<img height="1" width="1" style="display:none" src="https://www.facebook.com/tr?id=749646578535459&amp;ev=PageView&amp;noscript=1">

What is a false negative? (for dummies)

June 18, 2020



Someone asked me a question that I’m sure many of you also have:

Would you please address the issue of the reliability of the testing that is available? One of my friends tested “negative”, but she has many of the symptoms of COVID.  Her doctor said that 40% of the tests show a false negative.  She is being treated “as if” she has the virus. She feels that she indeed does have the virus, in spite of the negative result.

On the flip side, is there a high incidence of false positives?

Are some tests more reliable than others?

Yes. That doctor is spot on. The range I have seen is a 20-40% false-negative rate with the COVID PCR tests. This means a person HAS COVID but the test comes back negative. Another way to say this is if you have 10 patients who all have COVID when you test them, 2-4 tests will come back negative despite the patients having the disease. 

Ok, let’s break this down. Here is a review of a paper on this topic. Clay Smith MD, the author of the review, runs Journal Feed. We both trained and worked together back at Vanderbilt. He is excellent and it's no surprise that this is a great summary.  

From the paper:

If clinical suspicion is high, infection should not be ruled out on the basis of RT-PCR alone, and the clinical and epidemiologic situation should be carefully considered.

From his review:

It found that testing 3 days after symptom onset had the lowest false-negative rate (20%, 95% CI 12 to 31%). Testing on the day of symptom onset had a false negative rate of 38%. The study also concluded that testing after exposure but before symptom onset was almost worthless. Swabs collected during the prodrome had a false negative rate ranging from 67% to 100%.  Figure 2 from the article tells the story.

So, the meaningful variable from the paper seems to be how many days you have had symptoms. Three days after symptoms seem to be the most accurate. Pre-symptomatic seems to be of low value. And, the longer you have symptoms past day 3, the accuracy drops. 

Question: Why does it drop? 

Answer: The virus is thought to migrate down into the lungs after a few days so nasal tests are not coming into contact with the virus as frequently. 

I’m sure the collection technique and the test characteristics are important too. I just didn’t find as much on this but those two variables have face-validity (#learningdoctorspeak). 

To wrap this up, the KaiNexian who was admitted to the hospital for over a week and had a couple-day ICU stay had 2 negative tests before his first positive test. Doctors use Bayesian thinking when interpreting tests that aren’t perfect (almost no tests are perfect). So, if you have a very high clinical suspicion that someone has a disease then the test can influence your thought but not necessarily completely change your mind. Here is a nice article if you want to go deeper on Bayesian theory/thinking and how it applies to medicine vs a purely statistical explanation which makes my head hurt. The letter titled To Test or Not to Test incorporates some of this thinking.  

Stay emotionally connected and physically distant,


PS: If you would like to read more on the false-negative rate, this was also pretty informative. 

PPS: I have been unable to find an abundance of good information on the false-positive rate. I think the thought here is that it is much, much lower. If you get a positive COVID test and you are sick, you have COVID. Also, the false-positive rate would be very hard to reliably calculate because a number of patients are asymptomatic cases and/or never make antibodies. 

PPPS: After I wrote this (but before I sent it), another friend reached out to me and told me his daughter got two tests at the same time (one rapid and one send out). The rapid was positive and the send out was negative. I had to look that one up as it didn’t make sense to me. Here is what I learned from the Mayo Clinic site (third question down). The rapid tests (which are new) are Antigen tests (they detect proteins rather than genetic material). They are rapid but have a higher false-negative rate (low sensitivity) so they are often followed up with a molecular PCR test (higher sensitivity) which is a send out. So, if your rapid antigen test is positive, done deal - you have the disease and the second test is unnecessary and irrelevant.