CDC’s coronavirus report includes data that could artificially inflate testing

The Centers for Disease Control and Prevention’s latest coronavirus dashboard combines data on two different kinds of tests that experts say could be inflating key statistics that epidemiologists use to monitor the outbreak.

The agency’s recently released Covid-19 Data Tracker‘s dashboard on testing combines results for both diagnostic tests, which identify current infections, and serological tests, which detect whether someone has previously been infected. Epidemiologists caution that the tests are not designed for the same purpose and mixing the data could mask reality.

The CDC says it aims to separate data on the two types of tests, soon.

“[The dashboard] includes diagnostic and serology results, which includes repeat testing for individuals, as well as serology testing counts and additional serology test information from commercial laboratories,” CDC spokeswoman Kristen Nordlund said, in a statement to CNBC. “We hope to have the testing data broken down between PCR [or diagnostic testing] and serology testing in the coming weeks as well.”

The CDC did not respond to CNBC’s query about why the data was published combined. The Atlantic first reported on the CDC’s data.

Because serological tests are meant to be used for both asymptomatic and symptomatic people while diagnostic tests, or PCR tests, are currently being used often to confirm suspected infections, people are less likely to test positive with a serological test than with a PCR test, said Robert Bednarczyk, assistant professor of global health and epidemiology at Emory University.

“If we’re testing more people and not finding as many cases, that would be a sign that things are getting better,” he told CNBC. “But if that’s happening because we’re artificially inflating the test numbers, then we are not seeing the true picture.”

The diagnostic tests, most of which require a sample collected with a swab, are crucial for contact tracing and to stop the spread of the virus, Bednarczyk said. 

Serological tests, which use a blood sample, are used by epidemiologists to estimate how much the population has already been exposed to the virus, Bednarczyk said. He added that it’s useful from a research perspective, but shouldn’t be used to determine the state of the outbreak. 

“Without a clear apples-to-apples comparison of the number of tests per capita, it is hard to compare disease rates across states,” he said. “The test types should be separated.”

In lieu of federal data-collection guidelines, several state departments of public health have also reported their testing data in that combined fashion. 

Georgia, which made one of the earliest and most ambitious decisions to reopen large parts of its economy, currently combines data on the two tests, The Macon Telegraph first reported. The Georgia Department of Public Health confirmed to CNBC on Thursday that it is working to update its website to separate the data. A spokeswoman added that 57,000, or about 14%, of the state’s roughly 407,000 tests run were serological tests.

Virginia and Vermont confirmed to CNBC that they previously combined data on the two tests but have since updated their websites to separate the two data sets. 

Delineating the data made a “minimal” impact, said Vermont Department of Public Health spokesman Ben Truman, but he said the state made the move to avoid inflating the test-positivity rate, a key data point that indicates the number of people tested who turn up positive for the virus. Some states have tied their reopening plan to specific test-positivity milestones.

“To be a case you have to be lab-confirmed through a molecular test,” reads an explanation of the decision to separate the data from Vermont’s epidemiology data team, provided by Truman. “To allow for [percent-] positive calculation, we have to only include those tests that could potentially make you a case, so only molecular tests. If we include serology, we inflate the denominator.”