← Back to context

Comment by somenameforme

10 days ago

It's useful when done in good faith. During COVID there were numerous decisions that even if not intended to inflate mortality figures, then they did so inadvertently. In particular the CDC gave extremely broad guidance on what to classify as a death "of" COVID, and the government was giving hospitals additional funding per COVID death. So for the most ridiculous example of what this led to, in Florida some guy died in a motorbike crash and ended up getting counted as a COVID death because he also had COVID at the time. [1] He was eventually removed from their death count, but only because that case went viral.

Even in more arguable cases, preexisting conditions and extreme senescence are ubiquitous in deaths "of" COVID, and at this point there's probably no real chance of ever untangling the mess we created and figuring out what happened. For instance Colin Powell died at 84 with terminal cancer, Parkinson's, and a whole host of other health issues. His eventual death was flagged as 'caused by complications of COVID.' I mean maybe it really was, but I think the asterisk you'd put there is quite important when looking at these stats.

[1] - https://www.snopes.com/fact-check/florida-motorcyclist-covid...

I’m neither an epidemiologist nor a statistician (just a mathematician pretending to be a coder and/or butterfly), but I do not believe there are no mathematical tools to mitigate the statistical impact of comorbidities and accidental misreporting.

To contextualize this: my position is “weak signals are possible even with noisy data”; I read your response as “but the data is really noisy,” which, sure, agreed; the user I was responding to seems closer to the solipsistic position “there is effectively no data at all.”