Rapid Expert Consultation on Critical Issues in Diagnostic Testing for the COVID-19 Pandemic (November 9, 2020)

November 9, 2020

Robert Kadlec, M.D.

Assistant Secretary for Preparedness and Response

200 Independence Avenue, SW

Washington, DC 20201

Dr. Kadlec:

Attached please find a rapid expert consultation that was prepared by members and other experts on behalf of the National Academies of Sciences, Engineering, and Medicine’s Standing Committee on Emerging Infectious Diseases and 21st Century Health Threats: Jeffrey Duchin, Tara O’Toole, David Walt, Mary Wilson, and me. Details on the authors and reviewers of this rapid expert consultation can be found in Appendix C.

No dimension of response to the COVID-19 crisis is more dynamic than the field of diagnostic testing. Approximately 200 diagnostic tests related to COVID-19 have received Emergency Use Authorization by the U.S. Food and Drug Administration, and scores more are under development. Approximately 80 million tests have been performed in the United States. While the number continues to expand, the need and demand for testing have outpaced the growth. Some tests require centralized, specialized equipment, and others may be done at the point of care (POC) or in the home. Many difficult decisions beset employers, school administrators, public health officials, and clinicians as they come to grips with the availability and suitability of diagnostic tests as tools to detect, monitor, contain, and manage individual infections and the pandemic.

In this multi-faceted and dynamic situation, the Standing Committee on Emerging Infectious Diseases and 21st Century Health Threats was asked to examine four key topics that bear on the use and interpretation of diagnostic tests in the COVID-19 pandemic: (1) advantages and limitations of reverse transcription polymerase chain reaction (RT-PCR) testing for viral RNA; (2) the status of POC testing; (3) testing strategies, namely, considerations in the deployment of types and sequences of tests; and (4) next-generation testing that offers the prospect of high-throughput, rapid, and less expensive testing.

Regardless of when vaccines may become widely available, diagnostic tests will continue to play a vital role in coping with the COVID-19 pandemic. We hope the concise discussions in this rapid expert consultation will prove informative and useful.

Sincerely,

Harvey V. Fineberg, M.D., Ph.D.

Chair

Standing Committee on Emerging Infectious Diseases and 21st Century Health Threats

BACKGROUND

Goal of This Rapid Expert Consultation

The U.S. Food and Drug Administration (FDA) has granted Emergency Use Authorization (EUA) status to more than 200 COVID-19 diagnostic tests. New test types and formats are emerging rapidly. By the first week of October 2020, the U.S. Department of Health and Human Services’ Rapid Acceleration of Diagnostics (RADx) program awarded approximately $476 million in contracts to 22 biomedical diagnostic companies to support the development and manufacture of laboratory-based and point-of-care (POC) COVID-19 diagnostics using different technologies.^1,2

Since the start of the pandemic, diagnostic testing has been critical to the medical care of those infected with COVID-19, the protection of health care and other essential workers, and the efforts to contain the spread of the disease. This rapid expert consultation draws attention to four critical areas in developing diagnostic testing and strategies to reduce the number of COVID-19 infections and deaths as outlined in the Statement of Task in Box 1.

__________________

Testing Is Part of a Multi-Pronged Mitigation Framework

Extensive, strategically deployed SARS-CoV-2 diagnostic testing is an essential aspect of a multi-pronged mitigation plan necessary for the safe resumption of normal social and economic activity now and for some time after a vaccine against SARS-CoV-2 becomes available. In conjunction with physical distancing, masking, and hand washing, adequate availability of prompt and reliable testing is an important public health action needed to contain the spread of the virus through augmenting case and contact management and epidemiological surveillance. The global scope of the pandemic and continued spread of infection in the United States underscore the need for a sufficient scale of reliable and timely SARS-CoV-2 diagnostic tests.

The testing capacity required for COVID-19 diagnosis has overwhelmed the global diagnostics industry, including large, central testing laboratories; manufacturers of diagnostics instruments; and suppliers of test reagents and consumables. Never before has the world needed so many diagnostic tests for a new disease on a continuing basis. Although the number of tests conducted per capita in the United States has increased substantially over the past 6 months, shortages of vital testing materials, limitations on access to testing in parts of the country, and delays in obtaining results beyond the time when the information would have been most useful continue. These shortcomings impede the country’s ability to break the chain of viral transmission.

Testing Goals

In general, COVID-19 diagnostics answer one of two questions:

1. Is there SARS-CoV-2, or fragments of the virus, present in a person at the time a sample is collected (as a possible indicator of a current or recent infection)?
2. Does a person have antibodies in their serum that bind to SARS-CoV-2 viral proteins (as a possible indicator of an immune response from a past infection or from immunization)?

Two kinds of tests—molecular or genomic tests, also called NAATs (nucleic acid amplification tests), and antigen tests—are available to answer the first question. The second question is answered by antibody tests. The scope of this current rapid expert consultation will focus on critical issues in the use of testing to identify current or recent infections in order to control and mitigate viral transmission. The reader is referred to Appendix A for a brief background on antibody tests specific to SARS-CoV-2.

Clinicians use diagnostic testing to guide treatment decisions for individual patients. Public health professionals use testing results to monitor the incidence of specific illnesses in a community, and with communicable diseases, these data can also be used to determine when to implement control measures. Testing can be used to find infected individuals who may be unaware of their status in order to enforce isolation and to determine the infectivity of a pathogen and how easily it spreads from person to person. Testing can also be used to monitor the effectiveness of prevention measures. Institutions may seek to implement assurance testing to affirm the absence of infection in order to permit access, for example, to a workplace or a school.

For individual patient diagnosis, testing is important to determine who is infected, and also when alternative diagnoses should be considered. Accurate diagnosis will guide treatment and needed public health measures. Patients who are positive for SARS-CoV-2 will need to isolate until they are no longer contagious, and should also provide information to contact tracers to ensure that any close contacts are notified of potential exposure and can quarantine as needed. False positive tests can result in unnecessary isolation and anxiety in the patient and their contacts. A false positive may perversely expose an individual to the virus if that individual is in a congregated housing situation that isolates all positive cases together. False negatives deprive the patient of the optimal treatments and also may increase the spread of the infection if the patient and their contacts do not isolate.

Diagnostic tests may be used to screen individuals to allow them to participate in school, sports, work, or other activities. Tests for this purpose of assurance must provide rapid results and may be useful if repeated with sufficient frequency, even if each individual assurance test has lower sensitivity and specificity than a more definitive test.³ False negative results in screening contexts can result in the spread of infections if individuals inaccurately believe themselves to be uninfected and able to engage in these activities safely. Thus, people at high risk of

__________________

complications from COVID-19 should weigh the risks of taking part in these activities regardless of the test results of other participants. All positive antigen tests should be followed up with a more specific test as needed to ensure an accurate conclusion, and test results should be provided to local public health officials.

Diagnostic tests can also be deployed for public health purposes to perform routine or targeted surveillance of populations. Especially with a pathogen such as SARS-CoV-2 that frequently results in asymptomatic infection, routinely surveying populations can provide public health officials crucial information to determine the prevalence of infection and changes in infection rates over time. Surveillance can be performed in specific populations that might be at higher risk in order to detect an outbreak earlier and institute preventive measures. Public health laboratory surveillance can also be used to evaluate the effectiveness of control programs and to assist in epidemiologic research. There is an ethical obligation to let anyone involved in a surveillance program know if they are positive for the infection and to ensure appropriate medical and public health actions are available.

Recent Reports on Testing Strategies

There have been excellent reports elsewhere that cover the development of national and regional testing strategies, as well as comprehensive databases that summarize current testing technologies. These include reports from The Rockefeller Foundation and the Duke University Margolis Center for Health Policy that provide tools for policy makers to assess testing approaches and outline legislative and regulatory steps toward a national testing strategy, respectively;^4,5 an interactive tool from the Brown University School of Public Health and the Harvard Global Health Institute;⁶ and a frequently updated list of diagnostic tests that have received EUA.⁷

Many new diagnostic tests are in various stages of development. The RADx program at NIH, for example, is investing $1.5 billion in better diagnostic tests for COVID-19. Importantly, this program supports the full sequence of development and manufacture of improved diagnostic tests, more in the style of private venture, rather than stopping at support of fundamental or early-stage research. As the world learns more about SARS-CoV-2, and as current tests become more available and new tests are developed, testing strategies will change. This rapid expert consultation focuses on selected aspects of currently available laboratory tests, as listed below.

__________________

Available Tests for SARS-CoV-2

Worldwide, hundreds of molecular, antigen, and antibody tests have been developed, and they vary in their sensitivity and specificity, repeatability, reliability, availability, cost, and time to completion. In each kind of test, there are some that can be performed at the POC, while other tests require samples to be sent to a sophisticated laboratory. As a result of the rapidity of the SARS-CoV-2 outbreak, many of the tests described here have been granted EUA based on small validation studies with non-standardized samples. The establishment of reference testing data and need for field performance evaluation are discussed in Section 2. This section will give an overview of the different types of tests, and the reader is referred to Carter et al. (2020) for a detailed description of molecular, antigen, and antibody tests.⁸ In addition to the time required to perform tests, laboratories also vary in their ability to return results in a timely fashion due to variations in their workload, and the availability of the materials, machines, and personnel needed to perform the tests. Interpretation of a test result also depends on the prevalence of the population where the tests are carried out and relies on prior probability, or the likelihood of infection based on clinical assessment and recent patient history (e.g., participation in high-risk activities).

Molecular/Genomic Tests

Genomic tests detect the presence of portions of the SARS-CoV-2 genome, which is composed of RNA. Because the virus causes a respiratory disease (COVID-19), samples from a patient are typically taken from the respiratory tract. Most tests take nasopharyngeal (NP) swabs as their sample type, but samples from the anterior nares, mid-turbinate, or oropharyngeal areas are also accepted by many tests; these samples are taken by trained professionals in clinical settings or at testing stations. Tests have also been validated to use samples as simple to obtain as saliva or nasal swabs; these samples in some cases can be collected by persons at home, either alone or under supervision through a telehealth provider, and sent to a laboratory by mail or courier.

Antigen Tests

Antigen tests detect another portion of the SARS-CoV-2 virus, the protein coat that surrounds the RNA genome. As such, antigen tests are intended to detect the viral presence in symptomatic individuals. Confirmatory follow-up testing by more sensitive reverse transcription polymerase chain reaction (RT-PCR) methods is recommended in high-risk or suspected cases who receive a negative antigen test result. Likewise, the Centers for Disease Control and Prevention (CDC) currently considers a positive antigen test result in asymptomatic patients with low exposure risk as a presumptive case and recommends a confirmatory test by RT-PCR. Like molecular tests, antigen tests are performed on samples obtained from the respiratory tract, for the same reasons explained above. There are (as of early October 2020) five antigen tests authorized for use in the United States. All are performed with small instruments or devices that can be used at the POC or in laboratories, and all five take nasal or NP swabs as their input sample.

__________________

CRITICAL AREAS FOR CONSIDERATION

1. RT-PCR Tests Continue to Have an Important Diagnostic Role

Diagnostic tests based on RT-PCR were the first to be created and approved for COVID-19 diagnosis using samples obtained by health care workers or through self-collection.^9,10 RT-PCR diagnostic testing for a range of applications has been developed in commercial testing facilities, university laboratories, smaller biotech companies, etc. Overall, RT-PCR is a highly useful indicator of infection early in the infection, when decisions about care of the patient and limiting transmission are most important. RT-PCR and other genomic test technologies such as loop-mediated isothermal amplification and next-generation sequencing (NGS)¹¹ are highly specific, meaning a positive test is extremely rare in the absence of infection. In theory, genomic tests can return results within hours, although transport of samples, backlog due to high volume, and laboratory logistics can add hours or days before results are reported. Test sensitivity is variable in real-world settings, perhaps related to the vagaries of sample collection or the dynamics of viral load at different stages of the infection.

In general, a positive RT-PCR indicates a person has or has had infectious virus, but does not indicate how contagious the person is at the time that the sample is taken. Tracing of previous contacts may be needed even if the person who tested RT-PCR positive is not infectious at the time the test result is obtained. But fragments of viral RNA can persist long after viable virus has been cleared.¹² When a positive RT-PCR test is obtained 2 weeks or more after an initial infection but not corroborated with the detection of infectious virus in cell culture assays, this may indicate that either the virus level is below the detection limit of culture assays or that the RT-PCR test is detecting remnants of viral RNA rather than intact virus.¹³ Public health experts have questioned the relevance of RT-PCR tests conducted weeks or months after some individuals have recovered and remained symptom-free, suggesting that a large number of those who “tested positive” may not be contagious.¹⁴ In recognition of possibly persistent positive PCR results after an individual is no longer contagious, CDC guidelines no longer recommend that those infected should remain isolated until two sequential, negative RT-PCR results are recorded.

__________________

The current recommendation from CDC allows isolation to be discontinued at “10 days after symptom onset and resolution of fever for at least 24 hours.”¹⁵

Importantly, it is not possible at present to determine how infectious someone is from RT-PCR results. The problem is a lack of knowledge and data needed to correlate infectiousness to positive or negative test results: to do so, the amount of replicative virus in an individual must be quantified and the robustness of this measure as a surrogate for infectivity must be established, which has not yet been achieved. The relative amount of virus in a sample can be estimated from a RT-PCR test by the number of amplification cycles needed to increase the amount of viral RNA in the sample to a detectable level. The fewer cycles required (i.e., a lower “cycle threshold” or “Ct” value), the larger the viral load in the sample and the more likely the individual is to be infectious.¹⁶ However, Ct values are not routinely reported with clinical results, and the current CDC guidelines recommend against using Ct values to assess infectivity of individuals.¹⁷ Preliminary reports have provided good agreement between the correlation of Ct values and positive viral culture,^18,19 but there have not yet been enough studies conducted under standard clinical diagnostic conditions that take variability in sampling and testing into account to establish confidence. It is important to note here that RT-PCR assays may be qualitative or quantitative. Qualitative RT-PCR may be used as a detection assay, but Ct values alone cannot be used as a quantitative measure of viral load without reference standards of known viral loads.²⁰ The interpretation of quantitative RT-PCR is dependent on a robust standard curve, and the assay is susceptible to batch and instrument-dependent variations, necessitating careful calibration and complicating comparisons between tests. While Ct values provide valuable context to the diagnostic read-out, they should be interpreted with caution for all of the reasons noted above.

Despite its application and importance for individual diagnosis and population surveillance or reassurance so far, RT-PCR alone may not be able to handle the testing capacity needed in the long term. Notably, while large, national commercial laboratories have steadily augmented their RT-PCR testing capacity since the start of the pandemic,²¹ the private sector has found bottlenecks in increasing the centralized RT-PCR testing capacity to the degree needed to meet

__________________

national demands, which ultimately will require developing additional strategies.^22,23,24 The logistics of sample collection, transport to a central laboratory, and the testing and return of results lead to delays. Many essential testing materials (e.g., reagents, nasal swabs, transport media, etc.) are in short supply, and testing backlogs, especially in areas experiencing high rates of infection, cause days-long turnaround times for test results.²⁵ The relatively high cost of commercial RT-PCR tests further inhibit their sustained widespread use. These shortcomings impede the vital task of differentiating individuals who are actively infectious from those who are not currently infected. Without timely and accurate methods to identify infectious individuals, relaxing physical separation restrictions risks enhanced disease transmission and additional deaths, fueling an ongoing cycle of reopening and lockdown. While methods to improve the RT-PCR workflow and mitigate reagent and supply shortages are under active investigation,^26,27,28 complementing the use of standard RT-PCR diagnostics with other testing technologies, as appropriate to the use case, may decrease the bulk of the burden on the central testing laboratories in the short term.

Public health experts and epidemiologists are among many who have called for the development of a SARS-CoV-2 testing strategy aimed at infection control and containment. One approach is to shift from the reliance on laboratory-based RT-PCR diagnostics to deploying tests that are based on different technologies, can be performed outside of centralized clinical laboratories, and have the fast turnaround time needed for infection control in the community.²⁹ These rapid screening tests would complement the clinical diagnostic infrastructure in use right now. In late July 2020, FDA issued guidance for the performance of “non-laboratory” SARS-CoV-2 diagnostic tests.³⁰ At present, FDA is using RT-PCR as a gold standard of sensitivity and specificity thresholds for comparing diagnostic tests and requiring rapid screening tests, for

__________________

example, those that detect viral antigen, to be concordant with those based on PCR and detect nucleic acid.³¹ Rapid diagnostic tests often have intended applications that are distinct from that of laboratory RT-PCR tests; for instance, tests that are less expensive and provide rapid read-out may be advantageous for repeated, frequent screening purposes in community surveillance for infection outbreaks, whereas laboratory RT-PCR tests may be suitable for targeted or diagnostic uses.³² Therefore, it is important to consider the context of the testing strategy when evaluating their sensitivity and specificity.³³ A recent report, which has yet to be peer reviewed, suggests that the results of a rapid test that detects viral antigen may more closely reflect true viral load (i.e., replicative virus) compared with RT-PCR assays. As more evidence is needed to compare the results of antigen and RT-PCR tests with viral load, it is prudent to consider a cautious approach in using RT-PCR as the standard comparator for the approval of new antigen-based diagnostic tests.³⁴ Similarly, as highly sensitive methods such as bioluminescence may be developed for direct detection of viral RNA, results on current RT-PCR technologies may not be suitable as a gold standard.

2. Development of Reliable, Inexpensive Point-of-Care Diagnostic Tests That Rapidly Report Results Are Needed for Widespread Use

Types of POC Diagnostics Developed to Date

Two types of rapid read-out, POC tests have been developed. NAATs can be carried out at the point of service, such as in physicians’ offices, clinics, and nursing homes. Abbott’s ID NOW and Cepheid’s Xpert® Xpress tests were among the first POC NAATs to be granted EUA status. Both are instrument-based tests executed on mobile platforms. Tests that use the clustered regularly interspaced short palindromic repeats (CRISPR) technology are similar to conventional NAATs but use a different detection method that depend on recognition of the viral RNA by a CRISPR enzyme followed by collateral degradation of a signaling RNA sequence, rather than amplification of nucleic acids. The second type of rapid, POC tests is based on the detection of viral antigen. Antigen tests provide fast turnaround times and are inexpensive and relatively easy to manufacture at scale. Such tests may be useful for detecting asymptomatic individuals who may be carrying and able to transmit the virus. See Appendix A for an expanded discussion on the interpretation and limitations of molecular and antigen tests. FDA recently updated its EUA submission template for antigen tests to note that if a test is intended for POC use, the submitter should include data that demonstrate that non-laboratory personnel can use it accurately, and if it is intended to be used in asymptomatic individuals, the submitter should include a clinical study in that population comparing it to another assay.³⁵ The guidance provided through the EUA

__________________

template is not compulsory for submission, but signals the agency’s increased attention in POC and asymptomatic test use. Abbott recently received EUA status for BinaxNOW™, an antigen-based test that does not require an instrument to read and has been reported by the company to provide results in 15 minutes. The results can be displayed on users’ smartphones to enable those who test negative to display a “temporary encrypted digital health pass.”³⁶ As discussed below, the information value of a testing strategy builds on the performance of an individual test plus the frequency with which tests can be repeated. For example, the cumulative sensitivity of a sequence of regularly repeated tests can be superior to the performance of an intrinsically more sensitive test done only once.³⁷

Verifying the Field Performance of COVID-19 Diagnostic Tests

More than 270 different diagnostic tests for COVID-19, of which seven are POC RT-PCR tests that detect nucleic acid and six are POC tests that detect viral antigen, have become commercially available in the United States as of early November 2020.³⁸ On October 7, 2020, FDA announced that it will no longer issue EUA for laboratory-based diagnostic tests and will shift to prioritizing review for POC tests.³⁹ See Appendix B for a list of independently curated databases that track diagnostic tests in various stages of development and approval. Recognizing the country’s urgent need for testing capacity at the start of the pandemic, FDA initially granted EUA status to all diagnostic test submissions based on relatively scant performance data, much of which was based on only laboratory results and developed with non-standardized, contrived samples using SARS-CoV-2 material derived from a range of sources. The vast majority of PCR tests were validated with fewer than 100 samples, often fewer than 60. Many samples were taken from people known not to be infected and then spiked with known quantities of SARS-CoV-2 virus. Most validations of newer tests are performed against comparators that themselves have similarly scant validation data. Thus, jurisdictions, hospitals, and employers are buying and using COVID-19 diagnostics tests whose actual performance (i.e., sensitivity and specificity) are not well characterized. Several independent websites have been developed to display the available data associated with COVID-19 diagnostic tests granted EUA that detect either viral RNA or antigen.⁴⁰ Very few companies have provided FDA with data verifying field performance. At the same time, FDA worked to develop a reference panel as a tool for the precise measurement of relative sensitivity between NAATs, including RT-PCR diagnostic tests. This reference panel has been available to developers of diagnostic tests that are based on nucleic acid amplification, such as RT-PCR, since May 2020. FDA reviews the results and publishes the assay performance

__________________

on a rolling basis.⁴¹ While evaluation against the reference panel is not required to submit an application for EUA, the public assay performance data may be reviewed by health care providers, laboratories, or other institutions that use these tests.

As of today, there is no equivalent reference panel for antigen-based tests. In particular, the specificity and sensitivity of rapid antigen tests in the field are critical considerations for their use but are not currently reported to FDA, and the performance of these tests for “off-label” applications that often include the screening of asymptomatic individuals has yet not been evaluated in robust studies. Erroneous application and interpretation of rapid antigen tests that generate false results can confound public health efforts and may delay the recognition of outbreaks.^42,43

It is likely that some diagnostic companies whose COVID-19 tests have been in use have already collected field performance data. The Foundation for Innovative New Diagnostics (FIND) is a nonprofit organization that promotes access to diagnostic technology, particularly in low- and middle-income countries. It has compiled data on diagnostic test performance,⁴⁴ conducted independent evaluations of molecular tests and immunoassays,⁴⁵ and maintained a list of diagnostic tests that are commercially available or in development.⁴⁶ The Center for Systems Biology at Massachusetts General Hospital has also assembled an infographic that provides a brief explanation of different test types, a list of tests that have received EUA, and the reported assay performance data from the manufacturer as well as published field reports.⁴⁷ Establishing a registry of such data is important for clinical decision making and for selecting the type of tests to use and interpreting the results in screening situations.

Role of Government and Public–Private Entities in Expanding POC Testing

Government leadership and public–private partnerships are poised to support the large-scale procurement and implementation of rapid, POC tests in the community. After FDA issued the EUA for the BinaxNOW™ test, Abbott promised to produce up to 50 million tests per month

__________________

and the federal government entered into a contract to purchase and distribute 150 million tests.⁴⁸ Roche has similarly announced 80 million tests per month capacity by year end and Becton Dickinson will have 12 million test capacity by early 2021. There is additional capacity coming online with more than 100 smaller companies planning to have capacity of hundreds of thousands to millions of tests per month by spring 2021. This is a start toward reaching a national testing capacity, estimated to be at least 30 million each week by November 2020, needed to support a reopening of education, economic, and other social activities through quickly diagnosing infected individuals and their contacts.⁴⁹ Aggressive screening of asymptomatic individuals as a preventive rather than reactive measure may require up to 14 million tests each day, and frequency as well as strategic test use will differ based on the status of the outbreak control in each situation.⁵⁰

Additional steps include a bipartisan compact created among 10 states and The Rockefeller Foundation in an effort to induce diagnostic companies to produce sufficient numbers of antigen tests to enable large-scale screening programs that would enable the resumption of social and economic activities while avoiding large-scale outbreaks.⁵¹ The members of the compact have pledged to buy POC tests when they become available and to coordinate testing protocols.⁵² Colleges, universities, and private secondary schools have relied on RT-PCR testing and are also pioneering the use of rapid, POC tests, among other technologies, to detect COVID-19 infections before they trigger outbreaks.⁵³ Additional state testing compacts could catalyze the development of more POC tests; bulk purchasers could help to maintain reasonable pricing of such tests, encourage investment into expanding production capacity to increase overall supply, and avoid the detrimental aspects of states competing against each other while using tests to meet public health needs. It is important to support the expansion of diagnostic testing capacity with actionable public health measures such as contact tracing, wearing face masks, physical distancing, and vigilant hygiene practices and support mechanisms to enhance compliance with isolation and quarantine. Environmental surveillance methods, knowledge, and data sharing (e.g., testing protocols and practices) may also lead to more effective approaches to contain viral spread.⁵⁴

__________________

A major public health issue associated with the widespread use of POC antigen tests is the need to maintain an accurate understanding of the epidemiology of COVID-19 across the population through monitoring both positive and negative SARS-CoV-2 test results. Systems need to be in place to ensure reporting of rapid POC test results to public health authorities, especially if the tests are conducted by employers, universities, etc. In addition, a positive antigen test result is currently classified as a probable case under the current COVID-19 case definition, and there is no nationwide standard requirement for reporting probable cases. Thus, the numbers and location of individuals who are infected may not be documented, leading to an erroneous understatement of the geographic spread and extent of the pandemic. Such inaccurate situational awareness could, in turn, cause testing resources to be diverted from high-risk areas or allow infection transmission to go unchecked.⁵⁵ The lack of a shared data system that records the total number of tests carried out also precludes the ability to verify the effectiveness of infection control strategies by tracking the number and percentage of positive tests in different regions.

Finally, despite efforts to correlate live viral load to antigen or RNA level,⁵⁶ there remains a lack of knowledge needed to allow more accurate identification of the presence of infectious virus and inform regulatory standards for new COVID-19 diagnostics. Studies to fill this knowledge gap would require a BSL-3 laboratory and government oversight and verification.

3. Requirements for Testing to Meet Public Health Needs

Performance Trade-Offs in Screening Test

The goals for testing can be for diagnosis of individual cases, to provide assurance for participation in social activities such as school or work, or to perform population-level surveillance for the prevalence of infection. The target threshold for the sensitivity and specificity of the test may vary for each scenario, as discussed below. Both nucleic acid and antigen POC tests have been reported in developmental studies to have specificity comparable to that for standard RT-PCR tests, although the true specificity in field use may be lower due to factors such as variations in sample collection, operator skill, or off-label use.⁵⁷ Though they are generally less sensitive than RT-PCR tests, tests that have received EUA so far are designed to diagnose symptomatic individuals, who are likely to have higher viral loads, compensating to some extent for the test’s lower sensitivity. Their performance in the field can also be augmented by more frequent testing. For example, recent models have shown that the frequency and result turnaround time may be more important than test sensitivity for effective outbreak control when

__________________

testing is performed as part of community surveillance.^58,59 Such tests are also generally cheaper than PCR and provide rapid results (within hours)—performance traits that are favorable for screening applications. Indeed, the modeling of four testing strategies found that repeated population-wide screening using a test with modest sensitivity may decrease morbidity and mortality.⁶⁰ Such widespread and repeated screening of the population necessitates a substantial increase in testing capacity that may be difficult to meet. In the absence of sufficient national capacity, strategies to prioritize testing in different communities may be guided by multiple factors that include the current infection incidence rate. In general, when the prevalence of infection is extremely low in a community, test specificity is important in reducing the proportion of false positives; when the prevalence of infection is higher, better test sensitivity becomes more important as it avoids false negatives. In communities and subgroups or subpopulations with low infection rates, screening with tests that have inadequate specificity could lead to follow-up testing of large numbers of initial, false positive results and place a heavy burden on the diagnostic system, thus negating the purpose of screening in the first place.

Augmented Testing Strategies

Pooled testing is an approach intended to conserve test resources without sacrificing accuracy where infection prevalence is low. One procedure is called split pool testing and involves “halving” steps. In this approach, a cohort of samples that tests positive is split into two subpools of the same size that are each re-tested. The process is repeated until the positive individual sample(s) is identified. If the sample cohort tests negative, then the test is repeated once on the same cohort to confirm the negative result.⁶¹ This approach may be more accurate and efficient than a frequently cited “Dorfman Protocol” that specifies follow-up testing of each individual sample after a positive pool test.⁶² The appropriateness of pooled testing and the choice of protocol should take into account the test positivity rate in the area, as well as the technical feasibility of the responsible laboratory.⁶³

Split pool testing may be feasible and useful when laboratory technical talent is available and prevalence of infection is low. Reductions in test sensitivity are a concern with split pool testing. FDA noted that it has “seen highly variable results even on the same platforms in different laboratories. We believe the science is still evolving[…].”⁶⁴ FDA guidance recommends a

__________________

positive predictive value of 85 percent between pooled and individual tests and the implementation of a plan to monitor local test positivity rates. The RADx program has awarded funds to companies for the purpose of developing pooled testing protocols using NGS diagnostics.⁶⁵

Implementation of a national wastewater surveillance program may augment individual COVID-19 case testing by providing another, possibly cost-effective, layer of surveillance to provide early warning of the resurgence of virus in a congregate residential setting or a community. The recognition that COVID-19 infected persons shed virus in stool, even before symptoms manifest, is the basis of these proposals.⁶⁶ Wastewater-based epidemiology is not a new idea. The goal of the approach is to detect COVID-19 appearance and fluctuations over time, possibly offering actionable evidence to guide “reopening” of communities or to initiate more intensive testing of facilities.⁶⁷ It would yield limited value when there is a lot of virus circulating, but when levels are low, routine testing of wastewater of specific locations (e.g., dormitories, chronic care facilities, prisons, etc.) could provide reassurance, if negative. A positive result would trigger testing of individuals. It is a form of pooled testing.

It may or may not prove feasible to use quantitative measures of viral RNA, such as genome copies per liter (using RT-PCR tests), to estimate the number of people infected in these samples. Other countries and some states have reportedly initiated such programs.⁶⁸ Several U.S. municipalities and universities, as well as overseas jurisdictions, have employed wastewater analysis as a means of SARS-CoV-2 surveillance in communities with low infection rates. Wastewater surveillance has been used to detect and monitor the emergence of poliovirus as part of eradication campaigns, but that has built on years of method development. Protocol validation and data interpretation methods for SARS-CoV-2 are areas of active research. This surveillance approach for COVID-19, while promising, remains in the early stages of development.⁶⁹

4. COVID-19 Diagnostic Testing Using Next-Generation Sequencing Can Offer a Sensitive and Specific Test with High Throughput

NGS offers a highly sensitive and specific test modality with the possibility of providing extremely high throughput rates. Some companies and laboratories have developed COVID-19 testing capacity using NGS that can test up to 10,000 samples at a time with a turnaround time to obtain results in 24–48 hours. While NGS diagnostics must be carried out in a centralized laboratory setting, similar to RT-PCR-based diagnostic tests, their high throughput and rapid

__________________

turnaround may ameliorate some of the drawbacks of currently available RT-PCR tests. One benefit of widespread NGS diagnostic testing is the collection of viral genome data that may shed light on the epidemiology of the virus’ transmission and mutation rate.⁷⁰ The major logistical hurdles associated with NGS are the time and process needed to collect and deliver enough individual samples to make NGS testing cost-effective and to yield acceptable turnaround times. The transition of the bulk testing burden from traditional RT-PCR testing facilities to NGS testing systems has not yet occurred, and the promise of rapid turnaround time has not been validated. At this time, whether delays in traditional RT-PCR attributable to the transport, processing, and reporting of large sample volumes testing will similarly burden the turnaround time for NGS testing remains unknown.

One Example of an NGS Diagnostic Test System

The necessary NGS facilities are widely available, and a handful of companies (Illumina, Fulgent, Helix) have been granted EUA status for NGS diagnosis of COVID-19. Other companies have also submitted applications to FDA. NIH’s RADx program awarded grants to several companies to advance aspects of NGS-based testing services. One example of how an NGS diagnostic testing system can be implemented is the Concentric program at Gingko Bioworks. This program received funding from RADx for pandemic response activities that include repurposing Gingko’s existing research space into a Biosafety Level 2, Clinical Laboratory Improvement Amendments–certified laboratory that performs ultra-high throughput sequencing of COVID samples for sequencing and diagnosis.⁷¹ Gingko’s customers in this enterprise are large organizations that collect saliva samples at their own work sites and pack them in boxes for delivery to Gingko. It is then fairly straightforward to process the samples in large batches.

SUMMARY AND REMAINING GAPS

• The majority of diagnostic testing for SARS-CoV-2 is carried out using centralized RT-PCR-based tests that may not scale to the throughput, turnaround time, and cost-effectiveness needed for infection containment in the community.
• There are a number of POC diagnostic tests based on innovative technologies that are in various stages of development and may complement the existing RT-PCR diagnostics system.
• Early EUA for laboratory RT-PCR as well as POC tests has been granted based on validation with few, non-standardized, and contrived samples. There is little information on the field performance results for diagnostic tests. This needs to be collected and evaluated to obtain realistic sensitivity and specificity data that are necessary to guide decisions for deploying these tests in screening or surveillance purposes.

__________________

• New POC diagnostic tests developed for uses that are distinct from laboratory-based RT-PCR tests will need to be evaluated in the course of regulatory approval and in the post-market phase of use.
• Pooled testing and wastewater surveillance are two strategies that need further development but can provide advantages beyond individual diagnostic testing methods.
• Diagnostic tests based on NGS may offer a centralized testing system that can speed throughput and turnaround time for certain use cases.

APPENDIX A

Additional Notes on COVID-19 Diagnostic Tests

1. Interpretation and Limitations of Molecular and Antigen Tests

A positive molecular/genomic test result indicates that RNA from SARS-CoV-2 was present in the body at the time the patient sample was collected. A positive test is usually interpreted as indicative of active and transmissible COVID-19 infection. However, samples from some infected people may test negative, because they were tested before enough virus had replicated at the sample collection location in the patient to trigger the test. Also, sometimes samples are not properly collected, and results may differ when samples are taken from different sites in the body (e.g., NP swabs, throat swabs, and sputum may not all harbor the same amount of virus in a person). Molecular tests are highly accurate. If sufficient viral RNA is present and a sample is properly obtained, these tests will identify it 98–99 percent of the time. However, it is important to note that viral RNA can in some cases persist for weeks after a person has recovered from infection and viable virus can no longer be detected. At this time, molecular tests provide no other information about a patient’s exposure history, their symptoms, or their current immune status. Robust correlation of molecular test results with infectious viral load have not been reported and a lack of standardization increases the difficulty in comparing results across laboratories and methods. Therefore, while an initial positive result will trigger isolation measures, molecular test results should seldom be the sole basis for making clinical decisions such as treatment course.

Like a positive molecular test result, a positive antigen result can be theoretically interpreted as indicative of active and transmissible COVID-19 infection, and the same cautions in interpretation apply. Samples from some infected people may test negative, because they were tested before enough virus had grown in the patient to trigger the test. Also, sometimes samples are not properly collected, and results may differ when samples are taken from different sites in the body (e.g., NP swabs, throat swabs, and sputum may not all harbor the same amount of virus in a person). Due to specificity concerns of rapid antigen tests, presumptive cases should not be cohorted with confirmed cases before the outcome of the follow-up RT-PCR test. Likewise, CDC currently considers a positive antigen test result in asymptomatic patients with low exposure risk as a presumptive case and recommends a confirmatory test by RT-PCR. Molecular tests provide no other information about a patient’s exposure history or their current immune status. Like molecular tests, therefore, antigen test results should seldom be the sole basis for making clinical decisions.

Limitations of genomic and antigen tests. The value of a molecular or an antigen test result decreases quickly with time after samples are collected. A person who is positive for SARSCoV-2 but does not know it may spread the virus widely. Conversely, a negative result that returns many days after sampling is of limited value because patients who are negative when sampled may become infected soon after. Therefore, when choosing a test, total turnaround time (time between sampling and return of the result to the patient) must be a primary consideration. In areas of the country where COVID-19 prevalence is high, the demand for testing is correspondingly high, and supplies, instruments, and personnel resources may not keep pace with demand.

2. Antibody Tests

Antibody tests determine whether a person has developed antibodies against SARS-CoV-2 as a result of exposure or infection. The tests come in many forms; some antibody tests require complex machines installed in laboratories, other tests use less complex hardware and can be run in doctors’ offices, and some are as simple as home pregnancy tests and can be used by individuals. All of the tests use whole blood, serum, or plasma as their input sample. The simplest tests can be performed with a few drops of blood from a fingerstick.

Antibody tests are typically less expensive to manufacture and perform than molecular tests. However, they have many drawbacks. They are, generally speaking, less sensitive and specific than molecular tests.⁷² Their interpretation is also difficult. Even if an accurate positive result is obtained (“this patient has antibodies against SARS-CoV-2”), the significance of a result is currently not known. We do not fully understand the relationship between having antibodies to SARS-CoV-2 and resistance to infection, although there is now good evidence that neutralizing antibodies provide protection against infection.⁷³ The presence of antibodies to SARS-CoV-2 does not always correlate with better clinical outcomes. For example, there are documented cases of patients who have recovered from mild cases of COVID-19 who have had generated weak antibody responses, and patients who have had severe disease while mounting very strong responses. We do not know whether antibodies collected from a recovered patient can be given to another patient to treat a COVID-19 infection (those studies are currently under way). We also do not know whether antibodies confer resistance to re-infection.

However, the presence of antibodies is correlated with past exposure or infection with SARSCoV-2, and therefore antibody tests are valuable tools for surveying the population for that exposure. An important distinction to be made between serologic tests (detection of antibodies from host response) and diagnostic tests (detection of viral nucleic acid or antigen) is that serologic tests should not be used to diagnose active COVID-19 cases. Antibody tests are not well suited to diagnosis of new infection in exposed or symptomatic patients. Well-designed studies can help shed light on how the virus has spread geographically, what a minimum estimate

__________________

might be for the background rate of infection, and help identify patients whose immune responses could be studied to learn more about immunity to the virus. The low cost and ease of use of some antibody tests make them particularly well suited for these epidemiological applications. However, investigators in such studies must take care to use tests whose performance has been well characterized to minimize statistical errors in analyzing study data.

APPENDIX B

Sources for More Information About Tests

Additional information on COVID-19 tests and testing can be found in the following sources:

• CDC has drafted guidance for the use of COVID-19 tests.⁷⁴
• The University of Minnesota Center for Infectious Disease Research and Policy provides guidance on testing strategies for using molecular and antibody tests to detect the virus in both symptomatic and asymptomatic people.⁷⁵
• Carter et al. (2020) provide a detailed description of the features of molecular, antigen, and antibody tests in a recent publication.⁷⁶
• In a separate publication, Weissleder et al. (2020) further break down COVID-19 tests and their applications.⁷⁷

There are many COVID-19 tests in development in the United States and around the world at different stages of regulatory approval in the U.S. and other markets. There are compilations of data on COVID-19 tests that readers may find useful:

• The trade group news provider 360Dx maintains an alphabetized list of tests for diagnostic and clinical use and their regulatory status in the U.S., European, and Asian markets.⁷⁸
• In-Q-Tel’s B.Next developed and released a searchable database of molecular tests that have received EUA from FDA. The data were sourced from vendor information provided in documents to FDA.⁷⁹

__________________

• Massachusetts General Hospital prepared a downloadable infographic that includes background on the viral infection, diagnostic principles, and a summary of COVID-19 diagnostics with EUA from FDA.⁸⁰
• Arizona State University compiled a comprehensive database of tests that is searchable by a number of parameters, including regulatory status, detection target type (e.g., viral RNA, antigen, specific antibody, etc.), analysis location (e.g., laboratory or POC), company, type of specimen, and sensitivity and specificity reported in the EUA records.⁸¹
• The Center for Health Security at Johns Hopkins University maintains a COVID-19 testing webpage with a wealth of information, including a list of serology tests.⁸²
• The Joint Research Centre of the European Union has a compilation of COVID-19 test information that have received CE marking.⁸³

APPENDIX C

Authors and Reviewers of This Rapid Expert Consultation

This rapid expert consultation was prepared by staff of the National Academies of Sciences, Engineering, and Medicine, and members and outside experts on behalf of the National Academies’ Standing Committee on Emerging Infectious Diseases and 21st Century Health Threats: Jeffrey S. Duchin, University of Washington and Public Health Seattle & King County, Washington; Harvey V. Fineberg, Gordon and Betty Moore Foundation; Tara O’Toole, In-Q-Tel; David R. Walt, Brigham and Women’s Hospital; Mary E. Wilson, University of California, San Francisco, School of Medicine.

We are grateful to the Report Review Committee of the National Academies and to individual reviewers who provided many valuable corrections, comments and suggestions on an earlier draft. We extend gratitude to the staff of the National Academies, in particular to Lisa Brown, Julie Liao, Julie Pavlin, and Andrew M. Pope, who contributed research, editing, and writing assistance.

Harvey Fineberg, chair of the Standing Committee, approved this document. The following individuals served as reviewers: Keith Jerome, University of Washington, Fred Hutchinson Cancer Research Center; Alexander McAdam, Boston Children’s Hospital, Harvard Medical School; Jennifer Nuzzo, Johns Hopkins Center for Health Security; and Christina Silcox, Duke University Margolis Center for Health Policy. Bobbie A. Berkowitz, Columbia University School of Nursing; Ellen Wright Clayton, Vanderbilt University Medical Center; and Susan Curry, University of Iowa, served as arbiters of this review on behalf of the National Academies’ Report Review Committee and their Health and Medicine Division.

__________________

This activity was supported by a contract between the National Academy of Sciences and the U.S. Department of Health and Human Services’ Office of the Assistant Secretary for Preparedness and Response (75A50120C00093). Any opinions, findings, conclusions, or recommendations expressed in this publication do not necessarily reflect the views of any organization or agency that provided support for the project.

Copyright 2020 by the National Academy of Sciences. All rights reserved.