We now recognize that a greatly expanded capacity for mass screening will be essential to managing life and economic activity in an ongoing severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic. We also know that identifying asymptomatic people with acute SARS-CoV-2 infection requires frequent testing and results that are available quickly enough to act.1 Rapid antigen tests that can do this have been developed and their production is increasing, but the scale of capacity needed for effective screening is truly daunting. For context, consider that the National Institutes of Health estimates that the US could achieve capacity to run 42 million tests per week in 2021.2 Then consider that there are roughly 50 million children attending US public schools alone; even once-weekly screening in these schools would consume all the diagnostic and screening tests we can produce.
A parallel and complementary approach promises to immediately expand existing capacity in regional or national central laboratories by using high-speed automated pooling and polymerase chain reaction (PCR) to screen many specimens quickly. In the absence of strong outcomes data for this novel coronavirus, the US Food and Drug Administration, public health agencies, and industry laboratories have limited their pooling programs to using small (4-8 specimen) pools.3,4 Programs in Israel, China, and some US colleges have reported using larger pools. As yet, few laboratories have implemented the efficient (eg, 20 times capacity)5 algorithms that use large pooling matrices or sequential testing stages. The effects of complex algorithms on viral detection are difficult to know unless one can anticipate both the rate of positive specimens and the precise distribution of virus concentrations that will occur in the testing population.
Polage and colleagues6 revive a moribund discussion on pooling with a dose of hard data. They demonstrate a new online tool to estimate and compare outcomes of SARS-CoV-2 PCR testing strategies. Their approach samples a carefully measured and curated distribution of actual viral loads from 2984 consecutive SARS-CoV-2–positive individuals tested by diagnostic and screening programs in North Carolina. The results show clearly that for any given positivity rate greater than 1%, the number of false-negative results decreases as pool size increases; in other words, SARS-CoV-2 RNA PCR will be more sensitive at screening large pools than small pools of 4 or 8 (refer to Figure 2A and B in Polage et al6). This is counterintuitive because RNA from a single positive specimen is always diluted in bigger pools. Given the high positivity rates seen by SARS-CoV-2 screening programs,7 however, large pools that have 1 positive specimen will frequently contain 2 positive specimens, and with the wide distribution of viral loads, this usually adds sufficient RNA to rescue detection of any single low-viral-load specimen in a double-positive pool. This rescue effect dominates over the dilution effect in the data from both symptomatic and asymptomatic populations.
These are rudimentary considerations that force us to reconsider pooling strategies very broadly. The fact that large pools can be either more or less sensitive than small pools, depending on context, has startling implications. It tells us that multistage algorithms will perform with different sensitivity when testing different specimens. This is because these algorithms progress by stages, from bigger to smaller pool sizes; in this simple arrangement, whichever stage yields lower detection will determine the sensitivity for the whole algorithm. For SARS-CoV-2, this can change; when there are more than 1% positive specimens, a 25:5:1 algorithm will have sensitivity similar to that of a 5:1 minipool scheme. When the positivity rate decreases, the same algorithm will become somewhat less sensitive but ultraefficient.5 The ability of these algorithms to shift gears without external adjustment could make these extremely advantageous for large regional or national laboratories, which know that a stream of specimens from low-risk, frequent-testing programs will occasionally be interrupted by clusters of positive specimens from hot spots. As laboratories and regulators begin to reevaluate screening strategies, they will need to consider the impact of such case clustering on pooled testing performance. They should also consider viral load or antibody status criteria for acute SARS-CoV-2 for screening programs, because the low viral loads missed by pooling are nearly always from individuals with nonacute, antibody-positive infection.7
The most efficient pooling algorithms are not always easy to implement, particularly if laboratories are not already experienced with the high-speed robotic testing needed to generate results from complicated algorithms in a few hours. Just as important, laboratory speed will not translate to fast turnaround times for individual clients unless programs can receive, log, and prepare a massive volume of new specimens for them to test.
Although these challenges are very real, they are exactly what state, regional, and national public health programs know how to do best. Countries that are facing new waves of SARS-CoV-2 infection need ways to multiply their existing capacity and to do so quickly. With reliable data to guide implementation of highly efficient algorithms, the laboratories that already have pooled testing capability need only the right directives, funding, support, and regulation. A new mass screening solution is at hand and it is past time to act.
Published: December 10, 2020. doi:10.1001/jamanetworkopen.2020.31577
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2020 Pilcher CD. JAMA Network Open.
Corresponding Author: Christopher D. Pilcher, MD, Department of Medicine, Division of HIV, InfectiousDiseases, and Global Medicine, University of California, San Francisco, 1001 Portero Ave, San Francisco, CA 94110 (email@example.com).
Conflict of Interest Disclosures: None reported.
Additional Contributions: Michael Busch, MD (University of California, San Diego), and Christopher Holden-Wingate (undergraduate at University of California, Davis) provided helpful comments on the manuscript. They have no relevant conflicts and were not compensated for this contribution.
Pilcher CD. A Data-Driven Rationale for High-Throughput SARS-CoV-2 Mass Screening Programs. JAMA Netw Open. 2020;3(12):e2031577. doi:10.1001/jamanetworkopen.2020.31577
Customize your JAMA Network experience by selecting one or more topics from the list below.