Nacogdoches County on Thursday topped a milestone that none of us wanted to arrive. Confirmed cases of COVID-19 surpassed 100 and kept marching upward and then upward again.
Most of the other Texas counties with confirmed case numbers in the low 100s have populations far more vast than here. As of this writing, we’re equal with Ellis County — approximate population 185,000 — and right behind Smith County — estimated at around 233,000 people. Here there are around 65,000 people.
How do we unpack these numbers and make sense of the spread here? It’s hard to tell right now.
The novel coronavirus, and all communicable diseases for that matter, spreads faster in densely packed environments. Just look at New York City. It is the most densely populated place in America and had 141,754 cases of COVID-19 as of Thursday. More people are hospitalized there with the coronavirus than live in Nacogdoches.
Therefore, it comes as no surprise that Harris, Dallas, Tarrant, Travis and Bexar counties have the highest number of infections in the state.
Nacogdoches County is far more densely populated than some Deep East Texas cities, but not enough that our town should end up with numbers nearly equal to Smith County.
The infection rate — confirmed cases per thousand residents — in Nacogdoches and Shelby Counties are two of the highest in the state. Shelby County’s infection rate was third highest in Texas at 3.34 per 1,000 on Friday morning. The rate was the second highest for counties with more than 20,000 residents and the highest for counties with more than 25,000 residents.
Nacogdoches was at 1.68 per 1,000 — higher than both Harris and Dallas Counties.
Why are our numbers so high? The key to a meaningful answer could be testing rates. The more people who are tested for COVID-19, the more confirmed cases.
Unfortunately, the majority of that key is locked somewhere in black and white in the state’s data. We can only look at a tiny bit of it.
I was unable to find testing rates for other counties with similar confirmed case numbers. I couldn’t find the number of tests performed in Harris or Dallas counties from any public health entities there.
The most recent and best data I found anywhere for any of the previously mentioned counties was in the April 14 edition of The Dallas Morning News. Those numbers indicate that 9,500 Dallas County residents had been tested for COVID-19. That’s roughly 3.6% of their population.
Here, emergency management officials keep us up to date every day with plenty of data to bite into and chew on. As of Friday morning, approximately 810 tests for novel coronavirus had been given by the local COVID-19 Call Center, so around 1.3% of the total county population has been tested.
Our rate is far more than the one-sixth of a percent of all Texans who have been tested, according to state figures. But it’s also far lower than Dallas’ now-outdated 3.6%.
In other words, we have a problem, and based on the scant data available, it doesn’t appear that our higher than average testing rate is completely to blame.
State officials acknowledged the problem on Thursday as the infection rate continued to grow. County and state officials were set to meet Friday to discuss the ongoing spread of COVID-19. As of this writing, I’m not sure what their conclusions are.
Does the continued spread indicate that Nacogdoches County’s curve isn’t flattening and didn’t peak on April 19th as predicted? Not exactly.
The number of coronavirus cases grow in an exponential manner. You may or may not remember from high school that exponential growth involves numbers growing more or less by a constant factor every day. Exponential grow is very slow at first but begins to increase more and more until the rate of increase is massive.
I’m not sure how long it currently takes for COVID-19 tests to be confirmed by the state. But let’s say that each of the dozen or so people identified Thursday and Friday took 48-72 hours to become sick enough to require testing and testing takes 24 to 48 hours to return results.
That would mean that every patient confirmed with the disease on Thursday had become infected sometime around April 19. Remember, this is only conjecture because we don’t have the solid data right now.
Following the model of total confirmed cases on the Nacogdoches COVID-19 Dashboard we can see slow growth at the start but no true skyrocketing effect. There are only about two months worth of data in the graph, so there’s not quite enough for me to make any sort of longterm meaningful analysis, and new daily cases have not dropped in a long and meaningful manner.
In other words, only time will tell if we are successfully mitigating the spread of the virus.
Anecdotal evidence suggest Nacogdoches County isn’t doing the best job at socially distancing or staying home, and the collected data is horribly flawed and unreliable. It gives us no clear picture.
That’s not to say I haven’t noticed a significant downtick in traffic or people out in public, but when I go out grocery shopping, I still see people packed together in similar number and density as before the pandemic. A good portion of them are wearing masks, from what I’ve seen, so at least that’s encouraging.
So what kind of numbers can we put on social distancing? For this part of Deep East Texas, the numbers are statistical garbage. As my first statistics professor, Dr. Tommy Eads, used to say, “There are lies, damn lies and statistics.”
A company tracking cellphone GPS data gives Nacogdoches County a C- on its social distancing scoreboard. Much of this low grade comes from GPS data calculating the change in average mobility — basically how far people travel on a daily basis. In that category we received an F. Do we deserve an F? Probably not.
Let’s use the first principal of statistical analysis to prove that this is junk.
Good statisticians look for and try to mitigate what are called confounding variables — any unseen or unrecorded action that might have an impact on data. Analysis doesn’t start with math. It starts with critical thinking. Take this example out of Dr. Eads’ playbook.
Ice cream sales rise at the same rate as the crime rate. It’s true, and every statistical model shows this correlation. But ice cream sales don’t cause more homicides and house burglaries. The confounding variable in that situation is summer, which leads to higher crime rates and more people seeking out a cold, tasty treat.
The confounding variable in using GPS data is that the South is far more spread out and less densely populated than the North. A New Yorker might go a mile or less to the grocery store. A trip to the store for a Southerner might be 30 miles.
The low density also means that essential work is farther away from home. That New Yorker might spent more travel time going to work, but a Texan will on average go a greater distance.
This particular data is also severely skewed by Stephen F. Austin State University shutting down its dorms for the semester for the majority of students. It shows a major uptick in movement around the time of SFA’s spring break as students were leaving town and factors that into our overall “social distancing” grade.
For now, stay home and only travel for essential activity. The latest mathematical models released Thursday indicate that the coronavirus outbreak in America started much earlier than anyone realized. As of March 1, the five largest cities in the U.S. had 23 confirmed cases, but the model shows there could have been about 28,000 cases at that time.
We have no idea how many people get infected with the virus and show no symptoms. Antibody testing in New York this week found that 1 in every 5 residents had the virus at some point, though most didn’t know it.
Now more than ever, as we either finish our peak or continue to climb, stay home and stop the spread.
Josh Edwards is managing editor of The Daily Sentinel.