Hacker News Reader

Comment by gmarx | Hacker News Reader

Perhaps your April 2020 number is correct but I think it's irrelevant because it is too time-boxed. I did an analysis (matching those by others) using the OECD excess deaths data and if you extend the time period through last fall, Sweden did either best (according to other's analysis) or second best (according to mine). Excess deaths actually continued after the pandemic has subsided in many countries (US and UK at least I recall) and I think it reasonable to assume the pandemic and pandemic policy had something to do with it. To judge overall success of policies you need to look at the whole time of the pandemic and its aftermath and the Swedes did better than almost everyone

View on HN

Due to large numbers of immigration, Sweden has younger population than comparable nearby countries. Too many excess death analyses just take the average of 5 previous years (2015-2019) as the baseline, to compare yearly or weekly deaths to. But in countries with ageing population, number of yearly deaths have been on a slowly increasing linear trend already for a long time. If you ignore the trend, and only compare to the 2015-2019 average, all countries with ageing population would show continuously increasing excess deaths.

Also, Sweden has a peculiar bookkeeping system, and a surprisingly large number of deaths are recorded without precisely known date. If you only look at weekly death statistics, you will miss the "week 99" deaths from the yearly total.

All valid things to check but would also need to be checked for other OECD countries. I would not assume the OECD methodology does not take account of the aging. Also I would not assume this had much of an effect. The way it might would be if a lot of younger people moved to Sweden during the pandemic . In general cross border movement slowed during that time

> I would not assume the OECD methodology does not take account of the aging.

And here you would assume very wrong.

OECD does indeed estimate excess mortality simply by comparing to the 2015-2019 averages:

"The expected number of deaths is based on the average number of deaths for the same week over recent years (in this case the previous five years, 2015-19). This baseline could be considered a lower estimate of the expected number of deaths since both population growth and an ageing population would be expected to push up the number of deaths observed each year."

"Importantly, given the impact of COVID-19 to the overall number of weekly deaths in 2020, the average deaths for the period 2015-2019 continues to be used to calculate excess deaths in 2021, and still applies as the base for 2022 and 2023 excess deaths."

From their: Methodology_All-cause-Excess-and-COVID-19-deaths_OECD.pdf

Which you can find from https://stats.oecd.org/index.aspx?queryid=104676 when you try to find their notes for methodology.

> Also I would not assume this had much of an effect.

Also here you are assuming somewhat wrong.

For Finland, the 2015–2019 average annual deaths is 53723. 2022 deaths were 63219. Thus the simple, OECD-style, estimate for excess mortality is 9496.

Whereas an age-structured model from Statistics Finland estimated 56158 deaths for 2022 from pre-covid trends. So the age-structure-aware excess mortality estimate is 7061 for 2022 for Finland.

I haven't run the numbers, but it seems hard to believe just visually comparing excess mortalities for, say, Sweden and Denmark:

https://ourworldindata.org/grapher/excess-mortality-raw-deat...

April 2020 did a real number on that country...

I can always be wrong but I did a rigorous analysis and other people reported similar findings.

Curious. At the time, I also made a pretty picture comparing deaths in 2020 to the 2015-2019 average (ignore the drop off at the end, as the data was incomplete at the time):

https://i.imgur.com/X6BKdnt.png

Sweden was the only country where the result was clearly distinguishable from noise...

Early on Sweden had more deaths which led people to argue lockdowns worked, but then things subsided and other countries caught up and then surpassed it, leading to Sweden ending up near the bottom of the COVID death league tables (in Europe).

It's also important to remember in all this that the lockdown policy wasn't predicated on making a small difference you need powerful statistics to find. It was advertised as: anyone who doesn't lock down will experience mass deaths and full blown collapse. Epidemiologists claimed Sweden would experience double the usual death rate due to COVID, i.e. as many deaths from COVID as from all other causes combined! Their actual death rate:

https://www.statista.com/statistics/525353/sweden-number-of-...

There's a tiny bump in 2020, but at least part of that is simply noise due to 2019 having an unusually low death rate, so you'd expect it to be higher than normal in 2020 even without COVID.

Yes if you look exactly at the early pandemic then Sweden did worse than its immediate neighbors (but not worse than the rest of the OECD countries). But the real measure of success in policy has to include the rest of the pandemic and its aftermath. The others caught up.

Reply on news.ycombinator.com