More analysis of cases and deaths by population density

Summary: I do think social distancing measures are starting to work nationally and in many regions, but as I’ve mentioned before, this is such a huge patchwork and there are going to be places in the next 1-2 weeks that are really hit hard.

We thought we would update some of our previous analysis on population density. Now that more time has passed and the epidemic has developed further in more regions, I asked Aaron to make some graphs looking at the change over time in cases in different areas. These next two graphs show the change in cases and deaths in counties in the US by population density from 3/27/20 through 4/11/20, so about 2 weeks. The x-axis is population density on a log scale for each county, while the y-axis is 1) raw #s of cases (yellow) and deaths (green) and 2) cases and deaths per 1,000 population, still in yellow and green. Each arrow starts at the county’s 3/27/20 data and the tip points to the 4/11 data.

1) Changes in cases (yellow) and deaths (green) by population density for each US county. Each arrow starts at a county’s 3/27/20 data and ends at the 4/11/20 data. X- and y-axis both on a log scale.

2) Changes in cases (yellow) and deaths (green) per 1,000 population by population density for each US county. Each arrow starts at a county’s 3/27/20 data and ends at the 4/11/20 data. X- and y-axis both on a log scale.

Some of the harder hit big cities with high numbers of cases (in yellow) to begin with, towards the upper right of each graph, now have much slower relative growth in cases in the form of much shorter arrows (keep the log scale in mind though — these changes still can mean a fair raw # of new cases). Some less dense (so more rural) areas towards the left of the curve have had quite a bit of growth in cases. It’s clear that mid- and larger-cities have had the brunt of the deaths, both in terms of #s and in terms of deaths per population.

We highlighted the counties with the 10 longest arrows for deaths and for cases here, again by 1) raw #s and 2) per capita:

1) Changes in cases (yellow) and deaths (green) by population density for each US county, with the 10 longest arrows for eacd highlighted. Each arrow starts at a county’s 3/27/20 data and ends at the 4/11/20 data. X- and y-axis both on a log scale.

2) Changes in cases (yellow) and deaths (green) per 1,000 population by population density for each US county with the 10 longest arrows for each highlighted. Each arrow starts at a county’s 3/27/20 data and ends at the 4/11/20 data. X- and y-axis both on a log scale.

Then we graphed the curves for cases over time for each of these counties, starting from their first local case. Some places seem on track to be among the hardest hit areas in the country to date in the next 1-2 weeks.

Graph of cases by time since first local case for US counties with the top 10 increases in cases (yellow) and deaths (green) from 3/27/20 through 4/11/20.

I do think social distancing measures are starting to work nationally and in many regions, but as I’ve mentioned before, this is such a huge patchwork and there are going to be places in the next 1-2 weeks that are really hit hard. I’ll get the names of the counties listed here later today. Thanks to Aaron for the beautiful graphs and also for making the comment button not huge anymore. #thefamilythatblogstogether

Edited to add the counties in yellow above (cases) and green (deaths):

Cases (yellow lines/arrows above)

Lehigh, PA
St. Charles, LA
Terrell, GA
 Washington, LA    
 Mobile, AL  
 Tuscola, MI
Luzerne, PA
Madison, IN
Hall, NE
Hopkins, KY

Deaths (green lines/arrows above)
Arapahoe, CO
Bristol, MA
Essex, MA
Hudson, NJ
Miami-Dade, FL
Hampden, MA
Hartford, CT
Westchester, NY
Union, NJ
Philadelphia, PA   

We’ve also had a few questions/suggestions about things to add to the analysis that we will hopefully be able to update this week!

Previous
Previous

Some thoughts on reopening

Next
Next

Testing and social distancing measures by state