> The metric here was "probability that you will be hospitalized IF you are infected with COVID-19".

That's a good metric, but it's not the metric the numbers GP quoted were for. Those numbers were for the metric "probability that you would catch COVID-19 and be hospitalized with it over the past 1.5 years". That's fundamentally something different.

What's your alluding to is also known as infection hospitalization rate, and while numbers differ (it's hard to accurately count infections), in the unvaccinated population it's generally somewhere around 5%.

> That's a good metric, but it's not the metric the numbers GP quoted were for. Those numbers were for the metric "probability that you would catch COVID-19 and be hospitalized with it over the past 1.5 years". That's fundamentally something different.

Where did you see that? I looked up GP and didn't find that 1.5 year time anywhere. Furthermore, it makes no sense in the context of GP, where they said:

> > For example, do you know what the risk of hospitalization from Covid is when unvaccinated?

> If you said >50%, you would be in good company. It's the most common answer.

I don't believe there is any place on earth where you can poll people and reach an average response of >50% to the question "what is the probability that you would catch covid AND be hospitalized for it over the next 1.5 years, if you don't get the vaccine". Let's think for a moment what that would require: first off, it would require a large proportion of the unvaccinated population to get COVID. Second, a large proportion of those unvaccinated infected people would need to be hospitalized. So basically, it would require a place on earth that doesn't yet have significant natural immunity due to previous infections, and doesn't have anywhere close to herd immunity due to vaccinations, and somehow, a place like that would need to be ravaged by COVID in the future. Furthermore, it would need a strain of COVID that hospitalizes something like 70% of those infected, compared to current strains of COVID, which hospitalize something like 1% of infected (remember, most COVID infections do not cause symptoms and thus are not counted in infected counts).

This scenario is so outlandish it is very very clearly impossible. It's so outlandish that you will not be able to find a population where most people would believe such a scenario. Instead, what GP was likely referring to, was the "risk that you are hospitalized IF you catch COVID" (where, again, people overestimate the risks, but it's less outlandish).

> Where did you see that? I looked up GP and didn't find that 1.5 year time anywhere.

In the appendix of the Gallup link from GP, the <1% figure is calculated by dividing the total number of hospitalizations through August 9, 2021 (which is roughly 1.5 year after the pandemic started*) by the average unvaccinated population through that same timeframe. That doesn't result in hospitalization risk if infected, it results in hospitalization risk if infected multiplied by risk of infection.

* Actually, now that I'm looking at the actual CDC data instead of taking their word, the figure they used is for hospitalizations from August 1, 2020 through August 9, 2021, so it should be a year instead of 1.5 year (and their adjustment for reduction of unvaccinated population due to vaccinations is off, so the 0.89% is an underestimation). Doesn't matter for the point I'm making here, though (except to reinforce that we should leave statistics to the professionals).

> I don't believe there is any place on earth where you can poll people and reach an average response of >50% to the question "what is the probability that you would catch covid AND be hospitalized for it over the next 1.5 years, if you don't get the vaccine".

I agree, the people asked probably interpreted the question differently. This makes presenting the result of that question in opposition to the <1% figure misleading, since they are numbers for different things.

> current strains of COVID, which hospitalize something like 1% of infected

This is and cannot possibly be correct. The CDC reports that the US has had 3.1 million COVID-19 hospitalizations since 1st of August 2020, on a population of 329 million. If only 1% of infected is hospitalized, that means by now everybody must've had COVID. That's clearly not the case. Furthermore, if we add hospitalizations from before August 2020, clearly hospitalizations of the population already exceed that 1% number.

You're right though that it's hard to accurately calculate infection hospitalization rate, because counting infections is hard. It has been reported to be somewhere around ~5% in the unvaccinated population, but I can see it being anywhere between 2% and 8% (probably depends on the characteristics of the population under study as well). Since August 2020 the US also reported 39.2 million cases, which gives an upper bound of 7.9%, and assuming three-quarters of infections are asymptomatic gives a lower bound of 2.0%. Note that since a significant part of the population has been vaccinated during that timeframe, and the vaccines protect better against hospitalization than infection, the current infection hospitalization rate is lower than that in the unvaccinated.