Journal Club: The CDC’s Publication on Comparing Reinfection Events in Vaccinated vs. Unvaccinated.

(copyright belongs to whomever owns spongebob squarepants)

A friend of mine reached out to me the other day about vaccine mandates. He owns a local business and some of his staff have approached him about his plan. My first suggestion to him was to issue an anonymous survey to find out the current immunity rate among his employees. I advised him to include natural immunity in the survey to figure out total immune amongst his staff. If the immunity rate at the company was high (say 90%) there really wasn't an issue, if the immunity rate was low then an action plan should be considered to improve the immunity rate.

After receiving further counsel from other business leaders in the area he returned with some questions regarding natural immunity. He pointed me towards a recent CDC study that showed a 2.34x increased risk of re-infection with natural immunity compared to vaccine induced. He forwarded an article from USAtoday with several experts recommending the previously infected get vaccinated.

Later that night I was perusing social media and noticed several of my physician friends sharing the recommendation to get vaccinated regardless of previous infection status

Theoretically, this really doesn’t make much sense. Your immune system has evolved over thousands of generations to fend off foreign attackers. It is a finely tuned machine. It has multiple layers of immunity and memory in place to recall theses defenses should you be attacked again. This is the reason you don’t become deathly ill every time you catch a cold. Covid is just like any other virus, it’s just something your immune system has never seen before, so it can wreak havoc before your immune system mounts an adequate response. If you’ve recovered, you should have the blueprint to protect yourself again. The source of your immunity, whether natural or vaccine induced, should be irrelevant.

So, I had to check out this publication.

The CDC published their findings on their website under their Morbidity and Mortality Weekly Review.

It was a retrospective observation study, which means they were looking back through a database established by the state of Kentucky. It was set up as case-control. The study groups are below.

Cases – participants previously infected in 2020 that had a re-infection in 2021 that were not vaccinated.

Controls – participants infected in 2020 that had NOT been re-infected in 2021 and had the vaccine.

In these studies, the two groups need to be matched as best as possible for certain demographics. In this study they matched them based on age, sex, and date of original infection. They used logistic regression to obtain odd’s ratios to determine the comparative odds for re-infection between the two groups.

To calculate the OR you need a group of cases (individuals that had an exposure to something and became ill), and a control group (individuals that did not fall ill after exposure to some threat).

The formula for an odd’s ratio is fairly simple

(De / He) / (Dn / Hn)

De = Diseased and exposed

He = Healthy and exposed

Dn = Diseased not exposed

Hn = Healthy not exposed

The table looks something like this.

OR = (A/C) / (B/D)

Stay with me.

Here is their table.

They came to an odd’s ratio of 2.34. They used logistic regression, but we should be able to get close to that with a simple calculation from the numbers in the results section.

OR = (.72/.57) / (.20/.40)

Based on the raw numbers provided I get an odd’s ratio of 2.52.

Basically, they found close to 2:1 difference between natural immunity and vaccine induced immunity in the odds of reinfection.

Onto the appraisal.

One of the biggest issues I have with this odd’s ratio is that there is no context. The CDC continues to stub their toe on this important detail. During the 60-day window in this study Kentucky had roughly 27,000 infections. The reinfected unvaccinated represented 0.66% of cases. Kentucky had a total of 270,000 recorded infections from March 12th, 2020 – December 31st, 2020, the cut off point for inclusion in this study. The database found only 246 re-infections in a random 60-day window. If you apply this as a base rate to the other 4 months of the year you get 738 reinfections from a cohort of 270,000. That’s a re-infection rate of 0.27%. That’s pretty darn good. One caveat is that some of these individuals are vaccinated, but even with that, the reinfection rate is miniscule.

That’s not to say the reinfection doesn’t happen, obviously it does. Breakthrough infection after vaccination happen as well. We hear about it every night. However, if put into context, reinfections and breakthroughs represent a fraction of cases, hospitalizations, and deaths. Based on the 2:1 OR established from the CDC you’re looking at a ballpark of 0.27% and 0.13% breakthrough between natural immunity and vaccine induced immunity respectively. A controlled vaccine trial investigating effectiveness in the naturally immune would need to be in the 100’s of thousands per group to find any difference. The absolute risk of reinfection is very low.

The absolute risk is low because both natural immunity and vaccine induced immunity significantly prevent infection and transmission of the virus via neutralizing antibodies and shortened antibody response times. The immune are responsible for a tiny fraction of the cases in the current surge. The current data, when put into context, support this statement.

I don’t really get upset with the CDC recommending individuals with natural immunity get vaccinated. They want to end the pandemic and are probably under extreme pressure to do so. What upsets me is the label of this being strong evidence, which it is not. It’s weak, and it should be called as such. I’m sure the director even recognizes its weak, she’s been in research her entire career. Anyone that has reviewed medical literature would look at this study and recognize that it is not strong evidence. It’s two random cohorts plucked out of a database in one state.

It’s ok to make a recommendation on weak evidence, sometimes that’s all you have, but you should be honest. And you shouldn’t back up the recommendation with an odd’s ratio. It should be backed up by absolute risk. Real, raw, numbers. This is a pillar of evidence-based medicine.

This study meets no criteria for a strong observational study. One of the key elements to a case-control study is the circumstances for exposure (COVID) similar between cases and controls. We don’t know that. They only matched for age, sex, and date of initial infection. COVID doesn’t really care about any of that except maybe age. There are significant confounders not addressed like: county of residence, base infection rates, vaccination rates, urban or rural, race, income, occupation, and (at this point) political affiliation.

Second, the circumstances for determining the results needs to be equal in both groups. Controls were not tested for re-infection during the eligible months. We do not know if these individuals were truly negative. Many could have been asymptomatic, or simply didn’t seek out a test. So, the circumstances for determining the results were not equal between the two groups.

Finally, how strong is the association? Some may think 2:1 is a pretty strong odd’s ratio, but it’s really not. The odd’s ratio for lung cancer and smoking is around 14:1. That’s a strong association. 2:1 could just be noise. When dealing with < 1% event rate, the probability that this is just noise, is much higher.

I’m not saying that those with natural immunity should not get the vaccine. I am saying that natural immunity is robust and protective against reinfection. Those with natural immunity are in a different circumstance than the non-immune.

When we talk about risk and COVID, we should really be using the terms immune and non-immune instead of vaccinated of unvaccinated. In the end, my prediction is that natural immunity and vaccine induced immunity will be equivocal. The non-immune should be vaccinated and not take their chances against COVID. It is my opinion, that those with natural immunity have a personal choice to make, and should consider their current community, work, and home environment in that decision. Companies should have very strong reasons to mandate vaccines in the immune since the risk to benefit ratio is quite obscure at the very low reinfection rates we are seeing.

Reduced Risk of Reinfection with SARS-CoV-2 After COVID-19 Vaccination — Kentucky, May–June 2021


109 views0 comments