Iowa COVID-19 positivity rates may be even higher than the state has reported to the public, and the explanation given by state officials doesn’t make sense — even to biostatisticians.
School districts across the state are using 14-day positivity rates to help decide whether it’s safe for students to go to school in person or switch to online learning. These decisions affect students, teachers and families who have to scramble to find alternative care for kids.
The 14-day positivity rates reported daily by the Iowa Department of Public Health are computed with a method recognized by the U.S. Centers for Disease Control and Prevention. But since the state doesn’t report to the public some of the information it uses, Iowans trying to check the state’s work reach different conclusions.
For example, the state reported Monday that Linn County’s 14-day positivity rate was 22.8 percent. But when The Gazette used public data reported on the state’s COVID-19 dashboard and followed the state’s formula — individuals who tested positive divided by total individuals tested between Oct. 26 and Nov. 9 — the positivity rate was higher: 37.2 percent.
“The manner in which you’re calculating the 14-day positivity rate is correct,” Joe Cavanaugh, professor and head of the Department of Biostatistics at the University of Iowa, told The Gazette.
When The Gazette asked Cavanaugh if he had any idea how the state might be arriving at lower rates using the data available to the public he said: “Good question! It’s difficult to know, since the description on the IDPH dashboard indicates that the rates are calculated using the exact same approach that you’ve outlined.”
Paul Trombino, Gov. Kim Reynolds’s chief operations officer, last week told The Gazette he thinks the reason several media outlets have reported getting higher positivity rates is because of how the state tracks people who have multiple tests.
ARTICLE CONTINUES BELOW ADVERTISEMENT
“The biggest challenge in the 14-day window is the number of tests we are seeing, which is the repeat testing,” Trombino said.
Because of outbreaks at some state prisons and long-term care facilities, some people are being tested several times a week or even once a day. The state includes all of those tests in the daily data, which The Gazette downloads and reports on every day.
But when the state does its 14-day positivity rates, officials count only the one final test, which is often a positive test, removing the previous tests before doing the calculation, Trombino said.
“In order to do an individual (calculation), we can’t count an individual three times,” he said.
Trombino told KCRG-TV9 last month the number of tests being done each week in Iowa is twice as high as what is logged on the website, but the total testing numbers aren’t made public.
What doesn’t make sense about Trombino’s explanation, however, is if the state is counting fewer tests, then its denominator (total positive and negative tests) should be lower — which would make the state’s positivity rates higher than those computed by The Gazette and other observers.
“You should be getting a lower number if that were true,” Julian Wolfson, associate professor of biostatistics at the University of Minnesota, said of The Gazette’s computation.
Wolfson, who created the Minnesota School Opening Dashboard, which tracks COVID-19 positivity rates in all Minnesota counties, said he could see how the number of tests reported to the Iowa Public Health Department could be unreliable, given the data are coming in from various test sites and labs.
ARTICLE CONTINUES BELOW ADVERTISEMENT
“I could imagine a scenario where there are these straggling negative tests that people get around to reporting at some point, but it’s not nearly as consistent as the positive tests,” he said. “I could see a reconciliation process occurring behind the scenes.”
Wolfson and Cavanaugh agree Iowa should report the total number of tests given in addition to the number of people tested.
“Reporting counts at the test level provides a better indication of current transmission activity in the community,” Cavanaugh wrote in an email. “Moreover, if an individual tested multiple times will only be counted once, then whenever an individual who has been previously tested is tested again, their earlier result will be deleted and replaced with their current result. This leads to revisions to the historical record that can be confusing.”
The state also doesn’t report new county numbers each day, which limits Iowans’ ability to track positivity rates in their own communities that way.
Public health officials have argued some counties might have so few cases that individuals could be identified if the county numbers were made public. But other states, including Minnesota, do report county data and omit specifics if the count is five or fewer.
Trombino said state officials are “having dialogue” about providing information about the total volume of testing in Iowa, but there is no deadline for including that data on the public dashboard.
“Like anything else, we’re trying to make sure we’re representing it in a way that doesn’t cause confusion,” he said. “I expect we’ll continue to make improvements.”
Comments: (319) 339-3157; email@example.com
John McGlothlen and Katie Brumbeloe of The Gazette contributed to this report.
04:05PM | Sun, November 29, 2020
06:30AM | Sun, November 29, 2020
01:50PM | Sat, November 28, 2020