Slashdot is powered by your submissions, so send in your scoop

 



Forgot your password?
typodupeerror
×
Medicine Science

Coronavirus Antibody Testing Shows Lower Fatality Rate For Infection (npr.org) 164

Jon Hamilton, reporting for NPR: Mounting evidence suggests the coronavirus is more common and less deadly than it first appeared. The evidence comes from tests that detect antibodies to the coronavirus in a person's blood rather than the virus itself. The tests are finding large numbers of people in the U.S. who were infected but never became seriously ill. And when these mild infections are included in coronavirus statistics, the virus appears less dangerous. "The current best estimates for the infection fatality risk are between 0.5% and 1%," says Caitlin Rivers, an epidemiologist at the Johns Hopkins Center for Health Security.

That's in contrast with death rates of 5% or more based on calculations that included only people who got sick enough to be diagnosed with tests that detect the presence of virus in a person's body. And the revised estimates support an early prediction by Dr. Anthony Fauci, director of the National Institute of Allergy and Infectious Diseases and a leading member of the White House coronavirus task force. In an editorial published in late March in The New England Journal of Medicine, Fauci and colleagues wrote that the case fatality rate for COVID-19 "may be considerably less than 1%." But even a virus with a fatality rate less than 1% presents a formidable threat, Rivers says. "That is many times more deadly than seasonal influenza," she says. The new evidence is coming from places such as Indiana, which completed the first phase of a massive testing effort early in May.
Further reading: Antibody Tests and Accuracy Issues Leave Some Americans With More Questions Than Answers.
This discussion has been archived. No new comments can be posted.

Coronavirus Antibody Testing Shows Lower Fatality Rate For Infection

Comments Filter:
  • by DevNull127 ( 5050621 ) on Thursday May 28, 2020 @02:27PM (#60116880)
    "The current best estimates for the infection fatality risk are between 0.5% and 1%," says Caitlin Rivers, an epidemiologist at the Johns Hopkins Center for Health Security.

    So that's way higher than the 0.4% the CDC claimed as its "best estimate" for fatality rates.
    BR. Glad to see Slashdot is finally printing some more accurate numbers.
    • lol you thought the CDC was a reliable source of data?
      • by NFN_NLN ( 633283 )

        0.02%, 0.5%, 2%, 5%, 12%...

        All I know is COVID-19 is the cure. No one has died from general flu, pneumonia or old age since it struck. Now scientists just need a chance to study it so they can unlock the fountain of youth.

        In b4 a new article next month when they say it's only 0.02% and much less than the previously thought 0.5-1.0%.

    • by fermion ( 181285 )
      No it proves what the CDC has already stated. the anybody test is of little value right now in forming conclusions or using for decisions.
    • Everyone knew it was bad data which is why there were so many calls to have increased testing and not just test those who showed significant signs of covid-19. It's that that slashdot gave inaccurate numbers, it gave the numbers that the data that was available suggested. There were disagreements of course, some numbers were lower, some were higher, but there were all guesses built upon insufficient data.

  • False positives? (Score:5, Interesting)

    by Immerman ( 2627577 ) on Thursday May 28, 2020 @02:28PM (#60116888)

    The biggest potential problem with this approach is false positives - in particular detecting infections by any of the dozens (hundreds?) of other, harmless coronaviruses that circulate through the population. I've heard rumors that several of the current antibody tests suffer from this problem, but can't find anything definitive. Has anyone else heard anything on it?

    • Yeah everyone is rushing to push them out so it's bound to happen, everybody is fixing the airplane midair, as it were. It's hard without a reference standard.
      • by raymorris ( 2726007 ) on Thursday May 28, 2020 @04:23PM (#60117466) Journal

        False positives are a big issue and will always he a big issue unless a large percentage of the population is infected. That's because of something called prior probability.

        Suppose you have a test that's 96% accurate, which would be one of the better tests. By 96% we mean that if you test 100 people who are actually negative, it only has 4 false positives, saying that 4 people are infected when they really aren't.

        Suppose the actual infection rate is 4%. Your test will give 4 false positives and 4 true positives. For the people who test positive, HALF are false positives. That's with a test that's 96% accurate.

        Looking at it another way, if before the test you have a small chance of having it (4%), that's a 96% chance you don't have it. Then the test at 96% saying you do have it. Those two 96% cancel out and there is a 50/50 chance that you actually have antibodies if the test says you do.

        This is a recurring issue in criminal trials. The prosecution expert will say "there is only a 1/10,000 chance that a random innocent person would match". The jury thinks that proves that the defendant is the murderer. But wait! BEFORE the expert testimony, there was a 1/5 billion chance that the suspect is the murderer! 5 billion to 1 odds that they are innocent. 5 billion is much larger than 10,000, so deep the test result the defendant is probably innocent.

        Then you have the other side of that:

        If 1 of every 10,000 people match, that's a lot of people in the world who match. It's super weak evidence, right? Well, put of all those people that match, only ONE is the victim's abusive ex-husband. It's very strong evidence if the abusive ex matches, not a random person.

        False positives continue to be a major issue unless something like 40% of the population is actually infected.

        • I should add, researchers can somewhat adjust for this when calculating aggregate statistics. If they know that the false positive rate is 3%-5% and 8% of the people come up positive, that means that 3-5% of the people actually have it. Just subtract the false positive rate from the total positive rate to get the true positive rate.

          That does NOT do you any good for trying to find out if a specific person was infected. Aggregate stats can use the adjustment, specific cases can't. That's what led the FDA

        • Comment removed based on user account deletion
          • > the folks administering these tests and evaluating the statistical results are not fools. They know all of that

            That's a good point, and one I made in my follow up post. For statistics about the population as a whole, the tests can be useful. If you know with some precision what the false positive rate is.

            HOWEVER, you've also given an example of what can happen with someone such as yourself who has some understanding of it ...

            > If the false positive rate is 4% and the false negative rate is also

    • A lot depends on which test is used. Some are not very accurate. Roche's test claims to be very accurate, but I'll wait to see if that holds. Until then, knowing what test you are getting is very important.

      This site [cebm.net] can help explain how it works (no pay-wall)

      • >knowing what test you are getting is very important.
        Indeed. Given the serious issues that have already cropped up, it seems to me that all the COVID testing statistics should be grouped by the specific test used - maybe even down to the batch number. And they should have been doing that from day 1, *especially* during the period when they were allowing tests to be produced and sold with no evidence of efficacy. Then when it's discovered that Test A has a false positive/negative rate of X, the statisti

        • The individual labs track everything down to that level to be accredited; being able to aggregate individual's medical data for medical research is thorny and so will not be as widely able to be accomplished.
          • That's good to know.

            I don't see why you'd need to incorporate any individual medical data into reports for aggregation though - just break down test results by the particular test used rather than just by positive/negative results.

      • by aliquis ( 678370 )

        https://www.fda.gov/medical-de... [fda.gov]

        Very nice. Getting tested with that would be a good start (had fever late friday to early wednesday. (Sweden, not much interaction with other humans whatsoever + use mask at the food store but may have got unlucky by touching surfaces others use too + at-least meet people in a bit closer situations twice + been around bunch of kids once.)

    • They attempt to control for the false positives. Before the test is released, they test blood which is known to be virus free and record how many positive results they get. Later when they test unknown samples they assume they have the same false positive rate. They also put error bars around the rate using statistical analysis. Scientists are human and make mistakes but usually they try to account for the obvious stuff.

    • in particular detecting infections by any of the dozens (hundreds?) of other, harmless coronaviruses that circulate through the population.

      Try four different, known human coronaviruses in circulation [cdc.gov], probably now going up to five. The immune response to these is clearly different to COVID-19 because otherwise large numbers of us would already have immunity. The concerns I had heard about the tests, particularly the early ones, were more regarding false negatives: the number of antibodies produced could easily be below the detection threshold. Hopefully, this has been improved and these newer tests are now more accurate.

      • Four (or five) *known* - my understanding though is that until quite recently coronaviruses were generally considered generally harmless, and uninteresting research topics, so we haven't actually spent much effort identifying them.

        There's also the *minor* detail that not all diseases confer immunity to future infections, and there's already some (decidedly inconclusive) evidence that COVID19 may be in that class.

        • In general a disease that appears not to give immunity actually does. The problem being that the fucker then goes and mutates so what might appear to be the "same" disease actually is not.

          • Perhaps, but not always, as anyone who has ever been in the situation of repeatedly passing a disease around their household can attest. Happened to my family once when I was a child with strep (which is bacterial, but the point about immunity remains) That was a miserable few months. I mean, maybe it mutated several times within our household of 4, but that seems a stretch.

        • SARS and MERS were both coronaviruses. And both of those are hella old, and they had major deadly epidemics of their own. So this is not some new or obscure family of viruses. And no one thought they were harmless. If coronaviruses in general (As, of course, opposed to COVID in particular.) are poorly identified, researched, or understood; it's only from laziness and complacency following their respective outbreaks.

          So, if there are 5 different known viruses in the family; that leaves only 2 that aren't de

        • there's already some (decidedly inconclusive) evidence that COVID19 may be in that class.

          The evidence that initially suggested that in Korea was declared to be due to false positives by the WHO. It's still possible but so far there is no evidence to suggest it.

      • "The immune response to these is clearly different to COVID-19 because otherwise large numbers of us would already have immunity."

        Lots of people seem to suffer no ill effects...

    • by hey! ( 33014 )

      I looked for information on the test they were using, but all I found is PR releases -- no actual scientific paper describing the methods.

      Plaque neutralization is the gold standard test, and if that's what they're using I'd have a high confidence in their results. But rapid result antibody tests have had reported false positive rates from 2% to has high as 27%. It doesn't necessarily mean you can't use them this way, it just means you can't draw conclusions straight from the raw test results.

      • >But rapid result antibody tests have had reported false positive rates from 2% to has high as 27%. It doesn't necessarily mean you can't use them this way, it just means you can't draw conclusions straight from the raw test results.

        Agreed, provided you keep track of which individual tests were used in your statistics. If you're just compiling the number of positive or negative results, without keeping track of how many of which results came from which tests, then you can't meaningfully attempt to corre

    • The better tests are generally calibrated by testing them against all known human coronaviruses to rule out false positives. The number you look for is "specificity" that approaches 100% for some of the lab based tests. The one Quest (and UW) runs is 99.6% specific, for example. There are lots of really sketchy tests out there so it's worth doing some reading.

      Another problem is that there's some variability in how people's immune systems defeat the virus. They may not develop the antibodies the test looks f

  • Lower than what? (Score:4, Informative)

    by Njovich ( 553857 ) on Thursday May 28, 2020 @02:34PM (#60116914)

    A lower mortality rate than what? China reported 0.5% mortality rate initially in Guandong and most studies have pointed towards this range.

    • That's in contrast with death rates of 5% or more based on calculations that included only people who got sick enough to be diagnosed with tests that detect the presence of virus in a person's body.

      They're conflating CFR with IFR. What I don't know is if the writer is some dumb fuck who couldn't be bothered to learn the difference, or if he's trying to push an agenda- saying that it's less dangerous than we thought, when it is in fact more.

  • by DevNull127 ( 5050621 ) on Thursday May 28, 2020 @02:35PM (#60116922)
    Estimates of New York City's total death rate, 0.86% to 0.93%, are even higher than the CDC's worst-case scenario [buzzfeednews.com].

    Estimates from countries like Spain and Italy are also higher, ranging from 1.1% to 1.3%.
    • by DevNull127 ( 5050621 ) on Thursday May 28, 2020 @02:37PM (#60116936)
      A preliminary analysis of more than two dozen studies from Europe, China, the US, and elsewhere, conducted by Meyerowitz-Katz and colleague Lea Merone, suggests that the overall infection fatality rate is between 0.5% and 0.78% [buzzfeednews.com].

      Even the lower end of that range is higher than what the CDC says is its "best estimate" for the rate, which is about 0.26%
      .
    • by Calydor ( 739835 )

      And these higher numbers; are they from areas where the hospitals became overloaded with patients or the infected for other reasons could not receive care, eg. with ventilators?

    • New York's fatality rate is higher because they let their hospitals get overwhelmed, resulting in many people receiving insufficient or no care, and needlessly dying. Same for Spain and Italy (and now apparently the UK). This is what flattening the curve was all about - to prevent what happened in those places. Most of the rest of the U.S. (and most developed countries thus far) have flattened the curve sufficiently to avoid the inflated death rates seen in the areas which failed to flatten the curve suf
      • by ghoul ( 157158 )

        People might want to study what is unique about NYC that made it this bad. Tokyo is a bigger and denser city. Delhi has a bigger metro system and also more crowded. Moscow is just as cold and has an almost as big metro system. San Francisco and Atlanta are also Intrnational gateways for flights.
        What was it that made NYC such a bad case.
        This is important to know as many countries like Russia and Brazil and India are just starting their uptrend. They can learn from New York's mistakes.

    • 0.5% IFR means 5 in 1000 deaths. Lets say herd immunity is achieved at 70% so it will take 3.5 deaths per thousand to achieve herd immunity. The overall death rate in the US is 8.3/1000 each year. So over the next 1 year almost a third of ALL deaths in the US will be from Coronavirus - more than cancer, more than old age, more than car accidents, more than war and terrorism, more than heart attacks, more than all crime combined. At 0.5% IFR this will kill 1.05 million people over the next one year before h

      • The overall death rate in the US is 8.3/1000 each year.

        Do we really have a death rate that low?

        Looks like 8.3/1000 translates to a life expectancy of about 120. I'm pretty sure our average life expectancy is a bit lower than that....

        • Looks like 8.3/1000 translates to a life expectancy of about 120.

          You are assuming the population pyramid is constant.

          In 1940, the population of America was 132M, only 40% of the current population. Those are the 80-year-olds today. There are proportionally far less numerous than younger people. That pushes the death rate down. The death rate will rise as boomers age-out.

          • by ghoul ( 157158 )

            Interestingly enough the population of the Empire of Japan was around 120 Million. I keep reading how Japan never really had a chance against the US as the US was a much bigger economy but given similar levels of tech , economy is basically a factor of population so it was closer than we think today. if Japan could have got the oil it needed from Indonesia to run its war economy it could have defeated the US.

        • by porges ( 58715 )

          CDC data here [cdc.gov] has 8.6/1000 for 2017.

  • by magzteel ( 5013587 ) on Thursday May 28, 2020 @02:47PM (#60116982)

    There is no way to accurately calculate the death rate because no one knows how many people were infected.
    As more volume testing is done, and assuming the testing is accurate, the rate will likely keep going down.

    • There is no way to accurately calculate the death rate because no one knows how many people were infected. As more volume testing is done, and assuming the testing is accurate, the rate will likely keep going down.

      Well, yeah, that was kinda the point of the study, to more accurately determine the denominator.

      At the same time, I hope other people are working to refine the numerator. There have been all sorts of stories (and yes, data is not the plural of anecdote) of doctors using different criteria to count a death to COVID or not. I'm sure there are many uncounted cases and many other cases which got counted but really were something different. I have no idea of the magnitude of either.

      • by dgatwood ( 11270 )

        That's a good reason to use CFR and R0 and ignore the IFR. The measurement error on CFR is at least likely to be consistent from one disease to the next. :-)

        • by ghoul ( 157158 )

          Given that care for Covid is covered with stimulus funds and care for non Covid is not, hospitals have a motive to count every uninsured death as a Covid death. Given that the hospitals are a hotbed of Covid right now, anyone else brought in for something else probably has Covid by the time they die so the test can be positive.

    • Yes...but it may also go up if excess deaths are factored in as well. A lot of places are showing huge jumps (by an order of magnitude) in excess deaths for this time period, but these deaths haven't been tested. If those unexplained deaths are also factored it, the rate could go up. But as you've said 'we don't know', and there are a lot of unknowns still.

      • The excess deaths even if not directly caused, can still legitimately be put down to COVID-19. That is without it they would not have occurred.

        • The excess deaths even if not directly caused, can still legitimately be put down to COVID-19. That is without it they would not have occurred.

          That's a bit of a stretch. Some of those deaths might be, for example, increased suicides. I'd be stunned if traffic deaths didn't go down. I don't know enough about the magnitudes of any of these so I can't say whether one dominates the others. But I wouldn't use excess deaths to determine policy. I'd use it as a reality check of other measurements.

        • The excess deaths even if not directly caused, can still legitimately be put down to COVID-19. That is without it they would not have occurred.

          Really? That's going to make it hard to calculate risk for future pandemics if you count the deaths caused by preventative measures along with the deaths caused by the illness.

      • A lot of places are showing huge jumps (by an order of magnitude) in excess deaths for this time period...

        To be clear, you're saying the excess death rate, the rate above what you'd typically expect, went up by an order of magnitude, right? Not that the actual absolute death rate went up by 10x?

        Seems like a legit way to double check other measurements. Given that all our measures seem to have large uncertainty, I like the ideal of measuring several ways.

    • The death rate seems iffy also, as so much of that depends upon the quality of care being provided and comorbidities. I'd like to see the rate of getting seriously ill from it so that you're bed ridden. Maybe not the hospitalization rate because some people died while avoiding going to the hospital because they didn't want to cause a problem (very often you go from having a something like a bad flu to being dead in just a couple of days).

  • Infection fatality rates around 0.5% have been estimate for a while, but there is such widespread reporting of case fatality rates (which are 2%-10% in most places) that its been very confusing.

    A particular confusing issue is that even S. Korea with its very extensive testing program has a case fatality rate of 2%. Does that mean that they *still* missed 75% of their cases? Possible, but surprising.

    A similar mystery is the number of asymptomatic cases, which the CDC estimate at 35% but other studies ha

    • You are missing a very important factor that sways the fatality rate by an order of magnitude - the very best medical care versus zero care at all. People in third world countries, and Americans without insurance will often get zero care and even if they just need an inhaler, people can die without one. So far the results are looking like a ten fold increase+ in mortality if you don’t get so much as a single inhaler, bottle of oxygen, or even proper body positioning to outlast pneumonia symptoms.
      • Not disagreeing, but have you seen studies on the change in fatality rate with medical care. I thought venetlators didn't make a huge difference, but other care may. I haven't seen good data but am interested if there is some.

        Its a critical question for policy: how important is it to prevent hospital overload?

        • There have been no studies on this explicitly that I’m aware of. However, it does seem to be a general consensus that about 10% of cases wind up needing critical care and very few critical care patients would survive without it. That along with a handful of severe cases probably not making it either without care (with care severe cases are close to 0% mortality) and a base fatality rate of about .5-1% lead to that conclusion.
  • Told you so? (Score:3, Interesting)

    by Kokuyo ( 549451 ) on Thursday May 28, 2020 @02:57PM (#60117040) Journal

    You think anybody's gonna appreciate me going "I told you so"? No? Okay...

    Seriously though, the alst weeks were VERY frustrating to me. The problem wasn't even that people didn't agree with me but how they disagreed.

    To be put in the same league as a flat-earther by people whose opinion clearly is based on outdated science was vexing... to put it very mildly.

    Mentioning that my facts came from professionals usually resulted ad hominems towards these people (without actually consuming their information directly). In short, people would automatically assume I was either dumb or willfully ignorant on the basis that my opinion didn't match theirs.

    It was like talking to hardcore Christians.... I shudder to imagine a society in which I would have had to fear actual repercussions for my audacity.

  • Even the CDC is saying at this point antibody tests are giving about a 50% false positive rate [cdc.gov]. This is because of many reasons, but Tl;dr is the specificity is about 95% and with less than 5% of the general population having it means there will be about a 50/50 split in true and false positives. The test becomes more accurate the more people have it. This dosent even address the fact some antibody tests can show false positives for the wrong coronavirus strain, and in some cases is detecting if the pa
    • by dgatwood ( 11270 )

      Even the CDC is saying at this point antibody tests are giving about a 50% false positive rate [cdc.gov]. This is because of many reasons, but Tl;dr is the specificity is about 95% and with less than 5% of the general population having it means there will be about a 50/50 split in true and false positives.

      No, the CDC did not say that. The paragraph in question starts with the words "For example". Those numbers are entirely made up.

      For example, in a population where the prevalence is 5%, a test with 90% sensitivity and 95% specificity will yield a positive predictive value of 49%. In other words, less than half of those testing positive will truly have antibodies. Alternatively, the same test in a population with an antibody prevalence exceeding 52% will yield a positive predictive greater than 95%, meaning that less than one in 20 people testing positive will have a false positive test result.

      The current tests do NOT have 95% specificity. Several tests exceed 99% specificity, but some tests have as low as 84% specificity. In other words, you have no idea how trustworthy the numbers are without knowing which test was used. The ones with 84% specificity are basically useless, and the FDA really should not have approved them. The ones with 99% or 100% specificity are

      • The CDC used it as an example because it reflects the reality of the current situation. Even with 99% specificity, that yields results quite above what’s real at this early point. Thus the tl;dr the current antibody tests are useless at this point. And yes, the low specificity ones that are detecting other coronavirus strains are a problem, exactly as I pointed out in far less words because you are going to lose more people to that than a pandemic.
        • by dgatwood ( 11270 )

          Well, you actually said that the 95% didn't count the tests showing false positives, when in fact, that's exactly what it counts. :-) But no, the current tests are not remotely useless, unless you lump the really bad tests into the mix, which again, should never have been approved, and absolutely should not be used.

          At 99% specificity, 99% of the people tested who do not have the disease will come out negative. If we assume that about 10% of people have had the disease (a good guess, at least in cities),

          • You didn’t mention sensitivity, a test with 100% specificity may only have 80% sensitivity making your comment way off. No test has 100% in both. It’s quite likely we are way below 10% nationally. Without knowing accurately the sensitivity and the specificity of the tests given (which we don’t know as many have not even been approved) then we can’t know. It’s probably closer to 3-5% in which case the specificity is required to be quite high to be useful. There is at least [washingtonpost.com]
            • by dgatwood ( 11270 )

              You didn’t mention sensitivity, a test with 100% specificity may only have 80% sensitivity making your comment way off.

              The difference is that for an antibody test, sensitivity is likely to be fixable by testing people more than once. For a specificity problem, you're likely to be wrong over and over by detecting the same wrong antibodies.

              But you're right that if the sensitivity is only 80% and the specificity is 99%, then at 10% infection, you'd show about 9% infection instead of 10%, because the specificity and sensitivity errors go in opposite directions.

              No test has 100% in both.

              Actually, the Epitope IgG test claims 100% specificity and 100% sen

              • The National center for biotechnology information is indicating rates way off from what you are describing [nih.gov]. Do you have a source? I suspect every company claims near 100% specificity until it’s found out it’s really 88.7%.
                • by dgatwood ( 11270 )

                  source [evaluate.com].

                  • Thanks. The epitome test you mentioned is being reported at different rates, I’ll lend more weight to the independent organization unless something else comes up. Given that your source claims about a 5% infection rate in America a month ago, it’s unlikely we are near 10% and the accuracy of the epitome test you mentioned earlier is looking like it’s going to be giving over a 60% possibly even much higher false positive rate. The article also says we need independent verification even t
    • by ceoyoyo ( 59147 )

      Positive predictive value is not a synonym for false positive rate. You are better than most though, the usual statement is 50% accuracy.

      Your link doesn't say anything remotely resembling what you've claimed. As far as I can see it doesn't give any real numbers at all, and doesn't even make up any for false positive rate.

    • Even the CDC is saying at this point antibody tests are giving about a 50% false positive rate.

      Seems like a reasonably easy thing to correct for. If the numbers are as you cite, divide your measured result by two. I expect this is the sort of thing they teach in Epidemiology 101.

  • I am surprised anyone who has been following this is surprised. It is fairly well know that US testing was lacking (I know a few people who had symptoms but weren't allowed to be tested, and only one person who was able to get tested and was positive. Anecdotal, I know) so many cases went uncounted against the total, let alone the asymptomatic cases. Heck, even in a central US state, Indiana, antibody testing showed roughly 2.8% of the population had antibodies or positive for live coronavirus. ( https://ne [iu.edu]
    • So at 2.8% infected and 70% for herd immunity, Indiana can look forward to 165,000 new, diagnosed, cases and 11,000 new deaths. If the virus transmission gets worse with colder weather (and it probably does at least a little), this could be a very challenging winter for Indiana health care.

      • Might want to check your math there, buddy. Even with the high end IFR of 1%, 165,000 new cases would yield 1,650 deaths.

        • Indiana has had 1850 deaths so far. If you multiply that by (70/2.8) in order to get to herd immunity levels of infection, you get 46,250 deaths.

  • According to the CDC, the antibody test may be wrong 50% of the time [cnn.com] making any conclusions at all unwarranted.

    • by ceoyoyo ( 59147 )

      No. Stop saying that. That CNN article is an excellent example of how a lack of basic statistical knowledge can get people killed.

      • by sjames ( 1099 )

        Read this then. Jamming your fingers in your ears and yelling LA LA LA is another good way to get people killed.

        • by sjames ( 1099 )

          Sorry, I must have screwed up the link [medrxiv.org]

        • by ceoyoyo ( 59147 )

          From your link:

          Among specimens from SARS-CoV-2 RT-PCR-positive individuals, the percent seropositive increased with time interval, peaking at 81.8-100.0% in samples taken >20 days after symptom onset. Test specificity ranged from 84.3-100.0% in pre-COVID-19 specimens.

          They appear to have only tested PCR positive specimens so they cannot comment on sensitivity, and therefore not on accuracy either. So the article cannot support your 50% accuracy claim. On the specificity side, they cite 81.8-100%, so I'm n

    • According to the CDC, the antibody test may be wrong 50% of the time [cnn.com] making any conclusions at all unwarranted.

      Not at all. If they know half the positives are false, just divide the measured rate by two and you've got the real rate. I'm sure anyone doing this sort of study knows how to do this correction.

    • by dgatwood ( 11270 )

      According to the CDC, the antibody test may be wrong 50% of the time [cnn.com] making any conclusions at all unwarranted.

      See my comment above [slashdot.org] and the thread nested under it. That was an example with made-up numbers, not a statement about the current accuracy of tests. It was based on a really, really bad set of numbers that are far below the current state of the art for COVID-19 serological tests.

  • At least 10 times worse than it had to be.
  • Old news (Score:4, Informative)

    by WaffleMonster ( 969671 ) on Thursday May 28, 2020 @03:59PM (#60117338)

    IFR and CFR are different things. Saying that one number does not equal another is not news.

    IFR range from half to 1% has been arrived at by multiple studies going back months including Ferguson's from way back in mid March.

  • uh... olds ? (Score:5, Insightful)

    by Tom ( 822 ) on Friday May 29, 2020 @12:11AM (#60119246) Homepage Journal

    How is this news? Over here in Europe, the 0.5% to 1% fatality rate has been a familiar thing for at least a month.

    In fact, a bunch of the "we're against anything, just because" crowd cited that number in support of their protests (ignoring that 1% of 700 million is still 7 million).

    If you check the numbers at https://www.worldometers.info/... [worldometers.info] you notice that the severity rate (2% of cases are in serious condition) and the fatality rate (10-12% of cases end with death) have been rather constant for a very long time now.

    The numbers are pretty much in. Where's the news?

If all else fails, lower your standards.

Working...