Earlier this week, we learned that for several weeks, the Commonwealth of Virginia has been combining viral test data with antibody test data in their public reporting on COVID-19. Virginia officials suggested that other states were pooling these two types of results so that, in essence, Virginia was forced to combine them to keep up.
On May 13, 2020, we contacted 44 states and territories without clear statements about their antibody test reporting policy. As of May 15, we have confirmed that at least two states besides Virginia have been combining antibody and viral test results, though only Virginia's health department is on the record about combining their testing datasets. If more states are, in fact, combining viral and antibody results, we understand that they are making a choice that appears to increase their test totals. But we think it comes at much too high a price.
As antibody tests become more widely available across the United States, every state and territory will have to choose how to report this data to the public. We encourage all states and territories to standardize on reporting antibody and viral test results separately and fully, allowing government departments, journalists, and members of the public to maintain access to the most meaningful numbers throughout the COVID-19 outbreak in the United States.
To understand why it's critical to maintain separate counts for viral and antibody testing data, it's useful to understand how different the two kinds of tests really are.
How do we test for COVID-19?
There are two categories of tests currently used to identify COVID-19 cases in the United States: viral tests, which include PCR tests and the newly approved antigen tests, and antibody (or "serology") tests. Roughly speaking, viral tests identify evidence of current infection by looking at the virus directly, while antibody tests identify evidence of past infection by looking at whether our body has already responded to the virus.
| Test type | What it does | How it works | Who gets tested |
|---|---|---|---|
| Test typeViral tests - PCR - Antigen | What it doesIdentifies active infections | How it worksFluid samples from the respiratory system (like nasal swabs) are tested for the presence of RNA or antigens (molecules that provoke an immune response) specific to the SARS-CoV-2 virus. | Who gets testedTest availability remains highly inconsistent across the US. Viral testing capacity is increasing, but in many parts of the country, most people receiving RT-PCR tests are symptomatic. Current CDC guidance on viral testing gives highest priority to people who are symptomatic and sick enough to require hospitalization or who are symptomatic and in high-risk positions or situations, like healthcare workers or people in nursing homes and prisons. |
| Test typeAntibody (or “serology”) tests | What it doesIdentifies past infections | How it worksDetects an immune response in the patient. Measures antibodies in the blood that reach detectable levels between one and three weeks after the onset of a COVID-19 infection, according to the CDC’s serology guidelines. | Who gets testedAntibody studies are currently being conducted on many populations in the US, but by design, are never limited to symptomatic patients. The CDC explicitly recommends against using antibody testing to diagnose acute COVID-19 infection. |
Antibody and viral tests tell us different things
Viral tests (PCR and antigen) can only tell us about the present: who is infected right now. Antibody tests can only tell us about the past: who was exposed to COVID-19 weeks or months ago. A negative antibody test indicates that you probably didn't have COVID-19 more than three weeks ago, but says nothing about your present status. This means antibody tests can't boost our capacity for discovering current COVID-19 cases—which is something we must have to understand and respond to local outbreaks.
Different populations, incompatible positivity rates
Due to the difference in what is measured, antibody tests and viral tests are conducted on different populations. Viral tests are used in a clinical context, where doctors need to diagnose people with COVID-19 infection as they show up to the hospital. In many parts of the United States, this means access to viral testing is limited to people who are symptomatic and seriously ill, or who are part of a population known to be at special risk, like people in nursing homes or prisons or meat-packing plants. If you mostly test sick people, you'll get a lot of positives, so until access to viral testing is greatly expanded, we can expect to see relatively high positivity rates.
Antibody tests, on the other hand, are used as surveys (if you've been reading a lot of COVID-19 research, you may have heard the term "seroprevalence surveys") intended to discover the prevalence of past infections (termed "cumulative incidence" by epidemiologists) in a given population, like Major League Baseball employees or members of the general public. They are intended to measure overall exposure levels—and potentially immunity—and so by design, they are never restricted to symptomatic or seriously ill patients. As a result, we would expect to see much lower positivity rates from antibody tests.
Combining unlike tests with unlike positivity rates distorts critical data
State governments rely on testing data to help understand when they can safely reopen specific regions and sectors of the economy. Despite both being labeled as "tests," antibody and viral tests answer fundamentally different questions.
Dumping antibody data into the viral testing pool brings the overall positivity rate down, which looks like either:
an improvement in testing capacity, i.e., a state has increased testing while the number of people with COVID-19 infections is holding steady, or
a reduction in infection rates, i.e., a state maintains steady testing levels but the number of people with COVID-19 is going down.
But in fact, it's a deceptive misuse of the data—the true positivity rate for tests that measure active infections hasn't changed. This is crucial as we need increased capacity for viral testing before reopening to identify active infections even in the pre-symptomatic or asymptomatic stages.
The same thing is also true of the newer, still modest pool of antibody testing data. Combining antibody data with viral testing data erodes the true value of antibody data, which is to illuminate underlying exposure rates of a larger population. This data helps us assess the overall effectiveness of any developed vaccine, or to identify the proportion of people who have built some degree of immunity to COVID-19.
Antibody and viral testing have distinct and vital roles in responding to the COVID-19 outbreak. We need solid, clean data for both.
Quang P. Nguyen is a PhD candidate in the Department of Epidemiology at Dartmouth College.
Erin Kissane is a co-founder of the COVID Tracking Project, and the project’s managing editor.
More “Testing Data” posts
How Probable Cases Changed Through the COVID-19 Pandemic
When analyzing COVID-19 data, confirmed case counts are obvious to study. But don’t overlook probable cases—and the varying, evolving ways that states have defined them.
20,000 Hours of Data Entry: Why We Didn’t Automate Our Data Collection
Looking back on a year of collecting COVID-19 data, here’s a summary of the tools we automated to make our data entry smoother and why we ultimately relied on manual data collection.
A Wrap-Up: The Five Major Metrics of COVID-19 Data
As The COVID Tracking Project comes to a close, here’s a summary of how states reported data on the five major COVID-19 metrics we tracked—tests, cases, deaths, hospitalizations, and recoveries—and how reporting complexities shaped the data.
