Vol. # 2 Kentucky

In my prior post, I accidently left out discussion of the study done by the DOH in Kentucky and published in the MMWR. This study asks a different question than the ones I cited in my prior post, but I should have included it anyway.

The authors looked at the vaccination rate in pts, who were infected in 2020 and then again in May 1 – Jun 30, 2021 and compared them with the vaccination rate of pts who were infected in 2020, but not reinfected in May – June 2021. The rate of vaccination in the reinfected group = 20%. The rate of vaccination in the group who did not experience reinfection = 34%.

The authors calculated the Odds Ratio (OR) for re-infection between the Case group and the Control group. The OR is just one rate divided by the other. In this situation, the OR = the rate of reinfection of unvaccinated previously infected pts divided by the rate of reinfection of vaccinated, previously infected pts.

The OR = 2.34, meaning unvaccinated pts with prior infection were 2.34 more times likely to be reinfected than fully vaccinated pts.

So, this is a different issue than the other studies addressed. This study is trying to determine if post-infection vaccination reduces re-infection and is not, like the studies I cited, the rate of re-infection in pts with prior history of Covid.

246 Cases of reinfection were found. Each Case pt had a laboratory-confirmed infection sometime in March-December 2020 and again between May 1 and June 30, 2021. Reinfection had to occur during May 1 - June 30, 2021. That’s a small window.

The Controls were matched for age, sex and, in 2020, had to test positive for SARS-CoV-2 within a week of the matched Case subject. Each Control had been infected during 2020, but NOT re-infected during May 1 - June 30, 2021.The Control group was made twice as big as the Case group. Obviously, they had plenty of Controls to choose from and chose two matched Controls for each Case subject.

“Kentucky residents who were not vaccinated had 2.34 times the odds of reinfection compared with those who were fully vaccinated (odds ratio [OR] = 2.34; 95% confidence interval [CI] = 1.58–3.47).”

The data from Table 2 in the MMWR article are copied below with a modification to include the % reinfected by vaccine status. For instance, 38.7% of Never Vaccinated subjects was reinfected in May-June 2021 compared with only 22.8% of Fully Vaccinated subjects. In other words, the last column is new.

Table 2 with additional column on % reinfected.

NV- Never vaccinated; PV- Partially vaccinated; FV- Fully vaccinated.

To determine a rate of reinfection, the authors had to combine Control and Case subjects. All the Controls were not reinfected, and all the Cases were reinfected. They then compared the two rates to calculate an Odds Ratio. This rate determination is highly flawed. The number of Controls was contrived Specifically, the ratio of the Control group total to the Case group total, 2:1, was just made up. If one chooses more or less Controls, the Odds Ratio changes significantly.

Maybe I am missing something.

In the table below, I compare the reinfection rate in NV pts to the reinfection rate in FV subjects. However, I use increasing number of Control subjects, while keeping the number of Case subjects the same. Remember, all the Case subjects were reinfected. None of the Control was. The Control group: Case group ratio originally = 2:1. I did not include the Partially Vaccinated subjects from this analysis, since the article did not find any difference with this group.

In the Table below -

In Column C,

Row 2, the number of Controls used in the study.

Row 5 = Control group x 2.

Row 8 = Control group x 20.

Row 11 = the Control group x 0.2

As you can see, the percentage of reinfected pts drops, as expected. However, the ratio of the Never Vaccinated reinfection rate to the Fully Vaccinated reinfection rate varies based upon the size of the Control group. Add a bigger Control group, then the ratio increases; add fewer and the ratio decreases.

Maybe I am missing something.

But this article also suggests a much, much higher rate of reinfection after infection, even with vaccination, than other studies have shown. I am having trouble understanding the math done here, but I think the study is saying that the reinfection rate in unvaccinated subjects is well over 30% and the reinfection rate in vaccinated subjects is still over 20%. That’s another strange aspect of this study; they do not tell what the rates are, just the ratio of the rates of reinfection and That’s not right - https://www.youtube.com/watch?v=akTCiwEa59g

The previously cited studies on reinfection of HCWs was very low. These HCWs were not vaccinated. The rate of reinfection in unvaccinated Kentuckians was >> greater than the rates reported in those 5 studies.

The Covid vaccine studies reported protection of ≥90% compared with placebo. That doesn’t mean 10% of the vaccinated group was infected. It means that the percent of infected in the vaccinated arm was 10% of the percent infected in placebo arm. The infection rates of the placebo arms were always much less than 10%. For instance, in the 12/10/2020 NEJM article on the Pfizer vaccine, there were ~20K people in the placebo group and 20K in the vaccinated group. The number of Covid cases in the placebo group = 169; so, the rate of infection in the placebo group = 169/20,172 = 0.8% or less than 1%. The number of infections in the vaccinated arm = 9; the rate of infections in the vaccinated group = 9/19,965 = 0.045% or much, much less than 1%.

The reinfection rate for Kentuckians, who were fully vaccinated, was >100-fold higher than 0.045%.

Maybe I am missing something.

Addendum: The incidence of Covid from May 1st through June 30th was decreasing steadily and the lowest it had been since June 2020. In other words, delta hadn’t surged in Kentucky during the reinfection period.

SMS

P.S. – I emailed the corresponding author of this study and asked for an explanation of the math behind their calculation of the Odds Ratio. If I hear back, I will let y’all know. Maybe I am missing something.