After spending the last few weeks talking with students, journalists, neighbors and family members, I’ve decided that there might be some value in discussing some of the terminology that we use when talking about diagnostic testing.
Science communication and media relations folks always discourage “jargon” when describing scientific concepts to laypeople. However, technical terminology can be a key part of scientific discussions, especially when it’s important to be precise. Now, in the setting of a global pandemic, precision is particularly important, and I believe people can handle more complexity than the scientific community often gives them credit for. Increasing scientific literacy can empower people to better understand and digest current events. So, here are some of those definitions.
First off, is reagent. In general, a reagent is any substance that is a starting material for a chemical reaction. Do you remember finding the “limiting reagent” in high school chemistry class? That’s the chemical that runs out first; thus, limiting the amount of product that can be made by the reaction. Many outlets have reported that one of the reagents failed in the initial test that the CDC distributed for COVID-19. It is still unclear which reagent did not perform as expected, but the reagent was one of the parts of the test reaction that the CDC shipped out to labs around the country. This problem has now been fixed, and all of the CDC test components are working well.
The next term is assay. Assay is just the word we use for “test.” Anytime you hear someone say they are running an assay or assaying for something, they are simply running a test.
Controls and control material also come up often when discussing test design. Controls are needed to make sure that the test you are running is valid. Basically, controls are parallel experiments that you run alongside your testing to make sure that you didn’t make a mistake while running the test. We science folks are always skeptical and are always checking to make sure we didn’t make a mistake! A positive control, in the context of COVID-19 testing, is when we put some material in the reaction that we know will make the test come up positive. If that reaction does not come up positive, we know that we made a mistake someplace, or that some of our other reagents are not working properly. A negative control is when we set up a test that does not contain a sample or additional material, that we expect to come up negative. If a negative control comes up positive, then we know that we have some kind of contamination, or that something we did not expect is happening in our reaction. If this happens, we need to start over, and figure out what went wrong. In both cases, controls that do not work as expected render any test result invalid. These results are not reliable and should not be reported.
Now, we need to cover sensitivity and specificity. These guys are a bit more complex. Sensitivity measures how little of the virus you can detect using a particular test in a particular sample. A super-sensitive test can detect very small amounts of the virus. Sensitivity goes hand in hand with the concept of a false negative. A test that is not sensitive enough might come out negative for someone who is infected with the virus, but does not have a viral load high enough to make the test read positive. It is also possible that a false negative can arise from a swab being taken improperly.
Specificity is a measure of how well the test detects the COVID-19 virus, and not other things that might confuse the test. A very specific test is good at detecting COVID-19 and will not detect other closely related viruses. Specificity goes hand in hand with false positive rates. If a test is not very specific, it might show a positive result when someone is not infected with COVID-19, but with something else like a flu virus.
Both false positives and false negatives make it difficult for healthcare providers to interpret test results in the context of care. Low rates of false positives and false negatives make a test more trustworthy.
Now you know the basics of test design and metrics! We hope that this makes reading some of the science reporting a little more clear.