What Herd Immunity actually is (hint: it’s not what you think) and why it’s a good reason to be more careful.

Part of the confusion around herd immunity and it’s role in designing policy for the COVID-19 pandemic has been understanding the actual definition of herd immunity. It has been misconstrued or oversimplified in a harmful way by the media and even by some experts. The herd immunity threshold (HIT) is the percent of a population that is required to be immune in order for COVID cases to start to decrease, and it’s dependent on human social connectivity. The more people that the average person comes into contact with, and the more likely that that contact will result in an infection, the faster the spread and the higher the herd immunity threshold. The HIT will be different in crowded cities vs. farm lands and in different cultures and age groups. Cultures who greet each other with hugs and kisses (such as France and Italy), for instance, will have faster spread than cultures that don’t touch when greeting (Japan). Therefore “kissy cultures” will require a greater proportion of people to be immune than “non-kissy cultures” to control spread.

That is, unless they do something to change their behavior. Mask wearing and social distancing change the rate of spread of COVID through communities and thus lower the herd immunity threshold. Closing businesses, schools, and crowded indoor spaces also lowers HIT by making it less likely that people will get close enough to very many other people to contract COVID. Social connectivity is built right into the basic epidemiological model, the SIR model, in a variable called Beta. According to the Wikipedia article on SIR models (screen shot below), β is defined as “average number of contacts per person per time, multiplied by the probability of disease transmission in a contact between a susceptible and an infectious subject”.

The translation of this is, the more socially connected people are and the more likely they are to breath on each other and pass along COVID, the higher the Beta. If people are socially distanced and masked, they will be less likely to spread COVID even if they see the same number of people as in the case of open restaurants that have distancing and mask requirements. This is important because Beta is one component of R0, a variable that many people are familiar with. The higher the Beta, the higher the R0.

So all of this is to explain a key point, the herd immunity threshold is not a constant, it changes with social connectivity. This is actually really good news because it means that with behavioral interventions such as masks and distancing, we can slow viral spread and reach the herd immunity threshold (for that level of openness) with less people actually getting infected and dying, all the while keeping at least some aspects of the economy open. The US has actually reached a herd immunity threshold (for the level of openness) twice now, once during our partial lockdown in late March and April and again at the end of the Summer. Below are daily new cases in the US over time- the arrows show the first two herd immunity thresholds.

What many experts are calling herd immunity is the percent immunity needed to slow spread in a place that is completely open without behavior changes. For COVID-19, this threshold has been estimated at ~67% of the population being immune based on data early in the pandemic in France. There are many reasons why this is just a guess and why it might be different over time or in different places. For one thing, people know about COVID now and this definitely has changed their behavior. For another thing, France is a “kissy culture” and thus one likely to spread COVID faster than some other places. Although, I doubt French people are as likely to kiss each other now after all these months of a pandemic. So it’s likely that the herd immunity threshold now is lower in France than it was at the beginning of the pandemic.

So is the US close to reaching the next herd immunity threshold (for this level of openness)? Yes, actually we are. 45% of North Dakota has been infected for instance. I created the map below to show estimates of percent immunity per state. But remember, the HIT is actually just the percent of immune people required to have cases not spike. So despite ND being at 45%, it’s cases are actually continuing to rise during this current level of openness- so it is not at herd immunity at this time. This is in part due to their reluctance to impose a mask mandate and the giant super spreader event, the Sturgis motorcycle rally. ND may actually be spreading COVID faster (and have a greater R0) than France was in the beginning of the Pandemic.

Florida and New York, on the other hand, have seen only tiny upticks during this next phase of openness and now have leveled off, which means they ARE at herd immunity for their level of openness. They are at 36% and 27% immune, respectively.

Some people have said- “but they aren’t completely open though!” That’s correct. New Yorkers are at least partially wearing masks, businesses are requiring them, and schools are partially open. In Florida, the only thing not open is indoor bars I believe, and mask wearing is scorned and spotty at best. So this is a huge piece of information- herd immunity at 36% and 27% is possible under almost completely open circumstances. And 27% immunity is a level that many states are at or rapidly approaching in the US. So therefore, we could be almost entirely open AND not increase cases with some smart planning.

Some other states are not faring as well this time around. Which states those are was totally predictable (and indeed I did predict this)- at least if you understand herd immunity. As immunity levels go up, the amount of openness can go up without spiking cases. So states that aren’t very immune but have opened up a lot- like opening schools (crowded indoor spaces ie. super spreader environments), will see huge spikes in cases. In the US, this is the midwest and pacific northwest. See Ohio below for example. It’s at 16% and seeing a giant spike.

So what can we do with this information? For one thing, the level of openness to not spread COVID can be determined empirically from the percent immunity of a state, the number of active cases, and the willingness to follow mask mandates and other safety guidelines. People can use this information to determine whether it’s safe to send their kids to school and politicians can use it to determine what level of re-opening they should be enacting. This should also put the mask debate to rest — wear a mask — keep your business open and don’t spread COVID. My analysis of mask mandates showed that they related to less COVID cases in the US with a p-value of 10–16. So they work.

The converse of this is that some states could be more open than they are. There are real harmful effects of being shut down, on the economy, jobs, and people’s mental health. New York, for instance, could have been this open without spiking since late April. Think how many jobs could have been saved and how much better NYC residents would feel right now. I should make the caveat, though, that this type of evaluation needs to be done on at least a county level, because even within states there are areas that are more or less immune.

Lastly, I hope it has become clear that herd immunity is a value that changes with social connectivity and behavior. It is actually a good reason to wear masks and social distance. It is not a strategy. But it could be very useful in designing a strategy that is nuanced and saves the most lives while keeping the economy as open as possible. It could be used to create a win-win strategy that I believe most people would be ok with. It can also be used to predict people’s level of risk for activities in a particular county. The risk is not uniform in the US or even in any state. The more real, trustworthy information that people have to make decisions, the more empowered they will be over their own health. We need to understand and embrace the nuance in the science behind COVID-19 herd immunity. Because otherwise we are stuck between a rock and a hard place, unable to see that there is a perfectly serviceable path, right in the middle, that could save us.