The U.S. Centers for Disease Control and Prevention (CDC) acknowledged last week that it had lumped genetic tests that identify people with current COVID-19 infections together with serology (blood) antibody tests that measure whether someone has had the viral infection in the past.1 Combining the tally of the two tests overstates the country’s ability to test people who are currently sick with COVID-19 and provides an inaccurate picture of the state of the pandemic.2
Ashish Jha, director of Harvard’s Global Health Institute said such errors render the CDC numbers about how many Americans are currently infected with COVID-19 as “uninterpretable,” creating a misleading picture for people trying to make decisions based on the data.
William Hanage, professor of epidemiology at the Harvard T.H. Chan School of Public Health explained:3
It is incumbent on health departments and the CDC to make sure they’re presenting information that’s accurate. And if they can’t get it, then don’t show the data at all. Faulty data is much, much worse than no date. Combining a test that is designed to detect current infection with a test that detects infection at some point in the past is just really confusing and muddies the water.4
According to most public health professionals following the pandemic, viral test results that detect current or past COVID-19 infections should be reported separately. Making this distinction helps health officials understand how many people currently are actively infected, the percentage of people testing positive, and how those numbers change over time, which affects the setting of public health policy.5
The CDC obtains COVID-19 testing data from several sources, including state public health labs, commercial testing companies and hospitals.6 Several states are also blending data from the two tests in the same way, including Pennsylvania, Texas, Georgia and Vermont. Virginia and Maine also mixed their test results but changed their practice after several publications reported the inadvisability of mixing of the viral infection and antibody tests.7
Combining statistics for antibody testing with statistics for diagnostic testing of current infections could have the effect of reducing the number of tests that appear to be producing positive results, lowering the overall “positivity rate.” The World Health Organization (WHO) has recommended an infection positivity rate of 10 percent or less as a signal of whether enough testing is taking place in a population. “I suspect [combining the two tests] will artificially lower the percent positive,” says Caitlin Rivers, an epidemiologist at the Johns Hopkins Bloomberg School of Public Health.8 The White House’s guidelines say that for a state to reopen it must observe a downward trajectory in its number of daily cases or its percentage of positive cases.9
Adequate testing for current infections has been cited repeatedly as a key to safely reopening the economy, so the lack of clarity around how tests are counted raises questions about the true percentage of people with active infections. The American Medical Association said recently that antibody tests should only be used for research and deciding which patients can donate plasma, and not as a determination of natural herd immunity or a reason to lessen social distancing.10
COVID-19 testing capacity to determine the numbers of current infections in the U.S. has more than doubled over the past month, increasing from about 147,000 tests a day in mid-April to more than 413,000 tests a day as of May 20. Simultaneously, the proportion of positive tests recorded each week fell from 10 percent to six percent of total tests conducted. Both trends appear encouraging at face value, but given that results of tests for current infections and antibody tests for past infections are sometimes being lumped together, it’s impossible to know whether the data accurately reflects the reality of how many people in America are currently infected with COVID-19 or have recovered from past infections and have antibodies that give them protection.11
1 Stein R. Scientists Warn CDC Testing Data Could Create Misleading Picture Of Pandemic. NPR May 21, 2020.
2 Madrigal AC, Meyer, R. How Could the CDC Make That Mistake? The Atlantic May 21, 2020.
3 Smith MR. CDC, states’ reporting of virus test data causes confusion. Associated Press May 24, 2020.
4 Lanese N. The CDC is lumping positive COVID-19 viral and antibody tests together: Here’s why that’s bad. Fox May 26, 2020.
5 See Footnote 3.
6 See Footnote 1.
7 See Footnote 2.
8 See Footnote 1.
9 Johnson M. CDC acknowledges mixing up coronavirus testing data. MSN May 22, 2020.
10 Lee J. CDC faces more coronavirus testing questions, this time about how many diagnostic tests are being conducted. Barron’s on Market Watch May 23, 2020.
11 See Footnote 2.
12 See Footnote 4.