Pandemic in Western Illinois - April through July

In my previous post I recapped the data on the COVID-19 pandemic as experienced in Warren County, IL. In this post I’ll continue looking at the historical data but will shift to a wider focus and look at the Illinois and Iowa region surrounding Warren County. This analysis is made possible by the county level data provided by usafacts.org.

The Region of Interest

The collection of Illinois and Iowa counties I’m reporting consists of the recovery region that Warren County is grouped into along with a handful of counties in Iowa that sit on the Illinois and Iowa border. Only one county separates Warren County from Iowa. People that live here make fairly frequent trips to Burlington, IA and the Quad Cities area. Additionally, people working at Monmouth and Knox college as well as the Smithfield plant in Monmouth will commute to work from counties in this region. By looking at this set of counties we can get a feel for the risk of COVID-19 spreading due to regular travel to and from Warren County.

To get things started, let’s look at a map that visualizes the total number of confirmed cases for each of these counties as of July 31, 2020. You can mouse over the different counties to get their name and case counts.

Unsurprisingly, we see the most cases in the more urban counties with the Quad Cities, Peoria, and near Kendall County near Chicago really standing out.

Regional Totals

To see how we arrived at this picture we can look at a visualization of new cases on a day to day basis for the entire region combined. Like I did for Warren County, this graphic colors bars by recovery phase and includes a seven day average trend line as well.

In Warren County we’ve seen a resurgence of the virus come along with the relaxed social distancing measures. Thankfully, we have not reached daily new cases that compare to surge in April and May. Those case numbers were largely due to the outbreak at the Smithfield plant. Regionally, we see a smaller version of the national trend where the virus is not only surging again but daily case counts are surpassing the peak that put us into lock-down back in April. To understand why this is happening we new to look at what’s happening in each county.

Tracking New Cases in each County

Visualizing the county level picture on a day to day basis is tricky. Below you should see an animation that tracks new cases per county for every day of the pandemic up to and including July 31. There is a lot of information packed in there. Play and pause the animation as needed. You can also use the slider to pick and choose specific dates.

If you watched the animation, then should have seen the more populated, urban counties make their regular appearances at the top. If you didn’t take a look at the April to May time period, then go back and look for Louisa, Muscatine, and Warren County. They make brief appearances at the top of the charts. Like Warren County, Louisa County, IA plays host to a meat processing and packing plant. Their plan is Tyson chicken where Monmouth hosts Smithfield pork. Both plants saw an outbreak of cases at the start of the pandemic.

At the time, these outbreaks caught a lot of attention. Warren and Louisa do not have large populations. While the number of cases being reported were small relative to the major outbreaks in New York and Chicago, they were huge for the number of people living in those communities. Counties of that size probably lack the resources to deal with more than a few people sick with COVID-19. So, on a per capita basis, these numbers are objectively bad.

Comparing Counties with Respect to Population Size

The most popular way to compare counties with different populations is to normalize their case numbers per 100,000 people. According to the data hosted at usafacts.org, Warren County has a population of 16844. A single case in a population of that size is proportionally equivalent to about 6 cases in a population of 100,000 people. Similarly, if a county is home to 245,750 people and reports 25 cases in a day, then that would be equivalent to roughly 10 cases in 100,000 people. Looking at one actual case versus 25 cases makes you think things are much worse in the larger county but in fact the difference isn’t that great. On a per 100,000 person basis, we’re looking at six versus ten cases.

By re-calculating case number on a per 100,000 basis we can compare different counties on even footing. A group spearheaded by the Harvard Global Health Institute even provides general risk levels associated with cases on a per 100,000 people basis. One or fewer cases per 100000 people is consider low risk and on track for viral containment. Cases numbers between one and nine indicate that the virus is spreading in the community. Cases numbers up to 24 indicated accelerated spread and everything 25 or above indicates high risk, rapid spread and the need for increased social distancing or even lock down. Going back to our earlier example, that single case in Warren county indicates community spread where the ten cases in the larger county is on the community spread and accelerating spread border.

We can now go back and look at total number of cases in the region with case numbers normalized per 100,000 people. This map should let us evaluate which counties have had it the worst relative to the size of their population.

You can see Louisa and Warren county as having higher numbers and even outpacing or matching some of the more urban and densely populated areas. This is largely due to the outbreaks at meat processing plants in the early days of the pandemic. Both counties saw relatively large spikes in reported cases at the time. To see this play out we need to return the the day to day data.

By redoing the bars on a per 100,000 person basis really highlights the seriousness of the early outbreaks in Warren and Louisa county back in April. On April 14th, Louisa County is off the charts. Then on April 23, Warren County takes the top of the list. For how small those counties are, the number of cases being reported at the time were huge and dwarfed what was going on in other counties in the region. However, by late June and early July we begin to see fewer spikes and more general spread throughout the region. This is highlighted by the observation that more bars are of moderate length in July where earlier frames show a few long bars and lots of short bars. The cases are more evenly distributed across the counties.

Smoothing the Data to See Trends

Looking at daily case numbers on a per 100000 person basis lets you compare relative severity between counties but its still subject to the kind of variance that makes seeing the trends possible. For this we again turn to the seven day rolling average. This time you’re looking at the seven day average of new cases per 100,000 people presented as animated bars.

This is probably my favorite of all of these graphics. The smoother change in the length of the bars really lets you see important trends. In April and May we see just a few longer bars with all the other bars being relatively short. By July we see lots of moderate length bars and relatively few short bars. It certainly seems like we’ve traded localized outbreaks in a few places for slow, consistent spread throughout the region.

August and Beyond

My last two posts cover the pandemic in the county and the region from its start in April up until July 31. A few things seem pretty clear:

• There are more daily new cases in the region now than there were when we were put on lock-down in the spring.
• More counties are experiencing regular spread of the virus now than they were in the spring.

My plan moving forward is to continue to present longitudinal snapshots of the region through these graphics. At this point, I’m planing on doing three week looks at the data: one new weeks worth of data along with two weeks prior for context. Expect my first August post soon.