COVID-19 Data Accuracy Diluted by Different Ways States Count Cases, Deaths

By Petr Svab
Petr Svab
Petr Svab
Petr Svab is a reporter covering New York. Previously, he covered national topics including politics, economy, education, and law enforcement.
July 14, 2020Updated: January 24, 2022

News Analysis

The number of COVID-19 cases is increasing in the United States, harkening back to the weeks of hellish conditions in places with the most severe outbreaks, such as in New York and New Jersey. But this time, fewer people are succumbing to the disease and the case numbers are highest in Florida, Texas, Arizona, and California.

The difference indicates that case numbers are telling far from the full story. Given the myriad developments over the past several months, today’s data is hardly comparable to the data from the early months of the pandemic. The data isn’t only inconsistent over time, but also from state to state, undermining its validity on the national level.

There are at least six factors that affect the case or fatality counts:

Suspected Cases

The nonprofit Council of State and Territorial Epidemiologists (CSTE) instructed health authorities in April (pdf) to report not only “confirmed” cases based on COVID test results, but also “probable cases,” in which it’s sufficient at least to have some symptoms (such as a headache and a sore throat) and belonging to “a risk cohort” or having some risk of prior contact with other COVID-19-positive people, such as “residence in an area with sustained, ongoing community transmission.”

Some states make clear on their dedicated web pages how many of the local cases and deaths have been confirmed by laboratory testing and how many have been ascertained by other methods. Some don’t.

The inconsistent approaches may be partly caused by the changing guidance from the Centers for Disease Control and Prevention (CDC).

“For a time, CDC asked the states to count the probable cases among the overall counts for cases, hospitalizations, and deaths. More recently, CDC has asked states to separate that out again,” said Av Harris, spokesman for the Connecticut Department of Public Health (CDPH), in an email to The Epoch Times.

The Epoch Times contacted health departments in all 50 states and the District of Columbia. Among the 33 that responded, all except Georgia, Maine, North Carolina, Oklahoma, Nevada, New Hampshire, Rhode Island, and the District of Columbia confirmed that they currently publicly report “suspected” or “probable” cases of either COVID-19 fatalities, infections, or both. All except Arkansas and Hawaii also indicate so on their COVID web pages, though sometimes one has to dig around for the information.

None of the four most populous states (California, Texas, Florida, and New York) responded to The Epoch Times’ inquiries.

“Probable deaths from COVID-19 are tracked by epidemiologists but not reported publicly,” said Nancy Nydam, staffer with the Georgia Department of Public Health, in an email to The Epoch Times.

Some states, including Louisiana, Indiana, Minnesota, and Wisconsin, report probable COVID deaths, but lists them separately and don’t include them in their publicly reported COVID death totals. In the national totals, however, the CDC does include the probable deaths in these states’ figures.

Probable COVID deaths are generally defined as people who died while suffering from COVID-like symptoms, but weren’t tested for the virus. While COVID seems to be currently the dominant coronavirus disease, it shares symptoms with other viral diseases that may have caused at least some of the suspected cases.

“If someone who meets the CSTE case definition dies, then, in general, they are counted as a COVID-19 death,” explained Danyelle McNeill, spokeswoman for the Arkansas Department of Health.

“However, COVID-19 must be a contributing factor. If someone with COVID-19 dies in a car wreck, for example, they would not be counted as a COVID-19 death.”

But different states have adopted this approach at different times and it isn’t clear how far back in time each has gone to add the suspected cases and fatalities to their data.

New York, for instance, has added more than 3,000 suspected COVID deaths to its data. New Jersey added about 2,000.

The judgment calls on suspected COVID deaths can cut both ways—officials may mistakenly attribute COVID deaths to other causes. The issue can be resolved by testing deceased people for COVID, but it isn’t clear to what degree states have done so. The CDC issued guidance on June 15 that all people suspected of dying of COVID should be tested.

Moreover, the data is a work in progress, and sometimes cases get misclassified. Recently, North Carolina removed 10 deaths from the COVID count after determining that they were caused by other factors.

Past Antibody Positive

At least some states are including in their case totals people who’ve tested positive for coronavirus antibodies. People develop the antibodies within a few weeks after getting infected and retain them for months. Counting them in current totals skews the picture because they may have gone through the disease weeks or months earlier. In that case, they are probably no longer infectious and have little to do with the current state of the pandemic.

Arizona has seen a massive spike in cases. More than 22 percent of its tests were antibody tests, its COVID data website indicates. Florida also counts antibody tests, which comprised about 11 percent of its testing total, the state’s COVID website indicates. Texas counts antibody tests, too, though they make up less than 8 percent of its testing total, its COVID data website indicates.

In addition, Florida has had an issue with some smaller, private labs not reporting negative test results to state authorities. That has led to inflated positive test rates for those labs, sometimes as high as 100 percent, Fox 35 reported.

Deaths ‘With’ COVID

At least some states count fatalities of persons who’ve tested positive toward COVID deaths, even though the people may have died of other causes.

Sometimes, the primary cause of death is unclear because the patient suffered from multiple conditions. In others, the reason clearly wasn’t the virus.

Washington state has counted all deaths of people who tested positive for COVID. Since last month, it has vowed “to work on changes to provide more context to death reporting and report death counts that reflect deaths where COVID-19 caused or contributed to the death,” says its COVID web page. As such, it has removed 65 deaths. Of those, 56 were “natural deaths unrelated to COVID-19” and the rest were suicides, homicides, and accidents, according to the Washington State Department of Health July 14 report (pdf).

In Colorado, a local coroner objected to the state’s health authorities classifying a man’s death as COVID-19 after the coroner determined the man had twice the lethal dose of ethanol in his body and clearly died of alcohol poisoning.

A CBS4 investigation found at least three other cases in the state where deaths were counted under COVID-19, even though attending physicians determined other causes.

On its website, the Colorado health department reports both “Deaths Among Cases” as well as “Deaths Due to COVID-19.” But in the national data, CDC only uses the higher, “Deaths Among Cases” figure.

There is a similar situation in Connecticut. Despite being one of the smaller states, it has reported the 8th highest COVID death toll—more than 4,300. About 20 percent of those are marked “probable.” But even some of the “confirmed” cases may not have been caused by COVID, CDPH’s Harris indicated.

“COVID-associated deaths are those where the individual tested positive close to the time of death. This is not an indication of cause of death,” he said.

North Dakota details on its website which deaths were “due to COVID-19” and which “list something other than COVID-19 as the official cause of death.” It specifically says that its total represents “individuals who tested positive and died from any cause while infected with COVID-19.”

Michigan’s COVID web page says that “confirmed” COVID deaths include people “who die within 30 days of infection and their manner of death is listed as natural.” That suggests some room for including unrelated deaths, but, even in these cases, there needs to be “COVID-19 disease or SARS-CoV-2 as a cause of death or a significant condition contributing to death included,” said Lynn Sutfin, spokeswoman for the Michigan Department of Health and Human Services.


One major factor that determines the number of detected cases is the number of tests. At the onset of the pandemic, test kits were in short supply and generally only people with COVID-like symptoms were tested, usually upon admission to a hospital. Gradually, the testing was expanded to all people suspected of coming into contact with somebody infected. Finally, some states, such as California, New Jersey, Kentucky, and Tennessee, now allow anybody to get tested for free.

Experts generally believe that there have been many more infected people than those pinpointed by tests—possibly many times more. But since most people go through COVID with mild or no symptoms, it may not occur to them to get tested.

False Positive

Depending on the type of test used, there is at least a small percentage of false positive and false negative test results. Medical personnel is instructed to look out for false negatives, such as by double-testing and being suspicious of negative results in people with symptoms. False positives, however, don’t seem to be considered an issue, at least not from a better-safe-than-sorry health care perspective.

For example, the rapid tests developed by Abbott and popular in the United States have a false positive rate of about 0–3 percent, according to data from one laboratory and two hospitals submitted to the Foundation for Innovative New Diagnostics, a nonprofit.

Trace Positive

The tests may detect in one’s body an amount of virus that is too low to cause symptoms or make one substantially infectious. This issue, however, is difficult to quantify as there seems to be no solid data on how much of the virus is needed to make one infectious.

The CDC didn’t respond to a request by The Epoch Times for comment.

Update: The article has been updated with further information, including from the July 14 Death Category Report from the Washington State Department of Health and with further responses from states’ authorities.

Related Topics