Are there stats on % in US who have had Covid AND vaccine?
May 18, 2021 2:26 PM   Subscribe

If we want to calculate what percentage of the population are immune to Covid at present, to see, for example, where we are in terms of "herd immunity" (or, just because it's interesting), how do we deal with the following issue:

We are told that we can add to the percentage of people who have been vaccinated so far, the percentage of people who have (some) immunity due to having been infected with Covid. How can we determine how many infection-immune people to add to the vaccine-immune group, when there are plenty of people who have been infected but who have ALSO gotten vaccinated?

The point is: we want to know how many people (or what % of people) to ADD to the vaccinated group to reach the total percentage of Currently Immune People -- without counting anybody TWICE (once for having vaccine immunity and another time for having infection immunity).

Plus -- a further complicating factor is that there are so many more people who have had Covid than are recorded as "cases." (e.g. Youyang Gu says there are slightly fewer than 3x more total infections than the number of recorded as "cases" (his site). So -- how do these "invisible" people, who have been infected but didn't know it, get counted? If they get vaccinated, they're put in the "vaccine immune" group when, in fact, they had "infection immunity" but didn't know it possible long before they were vaccinated?

I'm not seeing this issue addressed in any of the articles I read or podcasts I listen to, etc. They just say, "To this vaccinated group we must realize that there are also people who have immunity because they had Covid" -- and then, sometimes, they talk about the percentage of people who are recorded as having had Covid, without mentioning that those SAME people may also have been vaccinated!

**In other words, they all seem to ignore the OVERLAP.** Why? They seem to be pretty smart scientists!

Is there anybody who is addressing this issue? Are there any estimates of what percentage of people have infection immunity who are NOT vaccinated and whom we may "safely" add to our percentage of vaccinated people to get an accurate percentage of TOTAL percentage of immune people in the US population? (I'd settle for just New York State or New York City!)

posted by DMelanogaster to Health & Fitness (8 answers total)
Given the data we have access to, I think the best we can do is look at the number of people who have not been vaccinated and estimate the number of them that had covid based on the percentage of the whole population that had covid. This assumes that people who had covid are equally likely to get vaccinated as people who haven't had covid, which is potentially a bad assumption and also might skew one way or the other depending on how common anti-vax sentiment and covid denial are in the area.

So, like, say the population is 100 people, and we estimate (or know) that 10% of the population had covid. Let's say 40 people have been vaccinated in total (and we have no information about whether each of them has had covid or not.) I would end up estimating that out of the 60 UNvaccinated people, 10% (so, 6 people) have had covid and have some immunity. So 40+6 =46 for a total of 46/100 with some immunity.
posted by needs more cowbell at 2:58 PM on May 18, 2021 [2 favorites]

I think the datasets of {vaccinated people} and {people who've had Covid} are non-overlapping; there's just not practical way to query the intersection without getting down to individual names. And therein lies all sorts of medical-records problems.
posted by Dashy at 3:21 PM on May 18, 2021 [1 favorite]

I wonder if there is a way to get at this, perhaps, from info on the forms people fill out to get their vaccines. I vaccinate in PA and one mandatory field is whether they have ever been told that they have tested positive for Covid. I don't know how much undiagnosed Covid there is among the self-selected people who do get vaccinated, nor do I know if this is a mandatory field in every jurisdiction. But among the group I've vaccinated in Philly - thousands, including hundreds of tweens and teens - it seems like about one out of every 10 or 12 people has previously tested positive. I'm in a relatively affluent area where testing is widely available - that matters, too.

This seems like a huge, moving target, if not Sisyphean in scope. But this is a small snapshot that might be extrapolated by a really, really clever bio-statistician.
posted by citygirl at 3:34 PM on May 18, 2021 [1 favorite]

Figuring out the overlap is an interesting modeling problem, but it's not on its own enough to decide if we've "reached herd immunity", which I assume is your goal.

People who've had covid are not "immune" to the same extent as people who've been vaccinated. So, even If you could figure out the overlap, and use that to calculate the fraction of people that have either 1) been vaccinated 2) had covid, or 3) both, that doesn't tell you the fraction of people who are "immune".

And even if you could figure out the fraction of people who are "immune", there's no agreement on what this number would have to be for herd immunity.

The only way to tell if you've reached herd immunity is to observe case levels go from the current pandemic to isolated outbreaks that fade out without restrictions on movement or behavior.
posted by caek at 3:42 PM on May 18, 2021 [3 favorites]

Best answer: If the number of unknown cases of covid is 3x the known cases, and half of each group also is vaccinated (current US percentage), then that leaves a group who were not vaccinated but have antibodies that is 2x larger than the group of known cases.

So it would be an easy simplification and conservative estimate, when talking to a reporter or lay audience, to not mention the complexity of overlap, or undetected cases, or variations in immune response to asymptomatic covid, and just talk about the number of detected cases plus vaccinated.
posted by joeyh at 4:32 PM on May 18, 2021 [1 favorite]

Best answer: Given the uncertainty intervals on how many have had COVID-19, the overlap with vaccination just isn't worth much time, as far as your final conclusions.
posted by away for regrooving at 11:47 PM on May 18, 2021

Best answer: I think the datasets of {vaccinated people} and {people who've had Covid} are non-overlapping; there's just not practical way to query the intersection without getting down to individual names. And therein lies all sorts of medical-records problems.

That's what public health informatics is for! It's not really a medical-records problem - Departments of Health routinely work with data like this, and in fact programmatically matching different datasets is fairly common. It's exactly how a lot of data on covid cases are generated in the first place.

Colleagues of mine, for example, are doing exactly this to study "vaccine breakthroughs." Vaccination data are in Immunization Registries, usually administered by local departments; case data usually originate with the State through laboratory reports but are often further developed (cleaned, analyzed, etc) by local departments.

But this particular question, on the overlap of cases & vaccinated, is an extensive process that's not a part of routine reporting from what I've observed.
posted by entropone at 4:48 AM on May 19, 2021

To one suggestions above: Assuming that 1/2 the people who had COVID were then subsequently vaccinated seems incorrect: COVID infections were distributed unevenly in the population, more heavily impacting for example Latinos, African Americans, and Native Americans. At the same time, vaccines are also unevenly distributed, with more vaccines received by white people, people with higher education attainment, and people in certain geographic regions.
posted by latkes at 8:17 AM on May 19, 2021

« Older How to print a zine in 2021   |   Most efficient window AC set up Newer »
This thread is closed to new comments.