Both King County Public Health and Washington Department of Health went 48 hours without updating the public on COVID-19 infections, then shared a confusing update with a surprising lack of detail. I dug into the numbers and found some issues with the most recent update.
First, there has been significant changes to the daily infection numbers, including 5 infections that don’t have a date associated with them (!?!?). Here’s a graph of the revisions: “Old” is based on the King County Public Health press releases over the past month, and “New” is based on the new Tableau-powered website:
A few takeaways based on King County’s revisions:
- First, it doesn’t give me a lot of confidence that King County can’t get the numbers right.
- Why the big differences? My guess is that previously, King County was reporting based on when the test results came in, and now they’re reporting on when the test was taken. This has caused infections to “shift back in time.” I do think this is a superior way to look at it, but why did it take a month to get here?
- The spike at the end of March was causing my model to predict exponential growth, while the revisions cause smoothing and no longer predict exponential growth (which is great).
- One other takeaway: Now King County is reporting with a 48 hour delay. Unfortunate.
Here is the new model built on the revised #s. It’s now linear and doesn’t predict exponential growth. The prediction for today is 162.
I also looked at the % of tests that are positive in King County. Very encouraging trend. It started out very high and has declined rapidly. It now sits at 13%.
Here’s the trend of COVID-19 tests conducted. My only guess on why the sharp rise and sudden drop: Social Distancing / Stay Home order must have been effective, not only to reduce Coronavirus transmission but also the transmission of flu/cold. Tests dropped, but positive tests stayed remained the same.