A top official with the Iowa Department of Public Health acknowledged that a flaw in its COVID-19 data system has led to inaccuracies in the reporting of new cases and, in turn, lowered 14-day positivity rates in counties across the state.
Rob Ramaekers, surveillance unit lead epidemiologist for the IDPH, disclosed the issue in an email exchange with Dana Jones, a nurse practitioner based in Iowa City. Jones later shared the correspondence with the Telegraph Herald.
“We recognize this is a problem and have been working on logic to handle it,” Ramaekers wrote.
In his email, Ramaekers admitted that the reporting error resulted in new positive cases being incorrectly assigned to prior weeks and months. Because many new cases have been coded as having occurred more than two weeks ago, the errors lowered state-tracked 14-day positivity rates for counties, the key metric regarding in-person teaching in schools.
The Telegraph Herald has been tracking inconsistencies in state figures for weeks, including positivity rates from the state that consistently were lower than what would be expected from the underlying data. Last week, the TH printed two stories regarding data inconsistencies — with varied partial explanations from state officials — that could be explained by the latest revelation.
But state officials had not made any public announcement about the flaw as of Monday afternoon, when the TH published an early version of this story. Reached by the TH on Monday afternoon, City of Dubuque Public Health Specialist Mary Rose Corrigan said the IDPH had not yet provided local health departments with any information about the data flaw.
“We hope it is resolved quickly,” Corrigan said. “We rely on the Iowa Department of Public Health website for accurate information. We use those metrics to guide our decisions and to help schools, health care providers and other organizations in the county.”
The Telegraph Herald reached out to the IDPH and the office of Iowa Gov. Kim Reynolds early Monday morning and repeatedly throughout the day. Neither entity responded to requests for comment.
Jones received the email from Ramaekers on Friday after spending multiple months tracking the state’s coronavirus data and noticing inconsistencies.
She sent an email to the IDPH general mailbox on Aug. 9. Five days later, she received a reply from Ramaekers in which he detailed the root of the flawed figures.
Once an individual receives his or her first COVID-19 test, Ramaekers explained, the system locks that person into the date of their initial test.
For example, a person could test negative for COVID-19 in March, then test positive for the virus last week. In this scenario, the positive case would be assigned to the date of the initial test in March.
Asked whether the positivity rates viewable on the website were incorrect, Ramaekers answered in the affirmative, noting that “a similar situation is happening with the percent positive calculation.”
This data plays a critical role in determining how local schools will function as they welcome back students.
State officials said late last month that school leaders only can ask for permission to close buildings or districts if their counties have a 14-day average positivity rate of at least 15% and at least 10% of students are absent, or if the county has a 20% positivity rate over 14 days.
‘NEED TO FIX IT’
Jones shared news of the IDPH glitch with Eli Perencevich, an infectious disease physician and epidemiologist at University of Iowa.
“My initial reaction was they need to fix it right away,” Perencevich said. “A lot of decisions have been made based on the belief that this is accurate data.”
Megan Srinivas, an infectious disease physician based in Fort Dodge, Iowa, said she also has tracked data inconsistencies over the past few months.
“The numbers presented on the state website didn’t seem to be matching up with the county numbers and didn’t match what we were seeing on the ground,” she said.
Srinivas had mixed emotions when she learned of the system flaw.
On one hand, she is hopeful that identifying the problem can lead to a solution.
On the other, she was chilled by the reality that the state has made key decisions based on inaccurate numbers.
“The number of positive cases they are reporting (for) each day is completely invalid,” she said.
This, in turn, taints the daily and two-week positivity rates that serve as a guide for reopening schools.
“Those positive cases are the numerator for community positivity rates, which has become the real vector of decision making,” Srinivas said. “It is very scary that we are using a completely invalid metric.”
TRACKING THE DATA
In recent weeks, the TH has tracked inconsistencies between state positivity rates and the underlying data used for such calculations.
Such trends were evident once again in the two-week period concluding on Sunday.
On its online dashboard, the IDPH reported 225 new COVID-19 cases in Dubuque County during that 14-day stretch. In a nearly identical two-week period — beginning at 5 p.m. Aug. 2 and ending at 5 p.m. Aug. 16 — the Telegraph Herald calculated 278 new cases.
The disparity in new cases was reflected in the two-week positivity rates.
A calculation by the Telegraph Herald — based on the number of new cases and new tests during the 14-day period — showed that the county had a 12% positivity rate during the two-week period. The state reported just 10.3%, however.
A similar phenomenon played out in surrounding counties.
Clayton County had an 8.6% positivity rate using the figures, compared to the state-reported 7%; Jackson County had a 10.5% calculated rate, compared to a state-reported 9.4%; Delaware County had a 8.8% calculated rate, compared to the state-reported 7.7%; and Jones County had a 5.1% calculated rate, while state data reported 4.4% positivity.
The TH also has started tracking results at midnight in recent days to fully overlap with the period tracked by the state.
From Thursday to Sunday, the TH, using state figures, recorded 75 new cases added to Dubuque County’s total. But the state website reported that the county had 17 new cases Thursday, 24 on Friday, 19 on Saturday and six on Sunday — a total of 66 cases. It is unclear what dates the other nine cases might have been assigned to.
The Telegraph Herald reached out to state officials last week in hopes of learning the cause behind these discrepancies.
State public health officials last Tuesday provided one explanation for how the positivity rates were calculated, then the governor’s office provided a different explanation two days later. Neither addressed the date inconsistencies found by the newspaper, but Pat Garrett, a spokesman for Gov. Reynolds, urged Iowans to take the state’s figures at face value.
“It doesn’t make sense for school districts and, frankly, newspapers to constantly calculate something and expect to get the exact same result as our website,” Garrett said last week.
As the state navigates an unorthodox return to school, discrepancies in county positivity rates could have a major impact.
As of Monday afternoon, four of Iowa’s 99 counties had 14-day positivity rates exceeding the key 15% threshold. Or at least that was according to the state’s website with the positive-case coding flaw still not fixed.
Another 10 counties, however, had 14-day positivity averages of 12% to 15%. And 12 more, including Dubuque County, were in the 10% to 12% range. It is unclear how much the coding flaw has driven down the rates in any of those counties that are close to the 15% threshold.
The Telegraph Herald’s analysis of two-week positivity in Dubuque County found that state-reported figures were often 3 percentage points lower than the underlying data indicated.
To many epidemiologists, the notion of a 15% threshold for returning to the classroom was problematic to begin with.
Perencevich said a more appropriate target rate would be somewhere between 3% and 5%.
“All of those cutoffs are dangerously high to begin with,” he said. “They are completely unsafe targets.”
Iowa Sen. Pam Jochum, D-Dubuque, said she was not surprised to learn of flaws in the data.
“There’s been some suspicion that (the state’s) positive rates were lower than what was actually happening in the community,” she said. “This substantiates the fact that the numbers have been off.”
She said these inaccuracies are particularly troubling at a time when students and teachers are preparing to return to school. She feels local educators are fighting an uphill battle.
“Schools are doing everything they can to ensure the safety of our children and employees,” she said. “The state needs to do its part to provide accurate information.”
She emphasized that ongoing concerns about the state figures have reinforced her belief that school districts should have more autonomy.
“I believe that decision rests with local school boards and districts,” Jochum said. “They know best what virus activity is in their community and how to deal with it from a public health standpoint.”
Mike Beranek, the president of the Iowa State Education Association, a union representing teachers, said Monday that school districts should put their reopening decisions on hold until the data is fully corrected.