Early USC COVID-19 Antibody Tests Shows Safer 0.1-0.3% Fatality Rate

Based on the results of the first round of COVID-19 antibody testing, the USC research team estimates that approximately 4.1% of the LA county’s adult population has an antibody to the virus. Adjusting this estimate for the statistical margin of error implies about 2.8% to 5.6% of the county’s adult population has an antibody to the virus — which translates to approximately 221,000 to 442,000 adults in the county who have been infected. That estimate is 28 to 55 times higher than the 7,994 confirmed cases of COVID-19 reported to the county at the time of the study in early April. The number of COVID-related deaths in the county has now surpassed 600.

This is about the level of fatality that was expected from estimates posted at Nextbigfuture in March. Nextbigfuture noted in March and notes again that having a 0.1 to 0.2% would put the fatality rate near the 0.1% of seasonal flu. However, COVID-19 still can get four times worse because ICUs getting overrun. The people who get extremely sick is far higher than flu. We need to protect vulnerable people and increase hygiene and improve procedures. The economy can be carefully opened but with masks worn by everyone, hand cleaning, surface cleaning and temperature checks and mass testing.

More on USC/L.A. County testing
With help from medical students from the Keck School of Medicine of USC, USC researchers and public health officials conducted drive-thru antibody testing on April 10 and 11 at six sites.

Gottlieb no Re-opening

We should have more activities outside. Having outdoor events is safer.

Antibody tests have higher false positives. They are good for general population studies, but they are not a tool for accurately determining who is safe to return to work.

Written By Brian Wang, Nextbigfuture.com

67 thoughts on “Early USC COVID-19 Antibody Tests Shows Safer 0.1-0.3% Fatality Rate”

  1. As, as an adendum- Sweden instituted no lockdown and they are approachinged levels of herd immunity [60%], with a mortality rate of only 2,000. So 6 million people infected and 2,000 dead.

  2. Causes of death are rooted in the underlying effect.
    If you were shot 30 years ago and your spinal cord severed and you are rendered wheelchair-bound, developed decubitus ulcers from that and die of sepsis from that then the cause of death is….the gunshot wound, manner of death is homicide. Because it caused the underlying chain of events.

  3. My last sentence could be taken in a way not intended. I was only saying that if the virus only attacks some small percentage, it would not be raging through nursing homes sickening dozens and dozens.

  4. That is my point. Saying the cause of death is not the virus, when they have other health issues but the bug took them out, is nonsense.
    People often have more than one thing wrong with them, but generally one cause puts them in the morgue when it happens.
    And this is not some bias just given to this bug. A relative had diabetes and leukemia…but it was the leukemia that killed him. Had he not had the diabetes and not been overweight maybe he could have had another course of chemotherapy, but that does not mean he died of diabetes.
    Obviously, if patients couldn’t get enough oxygen because of the virus, the virus did them in.
    If you don’t look at the immediate heath disaster in a person and look to some underling problem, where does that end? Is cause of death…birth?

  5. From what I’ve read, this virus is contagious while asymptomatic. Actually, if there’s a lot of undetected and asymptomatic cases, that more or less has to be the case. So apparently, just breathing is enough with this one. (And note that the denser an area is, the more infected people can be breathing near you simultaneously.)

  6. But the problem is more subtle than that.

    Dying of heart disease and heat stroke in the desert, Mindbreaker comes along and, being Thursday, he shoots you in the leg. Now there is no way you’ll ever be able to walk to safety and you dehydrate until the blood viscosity triggers a heart attack and you die.

    Was it heart disease? Dehydration? Being shot?

  7. You don’t necessarily need an autopsy to determine cause of death. One of the symptoms of covid is ARDS so your lungs would appear whited out on Xray- among other diagnostic tests that they can put.
    That being said, it would be a matter of which you want to put on line one under Cause of Death: Acute Myocardial Infarction or Complications of Covid-19; the other would go under line two.

  8. I sort of agree. That was probably the single biggest failing-aside for not closing the borders asap: trying to get widespread testing in place.

  9. If you are on the verge of dying from heart disease, heat stroke, drowning, whatever- and someone comes along and blows your brain out a minute before you would be declared brain dead then your cause of death is gunshot wound of the head. MAYBE the above would be other significant causes of death- MAYBE.
    No medical examiner would sit there and argue about medical causes of death with a gaping gunshot wound to a vital organ.

  10. So if I saw you in the desert, dry canteen, blazing sun, stumbling, deeply sunburned, 5 miles from the QuickyMart headed the wrong direction, it would be just fine if I shot you dead? You were going to die anyway. So I did not really kill you. It was the thirst, you see.

    Those with high blood pressure, diabetes, and obese might well be heading the wrong direction stumbling left and right. That does not mean they can’t change their lifestyle and correct these issues.

    And I have a surprise for you. Those with diabetes who are taking Metformin outlive people who are non-diabetic: https://www.ncbi.nlm.nih.gov/pubmed/25041462

    The reality is that 150 people normally die each day in New York. 661 positive for the virus died today. And the hospitals are reporting that the usual fatalities are reduced. Kids are not drinking the Prestone…because mom and dad are home looking after them. Freddy did not fall off his motorcycle racing between lanes on the highway. Burt did not drop a pallet of Nintendos on Cliff, Maria did not get food poisoning eating enchiladas from a vendor pushing a cart down Water St., Samson did not get hit by a taxi guessing when the “Walk” light will light…

  11. By that logic those that got the virus on the cruise ship should have just been thrown overboard, the instant that was verified. Any airplanes flying in from China shot down.

  12. Probably both since this thing is supposed to be highly infectuous. Having worked at an ME’s office, I can say that no medical examiner would attribute a cause of death by Covid without an actual, confirmed and ongoing disease; it would be like saying someone died of a gunshot wound of the head without any gunshot wound preent. Doesn’t mean a coroner or hospitalist wouldn’t do it- but not most ME’s.

    What may confound this, however, is what they are counting towards Covid deaths. Specifically, the other part that goes under a cause of death: “Other associated conditions”. Maybe someone had severe, chronic conditions, developed infection by Covid, tested positive but got over it and then died weeks after the fact?

    I don’t know.

  13. “False positives have a huge effect if you don’t have many infections. If you have 3% false positives, 3% false negatives, and 0.1% actual infections, then false negatives are 3% of 0.1% and false positives are 3% of 99.9%. You multiply your apparent infections by about 30X.”
    Which is why you use tests that have a high sensitivity for screening in low infected areas; you use high specificity tests for confirmation. False positivity affects one and not the other.
    So you shouldn’t be citing the specifity but the sensitivity in such screening exams as you are trying to rule infections out.

  14. That conclusion is what the best preliminary evidence I’ve seen so far supports. And the USA health system can’t handle more than about 6000 deaths per day from it. So without adjusting for a treatment or prevention breakthrough and assuming you go 4500 per day to be conservative, we have another 100 days or so of quarantine to go in the USA.

  15. Until there’s an autopsy. There are a lot of reports around of death certificates from different countries with “died of C19” and “died with C19”. They are counted just the same.

  16. If it’s true then it has basically burned itself out in the New York metro area and that’ll become pretty apparent in two weeks or so, immune tests or not.

  17. Iatrogenic deaths in Italy were attributed to coronavirus. Vastly overstated death rate. That is what an overwhelmed and panicked health care system gets you. Ventilator treatment is not benign. It is dangerous as hell to be on a ventilator.

  18. This is no good. Sampling errors are going to be huge. They contacted people and offered to test them. Obviously, people who had had something of some kind or were in contact with people who were sick would be more curious than others. You can’t force a random selection of people to take the test.
    Anything less than half of those contacted getting tested, I would consider extremely dubious. They contacted people with email. Who responds to emails from people other than their friends and businesses they have dealt with? And there were all kinds of scams about the virus. Everyone was warned not to respond.
    Smart people are not going to respond. Did they check these people’s IQs? Obviously not, but somehow I think less bright people are more likely to be infected. Not a prerequisite, just a probability thing. They are not going to follow the good advise or will do it inconstantly or wrong.
    And we have seen people in the nursing homes get it by the dozens.

  19. Iceland is 0.19% last I checked though?

    That’s the estimate based on the random testing. Are you just taking deaths divided by confirmed tests? Because that isn’t really a valid number when you have a good random sample to estimate how many people likely have/had it but haven’t had a test.

  20. Unless deaths are being falsely attributed to Covid or there are a lot of comorbidities in that city that falsely elevate the number of deaths from Covid.

  21. Depends on the sensitivity/specificity of the test and what you are testing.
    Some false positives are for a specific reason. Like testing for syphilis with RPR can be falsely positive with certain autoimmune conditions like lupus[lupus antibody]; it doesn’t matter how many times you test, it would be positive unless the titer just drops really low.

  22. Most of these people have died with other comorbidities on board as well. So the way the death certificate puts it is
    Cause of Dead: COVID
    Other Contributing Conditions: Heart disease[or whatever]

  23. If Samoa had it, there would be a lot of dead. They have all the wrong health features. 1/3 have diabetes. 47% are obese. Many have high blood pressure…
    We would have heard about it, if they were dropping like flies.
    Hmm over 35 years old obesity is over 70%.
    Then there is Nauru where 97% of men and 93% of women are overweight or obese.

  24. “A new car built by my company leaves somewhere traveling at 60 mph. The rear differential locks up. The car crashes and burns with everyone trapped inside. Now, should we initiate a recall? Take the number of vehicles in the field, A, multiply by the probable rate of failure, B, multiply by the average out-of-court settlement, C. A times B times C equals X. If X is less than the cost of a recall, we don’t do one.”

  25. Finally, turning a corner with this news story. Months of dismal projections – now clear evidence that it is basically as deadly as the flu. Sure, people are getting very sick, like my brother, but still – many, many more don’t.

  26. Here’s a simplie comparison a lot of people will make:

    If the death rate is 0.2% with a slow ramp in cases but 0.5% without, and you’re talking about 10 million people, 30,000 lives could be saved.
    If you say the quality of life is 20% worse for those 10 million for a period of one year, and each life is 50 life-years because it mostly hits the middle aged and elderly….
    2 million life-years of quality lost divided by 50 equals 40,000. That implies you could run it for 9 months or so before it isn’t “worth it”.
    This ignores the economic impacts that being poorer means you don’t tend to live as long. I think most voters are actually ignoring that “opportunity cost”.

    EDIT: reduced air pollution is also saving a bunch of lives…

  27. Virtually everyone who is tested for Covid 19 is tested because he has symptoms. Consequently, virtually everyone who is asymptomatic and has Covid 19 who dies of a heart attack is not counted as having died of Covid 19.

  28. Well if it is 200 jobs times 3 months that is 50 years. For one life. Seems like they’re hewing to the dismal math pretty well so far.
    Keep in mind the value of averted tail risk of not ending up like Lombardy – which is basically 100% if you do nothing. Does that double the death toll?

  29. another point – I highly doubt their findings from just another tack. Say that their death rate is correct, namely .2%.
    That means that ALL of New York is infected with sars-cov-2. They’ve had 15,000 deaths there alone, so far. 15000/.002 == 7.5 million.

    I strongly suspect that the researchers WANT this result to be true more than they are being faithful to the scientific method.

  30. I would take this study with a gigantic grain of salt.
    First of all, I can’t find the actual study. I see the summary of it, but I can’t actually go in and read the damn thing. Please point me to an actual source next time (USC).

    Second, this study shares the same problems with other antibody studies – namely non-specificity issues and test-reliability issues.

    1. non-specifity issues: the immune system is a messy thing, and it may just be that the antigen of the test triggers the immune system even THOUGH the person has NOT had sars-cov-2 at all. It may be that a small percentage of us are happily immune, or that some antibody in people’s bodies is confusing the test.
    2. test-reliability issues. all tests like this have large reliability issues, ie: they give false positives for a large number of cases. with a phenomenon that registers at less than 1%, bayes theorem comes into play in a big way.

    Say the actual percentage of cases is .001%, and the test itself gives 3% false positives. In that case, the percentage that you measure is 3.001% – ie the number of false positives far overwhelm the number of false negatives that you get.

    In any case without looking at the paper at all, we don’t know the methodology used in the USC study. But the tests in a similar Stanford study had an accuracy of around 87%.

    So I really wish that I could see the actual study. You can use 2 or 3 or 4 tests to narrow #2 down, but #1 is a big possibility.

  31. Preliminary evidence from Italy and New Orleans and New York is that death rates are higher in an overwhelmed hospital system. So those estimates are on the low side of deaths averted based on that very possible scenario.
    The flip side of that is most of the victims were relatively old, so the dismal math says each death probably isn’t equal to life-years of the average population member.

  32. You’re making the assumption that if you test a certain sample it will have the same false positive twice. That is rarely true.
    False positives you can deal with by testing twice. In fact many routine medical tests have only a 70% success rate so they are run more than once. In a large sample like this you could retest say 15% of them and get a very good idea.

  33. You can only really use the 0.1-0.3% number if there is zero stress on the healthcare infrastructure like in Los Angeles. Based on Italy it can easily blow out by 5x.

    EDIT: but yes, if it doesn’t turn out that New Yorkers are at like 33% positive for the antibody test, it would be time to revise the theory.

  34. “0.1% of seasonal flues” is not accurate language. 0.1% of what ? All the people that are tested with antibodies of flue ? In that case OK.
    But same probably happens with flu. Flue cases are under-reported and 0.1% is consequently on the high side.

  35. A CNS infection is unlikely anyway, because of the blood-brain barrier. But other peripheral nerves may be affected, if this is indeed neurotoxic and assuming the virus can reach those nerves. Obviously, it can reach the nasal nerves much more easily than other places.

    For the sense of smell, as soon as your nose is partially blocked (or just secretes more than usual), the sense of smell can already be affected just by the physical obstruction. It doesn’t have to be neurological at all. And there were experiments that showed a dependency of the sense of taste on the sense of smell. So if you loose your sense of smell, your sense of taste will be affected too, though probably not to the point of total loss.

    However, in this case, nasal congestion is 3 to 6 times less common than loss of smell, which does suggest a neurological cause.

  36. But there comes a point when the cure is worse than the disease, right? Here we are talking about suspending the activities of 200 people for some months for a grand total of 100 person-years lost.

    What if we were talking about 10 000 people, or 100 000 people or a cool million. Do you still think that it would be a reasonable trade off?

  37. There are other ways of estimating the death toll. The incidence of covid-19 is sufficiently high in some area to count the excess deaths. In the county of Stockholm, the year 2020 has been slightly lower in terms of daily deaths compared to an “average” year for the period up to about 1st of April.

    Looking at the deaths over time it is quite clear that there is a remarkable increase coinciding with the onset of official covid-19 deaths. But, the official count is twice lower than the increase in mortality. I.e. twice as many people die due to covid-19 than is officially noted.

    I have not compared to the estimated numbers of people in Stockholm county that have been exposed, since the antibody tests were faulty. But I am pretty sure it will end up significantly higher than 0.2% mortality given that more than 0.1% of the *total* population of Stockholm county have already died from covid-19…..

  38. I recently read an article about the different strains of coronavirus going around, and apparently there is at least some evidence (still early) that the strain in New York is far more virulent than what dominates most of the rest of the US.

    In the 1918 flu, the second wave was far more deadly than the first, after it mutated in the dense, unhygienic trenches to become more virulent. In conditions where people with stronger symptoms are isolated, evolution favors milder strains. In conditions where people come in contact anyway, evolution favors a faster dividing, harder hitting strain. In the modern day, crowded public transit systems may be to blame.


  39. ” (For the latter, he says “do you think loss of taste and smell is the only things that gets damaged?”) ”

    Loss of taste and smell are actually pretty common in viral respiratory infections. You’re talking about nerves that are right out on the front line, so to speak. Just losing your sense of smell with a respiratory infection doesn’t automatically mean you’ve got a CNS infection, too.

  40. You understand, the death rate from a viral infection is somewhat contingent. If you barely get exposed to enough of the virus to initiate an infection, it starts small, and there’s an excellent chance your immune system response will have time to stomp it before it gets serious. If somebody coughs straight into your face as you’re inhaling, and you get a huge dose of viral particles, the infection is accelerated, and can grow to overwhelm you before your immune system has time to respond, even if you’ve got a top notch immune system.

    It’s quite possible that people living in low population density areas typically get infected by a marginal dose of viral particles, and end up asymptomatic, while a lot of people in disease incubator NCY are getting hit with large viral loads that lead to severe cases. You can’t just treat heterogeneous areas as though they were uniform populations, the local circumstances matter a lot.

  41. “It’s almost as if some counties only have 3% of the infection rate of other counties.”

    Probably true. Cities are petri dishes for almost any sort of pathogen, and NYC is almost systematically designed to accelerate the spread of respiratory illnesses. NYC policy has only aggravated the problem.

  42. Shouldn’t which camp you’re in depend on the individual circumstances? I mean, otherwise healthy people die of heart attacks, so it’s quite possible for somebody with asymptomatic Covid 19, (Something like 80% of the people who test positive are asymptomatic.) to up and die of a heart attack unrelated to the mild infection.

    I suppose this sort of forensic pathology is too much to expect in the middle of a pandemic, but there are probably a lot of undercounts and overcounts.

  43. A very low fatality rate would be great but it does have a downside. Given the number of deaths we’ve seen in such a short period of time, low fatality would mean a very high R0 (the average number of people each person infects).

    The fraction of people who have to get infected before you have herd immunity is 1 – 1/R0. The R0 without social distancing has been thought to be between 2 and 3, though one study claimed 5.7. If it is actually 5, then 80% of Americans would get infected.

    At a death rate of 0.2%, that would mean 330M * .8 * .002 = 528,000 Americans die before it’s all over, unless we lower the effective R0 with some combination of social distancing, test and trace, and finally a vaccine.

  44. In addition to those additional concerns you mentioned, there possibly is another effect I don’t see mentioned very much. I have seen a couple of articles that say there is some evidence that the virus in some way diminishes the ability of the red blood cells to transport oxygen. I don’t know whether it is by attacking hemoglobin directly, or by some other means. I have only seen this mentioned a couple of times. I do not know whether the information is reliable. I don’t know whether it has been investigated properly.

  45. You have to determine how accurate it is. Go to Samoa or some other place that had zero cases and test 5k people. Also test 5k that you know had it. Without that, it is just guesswork.

  46. This study used the same test as the Santa Clara study, with a specificity between 90% and 95% according to the manufacturer. That means a false positive rate of 5 to 10%. They could have gotten the same results with zero actual infections.

    The Santa Clara study said they checked and found a max of 1.7% false positives. Let’s say we believe them. Trouble was, their actual measured infection rate was 1.5%. They “adjusted” it by demographics; e.g. they tested ten Hispanics, one was positive, so they said 10% of all Hispanics were positive. Obviously that amplifies random variations, and it gave an adjusted rate three times higher. After that they subtracted the false positive rate they thought most likely. It makes no sense to do this after the demographic adjustment.

    False positives have a huge effect if you don’t have many infections. If you have 3% false positives, 3% false negatives, and 0.1% actual infections, then false negatives are 3% of 0.1% and false positives are 3% of 99.9%. You multiply your apparent infections by about 30X.

    New York is starting a large antibody survey. If the death rate really is so low then they should get an infection rate far exceeding the false positive rate. Let’s wait on that before we get all excited.

  47. Just normal exhaling produces fluid droplets that can potentially carry the virus if it’s present in the airways. Not as much as coughing though.

    (re-posted in the correct sub-thread)

  48. Putting aside the CFR for a moment, there was a post by a medical professional (an epidemologist or virologist – I don’t remember) that there are other issues with this virus that don’t get enough attention. There is supposedly evidence of liver toxicity, heart muscle toxicity, and neurotoxicity with this virus. (For the latter, he says “do you think loss of taste and smell is the only things that gets damaged?”) And some patients supposedly have long-term lung damage. He also raises concerns about long term developmental effects in children who got infected, even those that have no short-term symptoms.

    I have no idea how much truth there is to any of that, but if true, this isn’t just a respiratory virus. There is more to it than CFR. Something to keep in mind.

  49. “we should have 20 million people in New York already infected.”
    Well I suppose that’s theoretically possible, though puzzling, I did a bit of math and the 8 counties with the highest death rates have a total population of 10.77 million with a death rate across them of 1223/million. All the other counties combined have a population of 8.73 million, but a death rate of only 182/million only 15% as high, I’m not sure how it can be that with everyone in NYS infected there’s such a huge difference in mortality rates between counties.
    And of course the differences between individual counties is even larger, in Queens (population 2.2 million) it’s 1513/million but in Onondaga (population 467,000) it’s only 42/million. Why is the death rate in Queens 36 times that in Onondaga if everyone’s been infected? It’s almost as if some counties only have 3% of the infection rate of other counties.

    *The total for NYS was only 14,828 deaths on the NYS Department of Health site I was using, this number being lower than public state wide numbers will be due to delays in processing the data – assigning county of residency to the deceased.

  50. Would you have crashed if you didn’t drive over the grease patch ?

    I get what you are saying though. If a person has a heart condition then contracts COVID19 and dies. You say he died of a heart attack. Others say he’d still be alive if he hadn’t contracted COVID19. I’m in the second camp.

  51. This contradicts all the large-scale data we have seen from countries with widespread testing regimes – Germany (3%), New Zealand (1%), Iceland (0.8%) etc. Interesting to see what the peer review throws out.

  52. If the fatality is really 1/1000 then given the 20,000 already dead in New York, we should have 20 million people in New York already infected. That works out to every New York resident. Hmmm… you think maybe it is wrong? Another way to consider this is that the false positive rate of the test may be 2%. Testing it on people who don’t have the disease at all will then yield a 2% prevalence rate?

  53. Brian, you don’t seem to emphasize enough that there has been a lot of criticism of the methods and interpretation of these results by people who are experts in such matters. Just two lines at the very end of the article, and not very clear about it.

    Maybe the results are a good indication of the actual rate of infection, but maybe they are not. We simply do not know yet. You should have emphasized that there is a lot of disagreement about the interpretation of those results.

  54. the fatality rates.

    At the moment anyone dying with COVID19 is said to have died of COVID19. This is total BS.

    Here’s an analogy for you.

    The headlights, brakes and steering in my old car don’t work very well. While driving my car at night I don’t see a grease patch on the road, lose control of my car and crash into a tree. My car is wrecked beyond repair.

    Now what was responsible for the crash? Was it the grease patch? Or the faulty headlights, brakes and steering?

  55. They should go further in their research and find out How many of those who develop antibodies were infectious at any stage. Probably most were not. Are there are two groups here, the smaller one that get infected and stays a symptomatic but still is infectious at one stage yet less than those who show symptom, and then there is another group, the much bigger one that gets exposed to very low level of the virus, develops immunity to it and does not get infectious, if it proves to be true than exposure to a low dosage of the virus by younger healthy people who are not very likely to get seriously sick from it can help accelerate developing of herd immunity at a very low cost in human lives. This question should be answered.

  56. From the ‘We’re Just Eye-balling it’ File:
    “…Using daily state-level coronavirus data and a synthetic control research design, we find that California’s statewide SIPO reduced COVID-19 cases by 152,443 to 230,113 and COVID-19 deaths by 1,940 to 4,951 during the first three weeks following its enactment. Conservative back-of-the-envelope calculations suggest that there were approximately 2 to 4 job losses per coronavirus case averted and 108 to 275 jobs losses per life saved during this short-run post-treatment period…” So that’s the statistic value of a life… huh


Leave a Comment