Skip to content

Raw data: Herd immunity from COVID-19

According to the CDC, fewer than 4% of Americans are completely unprotected from COVID. The other 96% are either vaccinated or have been previously infected:

These figures are taken from blood donors, who might be systematically different from the population as a whole. The study sample was weighted to account for this, but even so the 3.6% number should be taken as a rough estimate, not gospel.

The CDC did the same estimates for different age groups and the results were similar across the spectrum. The number who have never been infected or vaccinated are all below 4%.

But this is not entirely good news. COVID has become more infectious over time, and estimates for the Omicron variant suggest that to reach herd immunity the share of the population that's either vaccinated or previously infected needs to be 98% or higher. We're not there yet, as should be obvious from the fact that in the US there are still 10,000 new cases and 200 COVID deaths per week.

But we're continuing to make progress. About 30% of Americans are still not vaccinated, and if we could persuade as few as a tenth of those to get the vaccine our protection rate would go up to 99% or so. Would that provide herd immunity? Maybe, but COVID mutates rapidly and immunity appears to last only a year or so. Because of that, herd immunity might be permanently out of reach no matter how high our vaccination and infection rate is.

19 thoughts on “Raw data: Herd immunity from COVID-19

  1. reino2

    It would only matter if that tenth never had COVID. That's a very difficult group to both identify and persuade to get vaccinated. Realistically, we would need about half of the unvaccinated to get vaccinated, and that's actually probably not realistic either.

    1. xabicaj

      ?US Dollar Rain Earns upto *****50 to *****50 per day by google fantastic job oppertunity provide for our community pepoles who,s already using facebook to earn money 85000$ every month and more through facebook and google new project to create money at home withen few hours.Everybody can get this job now and start earning online by just open this link and then go through instructions to get started………: https://iplogger.cn/125t

    2. Jasper_in_Boston

      It would only matter if that tenth never had COVID. That's a very difficult group to both identify and persuade to get vaccinated. Realistically, we would need about half of the unvaccinated to get vaccinated, and that's actually probably not realistic either.

      Right. It seems nearly certain that the share of the unvaccinated population that has been infected is significantly higher than the 70% of the overall population claimed by this graph. Both because these people lack the sterilizing immunity protection of the vaccine (a weak protective effect, be sure, but not non-existent) and because "have already had covid" is one of the factors that people seize upon as a justification or excuse not to get vaccinated.

      So, I reckon we'd need to see well over 10% of the unvaccinated get their jabs in order to reach 99%, because this pool is not as "rich" a source of the "non-antibodied" as Kevin's post implies.

  2. bluegreysun

    I’m not sure what “herd immunity” means in this context. Do we speak of herd immunity to the flu or common cold? (Maybe? I dunno). Seems like the disease kills and injures older/sicker people most, and their vaccination/prior infection status is not that relevant - better to be vaccinated than not, at all ages and health statuses, obviously, but it’s better to be unvaccinated and 30 (and not obese/hypertensive/diabetic) than it is to be fully vaccinated/boosted and 80 and/or sick.

    I thought herd immunity was mostly used in the context of stopping transmission, if enough people were vaccinated/exposed, the virus wouldn’t find enough vulnerable hosts, and it would die out. Vaccines and prior infection only somewhat (to not at all) lower viral load, and only for a few weeks to months. Vaccinated people are still potential hosts, and still contagious once infected.

    1. D_Ohrk_E1

      I'm not so sure it's better to be sick and an unvaccinated 30 year old full of asterisks than to be 80 and fully vaccinated and boosted.

      The risk of any long-COVID symptom, regardless of age, is higher if you're unvaccinated, as is the risk of death.

      And since the risk of infection is lower if you're fully vaccinated and boosted, the more apt comparison is between a sick, unvaccinated 30 year old full of asterisks than a fully vaccinated and boosted 80 year old who isn't sick, n'est pas?

  3. tinfoil

    If I'm working the numbers right, the data imply that infection rates (not death or severity, of course, which aren't measured) for vaccinated and unvaccinated were the same in 2022, even though rates were 3x higher for the unvaccinated in 2021. FWIW the CDC authors claim a slight benefit for vaccines in 2022 because they didn't correct as I describe below. One could say theirs is a more liberal estimate and mine more conservative. More importantly, since the study didn't look at booster status, the 2022 number does not shed any light on the effectiveness of boosters. (Second observation: pretty much no unvaccinated people got a vaccine in 2022.)

    Here's the details: the data in the CDC link can be put in 2x2 tables:
    Apr-Jun 2021
    infection
    vaccine no yes
    no, 31.6, 12, 43.6
    yes, 47.5, 8.9, 56.4
    79.1 20.9

    Jan Mar 2022
    infection
    vaccine no yes
    no, 6.4, 20.5, 26.9
    yes , 39, 34.1, 73.1
    , 45.4 , 54.6

    July-Sep 2022
    infection
    vaccine no yes
    no , 3.6, 22.6 , 26.2
    yes, 26.1, 47.7, 73.8
    , 29.7, 70.3

    (the third row/column are the marginal values (sum of the columns or rows)

    The number of "new" infections in the unvaxed is the difference in "no/no" cell values over time adjusted for the change in the number of vaccinations, which would also move them out of no/no. The percent infected is the number of new infections divided by the initial number of unvax, again adjusted for those who left the group because they were vaccinated

    So between Q2 2021 and Q1 2022:
    new infected, unvax =
    31.6 - 6.4 - (43.6 - 26.9) = 8.5 ("no/no"#1 - "no/no"#2 - (unvax#1 - unvax#2)
    ... % unvax infected = 8.5/(31.6 - (43.6 - 26.9)) = 57%

    new infected, vax is simpler because the only way out of no inf/yes vax is by getting infected, so no special adjustments are needed:
    47.5 - 39 = 8.5
    ... % vax infected = 8.5/47.5 = 18%

    so 57% of previously uninfected unvax vs 18% previously uninfected vax got infected

    Doing the same thing for Q1 vs Q3 2022 has both groups at around 33%

    1. Jerry O'Brien

      I am ready to believe that getting vaccinated in early 2021 had little effect on your chance of getting infected in 2022, but it still reduced your chance of severe disease. Herd immunity won't be achieved. Annual boosters will be needed to keep the rate of severe covid cases low.

    2. D_Ohrk_E1

      If you recall, Omicron had sufficient antigenic shift, which created an issue known as antigenic "sin", where people vaccinated early (prior to the bivalent ones from fall 2022) had immune systems imprinted with a variant that had difficulty recognizing and responding to Omicron.

  4. civiltwilight

    Herd immunity for Covid is unachievable. The cat, as they say, is out of the bag. Respiratory Pandemics are tricky, especially with airplanes ferrying people across the globe. What would have happened if the CCP had not denied and downplayed the existence of the virus but instead stopped all flights to and from China in early Jan 2020? The CCP did block internal travel but continued to trade with the world

    1. D_Ohrk_E1

      What would have happened if the CCP had not denied and downplayed the existence of the virus but instead stopped all flights to and from China in early Jan 2020?

      It would not have mattered. The virus was already in the US by December 2019, and undetectable.

  5. rick_jones

    These figures are taken from blood donors, who might be systematically different from the population as a whole.

    People who willingly give-up some of their blood for the benefit of others? Yeah, they might be a little different from the population as a whole...

    1. Salamander

      And they are people who have never been to prison, have never engaged in sex for money, have never engaged in same sex sex, haven't been out of the country and definitely not to the UK, are not taking any of a range of prescription drugs, ...

      Well, the list goes on and on and on, as would be apparent if you've given blood recently. Excludes quite a few people.

      1. rick_jones

        The "never" has been removed from same sex sex, and replaced with a "specific form of sex within the last N months" proscription.

  6. cephalopod

    The most recent numbers here are from last August. I know many people whose first round of covid infection was after August.

    We may not have a lot of increase in vaccinations, but the percent of people who have never had covid must be miniscule by now.

Comments are closed.