Did you catch it ?

It would be more interesting if you add movement based on cell phone data.
There have been some pretty interesting analysis based on relative cell phone mobility. e.g. The first month Georgia re-opened saw a minor increase in cell phone mobility, hence even though Georgia was "open" no one was going out.

Tim

Yes that is a good idea. Indeed one of my collaborators on this paper is working on that.
 
Without a vaccine, if immunity is not for life, further "peaks" could be at least as large as the initial peak....

Future waves of infection are to be expected in the immediate future because very few individuals have yet been exposed, maybe only 5% nationwide. This has nothing to do with the persistence of natural immunity.

The persistence of immunity relates mostly to infections in the more distant future, say in a year or so. It is possible that any Covid-19 vaccine would need to be seasonal or biennial, depending on the persistence of immunity. We will only find out the answer to this after we track those who have recovered from natural infection , and the first wave of vaccinations are made and followed up. The current thinking, based on the immune response of recovered patients, is that some immunity might persist for at least a couple of years. More certain knowledge is yet to emerge.
 
Future waves of infection are to be expected in the immediate future because very few individuals have yet been exposed, maybe only 5% nationwide.

Love to know where you found an accurate “exposed” number.

That number doesn’t exist.
 
He did say "maybe" only 5%. That doesn't sound like a claim of accuracy to me. :dunno:

I took it as “might be as high as but probably less”. Versus “this number is fiction”.

But either way, it’s fictitious at this point. Not measurable beyond known cases.

Technically it’ll always be fiction. An unknowable number.
 
Future waves of infection are to be expected in the immediate future because very few individuals have yet been exposed, maybe only 5% nationwide. This has nothing to do with the persistence of natural immunity.

The persistence of immunity relates mostly to infections in the more distant future, say in a year or so. It is possible that any Covid-19 vaccine would need to be seasonal or biennial, depending on the persistence of immunity. We will only find out the answer to this after we track those who have recovered from natural infection , and the first wave of vaccinations are made and followed up. The current thinking, based on the immune response of recovered patients, is that some immunity might persist for at least a couple of years. More certain knowledge is yet to emerge.


What I was referring to by "Without a vaccine, if immunity is not for life, further "peaks" could be at least as large as the initial peak..." is the possibility that antibodies may only provide protection for a limited time period and the "spreading rate" has been reduced so that "herd immunity" isn't reached in that time period.

The thought exercise is pretty straightforward. If "spreading rate" is such that "immunity" wears off before "herd immunity" is reached, the cycle is endless until "artificially induced" herd immunity is reached via a vaccine or the virus mutates itself out of existence..

Note I have not seen anything bounding the length of time one might be immune from CV after getting it.

If it is a couple of years, then a "solution" would likely be reached.

If it is less than a year at the current "Identified case" doubling rate, likely not on its own.

As far as % "exposed" I've never seen a number but I would guess it would be greater than 5%.

Positive rate for PCR testing has been above 5% and one would have to guess each of these positives interacted with at least one other person. While there is certainly "selection bias" at play so the population of people given a PCR test isn't really random, it seems logical "exposures" would still be greater than the PCR positive rate.
 
I figure the current protests should both increase the exposure rate and help herd immunity along, at least among the young.

Rich
 
Love to know where you found an accurate “exposed” number.

That number doesn’t exist.

Random serology tests in NY state. Maximum exposure in NYC was 15-20%, depending on borough, and 2-3% upstate, about 15% overall. Additional survey testing of essential workers has continued to show similar results. Random testing with a well validated measure (very low false positives and cross reactivity with other coronaviruses) is essential for this kind of estimate. NY used an internally developed and validated test out of the Wadsworth Institute, not a commercial product, some of which were problematic. NY state data is likely an upper bound, as other regions have not been hit as hard. But more data from random testing around the nation would be very helpful for understanding the situation in more detail, and should be a priority. Unfortunately some of the extant serology testing was not random, and some was conducted with tests useless for this purpose.
 
Last edited:
What I was referring to by "Without a vaccine, if immunity is not for life, further "peaks" could be at least as large as the initial peak..." is the possibility that antibodies may only provide protection for a limited time period and the "spreading rate" has been reduced so that "herd immunity" isn't reached in that time period.

The thought exercise is pretty straightforward. If "spreading rate" is such that "immunity" wears off before "herd immunity" is reached, the cycle is endless until "artificially induced" herd immunity is reached via a vaccine or the virus mutates itself out of existence..

Note I have not seen anything bounding the length of time one might be immune from CV after getting it.

If it is a couple of years, then a "solution" would likely be reached.

If it is less than a year at the current "Identified case" doubling rate, likely not on its own.

As far as % "exposed" I've never seen a number but I would guess it would be greater than 5%.

Positive rate for PCR testing has been above 5% and one would have to guess each of these positives interacted with at least one other person. While there is certainly "selection bias" at play so the population of people given a PCR test isn't really random, it seems logical "exposures" would still be greater than the PCR positive rate.

The only way to determine population exposure is to do random serology testing. PCR tests only determine active infections, and they are extremely selection biased. Random serology testing in NY showed only about 15% statewide, but only 2-3% outside the city. Herd immunity is 55-85% depending on the exact value of R0, which is estimated to be somewhere between 2.3-5.4, I think. I'd have re-read my notes for the exact estimate range.
 
Last edited:
Here is a back-of-the-envelope calculation that is food for thought: The NY state data, as of today, is 380,000 confirmed cases. Fairly recent random serology testing conducted throughout the state suggests that about 15% of the population (19 million) has been exposed. That's about 2.85 million individuals. The multiplier is about 7.5 between confirmed cases from RNA testing and presumed total cases based on serology testing. That multiplier is informative about the fraction of asymptomatic or minimally symptomatic individuals.

If this multiplier is fairly representative, then the total exposure of the U.S. population is currently around 1.9 million x 7.5 = 14.3 million, more or less. That's about 4.3% of the US population. Of course, the multiplier COULD be different, but it would have to vary by a large factor to suggest that a large fraction of the US population has been exposed. It seems unlikely that the fraction of asymptomatic patients varies widely in the US population. If this estimate is off by a factor of 2-3 either way, it doesn't make much difference with regard to herd immunity.

Of course, what is needed is more well-controlled, random testing nationwide to be able to make better estimates of broad, national population exposure than extrapolated, back-of-the-envelope calculations. Knowing the extent of exposure is critical in knowing the risk of future transmission and waves of infection. But what little good data we have does not suggest we are near the end of this. We may be at the end of the beginning.
 
Looking at "your" 4.3% of the US population figure. Consider that the CDC estimates that there were a little more than 34 million cases of the flu during the 2018-2019 flu season (in other words, more than 10% of the US population, assuming only a small number of people got it more than once). Given all the stuff put in place to attempt to slow the spread of COVID-19, I suggest that a lower infection percentage for COVID-19 currently than for influenza would appear to be reasonable.
 
Random serology tests in NY state. Maximum exposure in NYC was 15-20%, depending on borough, and 2-3% upstate, about 15% overall. Additional survey testing of essential workers has continued to show similar results. Random testing with a well validated measure (very low false positives and cross reactivity with other coronaviruses) is essential for this kind of estimate. NY used an internally developed and validated test out of the Wadsworth Institute, not a commercial product, some of which were problematic. NY state data is likely an upper bound, as other regions have not been hit as hard. But more data from random testing around the nation would be very helpful for understanding the situation in more detail, and should be a priority. Unfortunately some of the extant serology testing was not random, and some was conducted with tests useless for this purpose.

And CDC says the serology test is “up to 50% inaccurate”. It’s a fools errand.
 
An active PCR test would indicate at some point in the future that case will have antibodies(or no longer be with us).

So while serology results would show who had it and survived, a positive PCR test would show someone that currently has it, more than likely will survive, and very likely recently "exposed" (though not necessarily infected) at least one other person.

So to calculate % exposed, wouldn't it be best to use the sum of PCR and Serology tests?

I.e. if 4 people out of 100 have a positive serology test(assuming accurate results which is of course questionable) and another 6 out of that 100 have a positive PCR result than the total "cases"that were either exposing others in the past or are currently exposing people is 10%.

A bit muddy if some of the 6 active cases got it from the 4 that had it in the past but.....
 
And CDC says the serology test is “up to 50% inaccurate”. It’s a fools errand.

There is no "one" serology test. There are many different products. Some are better than others. The main problem with serology testing for population screening is false positive rates. Tests with false positive rates above 0.5-1% will generate too many false positives to be useful for this purpose, where the rate of real positives is expected to be low. But note: false positives will OVERESTIMATE exposure. False negatives rates of a few percent are not a significant factor for survey screening when few positives are expected to be found.
 
An active PCR test would indicate at some point in the future that case will have antibodies(or no longer be with us).

So while serology results would show who had it and survived, a positive PCR test would show someone that currently has it, more than likely will survive, and very likely recently "exposed" (though not necessarily infected) at least one other person.

So to calculate % exposed, wouldn't it be best to use the sum of PCR and Serology tests?

I.e. if 4 people out of 100 have a positive serology test(assuming accurate results which is of course questionable) and another 6 out of that 100 have a positive PCR result than the total "cases"that were either exposing others in the past or are currently exposing people is 10%.

A bit muddy if some of the 6 active cases got it from the 4 that had it in the past but.....

The RNA test is a slice in time. It measures those who, for an approximate 2-week period, currently have an active infection. Serology tests reveal anyone who has ever been infected with and recovered from the disease, with a (hopefully) small addition of false positives and an even smaller number of false negatives. It is possible that immunity will fade, and some of these individuals will have negative serology tests in the future, but that does not appear to have happened yet. It is also possible that a very small fraction of individuals develop antibodies to viral components other than the ones used in testing, but validation studies suggest that is exceedingly low for most tests.The presence if IgG antibodies is the best marker we have to identify previously infected individuals. (The IgM tests, on the other hand, are another slice of time test.)

If you serology test a random sample of the population, you can measure an estimate of the fraction of the total population that has ever been exposed and recovered. The important piece is to choose a random (representative) sample as possible. You don't have to measure every person. The larger your sample, the more precise the estimate. If the sample isn't random, extrapolation of results may not be meaningful. It's a basic precept of analytical chemistry.

A random RNA test will only tell you what fraction of the population is infected NOW. It will not look back in time and integrate over the entire course of the epidemic. It is an integration over an approximate 2 week slice of time. It can't tell you what happened before that. But this measurement is really useful info to public health planners to get a snapshot of current infection rate changes withing a community.

Combining serology based exposure data with confirmed case data and confirmed deaths allows estimation of all sorts of other interesting statistics that are useful for making public health policy, such as infection fatality rates, population exposure, asymptomatic infection rates, and progress toward herd immunity. At this point, we may actually have more data about Covid-19 than seasonal influenza.

One of the surprising outcomes of the NY serology study was that there were apparently 7-8 times as many asymptomatic (or minimally symptomatic) Covid-19 cases than confirmed cases. That has significant implications for viral spread, in that it's not just obviously sick individuals that are a concern.

I hope this explanation helps. The whole testing thing is very confusing to the public, and although some of the print and other media are trying to educate, they don't always get it right, or make it very clear. It doesn't help when some public health authorities can't use their data right, either. Some states were combining their RNA and serology test data for some unknown reason.
 
A random RNA test will tell one about "active now" infections.

In a few weeks time, those cases should now measure as "positive serology" or no longer be in the population.

Take a random sample of 100 and today and 6 have an active RMA positive.

In a couple of weeks, test the same 100 people and, assuming the other 94 haven't been infected, the same 6 people should have a positive serology test.

Either way, it would give the appearance that 6% of the population were "exposed and infected"

But if the question is "exposed" and not necessarily "infected", it could be inferred that number is 6% or greater....
 
That's another thing I've wondered about. I asked one of my doctors about it over the phone. She went silent long enough for me to wonder if the call had been dropped, and then said, "That's very interesting." And like Poe's raven, only that and nothing more.

I think we're talking about the same line of reasoning, anyway: If we assume for the sake of argument that people who survive to advanced age are more likely to have robust immune systems, does that person's immune system's habit of aggressively responding to new pathogens make them more likely to experience bad outcomes due to hypercytokinemia -- the so-called "cytokine storm" -- that can cause respiratory distress and multi-organ failure?

The medical implications of that would be above my pay grade. But it does make me wonder, firstly, whether medications (or herbs, like the aforementioned "cat's claw") that seem to moderate immune response should be considered as routine therapeutics for symptomatic cases; and secondly, whether vaccines for this virus might be more risky than is generally the case. Could a vaccine trigger iatrogenic hypercytokinemia and kill a previously-healthy patient? I really don't know. But as a person like yourself who "never gets sick," I do wonder about it.

As for the aforementioned cat's claw, I started taking it many years ago for mild arthritis on the advice of a former girlfriend who was a physician with a keen interest in herbal medicine.

Fast forward to a year or so ago, when I was treated at an ER for a slip-and-fall accident. The follow-up exam revealed arthritis in my shoulder severe enough that three providers (two physicians and a PA) were flabbergasted that I had normal range of movement and zero pain. I could (and still can) feel the crepitus (as could they), but it is neither painful nor motion-limiting. I also know that if I stop taking cat's claw, I start experiencing pain in my shoulder and in the joints in my fingers after a few days to a week.

Their shock didn't surprise me as I've been taking and following research on cat's claw for decades. It's perhaps the most-studied herb in the jungle.

There are actually two species of cat's claw that seem to be of particular interest to researchers, Uncaria tomentosa and Uncaria guianensis, with hundreds, or maybe thousands by now, of published studies between them. It's been investigated for practically all inflammatory and autoimmune conditions (it contains pentacyclic oxindole alkaloids that are believed to help regulate the immune system and tame the autoimmune response), as well as osteoarthritis, HIV, Alzheimer's (where the research is especially promising), and most recently, cancer.

The only suspected side effect that I know of is a big one: Because cat's claw seems to have at least some anti-tumor properties, it's possible that it's also a teratogen, as many anti-cancer drugs that act as cell-division traffic cops are. So it's definitely not something that pregnant women or nursing mothers should take.

With those exceptions, however, I can't shake this nagging feeling that some researcher in some lab somewhere really needs to be looking at cat's claw as a possible therapeutic for the hypercytokinemia that seems to be the actual killer in many or most COVID-19 cases. I'm not a physician, so I can't go as far as to say they should be prescribing it; but I do think they should be looking at it.

And at least one scientist is. I read the article. But now I can't find it in the sea of search results. The problem is that since I read it, practically every science site in the world, along with altogether too many non-science commercial sites, have added COVID notices, links, or banners; so doing a Boolean search for "[literally anything in the world]" AND "COVID" yields bazillions of irrelevant results.

Rich
Thanks for the info.

returning to my original question, I came across this today: https://www.theatlantic.com/health/archive/2020/04/coronavirus-immune-response/610228/
It may be a partial answer to the puzzle.

Dave
 
I hope this explanation helps. The whole testing thing is very confusing to the public, and although some of the print and other media are trying to educate, they don't always get it right, or make it very clear. It doesn't help when some public health authorities can't use their data right, either. Some states were combining their RNA and serology test data for some unknown reason.

They combined numbers for one of two reasons. 1. They were dumb. 2. They were trying to pull a fast one and make it look like they have a lot more capacity online.
neither is a good sign.

Tim
 
There is no "one" serology test. There are many different products. Some are better than others. The main problem with serology testing for population screening is false positive rates. Tests with false positive rates above 0.5-1% will generate too many false positives to be useful for this purpose, where the rate of real positives is expected to be low. But note: false positives will OVERESTIMATE exposure. False negatives rates of a few percent are not a significant factor for survey screening when few positives are expected to be found.

The *best* one was the one CDC was referring to, supposedly.

Any numbers based off of them are pure fiction from an accuracy standpoint right now.

With the antibody stuff likely meaning little at this point also, it’s definitely a fools errand.

Widespread real testing and understanding won’t be available now until 2021.
 
And CDC says the serology test is “up to 50% inaccurate”. It’s a fools errand.
Key word: 'up to'. Several different tests have been developed. The accuracy of some are not great to say the least. The accuracy of others are on par with most other commonly used diagnostics. The trick is knowing which test you're getting I suppose.
 
Key word: 'up to'. Several different tests have been developed. The accuracy of some are not great to say the least. The accuracy of others are on par with most other commonly used diagnostics. The trick is knowing which test you're getting I suppose.

Not really. The one they’re claiming that leave of error on is the *best* one available.

They’re NOT “on par” with anything.

Maybe they will be in a year.
 
Not really. The one they’re claiming that leave of error on is the *best* one available.

They’re NOT “on par” with anything.

Maybe they will be in a year.
No intention nor desire to argue. I'm being told otherwise by people I trust who do that sort of thing for a living so I'll agree to disagree.
 
No intention nor desire to argue. I'm being told otherwise by people I trust who do that sort of thing for a living so I'll agree to disagree.

I’m sure they’re very confident of their bad numbers. LOL.

What a crock. If the public information is “these aren’t accurate” and your response is “I’ve got special private info” how can we even have a rational discussion?

It’s silly. Like most of the garbage based off of these useless numbers.

Not an argument but your “info” is useless. Let’s talk about what’s published, not what you heard at the bar.
 
The *best* one was the one CDC was referring to, supposedly.

Any numbers based off of them are pure fiction from an accuracy standpoint right now.

With the antibody stuff likely meaning little at this point also, it’s definitely a fools errand.

Widespread real testing and understanding won’t be available now until 2021.

I don't know what data you are reading, but the publicly available false positive rates provided by the CDC vary between 0.4-13% depending on the vendor. The data is also available, with some digging, from many of the manufacturers as well. Obviously the vendor with a 13% false positive rate is useless, but those with false positive rates of 0.5% or less will have good predictive value for surveying a population with a predicted 5% incidence. (That incidence rate BTW is what CDC used as a benchmark for calculating predictive values.) This kind of false positive rate (0.5%) compares favorably to other serology methods. For surveying populations with high incidence (e.g. recovered patients) higher false positive rates are more tolerable.

The data one can obtain from serology tests is certainly not "pure fiction" but you do need to understand the limitations based on what the testing is intended to accomplish and what the validation statistics say. There is considerable value in doing random serology and RNA testing. It is not necessary to do widespread testing to evaluate population statistics. No test is 100% accurate or specific.
 
I’m sure they’re very confident of their bad numbers. LOL.

What a crock. If the public information is “these aren’t accurate” and your response is “I’ve got special private info” how can we even have a rational discussion?
We can't. So let's not. Again, agree to disagree.
 
Agreed. Your info is rumor. Mine is published.

So we know which one actually adds value to the discussion.
You're making assumptions about where my info came from where it was or wasn't published. I've said agree to disagree twice already but I suppose you can just keep talking down to me if it makes you happier. :rolleyes:
 
If you want to evaluate the estimation accuracy of serology testing for yourself, here is a Google Sheet that can be used to show the over- or under-estimate of the presence of antibodies in a measured sample using the published positive and negative performance values for a variety of validated test methods. You will have to download the Google sheet as an Excel file or copy it to your own Google account in order to change the numbers. I pre-populated the sheet with the Wadsworth Center serology test values of 1.2% false positive and 12% false negative based on FDA testing. At a 5% incidence rate, it will overestimate numbers by a relative 11% (finds 5.5%). At 15 % incidence rate (which was what was found in NY state using this kit for random testing) it will underestimate by a relative 5% (finds 14.2%).

As I often told my research students, 100 hours of speculation (or laboratory work) will save you 1 hour in the library.
 
Last edited:

Go to the source for specific data: https://www.fda.gov/medical-devices...ices/eua-authorized-serology-test-performance

A news article is not data, and the particular news article you referenced did not provide very specific information. The link above provides detailed validation data for all of the tests that were evaluated. If you are mathematically inclined, you can figure out the relative accuracy and limitations of various serology tests at a variety of incidence levels for yourself using validation data.
 
Go to the source for specific data: https://www.fda.gov/medical-devices...ices/eua-authorized-serology-test-performance

A news article is not data, and the particular news article you referenced did not provide very specific information. The link above provides detailed validation data for all of the tests that were evaluated. If you are mathematically inclined, you can figure out the relative accuracy and limitations of various serology tests at a variety of incidence levels for yourself using validation data.

Are you saying the news article summary is inaccurate?
 
Go to the source for specific data: https://www.fda.gov/medical-devices...ices/eua-authorized-serology-test-performance

A news article is not data, and the particular news article you referenced did not provide very specific information. The link above provides detailed validation data for all of the tests that were evaluated. If you are mathematically inclined, you can figure out the relative accuracy and limitations of various serology tests at a variety of incidence levels for yourself using validation data.

The news article links directly to the CDC recommendations.

Did you notice?
 
Last edited:
Old Thread: Hello . There have been no replies in this thread for 365 days.
Content in this thread may no longer be relevant.
Perhaps it would be better to start a new thread instead.
Back
Top