Understand the Statistics of COVID – please?

Recently we have been hearing a lot of talk about the infection rate in Illinois. The growth in this number is quite shocking. Where it was 3% a month ago, it was 5.7% last Thursday, and it is 8% as of this writing!

Figure 1 New cases in Illinois (Bing's COVID Update)
Figure 1 New cases in Illinois (Bing’s COVID Update)

One might be led to believe that this means that 8% of Illinoisans are infected with COVID, but it does not. It means that 8% of those tested were positive. Those are wildly different things and the actual percentage of Illinoisans with the disease is something different. It could be higher and is likely much lower. Its rise could indicate a growth in the general infection rate or nothing at all. Residents of Illinois – more than almost any other state – need to be able to read between the lines recited by your officials.

I am not a conspiracist or an anti-science guy. Furthermore, I believe that the number of people catching COVID in Illinois in increasing rapidly and is cause for concern. However, my knowledge of science (and statistics in particular) leads me to worry that our elected officials are incorrectly interpreting “positivity rate” and ignoring more appropriate statistics altogether when making policy decisions.

Let us return to the early days of COVID. Initially the positivity rate throughout the Rush Medical System started in the 8% range. Within a couple weeks, that number had spiked to 25% This was consistent with what was reported in the press. The overall rate of infection in the state was unknown, but this number jumped because, given the shortage of tests, doctors began screening for symptoms before allowing a test to be administered. So, if a patient was asymptomatic or wanted a test to placate personal or professional curiosity, the request for a test was denied. Only the people who were likely sick or front-line were tested.  The infection rate of all Illinois was well below 1% at the time, but the 25% (and rising) infection rate measured meant that the tests were being used more effectively.

Post-it note captured at a Rush nurse’s station in April 2020

At some point the purpose of the this statistic was corrupted. The number is easy to track and regularly reported and it has come to be used in a way that was never intended. Testing issues have improved since then, people who are sick are still more likely to seek out testing and doctors are more likely to prescribe testing to symptomatic patients. Further complicating the issue, certain professional and demographic groups get tested more than others leading to overall results that do not match the population. In statistical terms, this is called selection bias. So, when you hear that the infection rate is 8%, understand that there is no scientific or even commonsense reason to equate that to the whole of Illinois, Chicago, or any geographic group.

Even though the positivity rate has risen 1300% since June, the fatality rate has fallen by 80% since the first peak

Still, there are important stats worth watching. My favorite (as macabre as this sounds) is fatality rate which – due to research, improved medical infrastructure, and improved treatment – has fallen consistently throughout COVID. On the first pandemic peak on May 13, 4100 people tested positive and 141 people died. On the second peak in October, 6100 people tested positive and 63 people died. This stat is not perfect either, but still, a positive test in May represented a 4.7% chance of death and a positive test in October represented a 1.0% chance of death. So even though positivity rate has risen 1300% since its low point in June, the chance of dying has fallen by 79% since the first peak. This is a reason to rejoice, not retreat further into our fears.

Figure 2 Fatal cases in Illinois (Bing's COVID Update)
Figure 2 Fatal cases in Illinois (Bing’s COVID Update)

Perhaps the best statistic available is deaths per 100,000 people. This stat cleanly identifies one’s likelihood to die from COVID and is calculated using the relatively bias-free numbers of population and COVID deaths while avoiding the sample bias of testing. Illinois’s current D/100K number is 78. That number sounds arbitrary but makes sense when used for comparison purposes. Remember our Mayor villainizing that COVID hotbed, the State of Wisconsin, a few weeks ago? Yet Wisconsin’s number is only 32. For whatever reason, Illinoisans have over twice the chance of dying from COVID than their neighbors to the north. Iowa’s number is 53, Indiana’s is 62, and Missouri’s is 47 – suggesting that all neighboring states are safer than Illinois. In fact, Illinois is and has been one of the top 10 most dangerous states as a function of COVID. Pile crime, politics, and taxes on top of that and start wondering why anyone lives here – but that is a subject for another day.

Figure 3 Death rates from COVID-19 as of October 28, 2020, by state (Statistica.com)
Figure 3 Death rates from COVID-19 as of October 28, 2020, by state.
For the complete chart, follow the link above. (Statistica.com)

I have been frustrated by misdirected, arbitrary, or politically motivated COVID policy since the beginning. I am not arguing that COVID is not dangerous. I am not arguing that people should not be diligent. I am arguing that Illinois officials are looking at the wrong data, looking at data incorrectly, and in too many cases expecting the public to accept “because science says so” without understanding the science themselves.

This issue has unfortunately been politicized. Please, do not reject my logic because it coincidentally aligns with the politics of others you oppose – some of whom you view as idiots. All the links to these numbers are included above and none of my sites have any political bias. If you wish not to believe me, click through and do your own research.

7 thoughts on “Understand the Statistics of COVID – please?

  1. Curt, you seem to be coming more and more professional. Is it true that in Chicago the health commissioner has been reporting Covid deaths while the death may be of another disease? I hope you are getting some responses from your work your blog work. How big is your following? There was one slip up you might want to try to avoid. Try not to use inflammatory words like idiot.This was an excellent blog as usual. Vic


  2. I don’t believe you can emphasize enough how the prescreening for COVID testing impacts the “positivity rate” in the same way that selection bias skews poll data. The general public assumes the rate is either % population with cases or a clean positives/tests with no bias impacting who seeks or gets a test.


  3. Its always nice to see someone on the opposite side of the aisle also shouting at their TV that that’s not what the math means. Despite a different political leaning I agree that it’s a very misleading statistic and that the information about hospitalizations is probably more accurate. I also think the first few months are not representative of anything because we really didn’t understand the basics. The statistics on vitamins out of our alma mater and Northwestern and the mask studies of Nebraska and Kansas seem to indicate that that’s all we need.


  4. Curt,
    You’re right that test positivity rate is not a good metric for guiding public policy or personal decision-making. What really matters is the locoregional prevalence of disease. I am beyond frustrated with our public health officials for not developing validated models for estimating prevalence. I have my own – based on daily deaths, an estimate of mortality rate adjusted for changing demographics/improving treatment over time, daily cases being diagnosed, and detection rate. It shows our prevalence creeping upward, but without antibody seroprevalence studies to validate the mortality rate, it isn’t something I have enough confidence to publicize.


  5. Curt, What would you say is going on such that people, who for the most part have served to make things better for the wider population, have got all of this so wrong? And separately, who would you say has been creating the politicization of Covid?


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s