Answering my own question

A possible answer to the previous post

Just moments after I clicked the “Publish” button of my last post , I thought of an idea to test regarding the unanswered question of the post. The question was how do we explain the differences in the relative number of COVID deaths in a US county, comparing their worst winter month in 21/22 to their worst winter month in 20/21. We were satisfied that vaccinations did not answer that question, at least not as the primary driver. An unexplored factor in that post was the role of the predominant variant on these numbers. In particular, this winter saw a quick shift from Delta being dominant to Omicron, with the latter considered much less lethal. After completing the post, it occurred to me that I could explore the question of specifically when did a given county experience their surge this winter.

I’m not aware of detailed data on variants over time at the county level, so let’s look at a high level view of variant predominance from the CDC:

From this we can see that Omicron went from being relatively rare the week of 12/11/21 to being overwhelmingly dominant the 1st week of January. If, in fact, Omicron is far less deadly than Delta, then we would expect counties that experienced their peaks in January to do better than those that experienced peaks in December. Peak in this case should be the peak of cases, since when you catch COVID is what matters in terms of what variant you have. Now, we can look again at a histogram of the counties’ relative COVID death rate in peak winter 21/22 vs. peak winter 20/21, this time we color the square based on the date of peak new infections of that county in winter 21/22:

Peak new infections ranged from Dec 26 (New York, NY) to Jan 29 (Alleghany, PA). Any county in green has a peak infection day in December, while blue is January. Darker green means earlier, darker blue means later. We see that 5 of the 13 counties that had the largest increase in COVID deaths this winter over last are green. None of the 30 counties that had the largest decrease in COVID deaths this winter over last are green. To my eye, seems that green and light blue have a greater probability to be on the right of this chart and darker blue a greater chance of being on the left.

Alternate look at the data with box & whiskers by week of peak infections:

The above looks persuasive in showing that the timing of peak infections of a county had a large impact on a counties relative COVID death rate from peak winter 20/21 to peak winter 21/22, in particular there is a jump from those counties that had the misfortune of catching their wave when Delta was still dominant (or at least prominent) in December to those that caught the Omicron wave in January.

Coming back to the original question: what can explain a county’s difference in relative COVID death rate between winters 20/21 and 21/22? At least part of the answer appears to be Omicron and being lucky/unlucky as to when you caught your winter surge.