U.S. Navy test shows 60 percent of carrier crew have coronavirus antibodies | Hacker News Reader

U.S. Navy test shows 60 percent of carrier crew have coronavirus antibodies

reuters.com

In a previous report by Reuters, all 4,800 personnel were tested for COVID-19. 600 tested positive of which 360 were symptom-free leaving 240 that tested positive and showed symptoms. 1 person died [1].

This report indicates that of a sample of 400 volunteers that agreed to take an antibody test, 240 tested positive and 160 did not. It also indicates that 1,100 people tested positive for COVID-19 back in April, possibly an update to the first report of 600.

At this point it appears that in this mini-testing lab of an aircraft carrier, about 25% of the crew tested positive for COVID-19 and 60% tested positive for antibodies, which seems to match with the earlier results showing that 60% of the sailors were symptom-free.

[1] - https://www.reuters.com/article/us-health-coronavirus-usa-mi...

I wish they would correlate the two datasets: Of those who tested positive how many also had antibodies.

This would answer the question on if antibodies are a valid way to check for prior exposure.

> The officials, speaking on the condition of anonymity, said that about 400 volunteers participated in the serology tests, lower than the 1,000 volunteers that were sought, but enough to provide statistically relevant data about how the virus spread aboard one of world’s largest warships.

There could be a correlation between the probability of getting tested and the probability of having had COVID-19-like symptoms or known exposure. In this sense, the Navy study resembles the ones that sampled from willing shoppers in front of grocery stores. Drawing conclusions isn't very easy here.

This probably under counts the true number of infection significantly.

If the sensitivity/specificity of the test was similar to the paper below, true prevalence for the ship probably approaches 100%.

https://www.biorxiv.org/content/10.1101/2020.05.13.092619v2

> This probably under counts the true number of infection significantly

Probably not. Ships are unusually concentrated social environments.

I don't understand, wouldn't a concentrated social environment suggest very high infection rates, which aligns with the GP?

It's not just infection rates, it's also dose/severity. See my linked paper.

Yes, higher dose/severity probably means more likelihood of detectable antibodies:

https://www.biorxiv.org/content/10.1101/2020.05.21.108308v1

> These data show that systemic IgA and IgG production against SARS-CoV-2 develops mainly in severe COVID-19, with very high IgA levels seen in patients with severe ARDS, whereas mild disease may be associated with transient serum titers of SARS-CoV-2-specific antibodies but stimulate mucosal SARS-CoV-2-specific IgA secretion. The findings suggest four grades of antibody responses dependent on COVID-19 severity.

But this does not eliminate the possibility raised by the GP, that some people could be clearing even a severe infection with antibodies that are not specific to SARS-CoV-2, and would never get a positive result on a highly-specific test.

This was a sample not a full test of all sailors on the ship and it was based on serology tests as developed by April.

> about 400 volunteers participated in the serology tests

> The Roosevelt has about 4,800 personnel on the ship.

This sample is statistically significant. Not sure what the problem is?

It’s not a random sample.

In the sense that they requested 1,000 volunteers, and 400 signed up? Do you mean it's a selection bias due to only a certain kind of person who signs up for these things getting tested?

People would sign up if they suspected they had coronavirus based off of their symptoms.

This comment thread is about antibodies.

There was a different line of testing for who actually had the active virus/disease which included far more sailors from the ship:

> As of Monday, the Navy had tested 94 percent of the crew for the virus, with 678 positive and 3,904 negative results.

[1] https://www.politico.com/news/2020/04/21/navy-extends-isolat...

Yes, correlation between suspicion of infection to volunteering to testing positive. If they wanted to know how many felt symptoms they would survey them. Wouldn't jump to conclusions.

This was not a random sample, with reasonable probability of a selection very skewed towards those suspecting a higher infection risk.

I'm surprised that it's not much higher. I would have expected over 90%.

I know a person who had mild symptoms, was tested positive in the PCR test and did not have any antibodies 6-8 weeks later. He then looked into the subject deeply. His conclusion was: You can develop antibodies after infections, but not everybody does. Also: a lack of ABs does not mean that you aren't immune. After all, many people were asymptomatic or showed only mild symptoms without any ABs... There is the second, much less known immune system called cellular response, which seems to play a strong role in this disease.

I'm under the impression that not much confidence can be placed in the results of a single AB test.

Isn't it also possible that the PCR test was wrong and it was something else, like a cold?

As far as I know a false positive in PCR is not very likely, but of course it is possible.

PCR is very unlikely to give a false positive if you never had the virus. The viral RNA simply isn't present to be detected. A false positive from PCR most likely came from contamination of the test materials.

PCR is quite likely to give a false positive if you previously had the virus but have recovered, including asymptomatic cases. There are still pieces of viral RNA that can be detected by PCR, even if they're inactivated by having destroyed the viral coating and you're no longer contagious.

So PCR is pretty accurate at saying if you ever had the virus, but not so accurate at saying if you currently have it at contagious levels.

The PCR test is a very specific test. A positive test means the RNA sequence of SARS-CoV-2 was detected. A negative test, however, would not be as strong of an indicator as false positive rates are quite high early or late in the infection (https://www.biotechniques.com/coronavirus-news/news_false-ne...).

Or the antibody test was wrong. Seems most of the tests look super noisy.

I believe this was the ship where the captain tried hard to intervene in the spread and got lots of the sailors off the ship to avert the spread.

Also worth noting is that this is the CDC performing serology tests in April, when the antibody tests were not performing terribly well.

Not everyone showed up for testing. Let's also not forget that the antibody tests are proving to be less than accurate. A high number of false negatives.

As far as I remember that's the case with PCR, but not with the AB tests. They rather show a high number of false positives...

I am fully aware that this is non-scientific, but I'm in a covid 'survivors' group of ~52k people on FB.

What I'm seeing there is very few people saying they tested positive for AB and a whole lot more who say they know they had it and tested negative.

Truth is, we just don't know either way. =(

Update: This was a good post from the covid group, sounds like people also just don't develop enough antibodies to show up on tests... https://medium.com/@mrina.nair/my-familys-covid-experience-2...

That’s really weird. But it fits the pattern my friend observed. Tested PCR positive with mild symptoms but negative for ABs...

Wouldn't that imply an R0 of 10 which is much higher than generally estimated? It should stop once it reaches her immunity and possible sooner if quarantines are successfully placed and the infection dies off. 60% sounds like about what you'd expect, or even a bit high.

R0 on a ship will always be higher than in pretty much any other situation. Most ships are packed to the gills, and in a gigantic ship like a carrier it will be even worse, since most of the crew will be in poorly ventilated areas rather than on deck.

Couldn't it reach such a percentage with a R0 being lower than 10?

If your R0 is 2.5, herd immunity will make the current R value go under 1 when 1 - (1 / 2.5) = 60% of the population got infected.

While that means that the epidemic will stop growing exponentially, the virus will not stop propagating, as there will still be a non-exponential propagation of the virus (e.g. if 10% of the population is still infectious at the time herd immunity is reached -- when R becomes 1 --, we should expect at least 10% more infections). Or am I missing something?

I'm not sure if it's the case but maybe 60% is the possible maximum?

I hope I understand it correctly how germs spread. Is it like this?

Imagine an infected person infects 4 other persons, and the incubation period was 1 year. Probably everyone would get the germ.

But if the incubation period was 1 second, at a certain point 1 infected person would find less than 1 person who is infectable (not already infected / had an infection that is gone now / has had an immunity? / etc.).

I posted elsewhere, but yes, there are thresholds for transmission when enough of population has developed immunity: https://en.wikipedia.org/wiki/Herd_immunity#Mechanics

In the french air carrier CDG , 1,081 out of 2,300 were found infected

https://en.wikipedia.org/wiki/COVID-19_pandemic_on_Charles_d...

Herd immunity halts transmission after a point. The flu is between 33% and 44%. Covid-19 has some wide estimates for the threshold, between 29% and 74% according to the two studies referenced here: https://en.wikipedia.org/wiki/Herd_immunity#Mechanics

Herd immunity and r0 would be significantly different on a ship

Reply on news.ycombinator.com