Remember when public health experts were warning us not to travel over the holidays? In case you don't, here's what Dr. John Brownstein, chief innovation officer at Boston Children's Hospital
said two days before Christmas:
Layering in the December holiday travel and gatherings means we will have a third jump in this wave, or a surge on surge on surge. Unfortunately, this increase will likely lead to catastrophic impacts on hospital capacity and the acceleration of terrible mortality milestones through February.
The chart at the top of this post shows new cases in Maricopa County per 100K people, seven day trailing average. It is my favorite chart from ASU's Biodesign COVID-19 Dashboard. Notice that things change right around January 1.
Harvard says the incubation period for COVID-19 is from two to 14 days, but symptoms normally appear after around five days. So there is a plausible story here that people traveled for Christmas, stayed a few days, then headed home infected with the virus. Some of them took an extra week or week and a half to develop symptoms, and infected others during that time due to asymptomatic spread.
How much did things change and how bad is it now? Here are two charts using the same data as above, split before and after January 1:
I fitted a trend line to each half (dotted line). The trend is downward on the December graph, but upward on the January graph. From the slope of these lines we can estimate how things are changing from day to day.
The trendline for the last two weeks of December has a slope indicating a decrease of about one new case per 100K people each day. The January trendline, on the other hand, has a slope indicating an increase of about 4 new cases per 100K people each day. That is a dramatic reversal. If the trend continues, Maricopa County could see 167 new cases per 100K by January 31, about double the rate on January 1.
These numbers have implications too. You need to have new cases before you can have hospitalizations and deaths. Since January 1 the Maricopa County trend lines (data from the Biodesign dashboard) show COVID-19 hospital beds increasing at a rate of about 51 per day, and total deaths increasing at a rate of about 78 per day. Both seem to show an increase in the slope of the trend lines around January 1.
Let me close by noting that without other data we can't claim that the change from December to January was
certainly due to holiday travel. Doing so would be to commit the
post hoc fallacy. There could be some other factor that just happened to come into play at the same time. But there is clearly an association, and given the expert warnings and timing holiday travel strikes me as the most likely explanation.
