From early in the COVID outbreak, the curtailment of the pandemic by attaining herd immunity has been dismissed because the consensus was that doing so would require 70-80% of the population to be infected.  That level of infection would result in millions of fatalities and widespread illness.  However, research has been emerging over the last two months that suggests the threshold to achieve herd immunity may be much lower than originally estimated.  If that is the case, it has enormous implications with respect to the future course of the pandemic.

Background

Prior to the invention of vaccines, every epidemic in human history was ultimately defeated by the human immune system when enough people acquired immunity through exposure to the pathogen that it could no longer reproduce and spread.  That  phenomenon has come to be known as herd immunity.  It is a miraculous natural defense mechanism.  But the downside to herd immunity is that in the course of infecting a portion of the population sufficient to achieve herd immunity, there will be some percentage of the population in which the pathogen will produce serious disease and fatality.

The miracle of vaccinations has been that they teach the immune system how to recognize a pathogen without the attendant illness that accompanies naturally acquired immunity.  Thus, vaccines accelerate the attainment of herd immunity and minimize the illnesses and fatalities.

Fortunately, reaching herd immunity does not require everyone to be infected or vaccinated.  At some point the number of available “targets” for a pathogen is sufficiently low that it cannot find enough of them to continue to spread.  The point at which that occurs is referred to as the herd immunity threshold (“HIT”) and depends initially on how contagious the disease is.

Scientists measure the contagiousness of a disease by estimating its “reproductive number,” more commonly referred to as the R-naught factor and notated as R₀.  R₀ is an estimate of the number of other people that one infected person will infect during the course of their illness.  It is fundamentally dependent on the basic characteristics of the pathogen and its method of transmission.   But the reproductive number is also affected by other environmental factors, including human behavior, such as the implementation of non-pharmacological interventions (i.e., lockdowns, social distancing, masks, etc.).  Mutations in a  pathogen can also affect its reproductive number.  The measure of the reproductive number as it changes over time is referred to as R_t.

The basic formula to calculate HIT is a function of the reproductive factor.  Specifically, it is the inverse of the reproductive factor (i.e., 1-1/R_t).    For example, if a pathogen’s R_t is 4, the formula would be = 1-(1/4) or 75%.  The early estimates of COVID’s R₀ were in the 3-4 range, leading many to conclude that we would not achieve herd immunity until 60-80% of a population was infected.

It should be noted that a lower HIT, based on reductions to R_t from non-pharmacological interventions (NPIs), will only exist for as long as the NPIs can be sustained which historically has not been for long.  That is why NPIs are generally considered as a tool to temporarily “flatten the curve” as opposed to being a long-term solution to end the epidemic.

The Susceptibility Factor

A growing number of epidemiologists have come to believe that the early consensus of a 60-80% HIT was flawed because it assumed that everyone is equally susceptible to the virus.  We have been told since the outset of the COVID outbreak that because this was a novel virus, no one would have any immunity, and everyone was at risk of being infected.  But it turns out that is not the case.

The reason is something called “heterogeneity of susceptibility,” which holds that not everyone is equally susceptible to a disease, even if there has never been a previous exposure to the particular pathogen at issue.  Most of us have seen examples of this in the current outbreak.  For example, I know three married couples where one spouse was infected but the other never got sick, despite living together.

The extent of a population’s susceptibility dramatically affects the herd immunity threshold.  The lower a population’s susceptibility the lower the HIT will be.

Here’s how the math works.  Let’s use our previous example of a pathogen with a R_t of 4.  The basic HIT formula says 75% of the population must be infected to achieve HIT.  Let’s further assume that 30% of the population is not susceptible to the disease.  In that case the pathogen only has 70% of the population available as targets.  So, when 75% of the 70% that is susceptible to the disease has been infected, HIT will be attained.  In this example, that would be 52.5% (75% x 70%).

The most important article I have read on COVID is this article in Quanta magazine about the math behind herd immunity.  This New York Times article recently followed up with some of the same researchers quoted in the Quanta article.  Both articles quote a number of epidemiologists who are now estimating herd immunity for COVID is something below 50%.  There are two studies (here and here) that have estimated that it might be as low as 20-25%.

The variation in susceptibility is not well understood.  It is believed that it is largely based on differences between individuals’ immune systems.  In a jaw-dropping example of this, a recent study found that a third of blood samples collected prior to the emergence of COVID had a T-cell immune response to COVID!  No one knows exactly why there would be an immune response in blood that had never been exposed to COVID, but some have speculated that a previous exposure to one of the common cold viruses, which like COVID are coronaviruses, may have provided some cross-reactive immunity.  The idea of cross-reactive immunity is not new.  The first vaccine administered by Edward Jenner in 1796 for smallpox was based on a different virus, cowpox.  The word “vaccine” is based on the Latin word for cow, vacca.

But even though such T-cell resistance to virus may protect a person from serious disease, its effect on transmission has yet to be determined. It may be that someone with cross-reactive immunity could still be “infected” with the virus and pass it along without showing any symptoms. One study did find that asymptomatic individuals transmitted the virus by at a 25% lower rate than those who were symptomatic.

Regional Variation

Herd immunity is mostly a local or regional phenomenon, among a population that regularly interacts.  Pathogens can be imported into a regional population after it has reached herd immunity and still precipitate outbreaks, but normally such subsequent importations will die out relatively quickly.

As a result, the more a region interacts with the “outside world” the more susceptible it will be to reintroductions and thus will require a higher HIT.  This is why islands and relatively isolated countries generally fare better in epidemics.

Some regions may have historical or shared genetic factors that decrease their susceptibility.  This study suggests that the COVID outbreak has been less severe in Asia and the Middle East because of their previous exposures to other recent corona-viruses.  Other research has indicated some Asians may have a genetic variation in their immune system that provides an advantage in resisting corona-viruses.

There are also other factors which will affect a pathogen’s R_t and thus its HIT.  For example, typically pathogens will have a higher R_t in urban areas than rural ones.  Also, cultural differences, such as the prevalence of multi-generational living arrangements and family sizes, affect R_t.  This suggests that HIT will vary, and perhaps significantly so, from one area or country to another.  That is why it is complete nonsense to compare the course of an epidemic between regions with dramatically different profiles, say the U.S. vs. Iceland.

This also means that herd immunity will not suddenly be reached all over the country or the world at one time.  Rather, herd immunity will spread gradually, basically following where the virus has gone.  In the New York Times article and in this Newsweek article, some scientists are now speculating that some of the hardest-hit areas (such as parts of New York, northern Italy and Mumbai) may have already reached herd immunity.

Lessons from History

An examination of the history of other pandemics also suggests HIT is very likely much lower than the initial estimates.  The Spanish flu circulated in the U.S. from 1918 until 1920.  It is estimated that its R₀ was 2-3, which according to our formula would require 50-66% of the population to be infected before herd immunity would be achieved.  But the historical consensus is that only about 30% of the country ever contracted the virus.  The same is true with the Asian flu pandemic in the late 1950s.  The CDC estimates that 25-40% of the U.S. population was infected before the epidemic subsided in 1958.

In the case of the Spanish flu, no vaccine was ever developed, so the demise of that epidemic was entirely due to naturally acquired herd immunity.  The U.S. government did direct the manufacture of a vaccine for the Asian flu but researchers have concluded that it was of limited efficacy and was delivered to the public after the epidemic had largely passed. Public Health and Medical Responses to the 1957-58 Influenza Pandemic, D. A. Henderson, Brooke Courtney, Thomas V. Inglesby, Eric Toner, and Jennifer B. Nuzzo, Biosecurity and Bioterrorism: Biodefense Strategy, Practice, and Science, Volume 7, Number 3, 2009.

I have not been able to find any historical account of a pathogen that has been estimated to have infected anything like 60-80% of a population, except for isolated communities in which there was very little genetic diversity.

History also reinforces that achieving herd immunity is not a red-line event.  Rather, it is an incremental process that comes about gradually.  Every person that acquires immunity becomes one less target available for the virus.  Even a small percentage of the population with immunity will begin to slow the spread of the pathogen.  A study by Michael Levitt, a Stanford medical school professor and Noble laureate, suggests that this “burnout” effect begins with the infection of as little as 20% of a population.  Justin Lessler, a Johns Hopkins epidemiologist professor, makes a similar argument in this Washington Post article, noting “that incremental benefits of slowly accumulating community immunity mean that as the pandemic goes on, control will get easier.” As painful as it is to read the daily toll the virus is taking, every case moves us one step closer to herd immunity.

What are the Implications of a Lower HIT?

So, what are the implications if COVID’s HIT is really something less than 50%?

As of August 27, the CDC estimated that about 5.8 million people in the US have contracted COVID.  But because so many cases present with no or mild symptoms, we know that is a small fraction of the total who have actually been infected.  The CDC has recently estimated that the actual number of infections is running 6-24 times the reported number.  If we assume a 10X factor, then about 58 million have actually been infected.  That is about 18% of the US population.

In addition, we are currently adding about 45,000 cases per day.  Again, assuming that only represents a tenth of the actual infections, then something like 450,000 new infections are added every day.  If that is true, and the new infections continue at the current pace, we would hit 50% sometime next April.  If some of the lower estimates are correct, HIT might be reached by the end of the year.

But to reiterate, herd immunity will not be suddenly emerge one day.  The growing immunity in the population will begin to slow the spread and new infections will begin to decline.  There will likely be waves in the decline.  For example, many are concerned there could be a wave of new infections this fall, perhaps combined with the seasonal flu.  But if HIT is really as low as the new research suggests, the overall trajectory is downhill.

Of course, the development of a vaccine would shorten this timetable.  Depending on how soon it is deployed, it will probably require a relatively small percentage of the population to be vaccinated to get us to HIT.

For example, let’s assume that we have a vaccine on December 1 and that the infection rate continues at its current level.  In that case about 100 million Americans would have been infected by then.  If HIT is 50% (165 million), then only 65 million Americans (about 20%) would have to be vaccinated.  In an Economist podcast, Bill Gates recently opined that herd immunity can probably be reached if only 30-40% of Americans are vaccinated.  So, the hand-wringing you see so often in the media about the delays and logistics of vaccinating the entire country is not really a concern.  We do not need to vaccinate anything like the entire country to reach herd immunity. Gates also thought the epidemic would largely be over sometime in 2021 through a combination of natural and vaccine acquired immunity.

Effect on Fatalities

As I previously mentioned, the problem with achieving herd immunity is the cost of getting there.  Even though the percentage of people who cannot survive an infection is very small, a small percentage multiplied by the millions who will be infected on the way to herd immunity represents a huge toll.  The lower estimates of HIT could put a cap on the number of likely fatalities.

The CDC currently estimates that percentage at about 0.2-0.4%.  If the US threshold is 50% (165 million people infected), that would imply 300,000-600,000 fatalities before we get to herd immunity.  That is about 10-20% of the total annual U.S. fatalities in recent years.  But the final toll will likely be somewhat lower because it is clear that the lethality of the virus has been steadily declining, as I described in this post.  If the fatality rate continues to run on the low end (say 0.2%) and if HIT ends up being even lower, say 40%, then U.S. fatalities would top out at around 250,000.

My guess is that with improving clinical care and the likely development of better therapeutic drugs, U.S. fatalities will ultimately total something in the 200,000-250,000 range.  Assuming all of those fatalities are incremental, that would represent about a 6-9% increase in annual fatalities for this year.

Still An Open Question

Not everyone in the epidemiological community agrees with the lower estimates of HIT.  You will still hear prominent epidemiologists continue to peg the threshold in the 60-80% range.

I have emailed several epidemiologists I have seen making these claims and asked why they did not agree with the lower estimates.  For the most part, those who responded conceded that HIT was likely much lower than the original estimates.  I got the impression they are reluctant to publicly support the lower estimates for fear of giving the public a potentially false hope and discouraging the continued adherence to suppression efforts.

But it seems to me their concern should be exactly the opposite.  If the public sees a light at the end of the tunnel, people will be much more likely to soldier on with precautions.  But if Americans are told that they are going to have wear masks for “several more years” as one Johns Hopkins professor recently suggested, the public is going to rebel and start ignoring the “experts.”

Of course, no one knows whether the lower estimates for herd immunity are correct or not.  Only time will tell.  In the meantime, let’s keep taking reasonable precautions and pray that they are.