3 ms·

Is there a lot of random statistical noise in these esimates of 'normal background mortality'? Prior to Covid you mostly saw these 'excess death' figures in re

Is there a lot of random statistical noise in these esimates of 'normal background mortality'? Prior to Covid you mostly saw these 'excess death' figures in relation to heat waves IIRC, e.g. if 3,000 people were estimated to have died from a heat wave, that figure was obtained by subtracting the actual mortality from the expected mortality over a time period of the heat wave, then reporting that as the death count. The attribution seems a bit challenging in retrospect. Do people make death forecasts, like with weather forecasts, and use that as a test of forecasting skill and the assumption of relatively non-random average outcomes under normal conditions? P.S. check out the CDC mortality dashboard: https://www.cdc.gov/nchs/nvss/vsrr/mortality-dashboard.htm https://www.cdc.gov/nchs/nvss/vsrr/mortality-dashboard.htm

mikeyouse 3y ago
It certainly is challenging, but there are a lot of statistical methods to gain confidence in those estimates. E.g; Heat-related mortality in Europe during the summer of 2022 - https://www.nature.com/articles/s41591-023-02419-z https://www.nature.com/articles/s41591-023-02419-z They looked at the 2022 outlier year in terms of heat across Europe, and then found strong correlations with the hottest weeks in the hottest locations in terms of heat-related deaths from the trends established over the prior 5 years. Seems pretty robust and they did a number of tests to ensure it was a reliable result.
- t0bia_s 3y ago
  To put in perspective, you should compare numbers with mortality caused by cold weather, which is significantly bigger (about 8 times more). It's 4 594 098 deaths by cold and 489 075 deaths by heat during years 2000-2019, source: https://www.thelancet.com/journals/lanplh/article/PIIS2542-5196(21)00081-4/fulltext https://www.thelancet.com/journals/lanplh/article/PIIS2542-5...
  - mikeyouse 3y ago
    Sure, but my original post was restricted to Europe and per that Lancet study, it's only maybe 3.5x more (660k cold vs. 180k hot) -- hence why the 2022 year stood out so much. The authors of the prior study estimated 70,000 excess deaths due to heat in 2022 alone which is nearly 40% of the total for the 20 year period from 2000-2019 -- or about 8x the 'background rate' of excess heat deaths. If we are seeing warming and years like 2022 are going to be more common, that's a really, really big deal.
PhilipRoman 3y ago
>death forecasts Sounds cool. I'd like to see that on the evening news TBH.

mikeyouse 3y ago

It certainly is challenging, but there are a lot of statistical methods to gain confidence in those estimates. E.g; Heat-related mortality in Europe during the summer of 2022 - https://www.nature.com/articles/s41591-023-02419-z https://www.nature.com/articles/s41591-023-02419-z They looked at the 2022 outlier year in terms of heat across Europe, and then found strong correlations with the hottest weeks in the hottest locations in terms of heat-related deaths from the trends established over the prior 5 years. Seems pretty robust and they did a number of tests to ensure it was a reliable result.

t0bia_s 3y ago
To put in perspective, you should compare numbers with mortality caused by cold weather, which is significantly bigger (about 8 times more). It's 4 594 098 deaths by cold and 489 075 deaths by heat during years 2000-2019, source: https://www.thelancet.com/journals/lanplh/article/PIIS2542-5196(21)00081-4/fulltext https://www.thelancet.com/journals/lanplh/article/PIIS2542-5...
- mikeyouse 3y ago
  Sure, but my original post was restricted to Europe and per that Lancet study, it's only maybe 3.5x more (660k cold vs. 180k hot) -- hence why the 2022 year stood out so much. The authors of the prior study estimated 70,000 excess deaths due to heat in 2022 alone which is nearly 40% of the total for the 20 year period from 2000-2019 -- or about 8x the 'background rate' of excess heat deaths. If we are seeing warming and years like 2022 are going to be more common, that's a really, really big deal.

PhilipRoman 3y ago

>death forecasts Sounds cool. I'd like to see that on the evening news TBH.