Newly released data from the state that shows coronavirus cases by ZIP code should be “cautiously” interpreted, say local and state health department officials, warning the numbers don’t accurately reflect where people contracted the virus and are skewed by testing availability.
The Arizona Department of Health on Sunday released data that in some cases provided more detail about the spread of COVID-19 than had previously been disclosed. That included locations of confirmed cases; some age, gender and racial demographics; and private and public testing numbers.
The dashboard was updated on Monday morning to show that 3,702 cases of COVID-19 and 122 deaths had been reported in all of Arizona’s 15 counties. That total included 668 cases and 29 deaths in Pima County.
More than 35 ZIP codes in the region had confirmed at least one case of COVID-19, with higher totals in Green Valley and Tucson’s south side.
In a blog post on Sunday, Arizona’s Health Director, Dr. Cara Christ, acknowledged that the numbers reflect anyone that tested positive in that ZIP code, “whether their infection occurred two months ago and has resolved or two weeks ago and is still active.”
“Thus, while ZIP code data is informative, it should not be used to determine whether or not residents of a particular ZIP code are or are not at risk of COVID-19,” Christ wrote.
Dr. Bob England, Pima County’s interim health director, reaffirmed his longstanding belief that the public should operate as if the virus is everywhere throughout the community.
“I understand that people wanted it, but it’s not useful to us and to the extent that it misinforms people to think that there is something inherently different about one neighborhood or another, that’s too bad,” he said.
Michael Worobey, head of the department of ecology and evolutionary biology at the University of Arizona, said he hadn’t gone through too much of the data, but that he already noticed a “few deficiencies.”
“I’m sure the people at the Department of Health Services are doing their best and trying to get all the information that they can and hopefully it will improve,” Worobey said. “But yeah, there are definitely some sort of gaps still in the picture.”
Data with a catch
The numbers released by the state show that the 85714 ZIP code has 67 cases of coronavirus, the most of any ZIP code in Arizona. Three other ZIP codes had at least 40 confirmed cases: 85706 (49), 85745 (45), and 85614 (42).
The data came with a catch, according to state health officials, who said that if they don’t know the home address of the patient, they include it in the ZIP code of the medical provider, followed by the address of the facility that reported the case.
England said the vast majority of the numbers reflect where the patient lives, not where they were infected.
He added that disease clusters, such as in long-term-care facilities, would drive up the numbers.
The 85714 ZIP code, which has the highest number of cases with 67, includes Sapphire of Tucson Nursing and Rehabilitation, which includes at least 27 known cases of coronavirus.
A spokeswoman for the facility previously told the Star that it couldn’t say how many people have since tested positive, although a source said that number is higher.
The 85614 ZIP code in Green Valley, where there are 45 cases, includes the Santa Rita Nursing and Rehabilitation Center, where at least six total residents and employees have tested positive.
But England said the more telling factor is that the numbers reflect the availability of testing, which he said has been more available in those facilities, to health-care workers and first responders, and for older people who have shown the most symptoms.
“Some of them are driven by clusters, more of them are driven by where more older adults happen to live because they were the ones getting the sickest and therefore being able to get tested,” England said. “Also where health-care workers and first responders live because they were able to get testing a little more than the general population.
“If there are 50 infections out there for every test result we know about, then this map really doesn’t tell us anything,” he said.
The data released by the state also included a running total of tests completed, which showed that 5,662 tests were done in Pima County as of Monday morning, with a 12% confirmation total. But testing appears to be decreasing week to week, with about 800 fewer tests last week than during the peak week of testing in late March.
Only 12 cases in Pima County were confirmed by the state’s Department of Health lab while the rest were confirmed by private companies.
In terms of the apparent decrease in testing, England said he’s not sure that’s accurate, as testing has become more widely available. He chalked those numbers up to some private companies either not reporting all negative tests or taking time to report their totals to the state.
“I don’t know how much credence there is to those numbers,” he said. “The numbers you get today reflect where it was spreading three weeks ago. ... It’s not a very reassuring thing to say, ‘We’re seeing this trend or that trend,’ because we’re always behind the curve.”
Death-rate differences
Meanwhile, the data did include some demographic information, which showed that 15% of deaths were among Native Americans, a far higher percentage than their share of the state’s population. However, 49% of deaths are listed as “unknown” and the data isn’t broken down by county.
Results for ZIP codes where Native Americans constitute a majority of the population were not released, though the Health Department provided no specific reason for suppressing the data other than they were waiting for tribal approval.
Worobey, the UA professor, labeled the number of deaths as the “most robust” and “most reliable” data point. But he said the lack of testing can impact those statistics if you don’t count people who died who could be suspected to have died from the virus.
“That still is a question in my mind about Arizona data is how accurate are those death statistics,” he said. “If they are accurate or fairly accurate, it’s still somewhat encouraging that we’ve got relatively low numbers compared to a lot of places.”
He said he was mostly concerned by the death and infection rates in Northern Arizona, including Coconino County.
Data shows Coconino County has a death rate of 14.64 per 100,000 people. That’s compared with Pima County’s 2.77 deaths per 100,000 people, which is still higher than Maricopa County, where it is 1.16.
England said he’s not surprised by higher rates of death of people of color, not only in Arizona but across the country, because of access barriers to health care and higher rates of preexisting conditions.
Speaking about the higher death rate in Pima County compared with Maricopa County, he said that can depend on a number of factors, including health disparities, age disparities or the location of an outbreak.
But ultimately he expects the ZIP codes with the highest populations to end up with the highest total of cases, and for the per-capita death and infection rates to end up evenly across the state, when further testing is available.
“A single long-term-care facility in either community that had a really bad outbreak could swing that either way,” England said. “I doubt that as we get further into this that there’s going to be a difference in death rates between communities.”
