Weekly Update: COVID Case/Death Rates by Partisan Lean & Vaccination Rate

For months I posted weekly looks at the rate of COVID-19 cases & deaths at the county level since the end of June, broken out by partisan lean (i.e, what percent of the vote Donald Trump received in 2020), as well as by the vaccination rate of each county in the U.S. (nonpartisan). This basically amounts to the point when the Delta Variant wave hit the U.S., although it had been quietly spreading under the radar for a few months prior to that.

More recently, I switched to posting the same data starting on December 15th, which is (roughly) the start of the Omicron variant wave (although this is fuzzier than the start of the Delta wave).

Now that the Omicron wave appears to be over, I've decided to switch things up again. Instead of using a specific wave of the ongoing COVID pandemci, I'm using the point at which every U.S. adult could theoretically have been fully vaccinated. Pinning down an exact date for this is a bit tricky since a) different populations were made eligible at different points in 2021, and b) it takes 3-4 weeks after getting your first vaccination dose before you can get the second one. I eventually settled on May 1st, 2021.

As always, here's my methodology:

• County-level 2020 Population data via U.S. Census Bureau's 2020 Census
• County-level 2020 Trump vote data via each state's Election Commission or Secretary of State websites
• County-level Case & Death data via Johns Hopkins University Center for Systems Science & Engineering for 47 states; NY Times COVID-10 in the United States github for NE & UT only; & the White House COVID-19 Team Community Profile Report for Florida only.
• BY POPULAR DEMAND, here's an updated Google Spreadsheet with all the relevant data.

Remember: "Decile" means 1/10th or 10% of the total population (all 50 states + DC).

First, let's look at vaccination rates. These bracket sizes change slightly from week to week as more people in every bracket join the "2nd shot vaccinated" population, but for the most part they're pretty stable relative to each other. As you can see, since May 1, 2021, the official cumulative case rate hasn't actually varied much whether a given county has a high or low vaccination rate.

This wasn't the case at all for most of the summer and fall, when cases in low-vaccinated counties were running 4x or more higher than high-vaxx counties...but the infection gap was almost entirely wiped out during the Omicron wave in December - February. At this point, the least-vaccinated tenth of the country is only running 7% higher than the most-vaccinated tenth in terms of case rates:

The same is pretty much the case when looking at case rates by red/blue partisan lean...they're only running 27% higher in the reddest tenth of the country than the bluest, which doesn't seem to signify much...

...HOWEVER, when it comes to death rates, there's still a clear and dramatic correlation between how much of the population has been 2-dose vaccinated and its COVID death rate since last May. The least-vaccinated decile has a death rate over three times higher than the most-vaccinated decile:

As always, what's even more disturbing is how closely the death rate by partisan lean matches the death rate by vaccination rate; they're virtually mirror images of each other:

The graph below shows how the ratio of case and death rates in the reddest and bluest deciles have changed over time. The gap reached a peak of 4.5x higher at the height of the Delta wave last fall before dropping significantly during the Omicron wave...but since then it's stabilized and is even starting to inch back up again slowly:

I've also added a new graph. This one shows the 10-day moving average of the ratio between the DAILY death rates in the reddest & bluest deciles, dating all the way back to January 1, 2021. As I predicted back in late December, after reaching near parity in late January, the death rate in the reddest parts of the country is again on the rise relative to the bluest parts:

The dotted lines are data gaps; these are due to a few absurdly massive outliers (either high or low) caused by one-time backlogged data dumps which skew the daily ratios (even on a moving average basis).