I mostly played with graphing the stuff shown in Table 2
It is lagging but I am going to assume that the proportion of all of these numbers (i.e. the percentages and distributions) is very similar to what is happening right now. I think this is a reasonable assumption.Table 2. Deaths involving coronavirus disease 2019 (COVID-19), pneumonia, and influenza reported to NCHS by age group, United States. Week ending 2/1/2020 to 4/11/2020.*
Most of what I am going to show is the following:
- Chance of death under age 45 is very low
- Older people are dying at outsized proportion to younger people but it turns out older people are more likely to die from anything and I show that COVID-19 is mostly proportional to the chance of dying from anything when looking at older demographics.
- Why should we compare COVID-19 deaths to all deaths? Because we need to understand the context of the deaths and not interpret them in a vacuum, especially when making judgements of risk or public policy/guidelines.
- I think it is right to talk about reopening the economy/society and probably socially distance the most vulnerable members of the population, as they are now.
I'll lead with this. The deadliness, taken in a vacuum, is vastly more deadly to older people and as I said before that's because they tend to have more underlying health conditions.
However, when we look at ALL deaths from any cause during this same time period, we see something very familiar. We basically see two things: 1) Over 45 (maybe over 35) the trend looks almost identical and 2) Under 45 (well maybe 35) the deaths from other causes significantly outperform COVID19 deaths, when compared to the 35+ or 45+ age groups.
Let's look at a few more from different angles and see if we can extract anything useful.
So on average, during the same time period, COVID19 was responsible for 2.25% of all deaths during this same time period. OK, that's the average and that feels fairly low to me but it does seem like it would be higher if we didn't do the lockdown, so that's fair. But now look at how it is distributed by age group. The rates don't even hit 1% of total deaths until the 25-34 age group. 35-44 is on its way to 2% which is still below the average. And then 45+ all hovers around the average. What does this mean? It means that COVID19 does NOT kill proportionally more people as age increases past 45 when normalizing for all other causes of death. It means you are no more likely to die of COVID19 at age 80 than age 50, it's just that 80 year-olds are more likely to die of something/anything in the first place. In other words, your chances of dying from something OTHER than COVID19 at 45+ are all about the same. Which I have graphed more directly later.
This is a variation of the third plot but stacks the COVID19 deaths on the same graph as all deaths, per age group. It's a good visual to show the death rate from COVID19 contrasted to other deaths. If no one was dying from anything, and then COVID19 came along, this would look way different. We originally thought it could be in the millions and if that was the data we had, we acted accordingly. But it turns out the models vastly overestimated the number of ventilators and stuff that we'd need - even as the models ALREADY TOOK INTO ACCOUNT the mass social distancing program. So that means one or more assumptions in the model was wrong, and that's probably the death rate/how likely the symptoms are to be severe. It think this will bear out when we do antibody tests.
In the Netherlands, a study revealed that 18x more people had the virus than the COVID19 test numbers revealed, lowering the death rate from 11.x% down to like 0.6%.
Source: https://www.reuters.com/article/us-heal ... SKCN21Y102
This matters because they were criticized for limiting their lockdown and trying to go for more of a herd immunity approach.
To me it seems like their hypothesis is probably more right than wrong despite all of the outrage and negative media coverage, which was based on the 10+% death numbers.
Do I think an order of magnitude times 2 should result in a different policy calculus? You're damn right I do and it should. If you're having problems squaring this hard data from your emotional state watching media coverage, that's because the media coverage has been overwhelmingly angled to frighten you, whether they have an agenda or just want clicks. It's likely both.
I took the same age groups I used in top 2 plots and calculated the multiple of likelihood that you will die from something OTHER than COVID19. This is just an inversion of the COVID19 death rate but different cuts help process the data I think. In this view even the older demographics are 40-43x more likely to die from something else. And under 35 it really spikes to over 160x.
Finally, another view but with all of the CDC age groups. Again what we see is the exponential likelihood that COVID19 is probably not a concern the younger (and of course healthier) you are. What I get from this is kids should be going to school and probably not interacting with their grandparents (they aren't doing that now so no change).