[zamalek]

> You don't, but if the death rates are generally stable [...]

For example: if a crematorium has historically dealt with 20 cremations a week and is now doing 100 a week, you can estimate that 80 people are dying from COVID per week. The estimate is likely to be pretty inaccurate, but it's the best source of information that there is.

[roywiggins]

Certainly in NYC, I recall that at the worst of the first wave, excess deaths were very correlated with registered COVID deaths and they spiked simultaneously. There were fewer registered COVID deaths than excess deaths, but during the first spike testing wasn't always done, and people who died at home didn't always get a test postmortem.

When there's a COVID spike, most of the excess deaths are people dying of COVID (some might be people dying due to fear of COVID, or hospitals collapsing- arguably still "COVID fatalities" even if they don't test positive).

[cobookman]

I'd expect in aggregate at the country level, its fairly accurate.

E.g. its unlikely deaths increased 100%+ across all of india due to existing death-causes.

[truth_]

You also have to take into account how many people are not dying of Covid, but some other diseases because the healthcare system is overwhelmed by covid, and they did not get treatment. Which is immensely more sad. Most of these deaths would have been prevented, if not for Covid.

[Spooky23]

Exactly it an order of magnitude number that helps you assess the official numbers.

[sgt101]

>>The estimate is likely to be pretty inaccurate

But let's be clear, inaccurate has different meanings - in this context if the official figure is 20 and it turns out that only 40 have died then that's 100% out on the official side and 50% out on the estimated side. In this context the high figure may have far more utility as well - by producing a more appropriate response from the authorities and society.