The CDC hits a new low

(I didn’t think it was possible either.)

But yesterday the Centers for Disease Control, America’s not-at-all-politicized public health agency, released a new study purporting to show that vaccination protects against Covid infection better than natural immunity. Of course, a wave of stories about the benefits of mRNA vaccination followed.

To do this, the CDC used some magic statistical analysis to turn inside raw data that actually showed almost four times as many fully vaccinated people being hospitalized with Covid as those with natural immunity - and FIFTEEN TIMES as many over the summer.

I kid you not.

Further, the study runs contrary to a much larger paper from Israeli researchers in August.

As my two-year-old likes to say, How dey do dat?

Well, the Israeli study drew on a meaningful dataset in a meaningful way to reach meaningful conclusions. It counted infections (and hospitalizations) in a large group of previously infected people against an equally large and balanced group of vaccinated people, then made moderate adjustments for clearly defined risk factors.

It found that vaccinated people were 13 times as likely to be infected - and 7 times as likely to be hospitalized - as unvaccinated people with natural immunity.



In contrast - how do I put this politely? - the CDC study is meaningless gibberish that would never have been published if the agency did not face huge political pressure to get people vaccinated.

Let’s take a look.

The study had a bizarre design.

The agency’s researchers looked at 200,000 people who had been hospitalized with “Covid-like” illnesses from January through August in nine states. Right away, this choice sets up the study in a problematic way; for most of that time, people who had received Covid vaccines believed (because the CDC and others told them) that they were at VERY low risk of getting Covid, and certainly symptomatic Covid. Thus they may have been less likely to go to the hospital at all, or be tested for Covid once they arrived.

But put that aside.

Then the researchers decided to compare two groups - people who had definitely had Covid at least 90 days before and received another Covid test around the time of their hospitalization and people who had been fully vaccinated at least 90 days (but no more than 180 days) before and received a Covid test around the time of their hospitalization.

This choice is also bizarre. Those of you who have been paying attention will know that this date range is designed to make the vaccines look as good as possible by testing in the happy vaccine valley, the short period when mRNA vaccines are at maximum effectiveness (in fact, they are probably starting to lose it by the sixth month).

But more importantly, this criteria excluded the VAST majority of the people hospitalized with Covid-like illnesses or tested for Covid.

Only about 1,000 people out of the 200,000 people hospitalized for Covid-like illnesses over the eight months had a previous documented Covid infection. (Given the fact that at least 20 percent of Americans, and probably more like 40 percent, had had Covid by the spring of 2021, this is a strikingly small percentage - and certainly doesn’t suggest long Covid is much of a threat.)

And only 89 of those 1,020 people with natural immunity tested positive. In contrast, 324 out of the 6,328 vaccinated people who met the study’s criteria tested positive.

But isn’t 324 more than 89?

It sure is. And the CDC didn’t have - or didn’t publish - figures on how many people were actually in the two groups - those with natural immunity and those infected. Instead it compared the PERCENTAGE OF POSITIVE TESTS in the two groups.

But why would the percentage of positive tests matter, when we don’t know how many people were actually at risk?

Great question.

But, amazingly, the statistical manipulation then got even worse.

The natural immunity group had an 8.7 percent positive test rate. The fully vaccinated group had a 5.1 percent positive test rate. So the natural immunity group was about 1.7 times as likely to test positive. (1.7 * 5.1 = about 8.7.)

With such a small number of people in the natural immunity group, that raw “rate ratio” may well have failed to reach statistical significance. (We don’t know, because the CDC didn’t provide an unadjusted odds ratio with 95% boundaries - something I have never seen before in any paper.)

Instead, the CDC provided only a risk ratio that it had adjusted with a variety of factors, including “facility characteristics [and] sociodemographic characteristics.”

And finally, the CDC’s researchers got a number that they could publish - hospitalized people who had previously been infected were five times as likely to have a positive Covid test as people who were fully vaccinated. Never mind that there were actually four times as many people in the second group.


By the way, buried at the bottom of report is some actual data. And it’s bad.

The CDC divided the hospitalizations into pre- and post-Delta - January through June and June through August.

Interestingly, the number of hospitalized people with natural immunity actually fell sharply over the summer, as Delta took off. About 14 people per month were hospitalized in the winter and spring, compared to six per month from June through August. (Remember, this is a large sample, with hospitals in nine states.)

But the number of VACCINATED people being hospitalized soared - from about three a month during the spring to more than 100 a month during the Delta period. These vaccinated people still were less than 180 days from their second dose, so they should have been at or near maximum immunity - suggesting that Delta, and not the time effect, played an important role in the loss of protection the vaccine offered.

Now that’s a finding worth pursuing.

Don’t worry, the CDC will get right on it.