On June 19, UAMS shared a very scary forecast. Arkansas would hit 18,000 positive tests about July 7, and peak with 150,000 active cases by the end of September. Viewed today, if anything, it underestimated how bad it could get. THREAD 1/12 https://publichealth.uams.edu/wp-content/uploads/sites/3/2020/06/Predictive-Models-for-Web-June-19-2020.pdf

Arkansas hit 18,000 positive tests* June 25, about eleven days early. Not good. At 22,257 total positive tests (the biggest increase in total positive tests ever today), we're off the charts. Literally. 2/12

The June 17 Arkansas model gave "very, very high numbers" according to Gov. Hutchinson at the time. 3/12 https://www.arkansasonline.com/news/2020/jun/24/state-gears-up-for-covid-19-surge/

It's pretty straightforward to use Excel to recreate the UAMS short-term forecast for today. (I didn't manually enter in all the data for March-May.) Thanks, as always, to @ArkansasCovid for keeping the data accessible. 4/12

If Excel did the arithmetic correctly, we're looking at about 6000 more positive tests in the next 2 weeks in Arkansas. A slight slowing - it only took 10 days (4 fewer days) to get the last 6000 positive tests. 5/12

However, Excel may not be right. 1) The more recent positive tests are pulling away from the trendline. 2) This does not address not having enough tests and contact tracing to identify most of the cases. 3) July 4 is coming, and we're already pretty germy in Arkansas. 6/12

I don't know how UAMS came up with 3,300 late-September hospitalizations in their forecast so I can't recreate it. (They didn't show their math.) I would guess that we'll have 3,300 fellow Arkansans in the hospital earlier, and still more if we don't change something. 7/12

If I understand this slide correctly, we're already ahead of the UAMS active infections chart simulation by 2 weeks. (My question is: what does "new infections" mean? I used active cases as a proxy. We know that 14 days after a positive test, the case is no longer "active.") 8/12

I also don't know why our active infections would decline from a peak of possibly 300,000 in November. What would we change in November that we aren't willing to change now? Or is it that only 10% of Arkansans will go to COVID parties? 9/12 https://twitter.com/ABC/status/1278430344392388611?s=20

n.b. * Positive tests aren't the same as positive cases. Not all positive cases get tested. I try to use the term "positive test" to help me remember that we're not testing anywhere near all the positive cases. 10/12

Stay home if you can. Wear a mask if you can. Advocate for those who can't. We can keep doing this. 11/12

Disclaimer: Not an epidemiologist. Not a statistician. Not a public health professional. Not even very good at using Excel. I hope UAMS will publish an updated forecast soon. 12/end

Sorry - one more graph. It works for Washington County, too. Only, 5% of our neighbors will test positive in early August. 10% in mid September. 13/12