Covid Prevalence vs Covid Vaccination

[M_T]

The first vaccines were distributed 6 weeks ago.  Although some attributed the decrease in cases even back in December to the vaccine, it was clearly too early.  Now, vaccinations may well be affecting the spread of COVID.  Moderna indicated their vaccine provided 95% protection at 14 days after the first shot.  One study suggested that in South Africa, the Pfizer vaccine provided 90% protection 3 weeks after the first shot (ie, at the time of the 2nd shot in the US).

I noticed an apparent negative correlation between the number of people with at least one vaccination and the prevalence of COVID per capita.  Visually:

However, when I calculated the correlation between the percentage of the total population with at least one vaccine to the cases in the last 7 days per 100K population, the result was just -0.13 and the pairs in a graph are

So, there may be a slight correlation, but it isn't obvious from this data.

[ChrisGreene]

This article summarizes protection from the Pfizer vaccine a bit differently from what you quoted:

At 14-20 days, documented SARS-CoV-2 infection risk reduction was 46% among vaccinated persons. At 21-27 days, risk reduction reached 60%. At 7 days after second dose and beyond, risk reduction was 92%.

I have to admit that reading newsy articles about this stuff is always dangerous. The actual nature of the comparison is so important. Did they use the Israeli database to match vaccinated against unvaccinated in the same risk group, age, neighborhoods? 

The numbers above make Pfizer look worse than the J & J numbers you quoted. 

My final point is that the variants are going to make the statistics harder to understand. J & J was evaluated during a period where variants are known to be among the population in large numbers. The P and M were evaluated when there were fewer variants.

[M_T]

This article summarizes protection from the Pfizer vaccine a bit differently from what you quoted:

At 14-20 days, documented SARS-CoV-2 infection risk reduction was 46% among vaccinated persons. At 21-27 days, risk reduction reached 60%. At 7 days after second dose and beyond, risk reduction was 92%.

I have to admit that reading newsy articles about this stuff is always dangerous. The actual nature of the comparison is so important. Did they use the Israeli database to match vaccinated against unvaccinated in the same risk group, age, neighborhoods? 

The numbers above make Pfizer look worse than the J & J numbers you quoted. 

My final point is that the variants are going to make the statistics harder to understand. J & J was evaluated during a period where variants are known to be among the population in large numbers. The P and M were evaluated when there were fewer variants.

Great NEJM report (and, recent data - as of 1 month ago).  Thanks for pointing to it.  The report is as, or more, readable than the news report.  (Yes, they did match across all you mentioned and more.)

I'm not sure how different the results are,  or whether it matters much.  One earlier study shows daily effectiveness numbers climbing from 0% to 75% for days 14-20; the latter says 46% over that period.  The 21-27 days period in the latter paper includes those that got doses at 21 (or 22 or ...) days.

[M_T]

The numbers above make Pfizer look worse than the J & J numbers you quoted. 

Sorry, I missed commenting about this before.  I was worried about the different yardstick confusing people.  I'm sorry if I stated anything that contributed to the confusion.

J&J vaccine primary claim is that it reduces moderate to severe, or any level of COVID by 66-67%.  It reduced sever/critical by 85%.  Both measured at 28 days post vaccine to end of study.   (If you want to negate most influence of the South Africa strain, the US measurements were 72% and 86%.)

The recent study indicates that Pfizer reduced any level of COVID by 94%.  It reduced severe disease by 92%  Both measured at 2nd dose +7 days up to end of study.

[Hmm... a worrisome number in the J&J trial.  For those that got the placebo in the J&J study, there were 11 severe/critical cases in day 14-27 but only 7 for day 28+.  For all cases of any severity there'd were 84 cases in day 14-27 and 112 in day 28+.   This suggests that the trial didn't really go a full 3 months, at least in the US.  I guess a falling case rate in the US could account for that, but I thought J&J data was mostly while case rates were rising.]