Is it true that everyone has had COVID-19?

For some time, we have been hearing many in the U.S. say that everyone has been infected with SARS-CoV-2 and already had COVID-19. I am guessing that those folks fall into one of three camps:

1. Those who assume this is the case because everyone they know has already had COVID. We need to urge caution with these kinds of observations because much to my chagrin, COVID mitigation measures have become pervasive social and entrenched ideological issues. In other words, we have seen many examples in which the social networks people have are increasingly ideologically aligned. In fact, so much so that, sadly, I am told or have heard about many examples in which families have been torn apart over politics and views on COVID. Those who have embraced COVID mitigation measures are more likely to have social networks that embrace these practices (and therefore know people who have not been infected yet) and those who have been emotionally charged and who have battled against mitigation measures are more likely to have similarly aligned social networks. Thus, it is not surprising that everyone one knows in the latter group has been infected. In my case, I have many friends and family members, as well as many acquaintances who have asked for my advice throughout the pandemic who, to the best of their knowledge, have not yet been infected.
2. There is a coordinated effort among some who have promoted from early on in the pandemic that everyone (some do make exceptions for certain high-risk groups) should get infected or that society would benefit from all of them getting infected. Some earnestly, but erroneously, believed that this would promote so-called “herd immunity” (if you want to read more about herd immunity, Dr. Epperly and I devote nearly an entire chapter in our new book about this concept and why it was flawed in the case of COVID-19: https://www.press.jhu.edu/books/browse-all?keyword=Pate%20and%20COVID-19%20), but others did this for less well-intentioned reasons, including some who are part of antivax campaigns and others who were funded by various groups to spread disinformation. It is beyond the scope of this blog post to get into all the details of folks in this group, but if you want to read more, but just a little more, here is one of many articles that discusses one organized effort (there are others such as America’s Frontline Doctors) https://www.respectfulinsolence.com/2023/06/02/censorship-the-word-disinformation-artists-use-when-called-out/, and if you are very interested in this topic and want to explore it more deeply, then here is a book that I can recommend (I am currently about half-way through it): We Want Them Infected by Jonathon Howard, M.D. https://redhawkpublications.com/We-Want-Them-Infected-p547021769. 
3. The final group comprises those who sincerely believed this to be true and based their belief on published data, but misunderstood what the data was telling us. The CDC and others do “seroprevalence” studies, by which they determine what percentage of the group sampled have antibodies against a particular infection, in this case SARS-CoV-2, and if the group is large enough, then they project what the “seropositivity” rate (the proportion of the population that does have antibodies) is in the general population. The problem is in how you interpret that data and how well the tested group represents the diversity in the general population. So, the remainder of this blog post will be based on interpreting the seroprevalence data.

First, a high-level review of antibodies. There are 5 major categories of antibodies (we abbreviate them with the prefix “Ig” for immunoglobulin (which is the biochemical name for an antibody) and then follow that prefix with a letter that tells us which major class the antibody or immunoglobulin falls into): IgA, IgD, IgE, IgG and IgM. For this blog post, we are only going to talk about IgG. (If you are losing your mind that I am not telling you about the other classes, here is a brief synopsis just to maintain your sanity):

• IgA – these are largely antibodies that reside in mucosal surfaces – nasal lining, gut lining, etc.), but they also are found in our tears and saliva to guard other parts of our body in which bacteria and viruses may try to make entry, and are even found in breast milk potentially helping protect newborn infants (who don’t yet make their own antibodies). When you hear me or others discuss mucosal immunity (which provides for a more rapid response to a pathogen we have already been exposed to previously, and for which we hope to be able to develop with vaccines of the future that hopefully will prevent infection and won’t require that you have previously been infected) IgA is an important component (but certainly not the entire armamentarium) of that mucosal immune defense.
• IgD – we don’t yet fully understand the role of this class of antibodies, but they appear to enhance the ability of cells that produce antibodies to do so and they do help in preventing respiratory infections presumably by activating certain types of white blood cells that you likely have not heard of before, such as basophils and mast cells.
• IgE – these antibodies are often the culprits in allergic and anaphylactic reactions. These antibodies are not always problematic for us, in fact, they play an important role in defending your body from certain multicellular parasites, such as worms.
• IgM – for most bacterial and viral infections or when vaccinating against these types of infection, IgM is the body’s first antibody response, but it is short-lived, so while most often, we test people for IgG in their blood as a marker of whether they have previously been vaccinated against or infected with a particular bacteria or virus (IgG is produced days later than IgM and persists longer than IgM), if we are concerned that someone may have just recently been infected (less than 2 weeks), we will often test for IgM and IgG because the IgG may return negative even though they are very recently infected, but often the IgM test will return positive.

Okay, back to IgG. So, when you read a report about a seroprevalence study, it is important to read exactly what did they test for. Almost always, these studies are based on IgG tests. If so, just understand that if there is a high level of infection going on in the population sampled, jumping to a conclusion as to the percent of the group, or a projection as to the percent of the total population that has been infected can be underestimated and there are other ways in which the data may overestimate the population immune status. How?

First, how might it be underestimated? In the situation I referenced, where there is still a high level of infection going on, realize that people who have been infected within the past week to 2 weeks are likely to test negative for IgG, either because their bodies have not yet had time to produce IgG, or they have started producing IgG, but it is early and those levels may not be high enough to be detected by the particular testing methodology being used.

We can also underestimate the percent of a population that is protected against infection based on seroprevalence studies in a couple of ways. First, antibody levels are dynamic, that is to say they are changing all the time. For most infections that we would be testing for, the body’s response is to make a large number of antibodies, but then reduce that production as the infection comes under control. And, that is a good thing.

As I mentioned, what we commonly refer to as antibodies are biochemically immunoglobulins. You likely, at some point, have had a blood test that measures your blood protein (commonly reported on your lab report as “total protein.” I didn’t contemplate going into math during this blog post, but here we go. Generally speaking, total protein = albumin + globulins. In other words, albumin is a protein (many of you have heard of that, and the level of albumin may be reported on your lab test report, usually right after the report of total protein).

Globulins are also proteins and they comprise a number of different types of proteins, many of which you likely have never heard of, but the largest component of globulins are immunoglobulins! Now, some lab reports will list out total protein, albumin and globulins, but most only list total protein and albumin. In those cases, I just use advanced calculus and subtract the value for albumin from the value for total protein and you get the value for globulins!

One last medical pearl and then I need to get back on topic, as a general rule, your globulin level should be less than your albumin level. When the lab report just lists the total protein and albumin, as long as the albumin level is normal, I usually just look to make sure that the albumin level is at least half of the value of the total protein.

Why would the globulin level ever be higher than the albumin? Well, we could have a situation in which the albumin level is low (decreased production of albumin, such as severe malnutrition or liver disease or increased losses of albumin through the gut or the kidneys). But, if the albumin level is normal and the globulin level is higher than the albumin level, that is when I look for acute infection (because the body is producing a large amount of immunoglobulins) or conditions in which the body produces high levels of globulins (inflammatory conditions that are not infections) in which the globulins that are elevated are not immunoglobulins (although patients with rheumatoid arthritis who produce rheumatoid factor may produce elevated levels of immunoglobulins (IgM) and non-immunoglobulin globulins) and blood disorders in which the types of cells that make immunoglobulins have gone rogue and are making excessively high levels of immunoglobulins such as seen in multiple myeloma and Waldenstrom macroglobulinemia.

My point in all of this is that if your antibody levels went up in response to infection, but never went down, we would be in real trouble from a health standpoint. Antibodies are immunoglobulins and immunoglobulins are a type of globulin, and globulins are proteins. Proteins are relatively large biochemicals in our blood compared to other things carried in our blood such as glucose (sugar), potassium, etc. In the disorders I mentioned above where the cell type that makes antibodies (B-cells) has gone wild making excessive antibodies (e.g., multiple myeloma and Waldenstrom macroglobulinemia, and especially in the latter), these proteins can thicken the blood and cause all kinds of problems, especially related to clogging up small blood vessels. Thus, the normal functioning of our immune system is to reduce production of immunoglobulins over time following either vaccination or infection and rely on so-called memory cells that have seen the protein from the vaccine or prior infection (in the case of COVID, the spike protein of the SARS-CoV-2 virus) that have retained the blueprint for making antibodies that are specific to that protein and have all the machinery of the body ready to go to pump out more and newer antibodies faster than if never before exposed to that protein in case of a booster dose of vaccine or a future infection.

In some infections, especially those that are mild, the body of some people will not produce much antibody and it may reduce to below the levels of detection of an antibody test giving a negative result, when the person has in fact been previously infected and may be protected by virtue of their memory cells as well as other components of their immune system that are not tested in these seroprevalence studies.

Another way that the percent of population immune protection may be underestimated is if we select the wrong group or wrong test for the seroprevalence study. For example, if we conduct a seroprevalence study of influenza antibodies among college students for past circulating strains of influenza, the results would underestimate the immune protection of an older population who lived when those strains were circulating and were previously infected by them. Conversely, if we tested these same college students in the first months of the fall semester for currently circulating influenza strains, we could anticipate getting a very high percentage of students who are seropositive and that might overestimate the seroprevalence of the general population.

But, here is the most important part. We have to know which antibodies specifically were tested for in the seroprevalence study. In terms of COVID, if the test is for antibodies to the spike protein, then people will be positive whether infected, vaccinated or vaccinated and then infected. Thus, when the CDC or anyone else states that 96.4% of those sampled at a point in time tested positive for SARS-CoV-2 antibodies, that does not necessarily mean that 96.4% of the general population has been infected.

Let’s dive into a study that was just published so that I can explain to you how I analyze this type of data. This study came out yesterday: Estimates of SARS-CoV-2 Seroprevalence and Incidence of Primary SARS-CoV-2 Infections Among Blood Donors, by COVID-19 Vaccination Status — United States, April 2021–September 2022 (cdc.gov)

So, here is how I analyze this report. I first look at the title to get an idea of what the authors are likely to try to show in their report: “Estimates of SARS-CoV-2 Seroprevalence and Incidence of Primary SARS-CoV-2 Infections Among Blood Donors, by COVID-19 Vaccination Status — United States, April 2021–September 2022.”

What impressions do I draw? First, the study is apparently designed to provide only an estimate, not the actual, exact measure of seroprevalence and incidence of infection. Second, the study reportedly is going to look at both seroprevalence and at infection rates, therefore, I already assume that they will not be simply relying on IgG assays against the spike protein (because you are exposed to spike protein both from vaccination and from infection. Therefore, to separate these out, I expect to see the investigators test for an antibody that you would only make if infected, e.g., IgG against the nucleocapsid (so-called anti-N) or another protein of the virus that is not contained in the vaccine.)

I also see that their study is going to be based upon blood donors (this makes sense because it is easy to get blood for testing from blood donors!), so we have to be mindful of how those who donate blood might differ from the general population, if they do.

Additionally, we already see from the title that they are likely to put these donors into groups according to vaccination status.

Finally, of critical importance is the time period – April 2021 to September 2022. When I see that, I conclude that first, this covered a time period when we went through a number of waves of different variants – alpha, then delta and then omicron. Secondly, it tells me that whatever that estimate is, it may not be the same today.

Now, having stated all this, there are occasions where an editor comes up with the title of the paper, not the researchers, so never rely solely on the title to draw your conclusions. But, for me, reading the title starts my juices flowing as to what am I going to be looking for and thinking about as I read the paper.

The first thing I do is look at the sample size – in other words, how many people were tested (the more the better). It was 72,748 people, so that seems like a large number, but my next question is how much of the total pool of blood donors does this represent (generally, the larger the better). It was 51%, which ordinarily would concern me as too low, however, in this case, it is actually better than what I thought it might be. This is a longitudinal study, which means that they followed the participants throughout that period of time. Many blood donors, unfortunately, do not donate regularly or frequently enough to provide all the data points needed for this study. Further, there needed to be data available during each period of the study as to the infection status and vaccination status of each study participant. This is data that is not generally collected on blood donors, so it took a lot of extra effort to collect all this data in all 72,748 participants.

As we read further, we see that, in fact, they do utilize antibody to the nucleocapsid protein to identify those previously infected and utilize the conversion of antibody to the nucleocapsid protein from negative to positive in a three-month interval in order to calculate infection incidence rates.

So, what were the findings from this study?

1. During the second quarter of 2021 (April–June), an estimated 68.4% of persons aged ≥16 years had infection- or vaccination-induced SARS-CoV-2 antibodies, including 47.5% from vaccination alone, 12.0% from infection alone, and 8.9% from both.
2. By the third quarter of 2022 (July–September), 96.4% had SARS-CoV-2 antibodies from previous infection or vaccination, including 22.6% from infection alone and 26.1% from vaccination alone; 47.7% had hybrid immunity.
3. During all periods, higher prevalence of hybrid immunity was observed among Black and Hispanic populations than among White and Asian populations.
4. Among persons with no previous infection, the incidence of first infections during the study period (i.e., conversion from anti-N–negative to anti-N–positive) was higher among unvaccinated persons.
5. From April–June 2021 through January–March 2022, the incidence of first SARS-CoV-2 infections among unvaccinated persons was 67.0%, compared with 26.3% among vaccinated persons (p<0.05). From January– March 2022 through April–June 2022, the incidence among unvaccinated persons was 21.7% and was 13.3% among vaccinated persons. Between April–June 2022 and July–September 2022, the incidence among unvaccinated persons was 28.3%, compared with 22.9% among vaccinated persons (p><0.05). ><0.005) [for the non-statisticians, p values are a statistical indication of how likely the finding is truly a significant difference compared to random chance; in this case, a p value less than 0.005 suggests this difference is truly and significantly different]. From January– March 2022 through April–June 2022, the incidence among unvaccinated persons was 21.7% and was 13.3% among vaccinated persons. Between April–June 2022 and July–September 2022, the incidence among unvaccinated persons was 28.3%, compared with 22.9% among vaccinated persons (p<0.05).
6. Incidence of first SARS-CoV-2 infections was higher among younger than among older persons.

What can we take away from this study?

1. The starting time period for this study was April 2021. This was after the first year of the pandemic when vaccines were unavailable and those who wanted to avoid infection largely had to rely on so-called non-pharmaceutical interventions (NPIs) – distancing, avoiding large groups of people, wearing masks, etc. The vaccine roll-out began for non-healthcare workers in January, but was prioritized for high-risk groups initially.
2. By April of 2021, the wild-type virus (original strain, if you will) had largely disappeared and was replaced by progressively more transmissible variants over the course of 2021, 2022 and 2023. In April of 2021, the new variant predominating in the U.S. was alpha.
3. It is striking to note, that at least among blood donors, that even as of June 2021, nearly 70% (47.5/68.4 x 100 = 69.4) of those with antibodies were from vaccination, who had no history or evidence of prior infection.
4. By the end of the study period (September 2022), with the emergence of far more transmissible variants delta and omicron, 27% of those with antibodies (26.2/96.4 x 100 = 27) still had no history or evidence of infection. With the development of progressively greater degrees of immune evasion by new variants, nearly half of those with antibodies had so-called hybrid immunity as a result of vaccination and infection (47.7/96.4 x 100 = 49.5%).
5. The prevalence of hybrid immunity is lowest in adults aged ≥65 years, possibly due to higher vaccination coverage and earlier availability of COVID-19 vaccines for this age group, as well as to higher levels of adoption of behavioral practices (NPIs) to avoid infection.
6. The authors conclude that “In this study, unvaccinated persons had higher rates of infection (as evidenced by N antibody seroconversion) than did vaccinated persons, indicating that vaccination provides some protection against infection.) Of course, it is also likely that those who seek vaccination may be self-assessed to be at higher risk than the general population and therefore were also more likely to continue NPIs in addition to getting vaccinated.
7. Statements to the effect that everyone has been infected by the SARS-CoV-2 virus and had COVID by now are likely exaggerated, but certainly were as of September of 2022. In fact, if the blood donor population is generalizable to the US population, this would suggest that as of September of 2022, almost 87 million Americans may not yet have been infected. Of course, with the continued loosening of COVID countermeasures and the increasing transmissibility of more recent variants, these numbers may have significantly changed by now. Further, this estimate assumes that infection rates in children are the same as adult blood donors, and there is reason to believe that in fact, it may be higher in children.

Nevertheless, it appears that there is a far greater number of yet uninfected Americans than has been commonly believed to be the case. This is good from the standpoint of the population risk for and burden of Long COVID (obviously, you cannot develop Long COVID if you have not been infected). On the other hand, with waning immunity from vaccines and lack of boosters that reflect the currently circulating variants, these previously uninfected persons may be at significantly increased risk if infected.