Quite a few of my friends have made claims lately that rural areas should be allowed to stop sheltering in place. After all, many rural areas have few to no documented cases of COVID-19, let alone deaths. But given the demonstrated virulence of this disease, I was not comfortable with their suggestion. Couldn’t we be seeing fewer cases and fewer deaths just because there are fewer people?
I decided to check it out for myself.
The data is pretty hard to argue with, but I’m happy to hear questions and feedback on my methods, which I outline below. The hard-to-miss conclusion is that COVID-19 is actually significantly more dangerous in densely populated areas. Obviously, it is still dangerous in sparsely populated areas, but significantly less so.
Now, I’m not encouraging anyone to take fewer precautions or to be careless—I’m a business analyst, not an epidemiologist. We would all do well to follow the laws and protocols relative to our locations. But after what I found, I am convinced that it’s important to have more localized risk assessment to assist governments in making decisions that are good for everyone.
So, what do I mean by “more dangerous in densely populated areas”? Of course there are more deaths among higher populations, right? But I am saying something different and more significant than that. I am saying that risk is actually higher per person in densely populated areas.
Here is a chart I put together that shows this, I think pretty definitively. The data is not just handpicked from one or two counties in each population density group to prove some point, but selected indiscriminately from the first 155 U.S. counties to have a COVID-19 fatality. Here are four important things to notice about this chart:
1. It does NOT show infections. Infection data has been shown to be wrong by more than an order of magnitude in some areas. While death data is not perfect, it is far more reliable as a measure of clear impact, so I have used public records of death by county by date.
2. This chart does NOT show number of deaths. If it did, it would be kind of obvious that higher population areas have more deaths. This chart shows a more significant data point. It levels the playing field for small and large populations by using a ratio as its basis: deaths per million.
3. Time on this chart does not start on the same date for all counties. Rather, Day 1 for a county is counted as the date of the first death in that county. This ensures that we are comparing similar impact curves.
4. Finally, this chart does not show growing death rates. Instead, it shows how death rates change over time. That steeply curving line you are looking at is showing acceleration. We see that higher density population areas experience rapidly accelerating death rates, while lower population areas have more static death rates.
Now make your own conclusions:
If risk is significantly lower in more sparsely populated counties (as this data seems to show), what conclusions would you draw about how governments should respond to the threat of COVID-19? How could localized risk assessment be useful? Also, what conclusions would you draw about how people should respond to rural vs. urban opinions about the threat of COVID-19? Is there a story about blind men and an elephant somewhere in here?
A native of Shasta County, Matt Peebles works remotely as a Business Analyst for Skyline Champion Corporation in Troy, Michigan. When he’s not writing software requirements and business use cases, he runs a local addiction recovery program and periodically chips away at bagging the 100+ local mountain peaks visible from his home in Anderson, California.
