I expected the second U.S. Omicron wave to start trending downward today. As you can see, that didn’t happen. New cases per 100K per day is steady at 29, down 7% in the past seven days.
But, while this wave seems to be lingering beyond all reason, that doesn’t appear to be due to any change in the ability of COVID-19 to re-infect individuals. At least, not since the original Omicron wave at the end of last year. That’s my (possibly surprising) conclusion from looking at New York State data on COVID-19 reinfections as a percent of total infections.
Data source for this and other graphs of new case counts: Calculated from The New York Times. (2021). Coronavirus (Covid-19) Data in the United States. Retrieved 6/23/2022, from https://github.com/nytimes/covid-19-data.” The NY Times U.S. tracking page may be found at https://www.nytimes.com/interactive/2020/us/coronavirus-us-cases.html
How important are reinfections in maintaining the current wave?
Today I’m going hunting for any recent data on the reinfection rate. I still have this hopeful notion that, eventually, COVID-19 is going to run out of people to infect, and this will finally die back to some low background level of new infections. But, of course, that’s nonsense if most of what we’re looking at is reinfections.
I’ve looked for the reinfection rate in the past, and only found scattered evidence from a handful of states. For example, I don’t think there’s anything about reinfections on the CDC COVID data tracker.
Last time I checked around (Post #1515, May 18, 2022), the only reinfection data I could find was for New York (this page) , which showed that 11% of new cases there were reinfections at that time. Prior to that, the only place I found it was Missouri, which at that time (months ago) showed 8%.
Now, as of the week beginning June 13, New York shows a 14% reinfection rate. Which is, sure, somewhat higher than the 11% rate shown a month ago in that state.
Assuming the situation in New York is reasonably representative of what’s happening in the U.S., the clear takeaway is that is still largely a pandemic of first infections. Almost all the cases you are seeing now are people who are getting COVID-19 for the very first time. It’s people who’ve managed to avoid it for more than two years, who are finally picking up an infection.
Reinfections do matter, but if NY state is any guide, only about one-in-seven current cases is a reinfection.
That said, let me now plot the reinfection data from New York, starting January 2021. I characterize it as weirdly interesting:
Source: Calculated from data from New York State. Read their methods from this page.
Before I even try to interpret the graph above, let me remind you of the rules. Mainly, reinfections are ONLY counted if they occur at least 90 days after an earlier COVID-19 infection. Presumably, this is to avoid the problem noted with PCR tests, which is that they can react to fragments of dead virus for something like 90 days after a person is fully recovered.
In the graph above, start by ignoring the early peak in the reinfection rate. That occurred when new case rates were miniscule. It might plausibly reflect the fact that some particularly vulnerable populations (e.g., immunocompromised patients) represented a large share of cases at that time, and were also inherently prone to reinfection (by anything, not just by COVID-19).
Now look at the area marked as A. This begins with the Omicron wave. As has been noted many other places, Omicron was associated with a much higher rate of reinfection than the prior variants. And you can see the reinfection rate rise right along with the build-up of new cases.
Note that all the reinfections captured in area A are, by definition, re-infections of individuals whose original infection is pre-Omicron. Those are all re-infections of people who had Delta, Alpha, or the original native strain of COVID-19, or some exotic pre-Omicron variant.
Anyone who was infected with Omicron, and managed to get re-infected within that window, simply didn’t count.
Area B, by contrast, finally begins to include reinfections of those who had a prior Omicron infection. Along with reinfections of individuals who were originally infected with some pre-Omicron strain.
Finally, Area B ends in the period when BA.4 and BA.5 are beginning to take over. By reputation, those are marked by an ability to reinfect those whose only immunity comes from an Omicron infection.
You have to interpret that up-slope with extreme caution. Basic arithmetic says that as as the number of people with some prior infection goes up, then, even if the probability of re-infection is constant, the total number of reinfections will rise.
(This is completely analogous to the initial rise of “breakthrough” COVID-19 infections during the first big push for vaccination. The increase in the count of breakthrough infections, at that time, was a natural consequence of the huge increase in the vaccinated population.)
And, recall, the Omicron wave accounts for a very large fraction of all cases.
So, reinfections as a fraction of total might be rising because these new strains (BA.4 and BA.5) are better at reinfecting those who had Omicron. Or it might be rising just due to the sheer increase in the count of those who had some prior infection, due to the size of the Omicron wave.
And the answer is that the size of the Omicron wave, by itself, is enough to account for that upslope. Here’s the math:
- Per the CDC, the cumulative count of COVID-19 cases in New York (including NY City) increased by 78% over the course of the Omicron wave (taken as 12/1/2021 to 3/1/2022).
- Per the graph above, reinfections as a fraction of all infections increased by 75% in Region B of the graph (from 8% to 14%).
Therefore, we don’t need to appeal to any change in the behavior of the virus (BA.4 and BA.5) to account for the increase in reinfections as a fraction of the total. We can account for it fully, so far, just from the huge increase in the fraction of the population that had some prior infection, due to the size of the Omicron wave.
Putting that another way, we’re seeing more reinfections, as a percent of total, now, in New York, because the number of people eligible to be counted for a reinfection has risen so rapidly due to size of the Omicron wave.
I believe that the funny shape of the reinfections curve above is mostly or entirely an artifact of the rules governing the counting of reinfections. That big lump of initial Omicron infections only slowly became eligible to count for re-infections, as it passed the 90-day post-initial-infection boundary that determines whether or not an infection counts in the official statistics.
The upshot is that it appears to be absolutely true that Omicron led to a higher reinfection rate than prior strains. And it is true that reinfections have continued to increase, as a fraction of all new cases.
But so far, BA.4 and BA.5 don’t seem to be causing a markedly higher rate of reinfections. At least, not at a rate that is clear from the available case counts. Instead, we can explain the higher count of reinfections, currently, merely from the sheer volume of people who have now survived some prior COVID-19 infection.
Finally, there is a little fudge factor in all of this. The CDC is total new cases (i.e., new infections and reinfections). That said, I don’t think the rate of reinfections, over this period, is high enough to change the results materially. Most — maybe all — of the upslope of the curve above can be explained by the sheer number of people who have now survived a COVID-19 infection.
