Missouri officially passed the grim milestone of 10,000 COVID-19 deaths Thursday when the Department of Health and Senior Services added 20 new fatalities as part of its daily update of coronavirus data.
But that official report is 1,180 too low, according to an analysis of city and county health department data by The Independent.
The 10,000 statewide total was likely surpassed sometime in May or early June, with the current total showing 11,182 deaths.
That difference of about 1,100 between state and local fatality data has been evident since March. When The Independent checked local health pages at that time, the difference between the state and local death report was 1,116 fatalities. A review in late July found 1,081 more deaths on local pages than on the state report.
The most important thing that determines whether a death appears on the state report is the type of test used to identify the COVID-19 infection. Only tests that detect the genetic material of the virus, such as the PCR test using a long swab developed early in 2020, or the more recent home test called a LAMP test, is recognized.
Deaths from cases identified with an antigen test, which detect virus proteins and can give results in minutes, are not reported by the state.
In the health department’s terminology, a PCR-identified infection is a “confirmed case,” while an antigen-identified infection is a “probable case.”
The state makes daily reports of both types of test. Antigen-identified cases make up about 17% of the total 713,113 cases.
But the state does not report deaths from antigen test-identified cases.
“For us, the answer is simple: we are reporting confirmed deaths because we believe this maintains the highest level of trust in the data,” Sami Jo Freeman, spokeswoman for the department, wrote in an email. “It’s consistent and straightforward, which is what Missourians expect.”
But in Linn County, the inconsistency between the local report and the state data led the local health department to stop reporting its count of COVID-19 deaths. As of Thursday, Administrator Krista Neblock said, her reports showed 43 deaths. The state dashboard reported 21, fewer than her department was reporting in March.
“I don’t understand the state’s stand,” Neblock said. “If it is good enough to count it as a case, why isn’t it good enough to count it as a death? I struggle with the inconsistency in messaging from the state. If the test is not good enough for them to report it, why are they using it or allowing people to use it to test, then?”
The department stopped including deaths on its regular updates via Facebook when it became a source of friction with the public, Neblock said.
“People were accusing us of lying about the deaths and looking at the state dashboard and asking about the difference,” Neblock said.
For a COVID-19 death to be reported to the public, Freeman wrote in an email, several things must happen.
One of the most important is that information about the patient – when they were originally tested and the result of that test – should be entered into the case tracking system called EpiTrax. The local health department where the patient resides is responsible for updating EpiTrax, including properly recording a fatality, Freeman wrote.
If everything in the tracking system is correct and a PCR or LAMP test was used, a reported fatality will be included in the regular daily updates to the department dashboard, Freeman wrote.
But 75% of the 4,483 deaths added since Jan. 1 have been reported on Tuesdays, following a review of death certificates received in the previous week. Death certificates can be received weeks or even months after a death occurs.
In June, the department’s weekly update still included deaths from 2020.
In order to be included in that update, the death certificate must report that COVID-19 was the primary cause or a major contributing factor. And as the Kansas City Star reported, elected coroners in rural counties are, in some cases, reluctant to list the coronavirus disease as the cause of death.
Macon County Coroner Brian Hayes told the newspaper that in cases where a family requests he omit, he accommodates the request if he can justify assigning the death to another cause, such as pneumonia.
“I won’t lie for them, it’s gotta be true, but I do what pleases the family,” Hayes said.
A death certificate that does report COVID-19 as a cause, but with no corresponding information in EpiTrax, is sent to the National Center for Health Statistics, Freeman wrote. If the center returns the correct death code for COVID-19 and the right test was done for the initial diagnosis, it goes on the state dashboard as a confirmed death.
“Deaths due to COVID-19 are supposed to be reported to DHSS as soon as they are known,” Freeman wrote. “This is often the result of investigations carried out by local health departments.”
Many local health departments, which operate with a handful of staff in many counties, are overwhelmed by case investigations, contact tracing and vaccine delivery, Freeman noted. As a result, “local health departments may not always discover an individual died of COVID-19 in a timely manner, or they may have received notification a person died, but they are unsure whether COVID-19 was considered as a cause of death.”
That makes the process of linking death certificates to identified cases important to maintaining an accurate count, she wrote.
The Independent analysis of local reports shows that the biggest differences with state data are in the St. Louis metro region and rural Missouri. In both, the state report is approximately 15% lower than locally reported deaths.
Of the 117 city and county health departments, 74 report more deaths than the Department of Health and Senior Services. There are 31 that are more than one-third higher than the state count, all in rural areas, with only four having a population of more than 20,000 people.
Many of those same departments are heavily reliant on the cheap, quick antigen test. It can be used at the health department instead of taking samples that must be shipped to labs in larger cities.
Of the 27 local health departments that have identified more than 25% of their cases using antigen testing, only two have populations above 30,000.
In Dallas County, where 45% of the 1,975 cases have been identified with antigen testing, Health Administrator Cheryl Eversole said she has been unable to understand why all of the 44 deaths from her county are not being reported. The state data shows 37 deaths in Dallas County.
“I asked them why that wasn’t accurate and the explanation they gave was about as clear as mud,” Eversole said.
In an email, Eversole said the difference between the state report and her data is the number of antigen-tested cases that have resulted in death.
“If that makes any sense???” she wrote.
The Delta variant surge that moved Missouri’s combined average of PCR and antigen cases from 396 per day on June 3 to 2,757 per day on Thursday has had a delayed impact on reported deaths.
Because so many deaths are identified based on death certificate reviews, the rate of deaths reported by the department does not indicate current deaths. However, Tuesday’s review reported more than 100 deaths — 109 in all, 105 from July, one from June and three from earlier this month — for the first time since April 13.
The weekly average rate of reported deaths, which peaked at 58 per day at Christmas, reached a low of 1.6 per day in the first week of April. It was 23 per day last week.
The state health department has recorded 440 COVID-19 deaths so far for July, the most since February. There have been 80 recorded for August.
Linn County was one of the first places in the state hit by the Delta variant. Along with adjacent Livingston County, it suffered in late May and early June what southwest Missouri experienced in June and July and the rest of the state is seeing now.
Along with 522 cases since May 1, the county of just under 12,000 people has experienced 20 deaths. Most have been reported since the start of July, Neblock said.
Patients hospitalized with severe COVID-19 spend weeks in the hospital before recovering enough to return home or dying. That is another reason deaths rise weeks after cases.
“We have a lot of people who were antigen positive who never got a PCR and when they expire that is a COVID death for our numbers,” Neblock said.
But the death may not ever be reported to the public by the state, she said.
“People can do a home test at home and we count that as a case,” Neblock said. “If they have a confirmed test and go to the hospital and end up expiring of COVID-type systems, that test is not good enough.”
One way to maintain accuracy, Freeman wrote, was for every person who has a positive antigen test to seek a confirming PCR test.
“If that process happens correctly, then there is no issue and it helps us maintain that high level of data confidence we are shooting for,” she wrote.
An accurate count of deaths from COVID-19 is important for understanding the disease, the Centers for Disease wrote in guidance issued early in the pandemic.
“In public health emergencies, mortality surveillance provides crucial information about population-level disease progression, as well as guides the development of public health interventions and assessment of their impact,” the guidance stated.
The health department is “not deviating from that guidance in any way,” Freeman wrote.
“Per CDC’s definition, COVID-19 can only be confirmed by a positive PCR test result,” she wrote. “Using that logic, a death that occurred after an individual tested positive on an antigen test is only enough evidence to consider the death to be a probable case of COVID-19.”
The data from the Johns Hopkins University COVID-19 dashboard has the world death rate for identified cases as 2.11%. The death rate for the United States, from CDC data, is 1.71%.
The Missouri death rate for PCR-identified cases is 1.68%. The death rate based on the local health departments totals compiled by The Independent for both cases and deaths is 1.55%.
The Columbia-Boone County Public Health and Human Services Department report and the state report were last in close agreement on May 18, when the local report showed 91 deaths and the state report showed 93.
The next day, Columbia reported a data reconciliation that added 25 deaths, occurring from November to April, from both PCR- and antigen-tested cases. As of Thursday, the Columbia data, showing 128 deaths, was still 21 more than the state report.
The difference is the antigen cases, said Scott Clardy, assistant director of health.
The information the department reported was actually found in the state’s records, he said. Cases were low in May, he said, and staff had time to catch up on paperwork.
“I think,” Clardy said, “it gives the community a better idea of what is actually occurring if we count those antigen deaths.”
Missouri Independent is part of States Newsroom, a network of news outlets supported by grants and a coalition of donors as a 501c(3) public charity. Missouri Independent maintains editorial independence. Contact Editor Jason Hancock for questions: info@missouriindependent.com.