Hacker News Reader

Comment by jbotz | Hacker News Reader

Since about a week ago data has been coming in from antibody testing in Europe... and the results are that IFR (infection fatality rate) is clearly less than 0.5% (0.37% in one city in germany[1]) and the number of actually infected people at least 15 times to 30 times higher than the confirmed number of cases.

[1] https://www.n-tv.de/wissen/Heinsberg-Studie-entraetselt-Coro...

View on HN

Note though that the study, its methods and especially its politically charged presentation by repeat social media offenders are still disputed [1]

[1] https://www1.wdr.de/nachrichten/offene-fragen-heinsberg-stud...

Today there's been a new result from antibody tests in the netherlands done on blood dononrs which showed about 3% infection rate. Extrapolated to the population that would imply 500k infections, 20x higher than the official stats.

The Heinsberg study isn't the only source pointing to this. Almost a month ago the NYT also ran a story on randomly tested Iranian airplane passengers which suggested that as of that time there may already have been 500k-5mil infections in the country.

https://www.straitstimes.com/world/europe/dutch-study-sugges...

Possibly dumb question, how do we know someone with antibodies was exposed? The whole idea of vaccination is based on multiple things producing the same antibodies, so isn't it plausible there could be related viruses or genetic factors or something that would lead to antibodies for some people that hadn't been exposed?

Yes, there is the chance of false positives with antibody tests, in particular for other strains of Corona, but scientists can do a sample of antibody tests on people they know to have been infected with covid-19 and get a rate for false positives and take that into account.

Netherlands is currently at 30000 cases with 3000 dead, cfr of 10%. 20x infection would put the IFR at 0.5% which is well in line with what is expected

I'm not so sure that's clear yet. Just going on blunt metrics, the virus has already killed 0.1% of NYC's entire population, probably on pace to hit 0.2% if the down-curve even only roughly matches the up-curve. So it would have had to have spread to 50% of the population for the fatality rate there to be only 0.4%, which seems unlikely (there's only a smattering of antibody data around the world so far but nothing has found spread even close to 50%, especially at the level of a metropolis)

When you include 3,700 non-positive deaths in your total death count, yeah - the numbers look scary.

The reporting we're seeing is CFR, not mortality rate. They are very different things.

Mortality rate is the number of deaths caused by a disease.

CFR (Case Fatality Rate) is the total number of deaths by those diagnosed with a disease.

They are very different numbers, and have very different meaning for different analysis and methodologies.

Saying that that virus has killed .1% of NYC's population is inaccurate.

Even w/o those non-positives, the death count for NYC is in the neighborhood of 8,000[0] (and still rising at a decent clip) for a population of 8 million. That's 0.1%.

You can also look at individual counties[1] and you get some variation but the same picture. Ex. 2.7k deaths in Queens for a population of 2.3mil, 0.4k deaths in Richmond for a population of 0.4mil

(That's 0.1% for the entire population and not even nearing the maximum yet, so obviously the IFR's and CFR's will be even higher.)

[0]https://www1.nyc.gov/assets/doh/downloads/pdf/imm/covid-19-d... [1]https://covid19tracker.health.ny.gov/views/NYS-COVID19-Track...

also while this may sound insensitive 0.1% of the population is not the same as X percent of old and sick people

it is not like the virus takes down everyone equally. The overwhelming majority of those affected are both old and fairly sick already.

Again I am not implying anything other than there is clear skew and the measures should target those that are at risk. It makes no sense to create generic, unworkable rules where we lock up all 20 year olds that have no risk whatsoever. It is not realistic.

There’s also the problem that healthy young adults are generally very stupid when it comes to assessing risk. Sure, they’re probably only at mild risk of days of debilitating hospitalization, but they’re also going to be fairly careless about people at greater risk around them.

Locking everyone up gives us a chance to gather data, prepare treatments, and find a way to protect those at risk. Letting spreaders run around carelessly will not make anything but the economy better.

So with a 1% IFR, that would mean 20% infected. Would you agree with that number? Has anywhere else found 20% infected at the level of a metropolis?

I think 20% is definitely in the range of plausibility. See 15% infection rate in a set of tested pregnant women in NYC[1]

I think the preponderance of evidence currently points to an IFR of 1% or maybe a bit lower.

[1]https://www.nejm.org/doi/full/10.1056/NEJMc2009316

Yes, very interesting, NYC will certainly be an important source of data for the foreseeable future.

This is massively informative. If anything pregnant women take a lot more precautions than other people. They are most likely under representing the overall infection rate.

If people on here are so keen on calculating CFR and not R0 to estimate how bad the virus is, they should also take into account logistic and demographic problems of every area. Hence, the CFR of New York might be so bad because of terrible medical infrastructure in the first place, and additionally patients who are not willing to go to the hospital because of fear of bankrupting themselves. Certainly not saying this virus is "just like the flu" because it is not and it is killing a lot of people. But epidemiologically this is much more complex than all of us are able to understand as non-epidemiologists.

Also, here's are actual estimations of R0: http://www.dkriesel.com/corona

As another data point in the Netherlands (officially 30k cases - only serious cases and hospital workers were tested, 30% of the tests are positives), actual seroprevalence is 3% ~500k people have been infected (~2 weeks ago), dividing the excess deaths, you get 1.3% mortality rate.

https://twitter.com/jankluytmans/status/1250769486174654464

That study has been widely critized as there a many false positives with this kind of test and the counting method seems flawed.

Reply on news.ycombinator.com