3 ms·

Your comparison points are great, but I think you are misreading the quote: "Death due to Covid-19 occurred in 65 participants in the booster group (0.16 per 10

by nkurz 5y ago

Your comparison points are great, but I think you are misreading the quote: "Death due to Covid-19 occurred in 65 participants in the booster group (0.16 per 100,000 persons per day) and in 137 participants in the nonbooster group (2.98 per 100,000 persons per day)." The death rate of the control group is .16 vs 2.98 per 100,000 people per day. If you annualize this by multiplying by 365 to make it comparable to your other figures, I think this means you get 65 deaths per 100,000 people per year for the booster group versus over 1000 deaths per 100,000 people per year for the double vaccinated but not boosted. Which is to say, the Covid risk is higher than all of the specific risks you mentioned with or without a booster. This might (or might not) suggest a different conclusion of the relative benefit of taking the booster. Or am I the one who is misreading the numbers?

timr 5y ago
That's fair. You can't annualize a daily risk by multiplying by 365 (volatility must be taken into account), but the broader point that annual vs. daily risks are different orders of magnitude is well-taken. Unfortunately, I don't have any examples of risks that low measured on a daily basis. > over 1000 deaths per 100,000 people per year for the double vaccinated but not boosted. We know this isn't even close to true by looking at aggregate numbers. For example, a truly conservative back-of-the-napkin calculation: 800k deaths in the US @ 50M reported cases (which we know is low by a factor of at least 2...) -> 1600 per 100,000. And many of those (half?) happened prior to vaccines; many now happen in the unvaccinated population. It's fairly obvious that 1000 per 100,000 is much too high an estimate. (...and this isn't even the correct number to compare...what you really want is absolute risk per person, which will be far lower than the risk if you get infected, which is what I've estimated above. If you do that calculation, you get ~240 per 100,000 in the US over the entire pandemic. That's somewhere between the annual risk of dying in an accident (1 in 700) and the annual risk of dying from cancer (1 in 500). Generalizing from the sample in the study is problematic for a bunch of different reasons -- not the least of which is that it's an older group of people -- but I didn't want to get caught up in a debate about IFR estimates.
- nkurz 5y ago
  > You can't annualize a daily risk by multiplying by 365 (volatility must be taken into account) Could you expand on this, or suggest the right search terms? My intuition is that it will be a little bit lower than straight multiplication, but that this a low enough probability that we can get a pretty good approximation by ignoring the "impossibility" of an individual dying multiple times during that year. Is there another complicating factor? >> over 1000 deaths per 100,000 people per year for the double vaccinated but not boosted. > We know this isn't even close to true by looking at aggregate numbers. I agree it seems extremely high, but I think this is the clear claim of the paper. There were 85,000 people who were never vaccinated, and over the 54 day period of the study, 137 of them died. Thus if you were to extend the study to a year, you should expect more than 1000 deaths. As I see it, the possibilities are a) the paper is lying, b) I'm grossly misinterpreting what it says, or c) the current death rate in Israel for those older than 50 is about 1% even among the doubly vaccinated. If you have time to read it more closely, I'd love to hear what your thoughts on the discrepancy. Edit: I realize I should've added another option d) the statistics are hopelessly confounded and the results don't mean anything. My current suspicion is that if people who contract Covid early are no longer eligible for receiving a booster, the apparent Covid incidence in the not-boosted arm may be drastically inflated, but I'm not sure if this is a large enough effect to negate the result.

by nkurz 5y ago

timr 5y ago

That's fair. You can't annualize a daily risk by multiplying by 365 (volatility must be taken into account), but the broader point that annual vs. daily risks are different orders of magnitude is well-taken. Unfortunately, I don't have any examples of risks that low measured on a daily basis. > over 1000 deaths per 100,000 people per year for the double vaccinated but not boosted. We know this isn't even close to true by looking at aggregate numbers. For example, a truly conservative back-of-the-napkin calculation: 800k deaths in the US @ 50M reported cases (which we know is low by a factor of at least 2...) -> 1600 per 100,000. And many of those (half?) happened prior to vaccines; many now happen in the unvaccinated population. It's fairly obvious that 1000 per 100,000 is much too high an estimate. (...and this isn't even the correct number to compare...what you really want is absolute risk per person, which will be far lower than the risk if you get infected, which is what I've estimated above. If you do that calculation, you get ~240 per 100,000 in the US over the entire pandemic. That's somewhere between the annual risk of dying in an accident (1 in 700) and the annual risk of dying from cancer (1 in 500). Generalizing from the sample in the study is problematic for a bunch of different reasons -- not the least of which is that it's an older group of people -- but I didn't want to get caught up in a debate about IFR estimates.

nkurz 5y ago
> You can't annualize a daily risk by multiplying by 365 (volatility must be taken into account) Could you expand on this, or suggest the right search terms? My intuition is that it will be a little bit lower than straight multiplication, but that this a low enough probability that we can get a pretty good approximation by ignoring the "impossibility" of an individual dying multiple times during that year. Is there another complicating factor? >> over 1000 deaths per 100,000 people per year for the double vaccinated but not boosted. > We know this isn't even close to true by looking at aggregate numbers. I agree it seems extremely high, but I think this is the clear claim of the paper. There were 85,000 people who were never vaccinated, and over the 54 day period of the study, 137 of them died. Thus if you were to extend the study to a year, you should expect more than 1000 deaths. As I see it, the possibilities are a) the paper is lying, b) I'm grossly misinterpreting what it says, or c) the current death rate in Israel for those older than 50 is about 1% even among the doubly vaccinated. If you have time to read it more closely, I'd love to hear what your thoughts on the discrepancy. Edit: I realize I should've added another option d) the statistics are hopelessly confounded and the results don't mean anything. My current suspicion is that if people who contract Covid early are no longer eligible for receiving a booster, the apparent Covid incidence in the not-boosted arm may be drastically inflated, but I'm not sure if this is a large enough effect to negate the result.