Different COVID-19 reporting unfairly affects whether Fayette schools can open

As you may be aware, a large discrepancy currently exists between state numbers and local numbers for COVID-19 cases in Fayette County. Last week, Lexington-Fayette County Health Department officials came out to say that the discrepancy is due to data delays statewide and that their data is the most accurate. The governor responded by pointing the finger at the LFCHD for being “one of the slower [health departments] in the state” on reporting, and the state public health commissioner stated that the difference isn’t significant because the state uses a 7-day average. So, why does this matter?

The debate revolves around measuring incidence rate of COVID cases per 100,000 people for return-to-school. This is calculated by taking the number of cases from Fayette County reported by the KDPH over the last 7 days, dividing by 7 (to average it out), and then dividing by roughly 3.25 (using 2019 Census data). The result is the incidence rate used for the state’s color-coded school safety map.

Now, what’s the issue? As of Saturday, the state reports 6,747 cases in Fayette County, while the LFCHD is reporting 8,673. We should take LFCHD’s number as most accurate, meaning the state lags by almost 2,000 cases. Per LFCHD statements, the county is roughly a week behind in its reporting to KDPH, leaving about 1,300 cases sitting somewhere in Frankfort that have been reported by the county but not yet by the state.

Will the state ever catch up? Maybe. Will they catch up in a consistent fashion? Probably not. So, what happens if one day an industrious data analyst decides to make a big push at this and reports 600 of the 1,300 lagged cases they have just sitting there? (this is not beyond belief—at the start of September the state appears to have quickly caught up on a reported 722 cases from Jefferson County, where over 500 were acknowledged to be a month old). Then, hypothetically, if the state reports 600 cases in Fayette County in a given day (old cases, not new infections), this on its own would jump the 7-day average incidence rate by 26.4 points. That means that, even if 0 cases were reported over the next 6 days, the county would remain colored red by KDPH for a week just because of a single “catch-up” day!

And something like this is happening right now. Thursday to Saturday, KDPH reported 159, 146 and 254 Fayette County cases, while LFCHD reported 70, 63 and 76. The highest three days ever from LFCHD are 131, 149 and 167. It’s clear that KDPH is reporting on a delay and using more than one days worth of cases a day currently.

Does this seem right? Should this metric form a single “line-in-the-sand” for determining if it is safe for our children to return to school? I would emphatically say, No. Under ideal circumstances, could this help? Yes. But given what is indisputable about the current data setup in Kentucky (no matter who’s to blame) this is a recipe for giving a skewed version of reality. Until people are aware of the subjective nature of the daily numbers they are being presented with, they will not understand how there might be a better way to move forward. I believe the LFCHD has presented an accurate view of the data to our board and superintendent. My hope is that they will see this, reject the guidance from the state, and act in a way that shows a better understanding of what’s actually taking place in Fayette County.

Todd Burus is a mathematics and statistics professor and free-lance data scientist. He is also the facilitator of the Let Them Learn Fayette County on Facebook.