Antigen Testing for COVID-19 Asymptomatic Screening: Why it Should be Preferred Over PCR Insights From Navy Region Japan’s Experience With a Comprehensive Antigen Testing Strategy

ABSTRACT

Polymerase chain reaction (PCR) is commonly used in asymptomatic screening testing, but is suboptimal for this purpose as it will identify many old persistent positives that are no longer infectious. This can result in placement of individuals that are not infectious to others into isolation. This results in substantial adverse impact to military manning and operations, without any benefit to public health. Antigen testing does not have this same drawback. Antigen testing, while less sensitive than PCR, will identify the vast majority of infectious positives, especially those with higher viral loads that are more likely to transmit to others. Importantly, use of antigen testing will also greatly increase the certainty of benefit from isolation, reducing the risk of isolating those individuals who are beyond their infectious period and pose no threat to public health. The literature on this topic is reviewed, with particular focus on studies that perform viral culture in addition to PCR and antigen testing. This allows for determination of sensitivity for infectious virus. Also, Navy Region Japan’s experience with a comprehensive antigen testing strategy is described. The challenges presented by persistent positive PCR test results are examined, as well as the real-world benefits from implementing widespread use of antigen testing.

There has been much debate regarding the role of antigen testing for coronavirus disease 2019 (COVID-19) during the pandemic. In Navy Region Japan, our unique pandemic situation has allowed us to experience firsthand the benefits of widespread antigen testing, particularly for asymptomatic screening. We have overseen a large volume of asymptomatic screening tests, which are mandated by various policies for Restriction of Movement (ROM) exit testing, ROM-Sequester, pre-travel, pre-surgery, etc. One centralized public health team has overseen all positive results directly. Due to lower case numbers than in the USA, we have never exceeded contact tracing capacity. Our public health team has conducted a very detailed contact tracing on every single case, numbering over 1,100 as of this writing. This has allowed us to take an aggressive public health posture, in line with our high level of risk aversion in an Outside the Continental United States (OCONUS) environment, with the largest forward-deployed force in the US Navy. We have also performed thousands of antigen tests since the winter surge began around the beginning of December. With limited in-house laboratory capacity for polymerase chain reaction (PCR) testing, 2-3 day turnaround times for samples sent to a local contracted lab, and a surplus of Abbott BinaxNOW COVID-19 antigen testing cards readily available due to a very large order by Fleet medical forces, the winter surge in cases drove us to embrace antigen testing out of necessity.

The ROM exit testing is mandated by policy and is performed near the end of a strict 14-day ROM period on all individuals after traveling from the USA (a high incidence area) to Japan (low incidence area). The ROM exit tests account for approximately 20% of our positive results since the pandemic began. On detailed history, many of these have been strongly suspected of being persistent positives. Examples include asymptomatic individuals testing positive on ROM exit with a prior positive result beyond the 90-day mark, a history of COVID-like illness several weeks prior with no testing performed at the time, a history of multiple friends or family getting COVID-19 in the last few weeks to months, or even a recent (weeks to last few months) history of being placed into quarantine due to close contact with a known positive, but with no subsequent testing. The switch from PCR to antigen testing for ROM exit purposes coincided with a steep reduction in the number of positives detected on ROM exit, without any suspected or confirmed cases of transmission from individuals released from ROM after a negative antigen test. This has effectively blunted the adverse impact of COVID-19 testing on manning and operations, although it is hard to quantify precisely how much of this drop is due to antigen testing alone. Mandatory pre-travel testing in the USA and an overall decrease in U.S. case rates have likely contributed as well.

Reverse transcription polymerase chain reaction (RT-PCR, subsequently referred to here as PCR) is considered the gold standard of COVID-19 lab tests, with generally excellent performance characteristics. However, a major limitation with PCR is its inability to accurately measure the infectiousness of detected viral RNA. Scientific evidence indicates that individuals with COVID-19 typically remain infectious for no more than 8-10 days,¹ but can test persistently or intermittently positive by PCR for much longer.² The U.S. Centers for Disease Control and Prevention have concluded that this can occur for 3 months post-infection,³ but it is unknown exactly how long infected individuals might continue to test intermittently positive. Persistent detection of SARS-CoV-2 RNA by PCR has been a significant problem for us in Japan. When we were still following the test-based strategy for ending isolation, during the early part of the pandemic, it was typical to see patients take about 4 weeks to obtain two consecutive negative results, with some taking beyond 6 weeks. This is in line with other studies.⁴ At one point we had a group of previously positive, recovered sailors that all underwent PCR screening testing in order to rejoin a deployed ship. Nine resulted positive again, although they were all at least 80 days past their initial positive result. Six of these were beyond 90 days, including one at 116 days and one at 124 days. They were all asymptomatic and without significant risk factors on contact tracing. This also came at a time when our baseline case incidence was extremely low. Some subsequently had a sample sent for viral culture, all of which came back negative for replicable virus. All were likely persistent positives, with no evidence to support that they might have been reinfected. This raised the specter of detection of persistent positives beyond 90 days during routine shipboard surveillance testing, which could dramatically impact operations on an otherwise COVID-free ship. While all of these active duty sailors were young and healthy, other studies have shown that persistent positivity is even more prevalent among older patients and those with comorbid conditions.⁵

It is also well-known at this point in the pandemic that many infected individuals have mild or no symptoms, and never seek testing. Serologic testing data indicate that there are likely a total number of infections that are far larger than the number of confirmed reported cases.⁶ The implication is that there are likely millions of unknown cases in the USA alone. In countries with less access to widespread testing, the proportion of unknown cases to reported cases is likely to be even higher. At any given time, many of these will be beyond 10 days but within 3 months of their initial infection. With the proliferation of testing requirements for screening purposes, such as before international air travel or before elective surgery, we should consider the potential unintended consequences and adverse impact, whether to military operations or to society in general, of such broad PCR testing.

With diagnostic testing in symptomatic patients, the date of symptom onset allows us to pinpoint the beginning of the infectious period with a high degree of accuracy. For asymptomatic cases with a known exposure, this accuracy is decreased but still reasonable. But for cases who test positive on a PCR screening test with no recent symptoms and no recent known exposure, it is completely unknown when their infectious period began. Given the long-term positivity seen with PCR, it is likely that most of these asymptomatic cases are already beyond their infectious timeframe at the time of their test. For example, one prevalence study among hospital staff from the UK utilized both PCR and viral culture. Nineteen positives were identified by PCR and subsequently cultured, but virus was isolated from only one sample.⁷ I am not aware of any research studies or statistical models that examine this probability in detail. However, these cases are typically managed the same as all others that test positive—a recommendation for 10 days of isolation from the date of test and 14 days of quarantine for their close contacts, even though there is a substantial possibility that this results in no public health benefit. Department of Defense policies may specifically require these actions for all new PCR positives, without flexibility for clinical judgment. As medical professionals, however, we should seek to attain a higher degree of certainty of public health benefit before directing individuals to isolate. Doing otherwise could inflict considerable and unnecessary harm in terms of decreased quality of life, operational impact for our forces, lost income and economic productivity for civilians, delayed surgery or other medical care, delayed travel, and negative impact on mental health.

A few potential solutions have been investigated. The PCR cycle threshold (CT) values could be used to help guide the risk of infectivity, as has been previously proposed.⁸ For various reasons, however, this idea has not generally gained the support of public health officials. Other factors besides viral load can affect how much virus is present in a sample, such as quality of the swab technique when obtaining the specimen. Other reasons are the lack of a universal CT cutoff beyond which infectious virus is not possible and the lack of standardization between labs (different labs may give different CT results on the same sample). Viral culture measures infectiousness directly, by testing the specimen to see if the virus is capable of replicating. These cultures are complex, time-consuming, expensive, and relatively few labs can perform them. This effectively renders them impractical for routine individual testing. An alternative strategy would be to shift preferentially toward antigen testing when performing low-risk screening tests. The most widely held argument against antigen testing is that it is less sensitive than PCR, especially in asymptomatic patients. The problem with this assertion is that sensitivity in this case is determined based on comparison to PCR alone, without factoring in infectiousness or the known limitation of persistent positivity with PCR testing. When calculating the sensitivity and specificity of antigen testing, a comparison with viral culture would provide a more useful determination, as it takes infectiousness into account. Identification of infectious individuals is our ultimate goal. There is no public health benefit in isolating persistent positives beyond their infectious period.

The authors of one study evaluated both antigen (BinaxNOW) and PCR results for all individuals, as well as viral culture for those with a positive test result.⁹ The sensitivity of BinaxNOW compared to viral culture was 93% in symptomatic patients and 79% in asymptomatic patients. Furthermore, among all results that were negative on BinaxNOW testing but positive by PCR, 91% were viral culture negative (113 of 124). This suggests that approximately 90% of asymptomatic persons with a negative antigen test but positive PCR test are no threat to public health. In other words, isolating all of these individuals results in nine unnecessary isolations for each one that might prevent further transmission. Some would likely consider this ratio to be unacceptably high. Another study reported antigen (Sofia) testing to have a sensitivity of 41% (7 of 17) when compared to PCR in asymptomatic patients.¹⁰ But on further inspection of the data, all 10 of the false negatives were also negative on viral culture, meaning sensitivity was actually 100% when compared to viral culture in these asymptomatic patients. When including symptomatic patients, 89% (16 of 18) of those with negative antigen results but positive PCR results had a negative viral culture. This mirrors the results from the previous study. These data suggest that antigen testing may, at least in some circumstances, provide a better public health value than PCR testing in persons with low pre-test probability for COVID-19 infection.

Another study compared antigen testing (BD Veritor) with both PCR and viral culture.¹¹ Antigen testing demonstrated a higher positive predictive value (90% versus 74% for PCR) for infectious virus as determined by viral culture, even though all patients reported symptoms. This disparity would be expected to grow substantially when testing only asymptomatic patients, due to the issue of persistent PCR positivity. Among those with positive PCR results, those with negative antigen results had a significantly higher cycle threshold value and therefore were likely to have a lower viral load than those with positive antigen results. An additional study also found that antigen testing (BinaxNOW) has good sensitivity (93%) in detecting PCR positives with CT value of 30 or lower, the ones most likely to be infectious.¹² They concluded that the test accurately detected infection with high viral loads in both asymptomatic and symptomatic individuals.

A recent study looked at several different antigen testing platforms and found a sensitivity (compared to PCR) ranging from 65% to 89%, with increasing sensitivity to at least 90% for most tests when the cycle threshold level was 25 or less.¹³ But all three antigen tests that were compared directly to viral culture were found to have at least 94.7% sensitivity for infectious virus, irrespective of cycle threshold value. This sensitivity for infectious virus is excellent. A simulation based on a very large volume of real-world contact tracing data, testing results, and cycle threshold values from the UK estimated that a lateral flow antigen test would identify approximately 90% of all cases leading to onward transmission.¹⁴

All of these studies support existing Centers for Disease Control and Prevention guidance on antigen testing, which is to accept negative antigen test results for those who are asymptomatic with no known exposure (i.e., those with low pre-test probability).¹⁵ Confirmatory PCR testing is only recommended on those with positive antigen results in this setting, although more recently updated guidance recommends confirmation only for congregate settings. This mitigates the risk of false positive results from antigen testing. While the extra tests are somewhat burdensome, the higher degree of certainty of benefit from isolation is a favorable trade-off. A few extra tests are certainly less of a burden than 10 days in isolation, especially when it may not actually result in any public health benefit.

Another reason that is sometimes cited for not using antigen testing for screening purposes is that the Emergency Use Authorization for most antigen tests specifies its use in symptomatic individuals within about 5-7 days since symptom onset. Some fear that doing otherwise may result in the laboratory being cited by the Centers for Medicare and Medicaid Services (CMS) for inappropriate use of the test. However, CMS issued a policy update in early December which clearly states that “CMS will not cite facilities with a CLIA Certificate of Waiver when authorized SARS-CoV-2 molecular or antigen POC tests are performed on asymptomatic individuals outside of the test’s authorization, when done so considering the information in FDA’s FAQ.”¹⁶ The referenced Food and Drug Administration language states, among other very useful guidances, that “It is not necessary to perform confirmatory high-sensitivity molecular tests on individuals with negative antigen test or other point-of-care test results if they are obtained during routine screening or surveillance.”¹⁷

In Navy Region Japan, our protocol is to confirm all negative antigen tests with PCR when the risk of COVID-19 is judged to be at least of moderate risk (i.e., a patient with typical COVID symptomatology). We do not confirm antigen negatives from asymptomatic screening tests with PCR, but we do confirm all antigen positives with rapid PCR (typically BioFire or Cepheid GeneXpert), regardless of pre-test probability or reason for test. Regarding the risk of false positive results, this has not been a problem for us here. To my knowledge, our large-scale antigen testing strategy has generated only two false positive results, both easily overruled with two consecutive negative PCR results. As for clinically relevant (infectious) false-negative results, we have had no cases suspected of transmitting infection after receiving a negative antigen result.

Having initially started using antigen tests out of necessity, our confidence with antigen testing has only grown over time. At this point in the pandemic, we use antigen tests for nearly all first-line testing, unless PCR is specifically required in policy. Line commanders love them since they are perfect for expeditionary purposes, with ease of use, reliable results, and without the concern for persistent positive PCR results, which can create a lot of churn on a ship. Shipboard medical staffs love them since they can be easily performed without access to a hospital lab and easily repeated serially as necessary. Medical leadership loves them since they save the hospital a large amount of money versus sending expensive PCR tests out to a contracted lab. The hospital laboratory loves them because they greatly reduce the pressure on the lab that comes with consistent high-volume PCR testing. The clinical staff loves them because they can tell patients their test results before they walk out the door. Patients love them since they do not have to wait hours to days anxiously wondering if they have COVID-19 or not. Public health officials love them because they eliminate the problem of persistent positives, which was a major headache for us earlier in the pandemic when we used PCR testing exclusively. Additionally, use of antigen testing during investigation of new case clusters allows us to rapidly test and trace, bounding outbreaks much faster than we could with PCR alone. We consider the adoption of widespread antigen testing as a major contributor to our pandemic success.

Antigen testing is by no means perfect. It is not suitable for mass testing of many people all at once, due to the strict timing requirement to read the tests. It is not the best choice when you have an extremely low risk tolerance for a missed positive, since PCR is more sensitive (although, as shown above, not by much for infectious virus). In particular, antigen testing is likely to miss more positives than PCR among those that are very early in their infection, when the viral load is still very low. But for most testing purposes, and particularly for general asymptomatic screening testing, antigen testing performs well. In any given situation, however, leadership would need to appropriately weigh the risk of a missed infectious positive against the risk of over-isolation of non-infectious positives. Determination of the proper level of risk acceptance goes beyond the scope of public health.

At this point in the pandemic, public health personnel must fully understand the trade-offs and potential burden from PCR testing for low-risk COVID-19 screening. PCR testing likely identifies far more non-infectious cases than infectious cases in this setting. For this reason, antigen testing is likely preferable to PCR for asymptomatic screening in most situations. Antigen testing will identify the vast majority of infectious positives, especially those with higher viral loads that are more likely to transmit to others. Importantly, antigen testing will also greatly increase the certainty of benefit from isolation, reducing the risk of isolating those beyond their infectious period who pose no threat to public health. This knowledge should improve the quality of our risk assessments as we advise Department of Defense and other Public Health policymakers.

FUNDING

None declared.

CONFLICT OF INTEREST STATEMENT

None declared.

Author notes

The views expressed are solely those of the author and do not reflect the official policy or position of the Uniformed Services University, U.S. Navy, Department of Defense, or U.S. Government. I am a military service member. This work was prepared as part of my official duties. Title 17 U.S.C. 105 provides that “Copyright protection under this title is not available for any work of the United States Government.” Title 17 U.S.C. 101 defines a U.S. Government work as a work prepared by a military service member or employee of the U.S. Government as part of that person’s official duties.