@Firefliess I think it works like this.
Take a million unvaccinated people. Ten percent get covid. That's 100,000.
Of that 100,000, 20% end up in hospital. That's 20,000.
Take a million vaccinated people. The vaccine reduces infection by 80%. So 20% of the 100,000 you'd expect without the vaccine get sick. That's 20,000.
Without the vaccine, you'd expect 20% of those to end up in hospital. That's 4000 people. The vaccine reduces that by 40%, so only 2400 end up in hospital.
The vaccine effects are cumulative. So by the time you reduce infection numbers, and then reduce hospitalisation numbers, 20,000 hospitalisations per million people turns into 2400 per million. Huge difference!
With those numbers, the overall impact on hospitalisation rates is 88%. My population is a million people, so you can see that the protection against hospitalisation isn't perfect.
In the American study, the total population was 32,000. Assuming half had the vaccine (haven't read the study yet) that's only 17,000 people. Entirely possible that the vaccine was sufficiently effective in that cohort that noone got sick enough to end up in hospital, just because the numbers were smaller. So it looks as if protection was 100%
This is, by the way, what confidence intervals are all about. If you read the actual study, you won't see that it's 80% effective. You'll see something like "80% (69-94%)". That means that they reckon the protection is around 80%, and they're 95% confident that it lies between 69% and 94%. As a rule of thumb, the bigger the population then the smaller the confidence interval and the more powerful the research, because you reduce the effects of chance.