COVID Crisis Post 41: Interpreting Test Results.
I think there is a lot of confusion about the accuracy of testing during this pandemic, and what a positive or negative result on a test actually means.
In terms of the pandemic, this could apply to a positive or negative nasal swab for COVID-19, or a positive or negative antibody result for antibodies against COVID-19.
Many people are testing negative for COVID-19 even though they seem to have obvious symptoms of it, while others are testing negative for antibodies even though they previously tested positive for COVID-19 weeks earlier.
So what gives?
First of all, these are screening tests. Second, no test is perfect. You may have come across instances where the sensitivity or specificity of a test is described, and a percentage value may have been associated with these terms. These terms describe the inherent characteristics of the test itself.
There also are terms called Positive Predictive Value (PPV) and Negative Predictive Value (NPV), which describe the clinical relevancy of these same tests.
Biostatistics was not fun in medical school and time has not changed that fact.
But understanding their general meaning is ESSENTIAL to understanding the usefulness of the tests. So apologies if this is painful and/or confusing, but it is SO important.
So if you care, please TAKE YOUR TIME trying to understand these concepts.
The basic gist:
Any test result you receive must be put into context. It is not as simple as you have the disease or you do not have the disease.
A highly sensitive test correctly identifies a patient who is a TRUE POSITIVE - ie it does not miss those with the disease, and therefore has few FALSE NEGATIVE results. This value is determined by taking KNOWN positive cases and determining how often the test comes up with a positive result. SnOUT - rules OUT disease.
A highly specific test correctly identifies a patient who is a TRUE NEGATIVE - ie it does not miss those without the disease, and therefore has few FALSE POSITIVE results. This value is determined by taking KNOWN negative cases and determining how often the test comes up with a negative result. SpIN - rules IN disease.
It is difficult to produce tests that demonstrate both high sensitivity and high specificity. Typically, if you have a highly sensitive test, specificity tends to suffer and vice versa.
Positive Predictive Value (PPV) describes the probability that a person has the disease given a positive test result. This is when you do NOT know the disease status of a patient. So if a test has a high PPV, a positive result is much more likely to mean a person is actually positive.
Negative Predictive Value (NPV) describes the probability a person does not have the disease given a negative test result. Same as with PPV, this is when you do NOT know the disease status of a patient. So if a test has a high NPV, a negative result is much more likely to mean a person is actually negative.
The prevalence of a disease, or the percentage of the population who has the disease, has a HUGE impact on both the PPV and NPV. A decrease in prevalence of disease decreases PPV and increases NPV, and vice versa. I hope this makes intuitive sense.
However, sensitivity and specificity are totally independent of prevalence since these values are characteristics of the test itself.
We do not know the true prevalence of COVID-19 overall in the US, nor in individual cities or patient populations. But if our test has very high sensitivity and/or specificity, it may mitigate the effect of prevalence somewhat and helps maximize the tests usefulness.
In the case of our current COVID-19 testing, if you test positive, you can feel pretty confident you have the virus because the test has a high SPECIFICITY. PCR testing should not come back positive if no viral DNA is present.
But if you test negative, this does not necessarily rule out having the virus because the SENSITIVITY is only about 70%, meaning 30% of the time the result is actually a FALSE NEGATIVE. This could be due to improper collection methods, early infection before viral shedding, etc.
The sensitivity and specificity of a test needs to be validated to confirm the values listed are in fact true. And with COVID-19, there has not been a determined "gold standard" in which tests can be consistently tested against.
The reliability issue is also big when discussing current antibody testing.
There are dozens of antibody tests on the market for COVID-19, but many of these tests have not been validated, and even if they are, they may not be very accurate due to a host of reasons. So many of them likely suck as tests.
This shit has always been confusing and caused a ton of headache for me while studying for Step 1 (and I'm sure many others).
This post may even have caused some of my medical colleagues to shudder.
But I hope this shed some light on the subject, and please reach out if any additional clarification is needed.
So again apologies if this was painful and/or confusing but...
Nerds gonna nerd 🤓.