The cohort I'm talking about is the number of days since a positive test result. When you have many new cases (loaves of fresh bread) and you use simple math, bread/moldy you get the wrong idea about death rate. Click-baity articles, and those with motivated reasoning for "it's just a flu" get breathlessly excited about this as if it were some sort of useful data. See ops assertion that the level of crisis "is independent of the number of people who test positive" for an example of what I mean.
People testing positive is a leading indicator for the size of the crisis.
In NYC, there were 358 hospitalizations and rising on 3/17, and 8 deaths. On 3/31, there were 383 daily deaths and rising. https://www1.nyc.gov/site/doh/covid/covid-19-data.page
Deaths in Texas have been trending upwards since June. https://www.us-covid-tracker.com/?field=newDeaths&state=Texa...
Of course, there's no guarantee that the course of the epidemic in Texas will look like New York City's. But a sharp rise in hospitalizations does seem to presage a lot of deaths.
It’s harder to fudge death numbers compared to case numbers.