Analysis of “Reduced Risk of Reinfection with SARS-CoV-2 After COVID-19 Vaccination inKentucky, May–June 2021"

This study was published on the CDC website on August 6th 2021 as part of their morbidity and mortality weekly report.

This is quite a significant study as it purports to be the first real world data to show benefit in vaccinating individuals who have previously recovered from Covid. Previous recommendations on vaccinating covid recovered individuals were based on elevations in antibody titers following vaccination. Given recent press reporting, it was immediately clear to me that people would use this study and likely misrepresent it to downplay natural immunity. I try to explain why natural immunity is far from finished by talking about limitations of the study below. Also what should be obvious is this study doesn’t really look at the relative strengths of naturally acquired immunity Vs vaccination acquired immunity, it is looking at whether vaccination can add further to the robust protection elicited by covid infection.

This study identified individuals in Kentucky aged 18 or over and tested positive on NAAT or antigen testing between March and December 2020, and then subsequently tested positive again in the period between May 1-June 30th 2021 — they identified 246. They then matched them with controls (similar for age, sex and week of first positive test) in a 1:2 ratio — thus 492 controls.

They then looked for vaccination rates in these two cohorts and found that in the reinfected cohort, 72.8% were unvaccinated, 6.9% were partially vaccinated and 20.3% fully vaccinated. In the non-reinfected cohort, 57.7% were unvaccinated, 7.9% partially vaccinated and 34.3% fully vaccinated.

Using conditional logistic regression, they calculated an odds ratio of 2.34 for the non vaccinated to be reinfected. If reinfections are rare, then you can use this odds ratio to give a reasonable estimate of the relative risk reduction after vaccination, which would be a 57% RRR for reinfection. That’s not far from what some vaccines achieved for primary infection.

So now two questions?

1. Is this reason to no longer give “immunity privileges" to covid recovered individuals? Absolutely not, there are a lot of studies showing risk reduction post infection is on par with that of the best vaccines — I link to my previous article.
2. Were the authorities right all along? Should convalescent individuals all get vaccinated? The matter is far from settled, first a reminder that this was a retrospective observational study with a short study period — not easy to infer causation.

Ok let’s say for now we are willing to consider causation. The study still had a few other limitations, they did not correct for testing frequency, if vaccinated individuals tested less often, this could overestimate the protection from vaccination.

As one of my readers on Reddit pointed out, some workplaces actually have a get vaccinated or get regularly tested policy. In addition, for some events, individuals need to either show vaccination status or show a negative PCR, thus it is possible that unvaccinated individuals get tested more. Hence, a lack of info on testing frequency could be a big confounder.

They only matched for age, sex and time of first positive test, not for ethnicity or occupation which have been shown to be independent risk factors for infection. There was no genomic sequencing to confirm whether the reinfections were true or not. However that said even though the majority of the primary infections occurred in November to December 2020, given that this was 5–6 months before the reinfection study period, it is likely that most of these were true reinfections.

I think the most important limitation however is that we do not know the vaccination rates by age group. Does the risk reduction really apply to every age group?

Even in the non-reinfected cohort, the fully vaccinated rate is “only" 34%. We know that vaccination rates aren’t uniform across age groups.

I do not have the exact vaccination rates between May and June 2021, however the above graph shows vaccination rates (at least one dose) in Kentucky by age group at the time of writing this article. There’s a clear increase in vaccination rates with increasing age. We can speculate thus that differences in vaccination rates between the reinfected and the non-reinfected cohorts are more likely to be influenced by the older age groups. We may find that perhaps there was no statistically significant difference in vaccination rates between the two cohorts in younger age groups. This study needs to show the vaccination rates by age groups. But considering the already small numbers we are dealing with, they may not be able to show statistical significance in any single age group. Still even a trend would be useful.

The frustrating thing in this study is that they made no effort to show the protection elicited by prior infection in the first place. 246 reinfections in 2 months in a state with population of 4.5 million where there were total over 465000 covid cases until June 30th 2021, (272000 cases from March to December 2020) suggests that reinfections are rare and that there is significant protection elicited by natural infection. Thus absolute risk reductions from vaccinating prior infected must be very small and numbers needed to treat (NNT) very big to prevent one case of reinfection.

We are also not told about the severity of the reinfections, if hospitalisations were rare in the reinfection cohort (which we have reason to believe as we know reinfections tend to be milder than primary infections), then NNT to prevent hospitalisation must be very big, weakening the argument to vaccinate convalescent individuals further.

Ultimately I still welcome this study, as I have been calling out for real world data to show evidence of benefit in vaccinating convalescent individuals. I would request further studies with longer follow up periods, with breakdown in severity of cases and vaccination rates by age groups. Whether convalescent individuals should be vaccinated is still open to debate in my opinion. Furthermore many of the same arguments I presented will apply for deciding whether currently “fully vaccinated" cohort require boosters.