I’ve been following my local hospital’s stats on Covid patients, and it’s been pretty steady at 15-20% of the overall Covid patient population being fully vaccinated, and about 10% in serious condition/on a ventilator. One thing I’ve never seen anywhere is info on which vaccine those vaccinated, hospitalized people received. It would be helpful information to know if, for instance, almost all of them had received X vaccine. Or if the hospitalization rate roughly followed the distribution rate. Or whatever it might be. Is that info out there?
I would argue that it would be more helpful to use different metrics and denominators and have additional data to protect oneself from implicit bias when one calculates one’s proportions.
Let’s look at an example.
An island has 10000 people on it, 100 are unvaccinated and 10000-100=9900 are vaccinated.
99 / 100 (99%) of the unvaccinated people are hospitalized. 100 / 9900 (~1%) of the vaccinated people are hospitalized. Using your denominator of 99 + 100 = 199 hospitalizations, 100/199=50.3% of the vaccinated people are hospitalized, which one can interpret as an ineffective vaccine or coin flip.
Your metric doesn’t take into account the proportion of the vaccination status of the subpopulation in your city. It assumes away many factors that contribute to vaccination status and hospitalization status.
This and other reasons are why epidemiologists do not use the calculations and denominator that you suggested. Instead, a better starting point is to take into account a different metric that uses the underlying population demographics to estimate vaccine efficacy. In the example above, one can calculate a relative decrease in hospitalization rate that accounts for the underlying subpopulations.
In my example:
Hospitalization Rate of unvaccinated: 99 / 100 = 99%
Hospitalization Rate of vaccinated: 100 / 9900 = 1.01%
Let’s calculate a crude implied efficacy metric to reduce hospitalization. Our reference is the 99% hospitalizations among unvaccinated people. We have:
(99% - 1.01%) / 99% = 98.99%
…which is the implied efficacy of vaccine with respect to hospitalization. In other words, that is the implied benefit (ie., staying out of the hospital) of being vaccinated in the toy example.
Anyway, to your question, JP Morgan Private Bank and JP Morgan Asset Management made a table that gets to what you are asking. As of this writing on 20 Oct 2021, it is the large table on physical Page 9 of this pdf. This report may get updated as the months go on. I am hesitant to paste an image of the table since it’s not my work.
It is implied that they are using the vaccine efficacy metric that I describe. They do not cite their data sources explicitly but at least give clues with respect to where to start. There is definitely data in the JP Morgan table that lives here, specifically the Israeli Ministry of Health data.
Perhaps I didn’t express my question clearly enough. I am in no way suggesting that the vaccines are not effective and good.
However, there are some “breakthrough” cases, where fully vaccinated people still catch Covid and end up in the hospital. This is to be expected, no one ever claimed that the vaccines were 100% effective.
My question is this: Of the breakthrough Covid cases, of fully vaccinated people, resulting in hospitalization, how many of those people got the Moderna shot, how many got Pfizer, and how many got J&J?
I didn’t interpret your question as implying that the vaccines weren’t effective - I was trying to clarify the metric and how to think about efficacy among the different vaccines and their outcomes as a prelude to the table.
Have a look at the pdf report and specifically Page 9. They break down the efficacy by vaccine formulation, including the lesser known ones. Their reference group (“Ref group” column) is unvaccinated patients in every comparison except one. They looked at 5 different outcomes of severity.
I’ve seen two studies on breakthrough infections resulting in hospitalization in recent months (Delta variant months) that show some loss of immunity with Pfizer and Janssen, but not Moderna.
The numbers are still very good though.
The other paper (data from Minnesota) also looked at breakthrough infections in general. There are limited data points so I’m not sure how good the data is. There seemed to be a drop in protection from Pfizer in the 5th month post-vaccination while Moderna seems to perform a bit better.
Bleach- Thanks for the clarification. I checked out page 9, and I do see what you’re saying. Lots of good info there.
Tfletch1, thanks, good info there as well.
It’s waaay more than than if you look at recent data instead of cumulative data. Minnesota, Michigan, Pennsylvania, etc are showing 30-50% vaccinated deaths or more. It’s in the 60% in England. Here’s the report on Sept vax deaths in Vermont:
The 76% figure in the headline uses the exact same biased statistic that was described upthread and is not useful for considering the efficacy of vaccines. The reporter is using the wrong denominator to look at vaccine efficacy; the Vermont Health Department spokesman tried to make this point, that this death fraction is a misleading statistic when you don’t take into account the underlying proportion of vaccinated people.
We can play this game too:
8/8 unvaccinated people died from an infectious disease
25 of 1000000 vaccinated people died from an infectious disease
25/33 deaths were in vaccinated people.
What group would you want to be in?
@bleach explained this already. You’ll obviously have a higher percentage of breakthrough cases in a highly-vaxxed population because most people are vaccinated. 89% of eligible people have at least one shot and 80% fully-vaxxed. Deaths are much higher in older populations. In Vt, 97% of > 64 yrs old are vaccinated. @bleach shows you how you can calculated the actual rates for vax vs unvax.
Another way to look at efficacy of vaccinations is compare the recent 7-day death rates in Vt vs an unvaxxed state like TN. At the peak of the delta surge, TN experienced daily death rates of 7 per 100,000. VT’s peak is more like 1.7 per 100,000. That’s a significant difference.
This is called the “Base Rate Fallacy”.