It’s not just high-density living, FWIW. It’s NYC, New Orleans, Detroit, and Chicago that was the first wave. Plenty of high-density “blue” areas didn’t get hit in the first wave because they weren’t centered around the spring 2020 hot spots.
If the COVID death rates are simply reflecting where a bunch of old people live after the first wave, this is telling you very little. Because, even after the vaccines, it’s mainly old people dying of COVID. The old people got vaccinated first (yes, even Trumpsters), but even after vaccination, old people have higher risk levels than young adults.
I bet excess deaths by county looks even worse, since most of the undercounting of COVID deaths is happening in red counties, through a combination of a higher likelihood of coroners to discount COVID as the cause and by requests from family members to have something else listed if at all possible.
In my mind, that graph is a fine illustration. If we know that republicans are less likely to be vaccinated (which I think is true, but I’m not sure?), combined with the fact that unvaccinated people are more likely to die, then that graph does a nice job of showing those two combined facts for everyone.
However, I don’t think it’s evidence. I wouldn’t look at that and think this is evidence that the Republican position on covid is worse than the Democratic one. This is because there are so many other systematic differences between Republican-controlled and Democratic-controlled states that can impact covid deaths. In that respect, it reminds me of those nonsense graphs showing covid cases between counties with and without mask mandates as evidence that masks do not work.
Go Actuary tells me Whiskey already posted this, but here is one article from before the last 20% of COVID deaths occurred:
I recall another that showed this by ranked by the redness of a county where the excess deaths were something like 90% explained by COVID in blue counties, and closer to 50% in the red counties. Something like overdoses I am sure skew towards poor rural red areas, but the increase in other causes of death that have not been counted as COVID doesn’t have a great explanation other than underreporting, at least so far that I have seen, and many of the causes that have had significant increases, like heart issues and diabetes are frequently cited comorbidities for COVID. In a typical year we have something like 50k deaths due to influenza, which has hardly registered the past two seasons. I am not sure that is even being factored in to a reduced baseline ex-COVID.
I am sure there are a variety of reasons for the excess skew in the red counties, but it is hard to deny they are going underreported.
It’s 100% true that Dems got vaccinated at a higher rate than Reps. I’m not going to post a link here but you can find a dozen. That seems to me to be the takeaway from the chart. Reps didn’t get vaccinated at the same rate as Dems, and that is driving deaths.
I didn’t say anything about the positions taken by either side, or masks, or whatever else. In my estimation, age is the most important variable in the mortality model. Vaccination is second. So that’s what’s going on in the chart.
I didn’t mention anything about the political positions taken, that horse has been beaten to death here already.
I gotta think the age difference is also hugely relevant.
And while I absolutely think that vaccination rate is also relevant, if you ignore the age difference then I think it’s both disingenuous and unconvincing.
I’d like to see the chart adjusted for age. But the story would basically be the same. To the extent that political party, or whatever, is a strong proxy for vaccination rates, you’ll get the same chart. Vaccines saved lives, cut that data however you will.
I assume those two variables are the most important for severity of disease.
But i thought population density was important for the number of people who get it. Maybe it’s not. That’s part of why im so hesitant to interpret that plot.
Since Dems skew urban, to the extent population density matters, it would artificially make the difference less pronounced, no?
I do think the chart is too simplistic to draw a ton of conclusions from. Which was my question, I think the takeaway is that vaccines work, and that’s the key driver here.
But this is political, so let’s all blame the other side!
So I’m working on something – I’m thinking I’ll do one analysis using the official COVID death count, one for all-cause deaths (and using the expected deaths from the CDC).
So the first will suffer from the potential miscounting of COVID deaths, both under- and over-counting
The second will suffer from all the drug overdoses, car crashes, homicides, and other excess deaths that aren’t directly from COVID. But dead is dead, so maybe it’s fair to use excess mortality from other causes as well.
I will want to use both crude death rates as well as age-adjusted death rates, for fair comparisons. That places with more old people have more old people die is not very enlightening. But also, I’ve found in relative excess deaths for the U.S. as a whole, the worst percentage increases have been for adults age 25-44, mainly due to non-COVID causes.
I’m going to try this out at the state level first to see how things look before I draw on the county-level data. I tried this with homicide data earlier, but homicides are too infrequent to get good results, and COVID by itself might have a similar problem.
I googled that (it’s been a while) and found that it’s only by a couple of years. And I know risk is an exponential function of age (or something like that), so I think it explains part of it but not all, but I don’t have a spreadsheet on that.
So it’s an interesting chart, but if it were adjusted for age, maybe gender (?), etc I think it would be more compelling. Then again, at this point, is anyone changing their minds about covid at this point in the game?
