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. Wynship W. Hillier, M.S.'s avatar

    I first met Professor Hoy when I returned to UC Santa Cruz in Fall of ’92 to finish my undergraduate…

  2. Justin Fisher's avatar

    To be worth using, a detector needs not only (A) not get very many false positives, but also (B) get…

  3. Mark's avatar

    Everything you say is true, but what is the alternative? I don’t think people are advocating a return to in-class…

  4. Deirdre Anne's avatar
  5. Keith Douglas's avatar

    Cyber security professional here -reliably determining when a computational artifact (file, etc.) was created is *hard*. This is sorta why…

  6. sahpa's avatar

    Agreed with the other commentator. It is extremely unlikely that Pangram’s success is due to its cheating by reading metadata.

  7. Deirdre Anne's avatar

New CDC estimates on COVID

Brian Leiter

35% of cases are asymptomatic (but still infectious), while the fatality rate for those with symptoms is .4% (so four times worse than the standard estimate for flu mortality); from a CNN article:

The fifth scenario is the CDC's "current best estimate about viral transmission and disease severity in the United States." In that scenario, the agency described its estimate that 0.4% of people who feel sick with Covid-19 will die.
For people age 65 and older, the CDC puts that number at 1.3%. For people 49 and under, the agency estimated that 0.05% of symptomatic people will die.
The article goes on to note one expert's skepticism about the fatality numbers as being too optimistic.

Leave a Reply to Charles Pigden Cancel reply

Your email address will not be published.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

17 responses to “New CDC estimates on COVID”

  1. Tim Maudlin

    These numbers make no sense at all.

    The population of NY is about 20 million. The seropositive rate is about 20%, indicating about 4 million who have been infected. If a third are asymptomatic, then that's about 2.7 million symptomatic cases. The number dead in NY so far is given as over 29,000. So that's over 1% fatality rate for symptomatic cases, over ten times the rate for flu.

    And this undercounts because 1) the number of Covid-19 deaths is undercounted (proven by the overall death rate compared to average) and 2) that does not account for the presently infected people who will eventually die of the disease.

    Reply
  2. Layman

    0.1% is a typical *case fatality rate* for the flu in the US. 0.4% is within the range of reasonable estimates of Covid-19's *infection fatality rate* for the US overall. Typical IFRs for the flu could be between 0.02%-0.0.5%. So the CDC's best estimate puts Covid-19 at 8-20 times more lethal than the flu. It will also infect many more people, making its burden on the population vastly greater.

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

    I'm continually surprised by how many epidemiologists fail to distinguish CFR (or sCFR) from IFR for the flu while distinguishing between them for Covid-19. Well into May, Ioannidis still hadn't figured out the difference.

    It should also be kept in mind that the CDC's "best estimate" of IFR is highly uncertain. Estimate of Covid-19 IFR vary widely. Here are two meta-analyses. This preprint meta-analysis' best estimate for Covid-19's international IFR is 0.75%: https://www.medrxiv.org/content/10.1101/2020.05.03.20089854v2

    Ioannidis' preprint meta-analysis (which seems to be getting rightly panned by his colleagues) estimates 0.2%.

    To speculate, I'd suspect IFR will be lower going forward than it was February to April. My reason for thinking this is that better knowledge of how to slow the spread will lower the amount of virus in a typical infection. More people will be able to get immune with less serious symptoms.

    Reply
  3. Jim Holt

    Tim, is the "seropositive rate" of 20% for NYS you cite based on antibody testing of a random sample?

    Reply
  4. Brian Leiter

    Seropositivity was 21% in New York City, but as low as 4% elsewhere in the state. https://www.nytimes.com/2020/04/23/nyregion/coronavirus-antibodies-test-ny.html

    Reply
  5. Robert A Gressis

    New York State has 19,450,000 people.

    13.9% of them tested positive for c19.

    That makes 2,703,550 New Yorkers who got c19.

    29,046 have been dead from c19.

    So, 29,046 divided by 2,703,550 = 1.07% IFR.

    I agree, though, that deaths have been undercounted. When I look to see estimates of how badly we're undercounting, I get estimates that range from the claim that there are actually 1.5 times the number of official deaths to 2 times the number of official deaths. So, if we foolishly assume that the actual number of deaths from c19 is 1.75 times the official count, then the IFR is 1.88%.

    Reply
  6. Josh Shepherd

    A large seroprevalence study run in Spain (N=roughly 70,000 sampled from all over the country) indicates an IFR of 1-1.2% in Spain. The study is in Spanish, but this report is in English:

    https://english.elpais.com/society/2020-05-14/antibody-study-shows-just-5-of-spaniards-have-contracted-the-coronavirus.html

    Reply
  7. Tim Maudlin

    They are randomish samples from people who are out and about…at stores, etc. So if anything, they overcount a bit. The people really staying a home will have a lower rate.

    Reply
  8. Kenny Easwaran

    Note that even taking the 29,000 number as the true number of deaths in the state, and dividing by the total population of 19 million in the state, yields a population fatality rate of 0.15%. The only way IFR could be as low as 0.4% is if something like 30% of the entire state has been infected.

    Reply
  9. Charlie

    The 13.9% number and 29K deaths occurs over a month apart though, and can't be equated.

    Reply
  10. Anonymous

    The death counts are certainly underestimates because of missing home/long term care deaths, but there are some counterexamples to IFR being 1%. With Diamond Princess, the official IFR is 1.6%, but it was an older skewing population. The USS Roosevelt carrier had 1 death out of 1,156 infected with a much younger population.

    I don't think Covid unchecked would eventually lead to 328k dead in the United States that a 1% IFR implies, because it's likely the most vulnerable will be disproportionately hit early on in the epidemic. The tragedy of our current policy is that those deaths were/are largely preventable, and will still be far higher than they should have been were proper precautions taken.

    Reply
  11. Anonymous

    Sorry, I meant to say 3.28m, my apologies.

    Reply
  12. Charles Pigden

    The New Zealand case is interesting because there is reason to believe that, with a ferocious lock-down regime (now relaxing) and a vigourous tracing and testing policy, the health authorities have detected the vast majority of cases. (The government believes that, as of now, there is *no* community transmission which means in effect that there are no undetected cases in the community.) Total confirmed cases stand at 1154: deaths at 21. That’s a death rate per infection of 1.8%.

    That’s probably a higher per capita death-rate than would have existed if more people had been infected since the virus got into a couple of rest-homes and decimated them.. But even if we cut it in half that’s a much higher death rate per case than the CDC suggests.

    Reply
  13. Colin Farrelly

    The antibody testing, as this stage, is not very reliable. Also, if, as some epidemiologists believe (Sunetra Gupta at Oxford), antibody studies do not indicate the true level of exposure or level of immunity then that would explain the situation of NY's appearance of a high case fatality ratio. Many more people in NY probably did have exposure to the virus, but were able to fight it off without producing antibodies. See:
    https://unherd.com/2020/05/oxford-doubles-down-sunetra-gupta-interview/

    Reply
  14. Lexington

    I suspect that in NYC as well as in a number of other places the current statistics on CFR and IFR are skewed by the high number of deaths in nursing homes. These seem to have been infected in a greater proportion than in the larger population. When infection has been introduced into a nursing home, it sweeps through the residents in, well, a New York minute. Even if in the larger population only 20% have been infected, it may well be that 40%-80% of nursing home residents have been. But this would imply that the CFR and IFR would be well less than the current rates if the disease were allowed to burn through the entire population: potential deaths in nursing homes have already been disproportionately counted.

    This may have a lot to do with why NYS seems such an outlier. NYS implemented the rather disastrous rule requiring nursing homes to admit patients who often carried Covid, much aggravating the situation, and, again, distorting the statistics.

    It would be useful to know what percentage of nursing homes in NYC have had to deal with infection, but I've never seen that number reported.

    Reply
  15. Robert A Gressis

    Yeah, I realized that after I posted. However … doesn't it take about 3 weeks to die from c19 after you get it? Moreover, I'm not sure too many more New Yorkers got infected over the last month.

    My guess is that the IFR will settle at about 1%, but I don't really have any particularly strong case for that.

    Reply
  16. Anonymous

    IFR is always tied to some particular population, or else it's meaningless. If there has been widespread transmission within nursing/care homes and hospitals – as seems to be the case *everywhere* – then the relevant population is going to be skewed to the more elderly and vulnerable and the IFR is going to rise potentially *significantly* higher than it would for the general population.

    Reply
  17. Charles Pigden

    Another country with a rigorous testing regime where they have probably detected the vast majority of cases is South Korea. Confirmed cases: 11225; deaths: 269; death rate per confirmed case: 2.3%.

    —–
    KEYWORDS:
    Primary Blog

    Reply

Designed with WordPress