If you have been listening to the news or reading the internet lately you have probably heard more about “testing” than you ever thought you would. Many people think that testing for COVID-19 is one of the most important pillars of any program that will ultimately solve this problem.
Before we jump in too deeply about testing let’s clear up a few terms:
Antigen test an antigen test is meant to detect the “thing” that is causing infection in your body. In this case when you hear antigen test you are usually hearing about a test that is supposed to detect pieces of virus from your nose or throat.
(notice I say “supposed to.” You will see why soon)
Antibody test: many people have heard of antibodies. They are proteins that your immune system makes to help give you resistance to infection. Antibodies are usually specific to a particular virus or bacteria. If you have antibodies in sufficient quantity to a particular virus, presumably you will have protection from the virus.
Prevalence: the proportion of individuals in a population that have the “thing” (antigen, antibody, disease) for which you are testing. For an antibody, prevalence would refer to the number of people in your population that actually have the antibody proteins in their blood.
You can imagine that testing people for either antigen or antibody may be useful. In an ideal world we would test everyone for antigen and treat / isolate ourselves accordingly until the virus just exited the population. In the ideal world the tests would all work perfectly also.
Unfortunately they don’t.
For better or worse, we live in the real world and tests aren’t accurate 100% of the time. Sometimes they turn negative when they should turn positive and sometimes they turn positive when they should turn negative.
I’m not going to tell you whether you personally should get antibody tested or not. What I am going to do is discuss some of the underlying basic issues with antibody testing in general and why the concept of having your antibody levels tested may leave you with more questions than answers.
I would like to thank Dr. Steven Strogatz of Cornell University for inspiring this post. He is a professor of mathematics and highlighted some of the basic math surrounding testing in some commentary over the weekend. Because of his outline I made a screencast expounding on this topic in detail.
The screencast is about 17 minutes in length. I made it lengthy because the concepts, while accessible to anyone who knows basic addition and multiplication, are often counter-intuitive. So I try to go through them slowly and with a few examples.
In the screencast you will:
- Learn why testing can be fraught with uncertainty
- That accuracy of the test matters (but other things matter too).
- The prevalence of the “thing” you are looking for in your population plays a huge roll in how well your test finds that thing
- When prevalence is low it has a strikingly out-sized impact on your testing results
- Whether I am getting the antibody test at this point in time or not.
If you decide to skip the screencast entirely here is the spoiler:
I am not getting the antibody test at this point in time. There is too much uncertainty. For the test to mean anything i would need to know more about the overall prevalence of antibody in my particular region of the world and know more about the accuracy of the test. Otherwise it’s not going to change my behavior. I’m still going to act with precaution.