's avatar
  2. Santa Monica's avatar
  3. Charles Bakker's avatar
  4. Matty Silverstein's avatar
  5. Jason's avatar
  6. Nathan Meyvis's avatar
  7. Stefan Sciaraffa's avatar

    The McMaster Department of Philosophy has now put together the following notice commemorating Barry: Barry Allen: A Philosophical Life Barry…

COVID-19 is much worse than flu (plus Diamond Princess, redux)

Brian Leiter

No doubt most readers already know this, but this article makes the point forecefully:

The CDC, like many similar disease control agencies around the world, presents seasonal influenza morbidity and mortality not as raw counts but as calculated estimates based on submitted International Classification of Diseases codes.2 Between 2013-2014 and 2018-2019, the reported yearly estimated influenza deaths ranged from 23 000 to 61 000.3 Over that same time period, however, the number of counted influenza deaths was between 3448 and 15 620 yearly.4 On average, the CDC estimates of deaths attributed to influenza were nearly 6 times greater than its reported counted numbers. Conversely, COVID-19 fatalities are at present being counted and reported directly, not estimated. As a result, the more valid comparison would be to compare weekly counts of COVID-19 deaths to weekly counts of seasonal influenza deaths.

During the week ending April 21, 2020, 15 455 COVID-19 counted deaths were reported in the US.5 The reported number of counted deaths from the previous week, ending April 14, was 14 478. By contrast, according to the CDC, counted deaths during the peak week of the influenza seasons from 2013-2014 to 2019-2020 ranged from 351 (2015-2016, week 11 of 2016) to 1626 (2017-2018, week 3 of 2018).6 The mean number of counted deaths during the peak week of influenza seasons from 2013-2020 was 752.4 (95% CI, 558.8-946.1).7 These statistics on counted deaths suggest that the number of COVID-19 deaths for the week ending April 21 was 9.5-fold to 44.1-fold greater than the peak week of counted influenza deaths during the past 7 influenza seasons in the US, with a 20.5-fold mean increase (95% CI, 16.3-27.7).5,6

Also interesting was this tidbit about the Diamond Princess cruise ship, which I had not seen before:

At present, the Diamond Princess cruise ship outbreak is one of the few situations for which complete data are available. For this outbreak, the case fatality rate as of late April 2020 was 1.8% (13 deaths out of 712 cases); age adjusted to reflect the general population, the figure would have been closer to 0.5%.

The age adjustment is needed because the age of people on cruise ships skews much higher than the population at large.

(Thanks to Dr. David Ozonoff for the pointer.)

Leave a Reply to s. wallerstein Cancel reply

Your email address will not be published. Required fields are marked *

26 responses to “COVID-19 is much worse than flu (plus Diamond Princess, redux)”

  1. Eric Winsberg

    I'm not a flu bro. I'm not arguing that Covid is no worse than flu by a long shot. But the comparison of numbers in that piece is very misleading. We don't *count* flu deaths nearly as well as we've been counting Covid deaths. But the CDC does *estimate* the total number of deaths from flu. And it estimates that in the 17-18 season we had about 80,000 deaths. I can assure you that we didn't get to that number with under 1,000 deaths per week. We almost certainly had to have had some weeks in there with multiple thousands of deaths. On the other hand if you look at the CDC's report of all excess deaths for 2020, it looks like we are doing a reasonable job of counting all the Covid deaths.

    Reply
  2. Brian Leiter

    I'm not sure I understand this response. Their point was precisely that comparing *estimated* flu deaths to *actual* counts of COVID cases is misleading.

    Reply
  3. Kenny Easwaran

    Eric Winsberg's point though is that comparing actual counts of COVID to actual counts of influenza is also misleading, because there has never been a year where hospitals around the country have been so focused on identifying influenza cases as they are this year on identifying COVID cases.

    None of the numbers we have for flu in other years or this year (estimated, identified, "excess deaths", test-based, symptom-based, etc.) is going to be exactly comparable to any of the numbers we have on COVID. But we can still try to do useful comparisons by looking at many of these numbers in context with each other and trying to estimate what those numbers would look like for the other disease on the basis of the numbers we do have for that disease.

    Reply
  4. Brian Leiter

    I see that, but it seems to me quite wrong to describe this article as "very misleading": what's very misleading is comparing estimates of flu to actual COVID counts. That was the key point. Actual flu counts are, admittedly, not as thorough as actual COVID counts (although even the latter are surely undercounts at this point).

    Reply
  5. Eric Winsberg

    Kenny more or less gets my point right, but still missing a key premise. The CDC is releasing weekly excess death numbers right now. They give us pretty good confidence that the *counts* of Covid deaths are going to match the eventual estimates. So I don't know, Brian, where you get the idea that "[covid counts] are surely undercounts at this point." Its in fact surely the opposite, if the CDC is to be trusted at all. And so yes, I think the article is very misleading. It says " suggest that the number of COVID-19 deaths for the week ending April 21 was 9.5-fold to 44.1-fold greater than the peak week of counted influenza deaths during the past 7 influenza seasons in the US, with a 20.5-fold mean increase" But this just isn't close to right. We *know* that if flu deaths had been counted in 17-18 the way covid deaths are being counted right now, they would be almost an order of magnitude larger.

    Reply
  6. Eric Winsberg

    sorry, by "its surely the opposite" I didn't mean that they are surely *overcounts*, I meant that we can be quite confident that the counts are quite close to correct.

    Reply
  7. Brian Leiter

    Thanks for the response. Is the CDC counting all the excess deaths as COVID deaths? Were the authors of the article?

    Reply
  8. Jason

    Worth noting that the peak death rates for the flu that are used in this comparison were not from a period of time in which much of the United States was under public health stay-at-home orders aimed at suppressing wide-spread contagion. Because, well, it's kind of obvious that covid 19 is far more lethal than the flu to begin with. What's misleading isn't the effort to compare the deaths that covid 19 has caused already with peak rates for the flu in past years. What's misleading is ignoring the context altogether!

    (That being said, it would be interesting to see what flu rates look like alongside covid rates for the same period of time since one might reasonably suppose that efforts to suppress covid would also suppress flu contagion.)

    Reply
  9. Peter

    Prof Leiter, I agree with Eric Winsberg. We want to compare out best (!) estimates for flu and covid mortality. What if the covid deaths are much better reported than flu deaths? In that case the comparison of these two quantities will be misleading. Just because the estimated flu mortality is an estimate that does not mean that is not a better estimate than flu mortality calculated from reported flu deaths. Why does the CDC estimate flu mortality rather than simply use reported flu deaths? The experts at the CDC must believe that estimated flu mortality is a much better indicator of the flu's death toll than flu mortality calculated from reported flu deaths. The key thing here is to realize that both estimated flu mortality and flu mortality calculated from reported flu deaths are imperfect indicators of the flu's death toll.

    The authors of the JAMA Viewpoint need to explain why we should prefer their comparison.

    Reply
  10. Peter

    The authors of the JAMA viewpoint conclude their article as follows: "comparisons between SARS-CoV-2 mortality and seasonal influenza mortality must be made using an apples-to-apples comparison, not an apples-to-oranges comparison."

    So they are suggesting that apples-to-oranges comparisons are invalid or worse than apples-to-apples comparison. This is not so. For instance, it is legitimate to ask whether eating apples provides more iron than eating oranges. Logically speaking, a comparison just requires that the compared entities share the feature on which we will compare them (it would be nonsensical to try to compare the heartbeat of a human with that of a stone). So we can compare apples and oranges along many dimensions (iron content, sugar content, thickness of their skins, number of seeds in a fruit, etc.). We could also ask is apples consumption more deadly than orange consumption. Just as we ask is the flu more deadly than the coranavirus?

    The authors of the JAMA viewpoint even seem to concede that comparing reported deaths from the flu with reported covid deaths is not an apples-to-apples comparison because influenza deaths are not like covid deaths insofar as influenza deaths are not reportable to public health authorities. Pitching the comparison of mortality from flu and from covid at the level of having to compare apples to apples is a red herring. What really needs to be done is to compare estimates of the two mortalities that are of similar quality, or if estimates of similar quality are not available then the comparison needs to take into account the differences in the quality of the estimates.

    I don't want to be misunderstood: I'm not saying that covid is about as lethal as the flu. I just find the JAMA viewpoint unnecessarily confusing.

    Reply
  11. s. wallerstein

    Isn't Covid much more contagious than the flu? And isn't that why it's so dangerous?

    Reply
  12. Brian Leiter

    COVID is more contagious than flu because no one (or hardly anyone at this point) has immunity to it: so it can spread more widely. But it also appears to be cause more serious symptoms than flu in mmany people.

    Reply
  13. Luca Ferrero

    Both Brian and Eric have a point here–and the author of the piece too.

    We must compare apples to apples…

    The most meaningful comparison is between what is called the 'burden' of flu and the 'burden' of Covid-19.

    BURDEN OF FLU: this is what CDC calculates and what is usually reported when talking about the mortality of seasonal flu. When we hear about 0.1% fatality rate and 61K deaths, we are talking about the burden of flu https://www.cdc.gov/flu/about/burden/index.html

    (incidentally, to Eric Winsberg, the 80K figure you quote above was from an earlier estimate of 2017-18 flu, which I've used myself in the past, it turns out that the current estimate is 61K see link above)

    The burden of flu is much higher than confirmed flu deaths and it usually gets pretty close to the excess mortality during the winter months.

    BURDEN OF COVID-19 – we do not yet know what this is going to be but using excess mortality is likely to provide a very good approximation. (Comparing confirmed deaths of flu to confirmed deaths of Covid-19 is not meaningful as pointed out by Eric because we are confirming covid-19 deaths much more than flu deaths. This is the mistake in the conclusion of the article linked by Brian).

    So how much excess mortality is there? and how much of this has not been reported as covid19? This is the main question.

    I have not been following the US data closely but I can give you the example of Italy based on the most recent study by the Italian CDC.

    Through March 31st, in Italy excess mortality: 25.3K, confirmed C19 deaths 13.7K, so there are 11.6K still unaccounted for deaths, most of whom I think should go into the covid19 burden.

    This excess mortality is 109% over the average 2015-19 mortality, so in Italy excess mortality is considerably higher that average seasonal flu burden. Some of the undercounted deaths are due to failures of the health care system in the northern part of Italy but I do not think that this accounts for more than a small percentage of those 11.6K deaths (short version: the ICU shortage which would be a good measure of this failure only lasted for a few days in Lombardy, when ICU was at full capacity–given that full capacity was in the order of 700 beds, I imagine that at most they would be turning away at most 50 patients a day for 5 days or so). Lombardy mishandled the infection in that it did not help to slow the spread but most of the deaths in Lombardy are, in my view, due to the direct effects of covid19 rather than the inability to hospitalize those sick with the disease.

    Also, in the parts of Italy were the incidence of the virus were not as high and the health care system did not crash, we have a similar percentage of possible undercounting (2.4K excess deaths, confirmed Covid 19 deaths 1.1K, unattributed excess deaths 1.3K)

    Finally, the lockdown in Italy did not produce an increase of non-covid 19 mortality (at least in the short term), in the South of Italy that was under lockdown but barely affected by the virus, the seasonal mortality actually decreased by 1.8%

    Reply
  14. Eric Winsberg

    The CDC is just telling you how many more deaths they think there have been so far this year than normal. That *happens* to be a pretty good fit with the counted number of covid deaths. That tells me that we are counting covid deaths pretty well. I'm actually pretty surprised by this, but there it is. The article is comparing counted covid deaths to counted flu deaths. But the CDC also, for past years, estimates flu deaths. And we know there were almost 80,000 of them in a bad year. So comparing counted Covid deaths to counted flu deaths is a bad camparison. It *happens* to be a good comparison to compare counted covid deaths to estimated flu deaths. If you do that for the worst recent flu year, you get about the same number. HUGE CAVEATS: covid is a shorter period, its not over, we have been taking mitigation measure of unknown impact etc etc etc. So im not saying the 17-18 flu was as bad as this year's covid. I'm just saying the comparison in the article is poor and misleading.

    Reply
  15. Eric Winsberg

    Thanks for correcting the 80k-60k figure. (It doesn't make that much difference to my point, since I'm not trying to make an apples to apples comparison, I'm just criticizing the article for failing to make one.)

    "Finally, the lockdown in Italy did not produce an increase of non-covid 19 mortality (at least in the short term)"

    I don't think anyone would expect the lockdown to cause non-covid mortality *in the short run* and *in the developed world*. But the economic downturn caused by the lockdown in the developed world will almost certainly lead to a large number of deaths in the developing world (in the months to come). Poor economic conditions lead to starvations and deaths from preventable diseases in the developing world in a way that we would never see in the US or Italy. And the deaths in the developed world will be downstream (in the years to come): missed cancer screening, missed stroke rehabilitation, drug addiction, poor neo-natal care, etc.

    Reply
  16. Layman

    Eric,

    Could you clarify what you're saying about CDC Covid-19 death counts vs. actual number of Covid deaths? You say "if the CDC is to be trusted at all" then their Covid-19 death counts are "quite close to correct."

    It sounds like you're not literally asserting this conditional but rather mean to say that the CDC's counts are quite close to correct. If so, what data do you base that on? All the commentary by statisticians or epidemiologists that I've been able to find suggests there's an undercount that could range from minor to significant. This is, of course, also the view Fauci recently presented to Congress.

    https://www.nytimes.com/2020/05/13/opinion/coronavirus-us-deaths.html

    https://www.npr.org/sections/goatsandsoda/2020/05/13/854873605/fauci-says-u-s-death-toll-is-likely-higher-other-covid-stats-need-adjusting-too (see comment by Sei Pen)

    Intuitively: between late March and late April there were over 20,000 more excess deaths than Covid-19 deaths. Of course, most or even all of these might be due to other causes, like people staying away from hospitals. But surely it's not unreasonable to think a significant number of these were due to Covid-19, is it?

    Or were you literally asserting the conditional ("if the CDC is to be trusted…")? I don't see why that would be true.

    (I do take your point that Covid-19 deaths are less undercounted than flu deaths.)

    Reply
  17. Luca Ferrero

    Eric, based on what you say about the CDC reports, I agree with you that we might not be undercounting in the US, but we should be aware that this might not be so in many other countries (see example of Italy)

    But see my previous comment about the 80K figure for 2017-18 flu season. The current estimate is 61K (80k was from a preliminary estimate that was available until recently but has now been updated)

    I also take this opportunity to raise an issue with how we count fatality rate for the flu – when we calculate 0.1% for flu this is not really IFR but more like sCFR, fatality over symptomatic cases (which are the numbers estimated by CDC based on hospitalizations and deaths). However, the flu has also asymptomatic infections although not as many as Covid-19. I have seen some rough estimates that asymptomatic infections for the flu could be as high as 20%.

    So we also have to make sure that we do apples to apples comparison here: the IFR for the flu is likely to be lower than 0.1%, possibly closer to 0.08% – we should keep this in mind when comparing with IFR for covid19.

    So let's compare Burden of Flu to Burden of Covid19, and IFR for flu (not sCFR) to IFR for covid19

    Reply
  18. Layman

    To the best of my understanding, the 0.1% flu "fatality ratio" is the typical *case* fatality rate in the US. On the CDC site you link to, Luca, it's stated that their estimates are based on a ratio of deaths to hospitalizations.

    This is worth noting, as people often insist on comparing Covid-19's infection fatality ratio to the flu's 0.1% fatality ratio. But that's apples-to-oranges in the bad way. There are many people who get the flu but never get officially diagnosed; these people push the flu IFR below its CFR just as analogous people do the same for Covid-19.

    According to Adam Kucharski, there are generally massive numbers of undiagnosed flu infections. These push the flu's IFR down to roughly 0.02%-0.05%. That makes Covid-19's IFR (for the US) roughly 10 to 50 times higher than the flu's.

    https://twitter.com/AdamJKucharski/status/1243466394991239170

    Reply
  19. Eric Winsberg

    We are probably taxing Brian's patience at this time. I didn't say ""if the CDC is to be trusted at all" then *their* Covid-19 death counts are "quite close to correct."" I said (or meant to say) if the CDC's overall counts *of total excess deaths* are correct, then the aggregated counts of covid deaths that you see on, e.g., the worldometer are quite close to correct. If you go to the worldometer, which aggregates all the state and county counts of covid deaths, you will see that the count is roughly 88K. If you go to the CDC page https://www.cdc.gov/nchs/nvss/vsrr/covid19/excess_deaths.htm, you will see that they project that when all death certificates are counted, the total number of deaths in USA year to date will exceed the usual average of recent years up to this date by a confidence interval of 68,774 – 92,742. It thus seems to me that the 88k deaths you see on the worldometer has to basically be in the ballpark of being correct. Even if its undercounting those deaths by, say, 25% (which i think is implausible given the evidence I just cited), it still shows that comparing covid counts to flu counts is extremely misleading for all the reasons I gave.

    Reply
  20. Lexington

    One way to get grip on how much worse Covid is than the flu is to look at the places where it is taking the biggest toll, and provides the most reliable numbers for that context: nursing homes.

    I gather that in at least a number of nursing homes, perhaps 1/3 or more of the residents have died of Covid, and have done so in the space of a month or two from the first infection. Now it seems pretty fair to include the vast majority of those deaths as Covid deaths, given the short time period over which they occurred. Certainly the ordinary death rate over that short period must be relatively quite low.

    It's pretty obvious that the numbers for the flu never reach those proportions, even within nursing homes. Otherwise there would be similar catastrophic waves of deaths in nursing homes whenever a new flu came to the door. And we have never heard of such waves. The point here is that this gives us pretty hard numbers as to the relative lethality of covid.

    Of course, this is only one segment of the population. But I think the default assumption should be that similar relative levels of lethality would be found in other contexts.

    Reply
  21. Layman

    Thanks for the clarification on the undercount point. I would say the sources I cited suggest we should accept the higher end of the CDC excess death estimate, which would mean Worldometer, is slightly undercounting. Which is fairly consistent with both your and Brian's statements on the issue.

    I think more caution is needed in discussing the effects of the lockdown on the economy. I don't think we know enough to say the lockdown caused a downturn. There are reasons to think the lockdown averted more Covid-related economic damage than it caused. At least a few weeks ago, this seemed like a widespread view amongst economists. Here's a nice overview of these reasons:

    https://medium.com/@benbayer_62236/no-one-has-shown-the-lockdown-isnt-worse-than-the-virus-9866a2d96f5

    (It's a critique of such arguments, though, to my mind, it just establishes the obvious point that they're less than conclusive.)

    I don't think we know enough to say lockdowns have helped or hurt the (US) economy. That will depend on a lot of things we don't know yet. One of the big ones is whether we make good use of the time the lockdown has bought us. If we end up just coming out of lockdown haphazardly and achieving herd immunity by accident, then the lockdown will have saved few lives and will only have delayed, rather than prevented, other Covid-related economic damage. If we largely come out of lockdown with a system like Korea's in place, with case numbers manageable and with people feeling safe going to work and stores, and if a vaccine will arrive within a reasonable amount of time after lockdowns end, then it could easily turn out that lockdowns were a net positive for the economy.

    I'm a lockdown agnostic because I have no idea where on the spectrum between these two outcomes we'll end up. And from what I can tell, no one does.

    Reply
  22. JDRox

    "But I think the default assumption should be that similar relative levels of lethality would be found in other contexts."

    I might be misinterpreting this, but while this might have been a sensible default assumption before we had any evidence, we now have a tremendous amount of evidence that the level of lethality is much, much less in other contexts.

    Reply
  23. Eric Winsberg

    "If we end up just coming out of lockdown haphazardly and achieving herd immunity by accident, then the lockdown will have saved few lives and will only have delayed, rather than prevented, other Covid-related economic damage. If we largely come out of lockdown with a system like Korea's in place"

    I should say that a lot of my view on this stem from the fact that I think the former has a 100% chance and the latter a 0% chance. The US gets a minimum of 500,000 new infections a day. Look at Boston. 2 months into lockdown and we find out today that 1 out of every 40 people with no symptoms in the whole city is *actively* infected. That probably means literally every person in Boston has come into contact with an infected person in the last 14 days. You cannot test and trace your way out of that.

    Reply
  24. Eric Winsberg

    Brian asked me to provide some support for those claims. The Boston one is easy.

    https://www.boston.gov/news/results-released-antibody-and-covid-19-testing-boston-residents

    The one about the US as a whole is harder. But I think its conservative. A typical number of confirmed new cases in the US is 30,000 (see the worldometer.) Everywhere we have systematic data, we see that somewhere between 30-50x cases are infected. We get that in all the Califonia studies, in the Florida study. We almost never see less than 20x cases. but you only have to assume 15x cases to get 500,000 new infections. But assume I am overestimating that by a factor of 4. It would still be 125,000 cases per day. Many of these people, when you caught them, would give you a list of 100 people they came into contact with in the last 7 days. Can the US track down 12M to 48M new contacts every day?

    Reply
  25. Lexington

    You may be misinterpreting me.

    I had in mind the *relative* levels of lethality in a context — that is, the level of lethality of Covid compared to that of the ordinary flu in a context. Thus if the level of lethality of covid in nursing homes is 10X that of the flu, then roughly that ratio would hold in other contexts. Of course, the absolute level of lethality might be vastly smaller for both covid and flu in some of these other contexts.

    Reply
  26. micheal pan

    I just want the whole world to know about this spell caster I met
    two weeks ago, wisdomspiritualtemple@gmail.com I cannot say everything he has done for me my wife
    left me 3 years ago left with my kids I was going through online
    when I meant this wonderful man's testimony online I decided to
    give it a try and my wife is back to me now and we ar1e happily
    married again cause is too much to put in writing all I can say is
    thank you very much am very happy .and does alot of spell
    including Love Spell
    Death Spell
    Money Spell
    Power Spell
    Success Spell
    Sickness Spell
    Pregnancy Spell
    Marriage Spell
    Job Spell
    Protection Spell
    Lottery Spell
    Court Case Spell
    Luck Spell etc. In case you need his help contact him on this email
    address wisdomspiritualtemple@gmail.com he is a good man
    thanks.whatsapp number +234813 648 2342

    —–
    KEYWORDS:
    Primary Blog

    Reply

Designed with WordPress