On Antigen Testing for COVID-19

By Gary W. Procop, MD, MS

The performance of antigen testing in microbiology is well understood as there is abundant literature and years of experience.  The principles, advantages and limitations that have been described for Group A Streptococcus, respiratory viral antigen testing, and other antigen tests for infectious agents hold true for antigen tests for COVID-19.  Critical to the performance and positive predictive value of these or any tests are the analytic test performance characteristics of the test (i.e. sensitivity & specificity) and the population that is being tested.  These factors are clear to the U.S. Food and Drug Administration, which acknowledged in a response letter to an ASCP recommendation that “test setting and patient population play an important role in determining test performance.”  FDA added that these factors are considered “when reviewing an Emergency Use Authorization request…”

Antigen tests for respiratory viral pathogens are less sensitive than reverse transcription polymerase chain reaction (RT-PCR) assays.  Unfortunately, the limits of detection (LoD) of antigen tests are not included in the FDA webpage that compares the LoD of numerous molecular SARS-CoV-2 assays (https://www.fda.gov/medical-devices/coronavirus-covid-19-and-medical-devices/sars-cov-2-reference-panel-comparative-data).  The FDA should include the LoD of these assays for comparative purposes and as a public service to patients and providers alike.

Optimal Use of Antigen Test: True Positives

Antigen tests are most useful to confirm a clinically suspected infections (i.e. a symptomatic patient).  The positive predictive value (i.e. the likelihood that a positive test result is a true positive) is high when the test is used in this setting.  Additional advantages include, ease of test performance, a quick time to diagnosis and a low cost.

Challenge #1: False Negatives

When the antigen test is negative on a symptomatic patient (i.e. a test result/clinical mismatch), then the patient should be retested with an RT-PCR assay.  Occasional false-negative test results are expected to occur due to the limited sensitivity of antigen tests compared with RT-PCR.

This recommendation is consistent with CDC guidance which has stated: “…it may be necessary to confirm a rapid antigen test result with a nucleic acid test, especially if the result of the antigen test is inconsistent with the clinical context (https://www.cdc.gov/coronavirus/2019-ncov/hcp/testing-overview.html).”  Otherwise stated:  If there is a mismatch between the clinical findings and the test result, then an RT-PCR should be performed.

Challenge #2: False Positives

As FDA staff noted during an October 7 FDA Town Hall “it is important to consider the positive predictive value when assessing how to react to a positive result.”  Whenever a test is used in a low prevalence setting then the likelihood of false positives increases (See example, below).  This includes using the test for screening asymptomatic patients.  If an asymptomatic patient tests positive with and antigen detection test, there is a significant probability that the test is a false positive, so the presence of SARS-CoV-2 should be confirmed with an RT-PCR test.  There are numerous examples of false positive SARS-CoV-2 antigen tests in the lay press.

This recommendation, again, is consistent with CDC guidance which has stated: “…it may be necessary to confirm a rapid antigen test result with a nucleic acid test, especially if the result of the antigen test is inconsistent with the clinical context (https://www.cdc.gov/coronavirus/2019-ncov/hcp/testing-overview.html).”  Otherwise stated:  If there is a clinical/test result mismatch, then an RT-PCR should be performed

There is a significant patient safety issue if these tests are used to cohort patients prior to RT-PCR confirmation, since uninfected patients with false-positives results would be placed in close proximity with patient who are infected, and thereby may become infected themselves.  If antigen tests are used to screen asymptomatic patients, then positive results should be considered presumptive positives, until confirmed by an RT-PCR test.  These patients could be kept in isolation, but should not be cohorted with other patients with COVID-19, until infection is confirmed.

There is also a danger of artificially inflated infection rates if positive results from asymptomatic patients are reported with RT-PCR confirmation.

Example:

This table is an excerpt from the Clinical and Laboratory Standards Institute report GP49, Developing and Managing a Medical Laboratory (Test) Utilization Management Program.  It demonstrates how the positive predictive value (i.e. the likelihood that a positive test result represents actual disease in the patient) of a very good test (i.e. 95% sensitivity/95% specificity) diminishes from 95% when there is a prevalence of 50% to 16% when the prevalence is 1%.  Otherwise stated:  At a prevalence of 1%, only 16 out of 100 positive results are correct, the remainder (84 out of 100) are wrong (i.e. false positives)!

Gary W. Procop, MD, MS, MASCP, is a member of the ASCP Board of Directors and also Chair of the ASCP Institute for Science, Technology and Public Policy. He is the director of Molecular Microbiology, Mycology, Parasitology and Virology Laboratories at the Cleveland Clinic, where he is also The Belinda Yen-Lieberman, PhD & James M. Lieberman, MD Endowed Chair in Clinical Microbiology.