As we learn about the new coronavirus and how long it’s been spreading in Oregon, more and more people are asking, I was sick this spring. Did I have COVID-19?
There’s one way to figure that out: antibody testing.
Antibody testing for the new coronavirus, also called serology testing, is still in the early stages, but public health officials say ramping it up could be a crucial part of the national and global response to coronavirus. But that doesn’t mean we’ll be testing everyone anytime soon.
Serology testing isn’t very good at telling you if you’re currently sick with the coronavirus. But depending on the type of antibodies it’s looking for, it can help indicate how many people have ever had the virus.
Let's talk antibodies
Antibodies are a vital part of your immune response. When we encounter a new virus, our immune system responds in two waves. First, the “innate” immune system kicks in. It’s old, really old — and found in some of the oldest vertebrates on the planet, and even in other organisms like insects and some plants. Our innate immune response includes things like mucus, skin, and hair, that can catch and sometimes kill microbes. It also includes physical reactions, like fevers, and a cellular response.
Cells called phagocytes, which is Greek for "cell eaters," are sent throughout your body, where they attack anything unfamiliar or novel. White blood cells like macrophages (also Greek, for "big eaters") are a type of phagocyte. There are great videos of phagocytes in action, and it can be oddly soothing to watch them eat away.
Sometimes that immune response is really effective. You feel sick and achy for a few days, but you’ve kicked whatever infection you have. But if you haven’t, your adaptive immune system kicks in.
Your adaptive immune system came about much later in the evolutionary process and is unique to vertebrates. It learns to identify different microbes and memorizes them. Some cells start to make antibodies, which are small molecules that look exactly like the letter Y. Antibodies are specific for certain types of microbes. When you’re sick, they flood your body and “tag” the viruses and bacteria that they recognize.That’s what antibody tests are looking for: these little flags that attach to the virus and signal to the phagocytes that there’s something important for them to eat.
The combination of cells that remember microbes and cells that produce antibodies are what (hopefully) make you immune to a disease. So detecting those antibodies can be a good sign of immunity.
What does the testing mean?
Unfortunately, at least right now, these tests probably won’t be able to confirm if you, yourself, are immune to SARS-CoV-2 (that’s how scientists refer to what the rest of us are calling the new coronavirus). That’s because the accuracy of these tests depends on how many people in a community are infected. That seems counterintuitive, because whether or not you test positive or negative shouldn’t (and doesn’t!) depend on the results of your neighbors. To understand why, you need to know a little bit about how testing works.
There are two main measures of accuracy for medical tests: sensitivity and specificity. Sensitivity is the percentage of people who are sick (or, in this case, have been sick) who get a positive result on a test, or a “true positive.” Specificity is the percentage of people who aren’t sick (or, in this case, have never been sick) who get a negative test, or a “true negative.”
Time for the numbers
So, for example, we can have an imaginary population of 100,000 people, and epidemiologists want to figure out. And let’s say 2% of them caught the fictional Oregon Virus — though researchers don’t know that yet. That means that there would be 2,000 infected people who could possibly have tested positive, and 98,000 people who would be uninfected.
Let’s say that these researchers had developed a particularly good test. It might have a sensitivity of 97% — that means that it would accurately diagnose 97% of people with antibodies. And let’s say it has a specificity of 95%, which would mean that it would accurately diagnose 95% of those without antibodies.
So, 97% of those 2,000 infected people would get a true positive. So that’s 1,940 positive tests that the epidemiologist would count, and 60 people would get a “false negative,” which means the test would incorrectly say they never got sick. That’s not too bad.
But there would also be 98,000 uninfected people. With a specificity of 95%, 93,100 of them would get “true negatives.” But the big concern is “false positives.” The epidemiologists would identify 4,900 folks as having the virus in the past, even though they were never infected.
Basically, if we were to test all 100,000 people in this town for coronavirus, only 1,940 of their 6,840 positive results would be correct. That’s a little over 25%, or just 1 in 4. In Oregon right now, if we're undercounting cases by a factor of ten, only about .3% of our population would have been infected, so the false positive rate would be even higher.
The good news is that epidemiologists can account for this mathematically and estimate the true spread of a virus. The bad news is that the test wouldn’t be very accurate for individuals. If COVID-19 does make you immune to reinfection from the virus, then identifying people who are immune can help us reopen parts of society and parts of the world. Immune people can work in higher-risk jobs or just have the peace of mind that they’ve had COVID-19 once and are unlikely to catch it again.
If one out of every four positive tests is right, then three out of four people who were told they were immune could still get infected with the virus.
But there are places with widespread viral outbreaks, and groups of people who are at high risk for contracting the virus, like healthcare workers. Let’s say these same epidemiologists test 100,000 healthcare workers, and 60% of them have been infected with a virus.
So, you’d have 60,000 people who have been infected. With a sensitivity of 97%, you’d get 58,200 true positives, and 1,800 false negatives. And of the 40,000 people who are uninfected, a test with 95% specificity would give you 38,000 true negatives, and 2,000 false positives. That means that just 2,000 out of 60,200 positive results would be wrong. Or, to look at it a different way, only about 3% of positive tests would be wrong.
It's not yet clear how public health officials plan to use serology testing. In Telluride, Colorado, public health officials planned to test the entire community. But results from their test were delayed, and the accuracy of the test was called into question.
In Oregon, scientists in the Providence-St. Joseph healthcare network have started a massive antibody testing study, where they plan to test healthcare workers to estimate the spread of the disease, and learn more about how our immune system responds to the virus. They’ve already tested over 2,000 people.