Editor’s note • This story is available to Salt Lake Tribune subscribers only. Thank you for supporting local journalism.
Health experts and elected officials have eagerly looked for case data that either supports or disproves the effectiveness of the masks being worn in the Salt Lake City School District — the county’s only district with a mask mandate.
Health officials and Salt Lake City Mayor Erin Mendenhall have pointed to the cumulative number of COVID-19 cases to show the mask order is working; Salt Lake City schools have had fewer cases per student than any other district in the county.
Meanwhile, a Republican member of the Salt Lake County Council, who voted down a countywide mask mandate in schools, pointed to the low reported number of active cases in the neighboring Granite School District to show that a mask requirement isn’t necessary.
So which side is right?
It turns out that a lot of factors are making districts’ case data difficult to compare — and they raise serious questions about the accuracy of the numbers coming out of some Utah districts. Here are some factors to consider.
1. Salt Lake City started classes a week later than other districts.
Salt Lake City schools indeed have had fewer total cases per capita than any other district in the county, as Mendenhall and county health director Dr. Angela Dunn both have pointed out.
One reason for Salt Lake City’s lower cumulative numbers may be that its school year has had fewer days of classes for cases to spread and be added to the total. Salt Lake City schools began on Aug. 24, while other districts began nearly all of their classes Aug. 16.
Adjusting the districts’ cumulative case rates to the number of days since classes began shows a different picture:
But that picture also is incomplete. There is a lot of evidence that some districts’ numbers are far less accurate than others.
2. Granite and Jordan school districts have been reporting drastically delayed case numbers.
In the first three weeks of school, some districts appeared to have severely undercounted the number of students and employees who actually tested positive, according to the county’s school case data.
For a case to be counted as a “school case” for any district, health and school officials have to verify the patient was in school while they were contagious — a step that can take up to 72 hours, county health officials have said.
That means a lot of positive tests don’t get included in school case counts until days after the results.
All districts’ case counts are at least a little delayed. But some districts are far more delayed than others.
Canyons District, for example, had recorded 152 new cases in the past 14 days as of Friday, plus 189 cases that were older than 14 days. But 14 days ago, the district was reporting just 157 cases. That means its case count at that time captured 83% of the cases now confirmed for that time frame.
And Canyons has been reporting the most complete case counts of any district in the county, followed by Salt Lake. Those districts’ case counts generally have been one to three days behind.
Case counts in Granite and Jordan, meanwhile, have been capturing fewer than half of the students and employees who have tested positive at a given time — and their case totals have generally been at least a week behind.
That means case rates in the four largest districts look very different in retrospect than they do with new daily numbers.
Here are the per-student case numbers each district was reporting through Sept. 7:
And here are the per-student case numbers we now know about for the same time frame, based on what districts have been reporting as “inactive” cases two weeks after the fact:
Salt Lake City’s per-student case count was notably lower than the other three large districts as of Sept. 7.
But again, the district started classes a week after the others. Adjusting the more complete, “inactive” case figures to the number of days after classes began shows that Salt Lake City’s early case numbers actually appear to be higher than those in other large districts.
That said, county data shows the school district was tallying cases a week before classes began, which could account for its higher case count on the first day of school. And while Salt Lake City’s case counts started out higher, its rate of increase in the first two weeks of school appears to be slower than the other large districts. Jordan and Granite’s numbers not only rose more quickly than Salt Lake City’s in the first two weeks, their biggest increases came shortly after that.
Meanwhile, we already have some idea how accurate (or inaccurate) districts’ counts have been, by averaging the percent of “inactive” cases a district captured in their daily case totals. By applying that factor to subsequent case totals, it is possible to estimate more complete numbers of positive tests in a district on the days since Sept. 3.
If each district continued to undercount cases after Sept. 3 to the same degree they were undercounting before then, Salt Lake City schools’ numbers likely will have fallen below other districts, even adjusting for its late start.
It’s possible that a district improved its verification processes after Sept. 3, producing more complete daily case counts. Salt Lake City, for instance, was only at the end of its second week of school on Sept. 3; its reporting may have become more efficient and accurate after that. But Canyons district had relatively accurate case counts before Sept. 3, and its subsequent reports show cases rising at about the same rates as Salt Lake City’s — and Canyons has no mask requirement.
With districts showing such a wide range of case-counting accuracy in just the first two or three weeks of school, it’s difficult to tell how Salt Lake City has compared to the rest of the pack — or how it will compare in the next month or two.
3. Active case counts also have been inaccurate.
Aimee Winder-Newton, a Republican who represents parts of West Valley City, Murray, Taylorsville and West Jordan on the County Council, raised eyebrows in early September when she tweeted a chart showing Granite School District, not Salt Lake City, had the lowest per-capita case counts.
Rather than cumulative cases, she looked at active cases — people who tested positive or developed symptoms within the previous 14 days. Salt Lake City’s late start date gives it far less of an advantage when active case counts are compared, rather than cumulative cases for the whole school year.
And as of Tuesday, Granite still appeared to be reporting the lowest number of active cases per capita:
But because the newest cases are the ones least likely to immediately appear in case counts, the degree of reporting delay also has implications for the accuracy of “active” case counts.
For example, Granite was reporting 375 “inactive” cases as of Friday — which means there were at least 375 total cases 14 days earlier, on Sept. 3.
On Sept. 3, Granite was reporting 27 inactive cases — people who tested positive on or before Aug. 20.
Unless the district identified a lot of cases more than 14 days after they tested positive — which itself would suggest a significant reporting problem — that means there were nearly 350 active cases at Granite District on Sept. 3.
Instead, the district was reporting just 98 active cases at that time.
Here’s how other districts’ “active” case counts appear in retrospect:
That delay is not just important to the question of whether active case counts support or disprove Salt Lake City’s mask mandate; it also can affect whether schools enact protections like Test To Stay.
For example: While Utah County doesn’t report school case data like Salt Lake County does, state data show the Alpine district’s Cedar Valley High School as of Friday had 65 cumulative cases for the school year, 15 of them active. That means there were 50 total cases on Sept. 3 — but at that time the state was reporting just 40 cases, three of them inactive.
Unless additional inactive cases were identified more than two weeks after the fact, there would have been 47 active cases on Sept. 3 — enough to reach the 2% case rate that triggers mandatory testing to attend school.
Instead, state data show the school was reporting just 37 active cases at that time, and protective action was not taken.
Figures reported two weeks after the fact aren’t the only evidence that districts’ real-time numbers are inaccurate. The county also tracks case totals for school-age children, often before those cases have been verified by schools. On Sept. 13, county health officials calculated “active” cases among school-age children for each district by cross-referencing positive tests with enrollment data, rather than waiting for attendance to be verified.
Again, the active case rates Granite and Jordan were reporting at that time were much further below the county’s “school-age” case numbers than other districts’ reported rates were, suggesting those districts’ case counts were more delayed. By looking at case rates among school-age children in a school district, Salt Lake City’s rates were the lowest in the county.
As of Wednesday, county health officials had reviewed unverified “school-age” and verified school cases, which they believe are now catching up with the backlog, said Nicholas Rupp, spokesman for the Salt Lake County Health Department.
“All of those variables are going to give you different rates, and in some of those circumstances Salt Lake is looking much better. In others it’s looking similar — but it’s not worse,” Rupp said.
4. More kids may be getting tested in some districts than others.
Even if districts could accurately count every positive test on the day of the result, there still likely would be a lot of infected kids who do not get tested.
Percent positivity can show how comprehensive a community’s testing is. The greater the percentage of people who test negative, the likelier it is that most of the positives are being identified. However, if a larger percentage of people who get tested receive positive results, that suggests not enough people are getting tested to get a very complete picture of how many people actually have the coronavirus.
And enthusiasm for testing varies from place to place. For example, there have been reports of parents in some schools discouraging each other from testing their children, Rupp said.
Parents from Canyons and Jordan districts, as well as charter schools, have told county health officials that “parents from my kid’s grade, my kid’s school, my kid’s football team, or whatever, are saying we should not test,” Rupp recounted. He noted he hasn’t heard similar reports from other districts, but, he said, that doesn’t mean it isn’t happening.
Meanwhile, in Utah County, Alpine School District last week sent out COVID response guidelines that reminded employees that they don’t need to get tested or quarantine if they are exposed to the virus because the health department can’t enforce its recommendations; district officials later claimed the emails were sent by mistake.
According to county zip code data, percent positivity among school-age kids has been lower in the Salt Lake City district than in the county’s other districts. In the zip codes that are within Salt Lake City district boundaries, about 7% of children ages 5 to 17 who have gotten tested since their school year began received positive results. That compares to about 9% in Murray and Canyons, 10% in Granite and 11% in Jordan.
“The testing rate definitely comes into play” when comparing districts’ case counts, Rupp said.
5. Some kids’ positive tests may never get confirmed for their school districts because their parents won’t talk to health officials.
Even if children did get tested, their cases may not ever be verified as cases in their school. That’s because some parents are refusing to talk to health officials who call to confirm the child’s school attendance, state health officials have said.
Those cases still are counted as school-age cases, but they may not ever be verified as school district cases — which means they won’t count toward district metrics that trigger Test To Stay.
Salt Lake County officials have said they planned to revamp their school case verification system to enable districts to more easily confirm an infected child attended school during their illness. But that doesn’t help previous school case data.
It’s not clear whether parent non-cooperation would affect some districts’ counts more than others.