Revisiting Arizona's Daily Coronavirus Case Count

|
Posted: Sep 02, 2020 1:00 PM
Revisiting Arizona's Daily Coronavirus Case Count

Source: AP Photo/LM Otero

Like a lot of states, Arizona's daily reported coronavirus case count data has painted a somewhat misleading picture of the true progression of the SARS-CoV-2 coronavirus pandemic within its borders.

That situation exists because of two factors. First, the state has periodically experienced relatively short-term shortages of testing kits and supplies, which has led to delays in Arizonans being able to be tested for coronavirus infections, particularly when demand for coronavirus tests was elevated. Second, Arizonans also experienced significant delays in being able to obtain results after being tested because of a prolonged lack of adequate capacity to process large volumes of coronavirus tests at labs within the state. Worse, the state's delays in reporting test results have varied over time, with a variable lag between test sample collection and reported results affected by test demand and the available supply of testing capacity.

In our previous analysis of Arizona's coronavirus pandemic experience, we've worked around both these issues. The latest update to our chart tracking the daily reported number of confirmed COVID-19 cases in Arizona, covering the period from 3 March 2020 through 31 August 2020, takes both factors affecting the timely reporting of positive coronavirus test results into account.

To see the chart, click here.

For reference, the following table summarizes the major events indicated by the letters in this chart.

Timeline of Events Affecting Rate of Spread of COVID-19 Coronavirus in Arizona
Event/DateDescriptionObserved Change in Trends for Hospitalizations 11-13 Days Later
A19 Mar 2020California imposes statewide lockdown orderSignificant change from rising to steady (bounded range) rate of hospitalizations. We think Arizonans effectively implemented practices to minimize their exposure risk to potential coronavirus infections, which then happened to show up as a change in trend immediately after Arizona implemented its own statewide lockdown order.
B31 Mar 2020Arizona imposes statewide lockdown order through April 2020Minimal change, new COVID-19 hospitalizations continue within bounded range. We think the main effect of the lockdown order was to standardize how Arizonans minimized their coronavirus exposure risks, which allowed the benefits to extend until the order was lifted, although that came at great economic cost. The lockdown would later be extended to 15 May 2020.
C15 May 2020Arizona lifts statewide lockdown orderSignificant change from steady to rising rate of new cases, hospitalizations, and deaths.
D28 May 2020 to 15 Jun 2020Large scale political protests (Black Lives Matter/George Floyd/Anti-Police)Change in rate of growth in rate of new hospital admissions as the protests greatly increased the risk and rate of exposure to the coronavirus for younger Arizonans, who are less likely to require hospitalization. Sharp increase in number of cases not requiring hospital admission.
E19 Jun 2020Governor Ducey's executive order allowing counties to require wearing masks in public venues begins to be implemented.Significant change as new COVID-19 hospital admissions peak and begin to decline. Upward trend in deaths stop as state turns first corner toward improvement.
F30 Jun 2020Arizona imposes 'mini-lockdown' orderSignificant change, with acceleration in decline of number of cases. Deaths peak before turning a corner and beginning a rapid decline.
G1 Aug 2020Expanded testing capacity comes onlineSignificant reduction in number of reported confirmed cases, as test labs speed processing and eliminate backlog. Continued downward trend.

Behind the scenes, we've wondered what picture would emerge if we could eliminate these factors from Arizona's reported test results. As it happens, Arizona's Department of Health Services has been posting the state's confirmed coronavirus cases by day on its COVID-19 data dashboard, which removes one of the factors. The resulting change is shown in our new chart tracking the number of confirmed coronavirus cases by date of test sample collection shows what our analysis looks like when we can eliminate the variable timing impact of test processing delays.

To see the chart, click here.

The first thing we observe is that the number of confirmed cases peaks at a higher level at a slightly earlier date than the chart tracking Arizona's daily reporting of confirmed cases. This makes sense because the lag in reporting test results became increasingly prolonged when the limited capacity for processing tests was reached, creating a backlog.

We also see Arizona's seven-day moving average "epi curve" as shown in our new chart looks more like a traditional bell curve than a misshapen bell curve with an unfortunate bulge in its right hand side as indicated in our first chart. In this case, the bulge is the result of a backlog of tests that were unable to be processed until testing labs in Arizona finally brought a significant increase in testing capacity online in early August 2020. After they did, the bulge quickly disappeared.

The epi curve still shows the effect of the delays related to shortages of test kits, but here, we also see greater consistency in the estimated lag from events marking a change in rates at which Arizonans were exposed to coronavirus infections and the timing of when they were tested. For example, when Arizona's statewide lockdown order was lifted on 15 May 2020, the time from that change to when a change in the trend for positive COVID-19 test results took 10 days. However, as the demand for tests increased as the number of cases rose sharply in June 2020, straining the available supply of test kits and other testing-related supplies, the lag between event affecting the incidence of viral exposure and the timing of a change in trend increased by three days.

That lag is confirmed by contemporary reports of delays Arizonans experienced in being able to be tested for coronavirus infections during June 2020.

Interestingly, those delays largely evaporated within a few weeks after the spread of coronavirus infections peaked in Arizona.

The "truer" epi curve also allows us to get a better estimate of the number of "excess" cases Arizona experienced as a result of the anti-police protests that took place in the state from 28 May 2020 through 15 June 2020. Here, we projected a trendline based on how fast the number of confirmed cases was growing following the lifting of Arizona's state lockdown order to approximate the trajectory that positive COVID-19 cases would have taken in the absence of the protests. Taking the daily difference between the actual trajectory of confirmed cases and that counterfactual, we conservatively estimate Arizona incurred roughly 31,000 more confirmed COVID-19 cases than it otherwise would had these protests not taken place. Since Arizona has reported a total of 202,342 confirmed COVID-19 cases through 31 August 2020, the protests would account for at least 15% of the state's total.

We can also confirm that the two most significant events in reducing the spread of coronavirus infections in Arizona were Governor Ducey's executive order allowing state counties to require residents wear masks in public venues (Event E shown in the charts) and also the closure of businesses associated with high risks of exposure to coronavirus infections, such as bars and gyms (Event F).

We also think the end of the anti-police protests contributed to the reversal, but mainly in the sense that fewer Arizonans were exposed in the super-spreader event the protests turned out to be, which is to say that the effect was similar to how a fire burns less bright if you throw less fuel on it. Here, we note that participation in these protests was greatest from 28 May 2020 through 7 July 2020, after which, participation levels dwindled to zero on 16 June 2020 as they ended.

The end of the heavier participation period for the protests coincides with a slight reduction in the rate at which new cases were being reported 13 days later, while the end of the protests altogether is associated with another similar reduction in the rate of increase, though neither was sufficient to reverse the upward trajectory of positive COVID-19 cases. The data suggests the events that made the difference was the governor's mask order, which reversed the increase, and the governor's order to close businesses with high risks for viral exposure among customers, which accelerated the state's downward trend in new cases.

Overall, Arizona's number of cases now being reported daily is similar to the levels that were reported prior to the lifting the state's full lockdown order. Though now, most businesses have reopened (though at lower capacity for many) and residents are no longer asked to stay-at-home. Compared to the full lockdown order, Arizona's current approach appears to be a lot smarter way to cope with the coronavirus pandemic.