Play Live Radio
Next Up:
0:00
0:00
0:00 0:00
Available On Air Stations

The Fuzzy Math Of Georgia's Percentage Of Positive Tests

Coronavirus testing  by a private lab at a Macon church in May included antibody and viral testing.
Grant Blankenship
/
GPB
Coronavirus testing by a private lab at a Macon church in May included antibody and viral testing.

Last week, in defending his decision to allow Georgia’s economy to reopen, Gov. Brian Kemp cited some milestones: Reductions in hospitalizations, ventilator use… and one other.

“And the percentage of positive cases continues to drop,” Kemp said from his now familiar press conference spot at the foot of the stairs in the Capitol rotunda.

But changes in how different types of coronavirus tests are recorded in the state have given the percentage of positive cases an unearned cushion of up to 3%.The Fuzzy Math Of Georgia's Percentage Of Positive Tests

The percentage of coronavirus tests that return as positive is a popular number because calculating it is seemingly easy. Plus, the state appears to be on the way to hitting World Health Organization targets for it. The thinking goes, the lower the percentage of positives, the better job a testing regimen is doing in finding sick people.

RELATED: Georgia’s Gaffe-Prone COVID-19 Dashboard Is Useful – If You Know Where To Look

To get the percentage of positive cases, start by dividing the number of positive tests by the number of total tests performed, and then multiplying by 100. On the night of May 20, that was (39,801 positive tests/402,940 total tests) x 100 = 9.8% positive tests.

“On its face, it's pretty simple, right? You don't need to be an epidemiologist or mathematician to work out what know what the percent what drives the percent positive,” said Ben Lopman, infectious disease epidemiologist at Emory University.

“That said, you do need to know what's generating the data that go into the into the top and the bottom…the numerator and the denominator of that equation.”

That’s where the trouble is.

Early in Georgia’s pandemic response, “test” only meant a test to see if you were actively carrying the virus. But now, there are antibody tests, too. Those are tests looking for signs your body has fought off an infection in the past but which say nothing about whether you carry the virus now.

Reporting by the Columbus Ledger-Enquirer and GPB shows when an antibody test is completed, the Georgia Department of Public Health records it on its data page as a “completed test” right along side completed viral tests. This began back in April.

The problem is that when antibody tests are returned as positive, they are not recorded with positive viral tests.

“If a patient has only serology/antibody results, they are not considered confirmed cases but rather a 'probable' case according to CDC guidance,” said Georgia DPH spokesperson Nancy Nydam via email.  “DPH epidemiologists keep track of these serologic 'positives' as probables but do not include them in our confirmed totals.”

Because antibody tests are counted as completed tests but are never recorded as "confirmed positives",  they count, in terms of calculating the percentage of positive tests, as negative test results.  That throws off the math, potentially painting an unduly rosy picture of both testing and infection rates.

Following on the Columbus Ledger-Enquirer's reporting, the AJC found out exactly how much the picture was off. The AJC found tallying antibody testing added 57,000 tests to the state's total.

That gave the percent positive number a 3% boost by May 20.

Epidemiologist Ben Lopman said he can’t guess whether fuzzy math is the state’s goal, but he can say he does not see much reason to combine numbers from two tests that serve completely different purposes for researchers.  

“For surveillance purposes, these are apples and oranges. And it does not make sense to put the two numbers together,” Lopman said.

For instance, positive viral tests lead to quarantine of the positive person and tracing of their contacts, important steps in containing an infectious disease.

Ben Lopman said antibody tests serve a different purpose.

“Typically in epidemiology what we'd be trying to do with an antibody test is to get a representative sample in the population, to do a so-called seroprevalence study, to work out who has ever been infected,” Lopman said.

Those studies can reveal important information about what communities have been infected by viral infections that otherwise went undetected. That can in turn inform future viral testing, followed by isolation and containment efforts.

As for the idea that the coronavirus data made public by the Georgia DPH depicts a downward trend in the disease, Lopman said he just doesn’t see it.

 

Copyright 2020 Georgia Public Broadcasting

Grant came to public media after a career spent in newspaper photojournalism. As an all platform journalist he seeks to wed the values of public radio storytelling and the best of photojournalism online.