Arizona Winter Case Outbreak Anomalies – 1/20/21

The Case data for the Arizona winter outbreak has shown some unusual characteristics that might be unique to the state. The chart below shows a few of these anomalies.

1. The cases began acceleration almost to the day that temperatures in Phoenix dropped from 15 to 20 degrees F overnight. The cooling trend only stayed for a few days (see the orange and blue bar chart below), but nighttime lows remained lower until November 8th when the temperatures plummeted about 20 degrees again. These drops plus the movement of the nighttime lows below 60 degrees seems to have triggered the case acceleration that has only recently began to slow.
2. The green line in the diagram below represents the 55-64 group. This plus the 65+ group makes up the bulk of Arizona’s annual winter visitor population. Arizona State University did a study a few years back that showed that the population in the Phoenix metro area swelled by 300,000 every winter. I have heard that Tucson’s population increases by over 100,000 every winter and areas in the western part of the state (such as Quartzite) are said to have over 2M visitors every winter. So why is the 55+ groups both accelerating faster than the 20-54 groups? My guess is that the winter visitors are bringing COVID cases with them, and since I’m normalizing by the population of these age groups that lives year round in Arizona, I’m showing a higher number of cases per 1000 “year round” residents. I suspect if I could correct the denominator of this ratio with the actual numbers of full-time plus winter residents each month, then the rates would equal the 20-54 groups. What does this tell us? Some decent percentage of our case counts during the winter outbreak are due to winter visitors. (edit: Using some back of the envelope math I arrived at about 10K cases in the 55 to 64 age group due to winter visitors. This would mean there have been around 90K winter visitors in this age group since November)
3. Just before Thanksgiving the case rate appears to have slowed a slight bit, but around the end of November, we start seeing strange case count disruptions and apparent acceleration. Looking closely (I’ve blown this range up in the graphic right after the one below), it seems like a short glitch that is probably due to cases being counted inconsistently during the holiday break. Some days had no cases reported and others had nearly as many positive cases as tests. Obviously that was the state catching up on counts they had accumulated when they were out of the office. I don’t see any strong evidence of a holiday surge for either Thanksgiving or Christmas. At best there could have been a slight acceleration, but it appears more likely that inconsistent data collection may have given the impression of a short uptick in cases.
4. The under 20 age group (blue curve) continues to show that the rate of infection in this group is significantly lower than the other groups. This may be due to our inability to measure these cases well (since so many are asymptomatic) or due to this group’s lower probability of getting infected. Or maybe it’s because this group is more sheltered than others because they stay at home, don’t need to go to work, aren’t buying groceries, etc. Or some combination of these. Remember from my recent excess death analysis that this age group had around 1/2 (on average) the deaths during 2020 than would be expected in a normal year.
5. For the last week or two, the cumulative curves for all the groups (less so the 55-64 group) show obvious deceleration. This coincides with the decrease in COVID hospitalization that I show below.

Below you can see the zoomed in area where the data glitches occurred. The overall case slope during these times appears to be constant, so the appearance of case acceleration during the holidays is probably due less to a holiday surge than to poor data collection.

Below is the hospitalization chart from the AZ DHS dashboard. It is a stacked bar chart, and the combination of the colored bar percentages will add up to 100%. The Red bars reflect COVID cases in the ICU (60% of all ICU beds today) whereas the dark grey reflect non-COVID ICU beds (about 30% of all ICU beds today). The light grey is the unoccupied beds (about 10%). As you can see the numbers of beds occupied by COVID patients has consistently been dropping for a week or two. This is consistent with how the Summer outbreak worked. Hopefully this means the hospitals are through the worst of it.