With the ability to order COVID antibody testing more broadly I think it is important to understand how to interpret this result if you do it especially if you don’t have a medical professional to discuss the result with. One of the weird things about interpreting some medical tests like this one is that the accuracy of the result is dependent on the likelihood you have the disease. For example, the Quest antibody test reports a specificity of 99%. (This has not been independently verified.) This is the probability the test will be negative when the disease is present. The higher the number the lower the number of false positive. For example, if we give 200 people the test and half of them are known to have the disease then we will get 100 positive results 99 from the group with the disease and 1 from the group that does not (false positive). This seems pretty good right?
The problem is this only really applies when the disease is present in about half of the population. As the prevalence in the community or likelihood someone has the disease goes down false positives become more of a problem because they make up a larger portion of the positive tests. (Currently, the best data we have suggests around 5% of Americans may have been infected though this data is very limited and is dependent on where you live. This would be considered the baseline risk you have the disease unless there is some other modifying factor.) To account for this, we use a test called the positive predictive value. This is the probability that the disease is present when the test is positive. For COVID antibody testing if we plug is a specificity of 99% and a prevalence of 5% we get a positive predictive value of 83.9%. This means that 1 in 6 positive antibody tests will be a false positive. So this increases the chance you had COVID from 5% to 84% but it by no means makes it certain you had COVID. We also don’t know for certain these antibodies are protective (they are in the short term, 1-2 years, for other coronaviruses) but this is a separate topic.
This does not apply if you were symptomatic. For example, if you lost your sense of smell in the last couple of months there was a ~68% chance you had COVID based on the best data we have. Using this as the likelihood you had the disease increases the positive predictive value of the test to 99.5% making it highly likely you had COVID. The percentage of people who have any one symptom is different, so this only applies for loss of smell.
I hope this helps some of you either deciding to order the test or figuring out what to do with the information if you do.
References
https://www.webmd.com/lung/news/20200424/more-data-bolsters-higher-covid-prevalence
https://www.livescience.com/smell-loss-confirmed-covid-19-symptom.html
Ryan Thomas, MD
Assistant Professor of Pediatrics
Director, MSU Cystic Fibrosis Center
Department of Pediatrics and Human Development
Michigan State University College of Human Medicine