Comment by WhompingWindows | Hacker News Reader

Comment by WhompingWindows | Hacker News Reader

WhompingWindows

Epidemiologist here, a few thoughts reading other comments:

-Death data is preferable to new cases in some ways, since it doesn't rely on testing, which ramps up/down from locality to locality and over time.

-To compare deaths from the Spring with deaths recently isn't apples to apples: we know now remdesivir and dexamethasone are helpful treatments, especially the latter, and these may help reduce mortality rate.

-Mortality now may be lower partly due to: if the virus killed off the more vulnerable populations (nursing home) early on, it has less frail individuals to infect now. These communities got ravaged, we had something like 40% of deaths in my area in nursing homes in our 6-week case peak.

-Trying to tease out immunity is very tough as well, early on there were few good tests and their specificity and sensitivity were less well known. At my agency, there were multiple tests we were collating, with different false positive and negative %s, and different reporting times (which all can influence R0, the variable number our behavioral interventions seek to tamp down).

-Comparing different countries is very tricky. Sweden is a favorite example for non-lockdown approach, but there we see it's very tricky to tease out the benefits. On the one hand, they didn't do much better epidemiologically than their neighbors. On the other hand, their economy was surrounded by locked down ones, so that may have tamped down any benefits their non-lockdown would've garnered due to neighbors' activities. Did it pay off? The verdict is still out, IMO.

Things are very complex in infectious diseases epidemiology. It's extremely hard to know anything for certain with this virus, it's behavioral tangled up with comorbidity with genetics and demographics and evolving treatment and viral dosage and strain of virus...it will take a few years of unpacking the mountains of data before we truly have a grasp of what happened here.

Comparing Sweden with its neighbours really highlights how poorly it has done: https://ig.ft.com/coronavirus-chart/?areas=fin&areas=swe&are...

Pretty much any way you slice the data, a lot more people died in Sweden than in its neighbours.

You can see the same pattern in excess all cause mortality too: Denmark 200, Finland 600, Norway 0, Iceland 0, Sweden 5500.

Now Sweden has a population slightly less than Finland and Norway combined, and figures can be difficult to compare, but still it's hard to avoid the conclusion that so far, around 4000 people in Sweden have died because of differences in policy compared to its neighbours.

The least you can say is that Sweden didn't manage to test as many as its neighbors did. It is claimed the influx of infected was greater etc, but lack of tests meant that it was hard to get it under control too.

The better question that Sweden may help answer is whether shutting down the economy was worth it. Was the deaths of 5,500 people worth keeping the economy open, in other words, could you have saved more lives by pursuing other health priorities and let covid run it's courses as Sweden elected to do?

This. The deaths directly attributed to the virus alone aren't the only relevant deaths. Suicides are up. Domestic violence is up. Child abuse may be up but schools are all "virtual" so good luck detecting it (as if it wasn't hard enough when kids were mostly in school).

Talk about throwing out the baby with the bath water...

Could you show any data that supports your assertion about rates of suicides, domestic violence, and child abuse?

in NZ suicides were down over the lockdown period.

NZ is an outlier in a lot of ways, to include population size and distance. The also cracked down on firearms ownership recently, too -- #1 choice for suicidal men.

huh. Thats such a weird place to go.

It was only the automatic weapons, Im pretty sure any suicidal men could probably still hit themselves in the head with a single shot from a distance of 0.2 feet...

I don't at all doubt that lockdown has had a number of second order problems (and benefits too), but the the figures I quoted at the end were excess all cause mortality, so those are all already included. And indeed, many countries with net 0 excess all cause mortality certainly had corona virus deaths, but lockdown and related benefits appear to have saved more people than the virus killed in those countries.

I think most people who are worried about secondary effects from lockdowns don't think that these will be visible immediately during the pandemic. But rather as a higher baseline of mortality in the coming years. This is just my guess. I'm not an epidemiologist, so I have no real expertise in how these effects usually manifest themselves in the statistics.

I didn't consider those second order effects. When I wrote my comment, I had in mind only the dollar cost of averting the disaster. When we spend whatever trillions of dollars shutting the economy down and limiting output, we necessarily take from other priorities, like strengthening emissions controls, or making cars safer, or re-wilding spaces, or spending money on providing people with affordable preventative medicine that could extend the duration of life.

Actually the figures I gave at the end there were excess all cause mortality, not coronavirus deaths. So far, they have definitely not saved more lives.

Just for some perspective: during the 2017/18 season 2822 people died from the flu in Denmark. https://www.ssi.dk/aktuelt/nyheder/2020/influenzasasonen-201...

Sweden had a high initial death rate but now their average daily deaths have been near 1 for several weeks. So whatever they're doing lately seems to be working well.

https://www.worldometers.info/coronavirus/country/sweden/

Look at the charts I shared, they took about an extra 120 days to get the death rates down to the same level as their neighbours. Now, it's not exactly comparable since they have a larger population, but even the average daily deaths of 1 is not an indication that they're doing well compared to the other nordics.

Playing devil's advocate:

The death counts are what they are, but on the other hand, Sweden is extremely unlikely to experience a 2nd/3rd wave, which is experienced now by the countries you listed - it's not even halftime yet fighting the virus, it's too early to say who did better.

The "any way you slice the data" is not so clear either. As someone pointed out, Sweden had fewer deaths (in the thousands!) in previous flu seasons, I've read the hypothesis put forward that part of the reason Denmark has done so much better is that their weakest already succumbed to the flu in 2017/2018, whereas Sweden's were spared. Nearly impossible to prove or disprove, but there ARE ways to slice the data that make the attribution to policy less convincing.

> Now Sweden has a population slightly less than Finland and Norway combined, and figures can be difficult to compare, but still it's hard to avoid the conclusion that so far, around 4000 people in Sweden have died because of differences in policy compared to its neighbours.

Another way to slice the data that is important: Sweden has, by far, more black skinned immigrants than Norway (and I think Finland too) -- which are faring worth everywhere, likely some sort of sunlight/vit-D related process. A significant number of those who died in Sweden are immigrants.

If you compare Sweden to Belgium (which had strict lockdown, but similar immigration patterns), the policy argument is less convincing again.

I find the flu argument pretty weak, but it ultimately boils down to a claim of demographic differences (that Sweden had more people at the beginning of the pandemic that were likely to die than neighbouring countries).

Belgium feels like a cherry picked example, having as it does a miserably poor record compared to its own neighbours, and being rather unlike the nordic countries in lots of other likely relevant ways - climate, population density, etc.

Perhaps a more likely demographic cause might be the age of the population in Sweden, which I believe does have a particularly old population.

The racial/skin colour angle is difficult to evaluate. There may be something to it, but there are so many potentially confounding factors - immigrant populations are unlike the native population in a number of ways. Some of the ways they are different are that they are overrepresented in high risk groups like front line medical workers or living in poorer quality accommodation or having more difficulty getting access to health care. If the primary cause that they are overrepresented in covid-19 deaths relates to any of those, then having a lower immigrant population may just redistribute the deaths rather than reduce them (the bottom socioeconomic 10% will likely be overrepresented in deaths, and that bottom 10% may feature more or fewer immigrants).

My point is that it’s not as clear cut as “any way you slice the data”.

Sweden’s immigrant population is not remotely like the US one. They are NOT over represented in healthcare work, AFAIK they are underrepresented (to a fault, of having almost no representation). They do live in lower quality housing, on average, but have access to nominally equivalent medical services (Sweden has socialized everything; once you are “in the system” you are entitled to everything, though rich people can buy some extras)

Also, It’s a little disingenuous to say “Belgium is cherry picked” without extending a similar courtesy to Sweden.

All in all, it’s too early to call winners and losers, and there are many ways to slice the data that don’t make Sweden look bad at all.

Well, its a bit unfair to say that since Sweden didn't reap all of the benefits of protecting their economy, because of neighboring countries' following a different path, that something was wrong with Sweden's approach.

If I had the right idea, but am surrounded by people with the opposite idea, that doesn't mean my idea became wrong. It only means I maybe wasn't the best at convincing my neighbors.

Where did they say something was wrong with Sweden's approach?

Because you exist within a system whether you like it or not, so the optimal/correct solution must account for those variables, including your ability to entice cooperation. So yes, you would still be wrong, just not for the narrow way you’re choosing to redefine “correct.”

Would't any preexisting immunity be accounted for by any decent estimate for R0 and the IFR?

Not really. If there is existing immunity then the measured R0 will be lower than the actual R0, but we don't know how to measure the true R0 in absence of the current population with their immunity so we can never know what the true R0 might actually be.

Note that true R0 is a concept I made up on the spot. I think it is clear what I mean but experts in this may have better terms.

A first-order SIR model with unmodeled pre-existing immunity will be accurate initially, but get gradually more wrong with time. For a simplified, non-representative example with fake numbers, let's say half the population already had perfect sterilizing (they don't get sick, and the virus doesn't replicate at all so they can't spread it to anyone else) immunity. Then an apparent R0 of 2.5 would imply an actual R0 of 5. But R0 = 2.5 would imply you need 1 - 1/2.5 = 60% of the population infected for herd immunity, while in that hypothetical case you'd need only 30% infected (so the 50% initially immune plus that 30% make 1 - 1/5 = 80%).

The above still assumes a well-mixed population, but with heterogeneous susceptibility (using the terminology from [1]). There's also heterogeneous connectivity (i.e., the fact that some people have more social contacts than others), which further reduces the herd immunity threshold.

To be clear, the above is a fake example. We know that almost no one has perfect pre-existing sterilizing immunity to the novel coronavirus, because we've seen highly-connected populations (homeless shelters, slums, boats, etc.) where almost everyone tested positive. But that doesn't mean everyone is equally susceptible! Immunity is a continuum, not a binary thing, and the 20-50%[2] of people with reactive T cells very likely get less sick given the same initial viral dose (or, stated in the other direction, would need a greater viral dose to get equally sick). They also might spread the virus less, though that's more speculative.

Finally, I'll note that the herd immunity threshold is the point where R becomes < 1, and the number of people currently infected starts to shrink instead of growing. People still get infected on the downslope, so the fraction of people infected can overshoot higher (and potentially much higher, especially if no precautions are taken to slow the spread).

1. https://www.medrxiv.org/content/10.1101/2020.07.23.20160762v...

2. https://science.sciencemag.org/content/early/2020/08/04/scie...

> We know that almost no one has perfect pre-existing sterilizing immunity to the novel coronavirus, because we've seen highly-connected populations (homeless shelters, slums, boats, etc.) where almost everyone tested positive.

We don't actually know that. Even if some significant % of the population has pre-existing sterilizing immunity, then by chance we'd still expect to have many groups of people where everyone is vulnerable.

Fair point. I still think pre-existing sterilizing immunity is unlikely, since (to my non-biologist's understanding) antibodies tend to be specific, and T cell immunity is broader but not usually sterilizing. The existence of those unlucky groups isn't conclusive evidence in itself, though, especially if pre-existing immunity is distributed non-uniformly (which it probably is, if there's some genetic basis, or if different strains of common cold coronavirus swept through different regions last year and primed people's T cells differently).

Of course non-sterilizing immunity could still be incredibly valuable. They're speculating that the current vaccine candidates may not cause sterilizing immunity[1], but immunity that turns a fatal illness into a bad cold (or a bad cold into a couple of sniffles) is great, whether naturally pre-existing or from a vaccine.

1. https://www.statnews.com/2020/05/22/the-world-needs-covid-19...

q for you: early on, there was a big variance (region vs region, one region over time, etc) in how mortality was reported, e.g. heart attack vs COVID, single cause vs multi-cause, etc. Is mortality more standardized now?

I've been wondering this myself; last I heard (2 months ago, fairly credible investigative journalism), there was a massive disconnect, state to state, how COVID-19 deaths were assessed and recorded. IIRC, Florida was on one end of the spectrum, tagging deaths as from any comorbidity present (only tagging as COVID if they couldn't find anything else), to California, which was tagging as COVID without test confirmation, if enough of the symptoms were present.

California and Florida have about the same gap between excess mortality and registered covid deaths, with Florida being slightly better.

You can see it here:

https://www.nytimes.com/interactive/2020/05/05/us/coronaviru...

Do you have any links/sources for this?

I think the CDC numbers say that, at least nationwide, 6% of deaths list only Cov 19 as the cause of death. 94% list other causes. This data is so messy, I don't think there is much of a take away. Of the 6%, could there have been other reasons which were missed? Certainly some. Of the 94%, would they have died anyway sans Cov 19? Certainly not all, and not then.

As others have pointed out, we should pay attention to the delta in deaths over some other year with an average flu season, and a year with a bad flu season. Someone will run these numbers at the end of 2020, I hope, and then we'll have a better idea of the effect of this particular pandemic.

They've already been running those numbers and they're available on CDC's website: https://www.cdc.gov/nchs/nvss/vsrr/covid19/excess_deaths.htm (Scroll way down for the chart)

We had a bad 4-6 weeks a couple months ago but we're already down in the range of a normal flu season as far as deaths go.

I'm pretty sure 6% was deaths where "only" covid was listed.

In most cases, doctors are more specific. So they may put covid & pneumonia. Pneumonia being any infection of the lower respiratory system. Which would in almost all cases, be caused by covid unless they also had a separate infection.

Doctors will also often list complicating factors like obesity (45% of Americans are clinically obese).

This is actually right along with the confusion of the public and media for the official coroner's report. The coroner listed cardiopulmonary arrest as the cause of death "complicated by" (in other words "brought on by") someone pinning him to the ground and blocking his airway. Drugs in his system and prior health conditions were listed in the report as any thorough coroner's report should. But they were not listed as complicating factors for his death.

Cardiopulmonary arrest is -not- a heart attack. (that's a myocardial infarction). It means your heart stopped pumping and your lungs stopped functioning. Which is how most people die.

Not GP. In my opinion, the only reliable measure is excess death (if you believe death count, which I trust for most countries, but not all countries), because it is invariant to these (often political) classifications. See e.g.[0] (it is up-to-date for 11-sep, despite the link date saying April 21). In fact, these graphs would look the same even if you didn't have access to any testing/diagnostics/causes-of-death.

It is easily visible which countries have gone back to baseline, which countries haven't, and which didn't even have a "first wave", which likely means that they are in great danger of one happening.

[0] https://www.nytimes.com/interactive/2020/04/21/world/coronav...

Where would I look of I wanted to follow geographicly diverse epidemiology news / publications / groups / individuals, not just covid related.

Thanks in advance.

Any thoughts on Sweden (and many other EU countries) saying masks have very little scientific evidence for the little benefit they supposedly offer. How can scientists in different countries have so different views. Some countries preach masks like holy water and they both say God (science) is on their side.

Reply on news.ycombinator.com