School openings will bring hundreds to thousands of individuals together for long periods of time from communities with substantial coronavirus levels. It is inevitable that numerous individuals, students and staff, will come to school while infected and contagious and expose numerous others. Given its fast spread from primarily asymptomatic individuals, if this potential viral spread is not effectively mitigated, these schools have the potential to become epicenters of super-spread events endangering the school and sparking or exacerbating community-level outbreaks.
To invoke a forest fire analogy, consider each infected individual as a "campfire". The key to safe school reopening is to ensure that the these campfires do not lead to a "wildfire" that burns through the entire forest. This can be done using two primary strategies:
Eliminate "combustible conditions" that increase risk of viral spread
"Contain" any local fires as they arise
This is why school districts are formulating and implementing various mitigation measures including reducing class sizes, imposing physical distancing in classrooms, requiring mask wearing, checking ventilation and HVAC systems, and eliminating or limiting large group settings where many students are close together, etc. These steps, crucial to reduce the risk of viral spread, remove "combustible" materials to ensure that the conditions are not ripe for the fires to spread.
While these measures will help keep the virus at bay, they are not sufficient. It is impossible to eliminate all risk of exposure, especially with groups of people together in enclosed rooms for long periods of time, and given the difficulty of achieving full compliance in distancing and mask-wearing (if we can't get close to full compliance among adults, do we really think we can with children and teens?). It is inevitable that some of these "campfires" will spread to the surrounding brush and produce "local fires". Thus, it is necessary to devise effective strategies to contain local outbreaks as they arise to prevent them from spreading throughout the school and into the community. To keep these "local fires" from becoming "wildfires."
The key tools of containment include testing, contact tracing, and isolation. Anyone with symptoms needs to be tested, and all individuals with whom they have had close interaction need to be contacted and also tested, and individuals at greatest risk for spreading the virus quarantined. Without fast testing, these containment steps become impractical or even infeasible.
Infected individuals typically demonstrate symptoms between 2 and 14 days after exposure, with a mean of 5-6 days. Gold standard SARS-CoV-2 tests are PCR tests measuring levels of SARS-CoV-2 specific messenger RNA levels, and require taking a sample and sending it to a laboratory for processing. The processing can take anywhere from 2 days to 2 weeks to get results, with longer delays typical in communities with higher viral levels. Thus, by the time someone finds out they are infected with SARS-CoV-2, it is anywhere from 4 days to 28 days (mean of approximately 10 days) after infection. This is far too late.
A recent study involving contact tracing of >15,000 individuals exposed to >1000 infected individuals showed that 95% of transmission events occurred between 1 week before symptoms appeared and 1 week after symptoms appeared, with a mean transmission time of 2 days before, and >60% of transmission occurring before any symptoms were observed. Thus, with the current testing lag time, by the time we know someone is SARS-CoV-2 positive, they are almost certainly already out of their contagious period, and the test results have no value whatsoever for surveillance or containment.
Under these conditions, the only way to contain the virus would be to isolate any individual with symptoms and all of those with whom they had close contact for 14 days, which would be, ummm, impractical. As we move into the fall, many other common ailments including seasonal allergies, cold, and flu will arise and have symptoms that resemble COVID-19. The quarantine of each person showing one of these symptoms and all of their contacts would produce rolling quarantines completely disrupting and eventually consuming a large portion of many schools. Stricter criteria for qualifying what combination of symptoms or how close a contact would trigger quarantine could be used and would improve feasibility, but at the expense of effective containment.
If fast tests were available at schools that could provide same-day results, the containment could be done in a much more targeted and effective way. A symptomatic individual could be tested and then their contacts only tested if they are positive. These contacts could subsequently be quickly tested, and only those individuals who are infected quarantined. This targeted strategy would be far less disruptive to the school, and have the potential to effectively contain the virus, allowing schools to stay opened indefinitely. Some presymptomatic individuals who were infected outside of school could still bring it into the school, but the rigorous contact tracing and testing would keep it from spreading too far from the source.
The good news is that we already have fast tests available for SARS-CoV-2 -- antigen tests. These tests measure active infections like the gold standard PCR-based tests, but are far less expensive and can return results very quickly on site without having to send samples out to a biological laboratory. They could even be adapted to be able to be used at home, akin to a pregnancy test, and give results in as soon as 15-20 minutes. They are broadly used for other infectious diseases including influenza, strep throat, and tuberculosis.
For SARS-CoV-2, there are currently two antigen tests that have been given emergency use FDA authorization, one by Quidel and the other by Becton Dickinson. They have been in use the past few months, primarily in hospitals and nursing homes, but unfortunately are not available for broad use. It seems the Quidel test has manufacturing bottlenecks, and the Becton Dickson device is broadly available but there are limitations in producing the testing materials. There are many companies with other fast tests under development, but the FDA has put up considerable roadblocks for approval that some have described as near impossible to get past, and removes all motivation for these companies to push ahead. The FDA has made it clear in their guidance documents that current PCR tests are the gold standard, and other tests will not be approved for use unless they are proven to be nearly as accurate as PCR tests.
The problem is that the PCR tests are inherently more sensitive than antigen-based tests and always will be. PCR stands for Polymerase Chain Reaction, which is a pathbreaking technology that allows messenger RNA to be multiplied and magnified, allowing accurate detection of even small quantities (but also requiring expensive laboratory-based processing). Antigen tests are typically based on measuring proteins, and there is no PCR-based analogs to magnify low protein levels; making measurement of proteins at lower concentrations inherently much more difficult. A recent medRxiv paper by Larremore et al. mentions that PCR tests can detect viral levels down to 1000 cp/ml (viral RNA copies per ml), while antigen tests have higher limits of detection, requiring on the order of 100,000 cp/ml to be detected, 100x higher threshold. For this reason, the antigen tests have been documented to have high false negative rates, with up to 50% of samples containing SARS-CoV-2 missed by the assays. This high false negative rate is the primary reason limiting its use.
However, Larremore et al. also mention that numerous studies suggest that samples need to have viral levels of at least 1,000,000 cp/ml before being considered infectious, so it seems likely that the 50% false negative antigen tests are from infected individuals who are unlikely to be contagious anyway. If that is the case, then antigen tests would do just as well as PCR tests in detecting infections in individuals with high enough viral loads to be contagious, while yielding results in minutes instead of weeks, bringing real-time surveillance into the realm of feasibility. This leads these authors to conclude "Surveillance should prioritize accessibility, frequency, and sample-to-answer time; analytical limits of detection should be secondary." It has been said that we want tests to be cheap, fast, and accurate, but you can only have two of the three at a time. In this case, it appears that for surveillance purposes, in a qualified fashion antigen tests can have all three, being cheap, fast and accurate ENOUGH. As we perfectionists know all too well, the perfect can be the enemy of the good, and I am afraid the FDA is caught up in this trap when it comes to SARS-CoV-2 testing.
The FDA needs to immediately consider revising standards for fast approval of antigen tests, perhaps using a criterion based on accuracy for detecting viral loads above a certain level considered most likely to be infectious. This will provide the motivation for companies developing these tests to get them quickly approved and on the market. Further, the federal government needs to get involved to assist these companies in the development, manufacturing, and coordination of the deployment of these tests around the country.
It would be incredible if every school could have the instrumentations and supplies in house for these antigen tests. If they could, then they could do real-time surveillance, and establish effective containment procedures to ensure any viral outbreak is immediately identified and contained. Given enough supplies, it might even be possible to do broader surveillance of students and staff, testing every week or two, and thus isolating infected individuals even before they show symptoms, preventing even asymptomatic spread. With this in place, schools could be safely opened and remained open, as these containment procedures would keep the virus from spreading out of control.
I believe that the development, approval, and deployment of antigen and other fast tests for active virus should be a number one priority in the USA pandemic response, and could be the key in the ability of our country to get the pandemic under control, and allow us to open up society to function at as close to normal level as possible.
I will conclude, however, by mentioning a few caveats:
Because of the high false negative rates, a negative test is not exculpatory, i.e. one can't be sure they are not infected just because they have a negative test. This is also true for PCR tests that have substantial false negative rates, but these for antigen test the false negative rate is even higher.
Societal-level pandemic management requires accurate measurements of relative infection levels over time and place, to identify emerging hot spots for more aggressive intervention. The advent of these fast antigen tests in locales such as schools, doctor's offices, and businesses could lead to an undercount of tests and cases if these results are not tied into the official counts. In principle, positive antigen tests should be followed up by a confirmatory PCR tests, which would then produce a "confirmed case" for viral tracking.
Alternatively, some suggest that positive antigen tests should be considered a "probable case", and that "probable" and "confirmed" case numbers be reported side-by-side, similar to what is done for "probably" or "confirmed" COVID-19 related deaths.
Either way, these would affect the interpretation of the testing positivity rate, so would add another level of nuance to interpretation of the testing, incidence, hospitalization and mortality data. They are already being used to some degree already, so it gives us another factor to take into account when assessing these data.
UPDATE: Over the weekend the FDA gave emergency use authorization to a new saliva-based tests developed at Yale University whose paper was posted (not yet peer reviewed) on medRxiv. This test does not require the nasopharyngeal swab used by other tests that is so invasive and technically difficult that trained healthcare professionals must administer it, and does not require any special equipment or supplies to administer. Their paper demonstrates >94% agreement with gold standard PCR tests, with detection limits of 6000-12,000 cp/ml that are reasonably close to the 1000 cp/ml of these tests and much better than the 100,000 cp/ml detection limit on antigen tests, and these tests are very inexpensive (<$5/test). They still require laboratory based processing, but do not require the nucleic acid extraction step whose extraction kits have been in short supply and a source of limited testing capacity in the past. The open source procedure outlined in their paper enable any reasonably complex laboratory to process the test using available reagents and supplies. While not being a point-of-care solution, the approval of this test could increase testing capacity to help with the surveillance and containment procedures once the number of processing laboratories are scaled up, assuming the accuracy holds up across multiple laboratories. I personally still think fast tests that could provide same-day results still have a role to play in this process.
UPDATE 8/26: The FDA just provided emergency use authorization for a fast antigen test produced by Abbott Laboratories that is $5, takes 15 minutes, requires no equipment, and has demonstrated sensitivity of 97.1% and specificity of 98.5%, that also works with a no-charge complementary phone app that allows people to display their test results to groups like workplaces or schools. They will ship 10s of millions of test in September with a remaining 50 million tests per MONTH starting October! Also, it appears that the Department of Health and Human Services (HHS), the parent organization of the CDC in the federal government, has purchased 150 million of these tests, essentially all of the supply for the remainder of 2020. It remains to be seen how they will be distributed ... my hope is that a large set of them are used for schools to enable fast testing for safe school reopening.
