The ex-Covid death rate in New York is significantly above its baseline. Part of this may be undiagnosed covid deaths, but with less people out doing dangerous things, a lot of it is probably people avoiding hospitals.
ex-Covid rate is more likely due to undiagnosed covid. Especially when you look at the excess death rate compared to previous annual rates. (at least last time I looked at the graphs).
The excess death rate (ex covid) is relative to previous annual rates. I don't think you can disambiguate between "deaths due to undiagnosed covid" and "deaths due to untreated illness due to avoiding hospitals"
You're right. But if the excess mirrors the diagnosed rate (rise and fall) one might reasonably infer that it is driven by covid. Compared to deaths resulting from/driven by people staying away from hospitals which would depend on when lockdowns were announced, when behaviors changed, etc.
Although both mechanisms would bear some relation, it looked to me more like it was driven by spread of the infection rather than changes to behavior.
The point that you can't really separate them completely is well-taken though.
EDIT: I'm clearly all over the place with terminology. Something along the lines of looking at the (all_cause_mortality - covid_deaths - historic_avg) residual and seeing how closely it mirrors say alpha*covid_deaths where alpha is some constant. If it mirrored it well (or for example preceded it and the lock downs in the manner that would be expected of infection) one might infer that those deaths were probably covid. If on the other hand they were strictly related to the time the news broke and lock downs and changes to hospital admittance rates, then it might be better explained as resulting from lockdown issues.
> But if the excess mirrors the diagnosed rate (rise and fall) one might reasonably infer that it is driven by covid. Compared to deaths resulting from/driven by people staying away from hospitals which would depend on when lockdowns were announced, when behaviors changed, etc.
Yea absolutely. I haven't seen yet seen any analysis that does this, but I'm sure one will come along soon
But regardless none of it is any good for anyone.