How to Tell If Utah Is Having a Bad Day Today

It's Saturday, September 19th. It's a couple of hours before today's case numbers come out.

How do we know if the recent  uptick is continuing to get worse? Here's a number for you: 634 new cases.

That's a lot less than yesterday's 1,089 new cases.

How could a big decline from yesterday be a bad thing? 

*****************************************

Here's how. The case count data is seasonal. 

Seasonal is the word used in economics (and other fields) for data measurements that have a repetitive pattern in them. For example, employment is higher every December due to retail firms hiring more workers for holiday shopping season. Seasonality is a generic name: it doesn't always have to be related to seasons. What's important is that the pattern occurs regularly.

Case data in epidemiology is usually seasonal too. Importantly in this case, there are weekly seasonals in case data for most diseases. The reason is simple: the weekend. Doctor's offices are open less. Labs have fewer people working. Sick people who might have been out and around a bit and able to pop into a testing center on a weekday decide to hunker down at home for a couple of days, and so on. 

With positive test results, they're usually higher early in the week (as the labs catch up from the weekend), and lower on the weekend when only the most urgent tests are completed.

A particularly simple way to deal with this is to look at the growth rate between the same day each week.

This is really bad folks. That's a multiple that's plotted. When it's above one and level, we'll get more cases on a day than we got the same day last week. An outbreak, could even show up in something like this: if it bumped up to a new level, and then stayed horizontal (say, we'd been getting 5% more cases each week, and suddenly we start getting 6% more). 

But this multiple hasn't even leveled out yet!

Do the math. Yesterday's total of 61,775 cases was about 1.09 times larger than the previous Friday's 56,693 cases. So, if the outbreak is the same today and is was yesterday, we should have 1.09 times as many as last Saturday's 57,275. That's 62,409: an increase of 634 from yesterday.

Anything more and that upward sloping line has taken another step upwards.

Here's some speculation on where this outbreak came from (credit to MJ). The uptick starts a week ago Thursday. That's 3 days after Labor Day. The incubation period for COVID-19 is about 4-5 days. All of this is consistent with students away at college mixing with family back home over the holiday weekend. Given that the outbreak is mostly in the population center Salt Lake and Utah counties, and has skipped over some other counties with big student populations (Cache and Weber), I suspect the direction was from the colleges into the Salt Lake metropolitan area.