The accuracy of COVID-19 tests

619 Shares

The current COVID-19 epidemic has highlighted the availability and the results of medical diagnostic tests.  As a physician, I treat the results of lab tests like I treat movie recommendations from a friend – I am always skeptical.  For the movies, I need to consider my friend’s suggestion in the context of their mood, tastes, current life crises,  and track record. A context also needs to be created for ordering and interpreting test results, including the patient’s current complaints and symptoms,  recent activities, and medical history combined with findings on exam, my personal experience and the expertise of my colleagues, and information from published medical studies. My friend’s movie judgments are occasionally biased and off-kilter.  In the same way, medical diagnostic test results are not perfect. There is always the chance that they provide incorrect information.

Medical professionals, policymakers, and members of the general public may overestimate the accuracy of diagnostic tests.  The usefulness of any test depends on how likely the patient has the disease, the ability of the test to correctly identify the disease, and the capability of the test to correctly confirm the condition is not present. Unfortunately, test results will be negative for some people that actually have the disease, and some people without the disease will have positive tests.

Other factors, such as when the test is taken during the course of the disease, how and from where the sample is obtained, and how the sample is transported and processed also affect the results.

The accuracy of the current COVID-19 tests is not precisely known.  Reasonable estimates, based on test performance in China and the performance of the influenza tests, are that the tests will correctly identify around 60 percent of the patients with the disease and correctly identify 90 percent of the patients that are disease-free.

Assume that the physician thinks there is a 50 percent of the patient having COVID-19.  Given the above numbers, if the patient has the disease, the test will be positive 85 percent of the time.  Fifteen percent of the infected patients will incorrectly be diagnosed as not having the disease.  If the patient does not have the disease, only 70 percent of the patients will have a negative test.  It would take four consecutive negative tests to conclusively prove the patient did not have the disease.

Just as people should not demand a prescription for a medication based on TV ads, they should not seek unnecessary medical tests. Physicians and patients need to discuss whether any test, including the COVID-19 test, needs to be ordered and the ramifications of the results.  Policymakers need to define the importance of testing in their plans.

Test kits for COVID-19 will be important tools for the containment and elimination of the virus, but their use needs to be incorporated into the overall strategy for combating the disease. A comprehensive approach, including the judicious use of testing, will be essential.

Richard L. Hutchison is a plastic surgeon.

Image credit: Shutterstock.com

619 Shares

View 2 Comments >

Most Popular

✓ Join 150,000+ subscribers
✓ Get KevinMD's most popular stories