The world is enduring a very painful pandemic and everyone wants to get back to normal life as soon as possible. But the tension between the desire to open up the economy and the twin goals of preventing deaths and not overwhelming our health care system requires a careful plan armed with powerful tools.
One of those tools is measuring antibodies to COVID-19 in peoples’ blood. In the course of most infections, people’s immune systems produce antibodies. For some, but not all, viruses the presence of those antibodies indicates partial or full immunity from reinfection.
The hope is that if we could test everyone’s blood for antibodies, it would tell us who can safely return to work and to social activities and who needs to remain protected. But the utility of such tests depends on two assumptions: The first is that the presence of antibodies grants immunity to infection, and the second is that when an antibody test is positive, it truly means the person has antibodies in their blood.
‘False positive’ tests
In a setting where the prevalence of a disease like COVID-19 is small, then even with serologic tests that are very specific (but falling just short of 100%), many people who have a positive test will not actually have the antibodies. These are termed “false positive” tests.
If those with false positive tests (and no antibody protection) reintegrate, they won’t be safe from acquiring COVID-19. For example, let’s say a test 90% accurate (which is deemed to be pretty accurate in many clinical scenarios) is deployed in a setting where COVID-19 prevalence is only 1%, which in American today would be about 3 million people.
In this setting, assuming every person was tested, 30 million Americans would receive positive results, and of those positive results 27 million would be false positives.
While it seems counterintuitive that a positive result from a test deemed accurate leads to so many false positives, it is because even the slightest degree of misspecification is magnified by the fact that almost no one in the population really has antibodies. The only way to negate the low prevalence is to have a specificity of 100%, which is usually difficult to achieve.
Why serologic test is not useful for now
The question is: What is the actual prevalence of recovery from COVID-19 in the population? If we look at the United States as a whole, as of April 18, there were about 700,000 documented cases, most of whom will recover. We know that this is a drastic underestimate of the true number of cases, but even if it is 10 times larger, the total number is 7 million.
There are 328 million people in the USA, so the prevalence would be 2.1%. With a rate of false positives at 10% or greater, these serologic tests are not as helpful as we would like them to be.
Of course, the prevalence is not uniform across America and Canada. In New York City, the prevalence might be as high as 15-20%, in which case serology will be more useful. But in Montana, the prevalence at this time is likely well below 1%, making the tests not as useful there.
More on COVID-19:Could an old vaccine be a godsend for new coronavirus?
What do we do?
First, to be useful, COVID-19 antibody tests will have to demonstrate that they are extremely specific — in the range of 99% or higher. Some serology tests are performed in laboratories, while others are being marketed for point-of-care use at home or in doctors’ offices. It is likely that the former will have greater specificity and more meaningful results compared with the latter.
Second, for most of the USA at the present time, COVID-19 serology will not be useful for individual counseling. This situation will change as the prevalence of infection grows, but that will take many months.
Rushing point-of-care tests to market is a mistake until we can document and validate that they are very close to 100% specific. And when they are used, the results must be interpreted in the context of the prevalence of prior COVID-19 infection in those communities and settings.
Dr. Allan Detsky is a professor of Health Policy, Managementand Evaluation and Medicine at the University of Toronto. He was physician-in-chief at Toronto's Mount Sinai Hospital during the SARS outbreak of 2003. Dr. Isaac Bogoch, an infectious disease physician and scientist at Toronto General Hospital, is associate professor of Medicine at the University of Toronto.