How common are COVID reinfections?

A version of this article first appeared in Emily Oster’s newsletter, ParentData.


The latest talk in the world of COVID is BA.5. In combination with BA.4 and BA.2.75 (which inexplicably someone gave the name “Centaurus”), these are the latest omicron subvariants. They share many features with omicron but have variations in their genetic code that render them distinct.

BA.5 has become the dominant variant in the US (we think, based on fairly limited data). This fact implies that it must be either more transmissible than earlier variants or better able to evade existing immunity, or both. Without these features, it wouldn’t become dominant—a new dominant variant needs some advantage. Beyond the survival advantage, our (again, overall limited) evidence suggests that the variant is similar to the original Omicron (or BA.1, or BA.2) in terms of severity.

The presence of BA.5 has raised significant concerns—in some of the messages I get, I would say people are bordering on panic—about reinfection. Omicron in general, but perhaps the new variants even more so, show some immune evasion. We already know that COVID reinfection during the omicron wave was much more common than during the delta or alpha waves. This has translated, in popular discussion, to questions like “Are we all just going to get COVID every three weeks now?” and “I heard that Vaccines and prior infection do not protect you at all anymore.”

You know how we can better answer these questions? Data.

First: Yes, reinfections can happen. Although before getting into data, we should be clear on what we mean here. By “reinfection,” I mean a second (or more) COVID infection. A first infection after vaccination is a “breakthrough.”

Both post-vaccine breakthrough infections and reinfections were much more common in the omicron wave than in previous waves. The evolution of the virus meant that antibodies developed in response to earlier variants, either through vaccine or infection, provided less immediate protection. However: both Vaccines and prior infections continue to provide very good protection against serious illness and death.

How common is this reinfection? It’s difficult to know for certain given the data we have. I’ll talk below about data on the protective effect of prior infection, but that’s a slightly different question. To get a vague sense of the extent of reinfection in at least one population, I surveyed my newsletter Readers and social media followers.

I got more than 21,000 responses. The graphs below show reported infection counts—for individuals and a sample of their children—in this group.

Of course, a random sample would give us different numbers. This group probably leaves out some demographic diversity and range of vaccination status. But what I believe is useful here is to illustrate the balance of one versus multiple infections.

A pie chart showing that, of respondents, 5.7 percent had had COVID twice or more, 54 percent once, and 39.8 percent never.
Emily Oster
A pie chart showing that of respondents' children, 3.7 percent had had COVID twice, 48.8 percent once, and 47.3 percent never.
Emily Oster

About 60 percent of adults who responded here had had COVID at least once, but only about 6 percent reported having it more than once. When asked about their children under 18, about 53 percent of kids had it, but only 4 percent more than once.

I do not want to suggest that these data are a representative random sample. In this group, though, we see that while repeat infections have happened, they are not the mode.

The next question is: How much protection is provided by prior infection/vaccination? I hear from a lot of people, often somewhat terrified, asking, “Am I totally unprotected even if I had COVID before?”

My sense is that some of this fear stems from the feeling that everyone has already had COVID, so all the infections we hear about must be reinfections. That’s not true, though. When we dive into the data we have, it becomes clear that prior infection does provide a lot of protection, even in the current wave.

One way to see that is using data from the UK and from New York state, both of which report the number of first infections and reinfections separately.

Let’s take the UK data first. On July 8 (the most recent fully reported day as of this writing) the UK Health Security Agency estimated 17,179 first infections and 4,607 reinfections, so about 21 percent of total infections were repeat infections. This number isn’t very meaningful on its own; to interpret it, we need to know how many people are at risk for the two categories.

Imagine that in the population, at the moment, 21 percent of people have had COVID sometime in the past. If we then—this week—saw that 21 percent of cases were reinfections, that would mean that infection this week was equally likely regardless of whether you had COVID before. This 21 percent of the population makes up 21 percent of infections.

On the other hand, if (say) 90 percent of people had COVID before, then the fact that only 21 percent of infections were among this population would suggest that prior infection is very protective.

In fact, estimates from the UK indicate that about 90 percent of the population have had at least one infection. Using an analog to a “vaccine efficacy” calculation, this would suggest that having had COVID before reduces your risk of infection by about 95 percent. It doesn’t reduce it to zero, but it’s a huge change. The New York data shows a similar number—about an 88 percent reduction in risk.

These numbers are back-of-the-envelope calculations that are subject to concerns about who has been infected before, their risk factors, and so on. But we see similar things from new published work out of Qatar. This paper used the entire population of Qatar to analyze the protection provided by prior infection and vaccination against COVID. It is important to note that the country has a very young population—only 9 percent of people are over 50, and 89 percent are expatriates from other countries.

This paper predates BA.5, so of course things could be slightly different now, but overall, the authors observe that prior infection is quite protective, and even more so with two or (better!) three vaccine doses. They estimate that prior infection plus three doses is 74 percent to 77 percent protective against symptomatic COVID, and 100 percent protective (again, young population) against severe infection. Of note here is also that two doses of vaccine—which at this point would have been quite far in the past—are protective against serious illness but not against symptomatic illness at all.

What I take from this is that if you have had COVID before, and especially if you’ve been fully vaccinated, you have significant protection against symptomatic COVID. It’s not 100 percent, but it’s really quite high.

However, there is much we do not have detailed information about. Based on some preliminary data out of Denmark and from our general understanding of illness, we expect reinfections to be on average less serious. (There was a preprint about the VA that suggested otherwise, but it is problematic; see discussion here.) However, this understanding is incomplete, and for questions about, for example, long COVID, we are in the dark.

We do not understand well the likely time gap between COVID infections if people are reinfected. There are a lot of assumptions made that people are protected for 90 days, but there seem to be exceptions to that, although they are probably rare. We also don’t have a good sense of variation across individuals. If you haven’t gotten it yet, are you somehow magically immune? (Probably not.)

These are all questions I wish we had better data to answer, and I remain amazed that two years into the pandemic, the US still has not managed to create a better data infrastructure. For a long time I reacted to that by saying I hope they develop it soon (like here, here, and here). But at this point, I’m going to just say it’s too bad we don’t have it and leave it at that.

The bottom line, with our Imperfect data: Reinfection is possible, and has become more common over time. However, prior infection does provide some really substantial protection. Most infections you are hearing about are still first infections.

Future Tense is a partnership of Slate, New America, and Arizona State University that examines Emerging technologies, public policy, and society.


What do you think?


FDA weighs oversight changes after formula, Juul troubles

GOP fake electors ‘targets’ in Georgia election fraud inquiry