There’s maybe an amusing counter point from this study. Here they ran a trial with a highly at risk population (due to comorbidities) and reduced hospitalization by 50%. I guess it’s possible the drug is 100% effective and the hospitalization of the remainder were driven by other factors, but even in that case you’d be overestimating hospitalization by just a factor of two. And since none of the people who received the drug died, I guess it’s good indirect evidence that we’re counting covid deaths among the high risk, diagnosed population reasonably well?