Excess mortality from March to December of 2020 | Hacker News Reader

Excess mortality from March to December of 2020

ourworldindata.org

142 pointsaronowb14111 comments

Excess mortality seems like the best way to track the pandemic because determining whether someone died of covid can be debatable sometimes, but there can be no debate over whether someone died.

I know at least one person who’s having problems exacerbated by lack of mobility. Atrophy in old people can be a big problem.

I would not go so far as to support anyone claiming this is why we should be breaking quarantine, but can’t quite support racking them all up as COVID deaths either. Reality is somewhere between the two (but close to “most” than “few”)

On the other hand there are other deaths that were reduced, like automobile accidents and just accidents and general.

I am confused by this. Either this person is only mobile in indoor spaces for some reason or this person lives somewhere they were doing police checkpoints if you leave your house (like Bangalore). I've spent at least 2x more hours outside, walking, riding bikes, hiking, hitting golf balls, etc. There's nothing else to do but be outside. The fact that I don't have to go into the office and no longer can or want to visit breweries and restaurants has been a great boon to my health and Vitamin D intake.

Hm, interesting. I’m guessing it’s the lack of socialization more than a lack of movement that’s affecting people’s health.

Changing routines is hard. Just because people can spend time moving about outside doesn't mean they will. Just like because people can eat healthy food in appropriate amounts doesn't mean we don't have an obesity problem.

For example, someone could have used to walk to work, or walk to lunch. Or just spent more time walking about an office building than their home office. Or had a routine of hitting the gym after work, but be unable to find the same motivation to work out at home.

"Or had a routine of hitting the gym after work, but be unable to find the same motivation to work out at home. "

I had a routine of hitting the gym between 13.00-15.00, when it is at its emptiest. When the deadlocks came, I bought some rudimentary equipment home, but the experience sucked... Polished hardwood floors are just too slippery.

Fortunately, they opened the gyms on Thursday again and I am going back to my previous routine.

I get the motivation thing but what else is there to do on a Saturday besides go hiking or take the dog on a long walk. We’ve been through this for eight months now and people should be getting outside just for sheer boredom.

It really depends. My outdoor activity levels are way down. I used to hit 2-3 miles a day just going to work, lunch, the grocery store, meetings, etc which I don’t get anymore.

It’s not that I don’t have the Time to make it up, but there’s a lot of days where I only left the house because I had somewhere to go (cold, rainy, busy, etc). There’s a huge difference between “I have to walk to the store because we’re out of bread” and “I have to walk around the block because I need exercise”

If you start looking at data more broken down by cause, there are huge surges in (not-confirmed-Covid) “pneumonia”, “stroke”, etc. deaths which very closely follow the distribution in place/time of confirmed Covid deaths.

The pattern is very clear and obvious, and not at all what you would expect if the primary cause were just something to do with being in lockdown.

There are quite a few risk increases associated with general shutdown. This is one with large impact. Another large impact is mental illness, in particular presenting in suicide and substance abuse. On the other side, many other common risks are reduced: workplace injury, other transmissible ailments.

If the purpose of the analysis is to determine whether the preventive measure is effective and not overwrought, an excess deaths analysis is a good one because it helps infer what the toll would be in absence of the measure.

If the purpose is to attribute cause of death, it’s just a heuristic. But a useful one when other methodologies are (rightly or wrongly) suspect.

>If the purpose of the analysis is to determine whether the preventive measure is effective and not overwrought, an excess deaths analysis is a good one because it helps infer what the toll would be in absence of the measure.

How do you know what the toll would be in the absence of the measure? Excess deaths tells us how many deaths are from virus+shutdown. I don't see how we can use this data to find out how many deaths there would be in a virus (no shutdown) scenario, or in a shutdown (no virus) scenario.

> workplace injury, other transmissible ailments.

For white collar workforce, it just becomes at-home injury, whatever it is. Transmission generally might not be so greatly reduced either - there are jobs which have to go to work (like my wife who is a doctor) so whole household is exposed, there are schools/kindergardens still open in many countries (so our baby can easily bring covid from it while I haven't been sitting in the office since mid march).

Using face masks and washing hands might be helping more with spread, something that was ridiculous and laughed at in western culture still 1 year ago. I was a fan of it even for regular flu as done in south east asia for a long time, but no way you could have done it here.

> Another large impact is mental illness, in particular presenting in suicide

At least one study didn't find an uptick in suicides as a result of the lockdowns. Situation may differ in America because it's run by idiots that cut off state income support and didn't commit to locking down till the situation was under control.

https://www.thelancet.com/journals/lanpsy/article/PIIS2215-0...

As a counterpoint, Japan seems to be undergoing a startling rise in suicide rates.

https://edition.cnn.com/2020/11/28/asia/japan-suicide-women-...

Japan and China always had problems with suicide, because of culture.

That doesn't explain the rise in the graph this year, does it?

Interesting quote from your article:

"We didn't even have a lockdown, and the impact of Covid is very minimal compared to other countries ... but still we see this big increase in the number of suicides," said Michiko Ueda, an associate professor at Waseda University in Tokyo, and an expert on suicides.

A bit further down:

> Women make up a larger percentage of part-time workers in the hotel, food service and retail industries -- where layoffs have been deep.

> In a global study of more than 10,000 people, conducted by non-profit international aid organization CARE, 27% of women reported increased challenges with mental health during the pandemic, compared to 10% of men.

> Compounding those worries about income, women have been dealing with skyrocketing unpaid care burdens, according to the study.

Doesn't exactly sound "very minimal" for women.

This is not an article in an academic journal and we don't know how scientifically valid the CARE survey was - which btw wasn't specifically about Japan. It would almost certainly cover countries that were unable to effectively subsidise their shutdowns. Having said that, it does intuitively make sense that the economic down turn would place additional stress on those most affected by it.

The lockdown, however, was not the source of the economic down turn, but rather the virus roaming about killing people was the cause. An effective lockdown, supported by government subsidies and contact tracing, has been proven to be beneficial to the economy. After the shutdown, consumers can roam about in the comfort of knowing that they have almost zero chance of contracting the virus. Australia and New Zealand are now so confident in their success that they can fill entire stadiums with people as though a vaccine had already arrived.

That's not a great source.

https://www.medrxiv.org/content/10.1101/2020.10.06.20207530v...

> By August, the total number of suicides was 7.72% higher than the average number of suicides in the same month of the previous three years.

Still though, as noted by the paper, there hasn't been a shutdown in Japan:

> Without introducing lockdown measures or strict domestic movement restrictions during the state of emergency, authorities requested non-essential businesses to close or opt to work remotely. Stores and restaurants were asked to operate for reduced hours. The state of emergency was lifted on May 25, 2020.

Yes economic down turns can increase the suicide rate shock, but shutdowns are obviously not a factor for Japan as they haven't imposed nearly as onerous containment measures as other countries. The main reason for Japan's current plight is that its economy is heavily susceptible to external economic shocks such as decreased global consumption due to an ongoing pandemic.

https://www.businesstoday.in/current/world/covid-19-effect-j...

I'm not sure there are -net- additional risks with shutdown though. Fewer miles driven means fewer road deaths. More isolation and precautions also lead to fewer non covid deaths like flu and other more serious communicable illnesses, etc.

> There are quite a few risk increases associated with general shutdown. [...] Another large impact is mental illness

Although there clearly is a harmful impact on mental health caused by "general shutdown", it's not clear that it's a net negative. Compared to the potential alternative (higher mortality, overwhelmed health systems, and anxiety about insufficient measures being taken) it may be a net positive.

Don’t you think the evidence for this level of infringing on peoples basic rights should be less flimsy than ‘may be a net positive’?

Yes, but we're not limiting people's freedoms to avoid a negative mental health impact. That's a separate observation.

We have these restrictions to avoid collapse of our health systems. The hospitalisation rate of an uncontrolled spread of the virus far exceeds any country's healthcare capacity. The maths is inescapable.

Clearly the example of Sweden shows that that isn’t true.

Are you arguing that Sweden is allowing an uncontrolled spread of the virus? They've had strong voluntary measures in place all year.

In the past month they've also imposed several enforced measures to restrict the spread of the virus, including a limit on gatherings to 8 people, a bar curfew, and they've just closed secondary schools for at least a month.

"Don’t go to gyms, don’t go to libraries, don’t host dinners. Cancel." - Swedish Prime Minister Stefan Lofven.

Despite these measures they are seeing rising hospitalisations. Further measures look inevitable.

Or if not, what are you arguing?

I love how the No True Swedish Man, when the numbers are going down, is following the voluntary measures to the dot but when the numbers are going up retroactively is not.

By the way despite all the drama hospitalizations are still not at the level they were beginning of the year. But I’m sure that’s because they are following the rules again?

What about all the traffic accidents which didn’t happen due to shutdown?

Shhh we don't talk about that! The car industry brings lots of jobs and is nothing but good.

domestic violense is up which hurts kids too. Millions will die of starvation around thec world because of extended lockdowns in te west and we could go on.

> because of extended lockdowns in te west

I'd rather phrase it as "because of inadequate social support during lockdown". A few countries paying low support had money to do more. Blaming that just on long lockdown is missing all the other possible solutions.

As for the effect on mortality of the lockdowns themselves, it will be interesting to look at Australia in a few months where the infections stayed relatively low, but lockdown lasted months. Especially comparing Vic to NSW.

I agree but I am talking about the consequences on third world countries of a slowing economy.

The economy would have slowed down anyway, lock down not though.

> What about all the traffic accidents which didn’t happen due to shutdown?

That represents only a few hundred lives, it seems. Deaths per vehicle mile traveled are actually up.

https://www.caranddriver.com/news/a34240145/2019-2020-traffi...

We don't rack them all up as covid deaths.

We combine excess mortality with other stats.

One of the reasons excess mortality has attracted so much attention is because there are a bunch of people saying that covid just isn't killing many people and we point to the pile of corpses and ask "what happened to them then?"

The main problem is that different people have different definitions what pile of corpses actually is.

For me covid is not a big deal and it should have been ignored. And I have came to this conclusion looking at the official data. For a lot of people that are very passionate about masks and lockdowns this is incomprehensible - that I could look at the data and come to different conclusion.

That is the point - for me covid isn't killing many people. It is killing tiny amount relative to the global population and it is usually people that were on the way out.

Which ones again shows how difficult is to argue with someone when your basic axioms, values and definitions mismatch.

> and it is usually people that were on the way out.

You haven't looked at the data closely enough if you believe this.

Covid is killing people an average of 10 years early. Most of these people needed no assistance in their day to day life before they died. This isn't the frail, very elderly, population you think it is.

I have seen extremely strong amplification of risk by stuff like obesity, diabetes and the rest of the metabolic syndrome buddies.

A lot of the dead were having chronic conditions or were already old.

Turns out a lot of people have heart conditions or diabetes. Doesn't meant they were going to die.

> For me covid is not a big deal and it should have been ignored.

In that case, how do you avoid your health system collapsing? The hospitalisation rate for an uncontrolled spread of the virus far exceeds the capacity of any country's health system.

The only option in this case is either to introduce drastic measures to reduce the spread of the virus or accept that very large numbers of people will die at home without oxygen or other medical care.

> for me covid isn't killing many people

If you're doing x to suppress y, you can't claim that low y is a reason not to do x. You have to consider how much y would be if you weren't doing x. And that y would be far higher.

Consider the argument that we should not bother with antibiotics because bacterial infections don't kill many people.

> Which ones again shows how difficult is to argue with someone when your basic axioms, values and definitions mismatch.

It's interesting that not many people are arguing that huge numbers of people would die without measures and that that would be acceptable. People tend to fall into two groups: either restrictions are necessary to avoid large numbers of death, or that restrictions are not necessary because you wouldn't have large numbers of deaths.

This suggests that the argument is not a mismatch of values, but one of whether the data are correct.

Assigning blame for various components of the overall mortality is important for those who can change the policy, but for lil' old me it's very great to see at least one indisputable number (excess mortality - hard to fake the fact of death) that can be trusted to represent the magnitude of what we're dealing with in 2020, and which, yes, includes all deaths caused by (or avoided by) various policies. Everything else quickly becomes a jumbled opinionated mess which depends on how deaths are attributed and quickly loses any objectivity and meaning.

The problem is that due to significant and pervasive behaviour changes, other forms of mortality and altered.

In some places, there are apparently many fewer people dying from the flu, because, well, we're isolated from one another!

Similarly, fewer car accidents. Suicides are probably altered, as well as violence in some cases, in particular domestic violence.

People forgoing treatments for any host of disease due to either fear of going to the clinic, or delayed treatments.

etc..

The other factor little talked about in Europe is aged carers - usually from Eastern Europe and the FSU - rushing to return home to their families in the midst of border closures, lockdowns, and flight cancellations.

This lead many residents to die of neglect, or for COVID to sweep easily through many centres if the remaining staff worked whilst symptomatic and took shifts across multiple locations.

This is part of the reason why countries in EE and the FSU, as well as those with less strict lockdowns (ie. Sweden), had lower mortality rates than in places like the UK, Spain, Italy, France.

It might be worth reconsidering the whole aged care model, and advancing the idea of voluntary euthanasia for the elderly.

What do you mean by that last paragraph? Because on first read it sounds like you're suggesting something horrifying and I hope there's a more generous interpretation I've missed.

People should be able to go out on their own terms, instead of deteriorating in an aged care centre. It should be completely legal for those aged 80 and over, for example, to seek to end their life - if they decide they have lived all they want, and don't want to suffer the problems associated with old age. This also allows them to choose their time of death, and for their family to remember them in a state of relatively good health.

This data is BS.

Firstly, it's nonsense to take "average amount of deaths" over the past few years and extrapolate it to the current year without accounting for population growth changes, and rate of population growth acceleration.

Secondly, if you look at the "expected U.S. Death Rate" data[1], it shows that they expected U.S. death rates (per capita) to rapidly accelerate about 7 years ago, but the rate of deaths didn't really begin that acceleration cycle until ~2 years ago:

2017: 858.0 deaths per 100,000 predicted

2018: 868.5 deaths per 100,000 predicted

2019: 878.2 deaths per 100,000 predicted

2020: 888.0 deaths per 100,000 predicted

----------------------------------------

2017: 731.9 actual deaths per 100,000 [2]

2018: 723.6 actual deaths per 100,000 [3]

2019: 867.8 actual deaths per 100,000 [4]

2020: Could be anywhere between 835-910 depending on the population data and total death numbers you use to calculate

----------------------------------------

Not sure why people aren't dying as fast as they initially thought (probably modern medicine, etc), but that simply means there were more old people on the verge of dying coming into 2020. So even if 2020 shows a per-capita increase in deaths, it's to be expected.

[1] - https://www.macrotrends.net/countries/USA/united-states/deat... [2] - https://www.cdc.gov/nchs/data/nvsr/nvsr68/nvsr68_09-508.pdf [3] - https://www.kff.org/other/state-indicator/death-rate-per-100... [4] - https://www.cdc.gov/nchs/fastats/deaths.htm

Did you look at the "Excess mortality using raw death counts" chart? It's very very clear that 2020 is a huge outlier. Is it possible some of their numbers are a bit off? Sure. But the size of the difference in 2020 dwarfs that.

I'm more inclined to believe the organisation literally called 'our world in data' as an authority for data research. Your analysis just seems incomplete and I highly doubt they would just leave something out or use bad data. Look at their formulas to get a better idea why.

The CDC has an "expected death" curve which reflects the time trend (which thus includes population growth and aging) and seasonality of death.

U.S. deaths this year are well above that curve: https://imgur.com/p4kBgXF

We are on track to have 3.2 million to 3.3 million deaths this year. That's a crude death rate of 970 to 1000 deaths per, well above the megatrends (U.N.) forecast of 888.

True.

So take a look what this guy found out: http://market-ticker.org/akcs-www?post=240767

Especially in the second peak, excess mortality also include patients with other underlying health conditions that couldn't, or didn't want to, access health services. In the UK cancer referrals have been at an all time low for example.

To be accurate, we would also need to know the number of deaths attributed to the policies established to prevent COVID deaths and increase hospital capacity. What were the number of excess deaths from other causes - due to fewer doctor visits, fewer diagnostics, fear of going to the hospital, discharging early, etc?

But what are you going to do with that information? It's important to know if someone died of Covid or because of our measures against it. Or even of Covid because of our measures against it (stress weakens the immune system + getting infected anyway).

> there can be no debate over whether someone died.

Sure, but their deaths can be attributed to lockdowns. And when the vaccine is out, their deaths will probably be attributed to the vaccine.

Measuring excess mortality is not going to convince a covid denialist.

Excess Mortality is a tool used to attribute deaths to _the pandemic_. It is helpful for answering questions like "How well is <country> handling the Pandemic?" and for comparing performance between countries and comparisons over time.

It conflates "deaths caused by pandemic" with "deaths caused by reactions to pandemic" with "deaths caused by random things that accidentally happened during the pandemic".

Imagine a parallel universe, where COVID-19 is actually just a flu, but for some reason the oppresive governments execute anyone who says so. In the parallel universe, they also have excess mortality (because of the executions) during 2020. So you cannot use the mere fact of excessive mortality to prove that we do not in fact live in this parallel universe.

> It conflates "deaths caused by pandemic" with "deaths caused by reactions to pandemic" with "deaths caused by random things that accidentally happened during the pandemic".

It is not "conflating". That's the actual point of the exercise here. It's asking, how would things have been had there not been a pandemic? How much impact did the pandemic have on the society, in TOTAL, OVERALL.

A great data source for EU excess deaths is Euromomo. They publish official national mortality statistics weekly from the 26 European countries.

Graphs and analysis as well.

https://www.euromomo.eu/

Yes. I wish they would provide the raw data though.

I'm very confounded by the "Excess mortality by age group" data in the US. It _appears_ as if the mortality increase is _almost the same for all age groups_, especially at the beginning and after the initial hump. 15-64 yr olds had a 32% higher chance of dying at the hump and stayed above 11% for most of the year. Wow.

All the previous findings were that COVID is _extremely unlikely_ to kill you if you are under 50, and pretty unlikely-ish under 70 -- so where are these deaths coming from?

I can't answer your question, but I do happen to know that there has been a large increase in drug overdoses and suicides according to the data I've seen for a specific region I was looking at recently. No idea if it's a national trend.

It's proportional. The excess mortality increases by ~10% across the board but the base mortality rate for younger demographics is much lower.

The excess mortality in the United States doesn't make any sense at all. I followed the link to the article that contained the data regarding the United States, however it was an article that only dealt with the Eurozone and made no mention of "us" "united" "the united states" "America" or "usa" anywhere in the article, so I'm not really sure what evidence supports their claims about the USA?

I made this table: https://imgur.com/PZqKWk2

It shows the point melolife made, that the percent excess (excess deaths as percent of expected) can be high, but the expected number is fairly low, so the total number of deaths is still relatively small.

The data is from November 9, so not up to date.

> All the previous findings were that

That was first wave data. Second wave data has shown that the risk for younger people is higher than we thought.

I don't think this explains all that increase though.

As early as Spring last year, mortality increase was estimated to be about the same for all age groups.

In the New Zealand winter less people died than usual by more than 10% in some months - I suppose less flu cases

Yup. This also was also reported to have provided a weaker than usual signal for the seasonal Flu vaccine in the North.

Each hemisphere uses data collected during the winter in the other hemisphere to guess what Influenza variants will be most common in their countries when it's their turn, then it takes several months to turn those guesses into bulk vaccine for immunising the elderly and vulnerable and give them a jab. Sometimes the guess is right on the money and it's extremely effective, other years not so much.

Less signal may make this year's winter 2020/21 vaccine less effective.

> Sometimes the guess is right on the money and it's extremely effective, other years not so much.

Oh thats why some years it seems to work, others not, TIL.

Overall mortality tends to decline in recessions, so that could contribute too.

Would total deaths be a more useful data point? CDC states that 2,839,205 Americans died in the year 2018. https://www.cdc.gov/nchs/fastats/deaths.htm

I would be curious to know how many Americans died in 2020, once the year concludes.

This graph shows 357,335 excess US deaths from January 5 - October 25:

https://ourworldindata.org/grapher/excess-mortality-cumulati...

There have been 2,000 - 3,000 known US Covid-19 deaths per day for the past few days. So, devastatingly, I think that number has gone up a lot since October 25.

https://www.google.com/search?q=us+covid+numbers

The death data that the CDC has received and processed by yesterday, December 4, have a total of 2,893,571 deaths for weeks 1 through 48 (ending November 28). This number is incomplete because of a lag in reporting. The CDC estimates that there are at least 73,000 more deaths in weeks 1 through 48 that haven't yet been received and processed.

(Note that this data includes the last 3 days of 2019.)

But even with the incomplete count, we have already surpassed 2018 and 2019.

It's difficult to make predictions, especially about the future, but it looks we're on track to have 3.2 million to 3.3 million deaths this year. We won't have a firm grasp on the numbers until March.

This study does not account for the statistical significance of changes in population structure. This study compares 2020 to data including 2015.

The 65-and-older population in the US grew by over a third (34.2% or 13,787,044) during the past decade, and by 3.2% (1,688,924) from 2018 to 2019. The world is aging in all of the countries reported in this study.

The peeks in March are interesting.

This is a great point and could be modelled providing that you have access to prospective mortality tables. I fear they are not easily available for most countries and some might be guarded by actuarial societies. For a bias that might balance things out: measuring covid by excess mortality also fails to account for less deaths by traffic and other infectious diseases. I've never come across mortality tables that are broken down to causes of death.

I assume someone posted this link in the discussion and it has faded, but the US CDC has a fine webpage addressing excess deaths in the US, with time series data for the last 3 years and an ability to look at individual State-level info

https://www.cdc.gov/nchs/nvss/vsrr/covid19/excess_deaths.htm

I’m not a fan of the automatic rescaling and truncating of the y-axis on charts by the original link. It is associated with a “lying with statistics” attack on the meaning of the graph. The CDC graphs don’t suffer from that problem.

Am I the only one who wished they showed cumulative deaths over the course of the year?

https://ourworldindata.org/grapher/excess-mortality-cumulati...

Thank you

A puzzle I'm facing right now professionally is in what way to take Corona into account in actuarial longevity / mortality models. Current thinking is it doesn't impact the long term best estimate of longevity. We will probably resume the ever upward trend in lifespans after we've successfully managed this storm. But what to think of the 1-in-200 year shock scenario? Do we expect another pandemic in the next 200 years (:: Yes, return frequency of pandemics is about 20-30 years)? Is this the biggest we can face or are there even bigger shocks (Black Death-like scenarios)? How to account for differences in policy response (these models use data from many countries as a baseline) and expected future responses (will we learn something from this crisis). And then the kicker: How to model this rigourously enough to get your model past supervisors.

>> We will probably resume the ever upward trend in lifespans after we've successfully managed this storm

US life expectancy was lower in 2019 than it was in 2013. Pretty unprecedented outside of things like the collapse of the Soviet Union. That's probably a bigger issue than the pandemic long term.

There's large variance there depending on demographic factors & income quintiles. Life expectancy is going up for some, while down for others:

https://fas.org/sgp/crs/misc/R44846.pdf

"Their most relevant finding for Social Security reform is that life expectancy increased continuously with income and that, according to them, “[t]here was no dividing line above or below which higher income was not associated with higher life expectancy.” At increasingly higher levels of income, they report that an increase in income of a given dollar amount produced positive but smaller gains in life expectancy. "

The data for the bottom chart of US mortalities comes from this source, updated each weekday: https://www.cdc.gov/nchs/nvss/vsrr/covid19/index.htm