This story has been updated to reflect a change in policy since the story was originally published.
Kansas Governor Laura Kelly announced on Thursday, May 15 that the state would slow its reopening plans, citing COVID-19 trend data that was revised on Monday to include asymptomatic cases.
Kansas Department of Health and Environment spokesperson Kristi Zears explained in an email to KCUR on Monday that asymptomatic cases would be added to the state’s tally of cases based on the date that they were diagnosed.
The addition of the asymptomatic cases appeared to immediately shift the trends shown in the state’s data. While case data from the previous week, which omitted asymptomatic cases, suggested a decline in new cases, data since the policy change shows cases have roughly plateaued after increasing in late April.
Governor Kelly acknowledged the trend on Thursday when she announced the state would take a slower approach to reopening.
“Unfortunately, the rate of disease spread has not shown the downward trajectory necessary to move fully into Phase 2,” Kelly said on Thursday.
The original story continues below.
As leaders in Kansas have plotted the state’s response to the COVID-19 pandemic, they’ve pointed to a set of data-driven guidelines and a website filled with what appears to be encouraging news about the virus.
The state health department’s reopening metrics show that the virus has been declining for weeks, but the numbers of newly diagnosed cases suggest just the opposite: transmission of COVID-19 has been picking up rapidly in Kansas.
A chart on the health department's website on Wednesday, for example, showed a total of 1,294 new cases by "onset date" (the date a patient first reported symptoms) from April 16 to April 29.
However, a total of 2,673 new cases were actually diagnosed during that time.
A major reason for the difference: The state's optimistic figures are leaving out a key data point — asymptomatic cases, in which people have the disease, and can spread it, but don't show any symptoms
TJ Muehleman, founder of the COVID Mapping Project, which analyzes COVID-19 public data sources, says the state's data and its public information could be misleading
“The way the data is presented is, it’s very confusing,” says Muelhleman. “It leads to a lot of people saying, ‘We’re trending down. Everything’s good. I can go to the mall’ or what have you, and the reality is that the data for the last 10 to 14 days is just not reliable enough to make those kinds of decisions.”
Days after phase one of reopening, Kansas Gov. Laura Kelly said at a press conference that she and state officials would take a cautious, evidence-based approach to reopening business and relaxing stay-at-home restrictions.
“We’ll continue to reevaluate Kansas’s disease spread, testing rates, hospitalizations, personal protection equipment availability and other factors before moving from one phase to the next,” Kelly said. “Science and data will dictate how quickly or how slowly the reopening unfolds.”
The plan introduced by Kelly in April, titled “Ad Astra: A Plan To Open Kansas,” calls for three COVID-19-related benchmarks for the state and communities to reach before they move from one phase of reopening to the next.
The benchmarks include 14-day decreases in COVID-19 deaths and new COVID-19 hospital admissions, as well as stable or declining numbers of new COVID-19 cases over a 14-day period.
On the Kansas Department of Health and Environment website, the “Reopen KS Metrics” page features charts that show the state has been achieving all of these benchmarks.
One chart, labeled “disease spread,” shows a strong and steady decline in new cases during the last half of April.
However, testing labs and health departments in Kansas saw a dramatic surge in reported cases in late April that is reflected nowhere on the graphic.
The Ad Astra plan calls for the evaluation of trends in cases based on their onset date.
However, the health department's data does not include cases if certain information is unknown, such as the date that symptoms began, said Kristi Zears, director of communications for the state health department.
“If cases are still being investigated by the local health department, or if an illness onset date was not included in the case investigation form, then these cases are not included in the chart,” Zears wrote in an email to KCUR.
That means asymptomatic cases aren't included. Meanwhile, the numbers of asymptomatic cases have often far exceeded the symptomatic ones.
Zears says that the health department’s COVID-19 data practices were developed with “university partners” and aligned with federal reopening plans.
But experts say that asymptomatic patients or people with only very mild symptoms can transmit the disease to others, and these “silent spreaders” are one of the major challenges of mitigating COVID-19.
One recent study showed asymptomatic patients accounted for 44% of COVID-19 transmissions.
Sarah Bass, a researcher in public health communication at Temple University, says omitting these cases leaves important information out of the picture.
“From a public health perspective, that is problematic, because all of those people who are asymptomatic are still able to spread the disease to other people,” says Bass. “So having a more realistic count of what the caseload is would be important.”
Without the large numbers of asymptomatic cases included, the COVID-19 data presented to the public and used by public officials paints a dramatically different picture of the spread of the virus when compared with the full data on all diagnosed cases.
For example, one chart on the health department’s website on Wednesday, meant to depict the spread of the disease, showed 33 COVID-19 cases on April 29 as measured by onset date. However, data presented on other parts of the website show 452 cases were diagnosed that day, making it one of the biggest case days ever for the state.
The same chart showed a steady decline in case numbers as listed by onset date across the last half of April. But the numbers of new diagnosed cases actually surged during this time.
The charts and graphs on the health department website also show other inconsistencies. For example, two different “disease spread” charts that are both labeled as showing cases by onset date show slightly different case numbers for the same days.
Some of the differences and inconsistencies are due to the often slow nature of public health reporting. Zears says that the data and graphs are revised and updated as the information is processed.
“KDHE staff are continually cleaning this data which includes going back to local health departments [that] consistently leave this information out of the surveillance system to request any paper investigation forms they might have kept so that we can pull out the relevant missing information. It is an ongoing process,” Zears wrote.
In fact, later in the week, the same chart had been updated to show 55 cases, rather than 33, on April 29.
These delays and the sometimes incomplete data they provide are common in health data reporting, says the COVID Mapping Project's Muehleman, but they could be particularly confusing to people who are not used to reading it.
“From an epidemiological perspective, there’s nothing wrong with the way they are doing it,“ Muehleman says. “That is a sort of way of tracking and backwards-revising data. That’s fine. The challenge is that professionals understand that, but laypeople may not.”
The delays in reporting and the omission of many cases make the data in the disease spread chart appear far more optimistic about the spread of COVID-19 than would be suggested if the chart showed all diagnosed case data.
Despite the surge in cases documented in the diagnosed case data, Kelly pressed ahead with reopening.
One day after the April 29 case surge, Kelly announced that the statewide stay-at-home order would be allowed to end on May 4.
Executive Order No. 20-29, which Gov. Kelly signed to authorize phase one of the “Ad Astra: A Plan To Open Kansas,” read:
“I have determined that the evolving public health and economic threats posed by COVID-19 require a proactive approach to both incrementally re-opening businesses and activities while still mitigating the further spread of COVID-19.”
Kelly and public officials say they will watch COVID-19 cases trends as they determine when to enter the next phase of opening, which could happen as soon as May 18.
However, the onset date data that Kelly and other officials are using is increasing diverging from the full diagnostic case data.
In the meantime, Muehleman says that, as businesses are reopening and many people are weighing their own risks, the incomplete and inconsistent COVID-19 data provided by the state could lead the public to make misinformed decisions.
“You’re doing a disservice," Muehleman says, "by presenting data that is confusing and misleading and potentially harmful.”
