{"paper_id":"4a4ab1f2-aa96-493f-a9ff-19aca13a34b9","body_text":"1 \n \nSARS-CoV-2 Viral Load in Saliva Rises Gradually and to Moderate \nLevels in Some Humans \n \n \nAlexander Winnett\n1+, Matthew M. Cooper1+, Natasha Shelby1+, Anna E. Romano1, Jessica A. Reyes1, \nJenny Ji1, Michael K. Porter1, Emily S. Savela1, Jacob T. Barlow1, Reid Akana1, Colten Tognazzini2, \nMatthew Feaster2, Ying-Ying Goh2, Rustem F. Ismagilov1* \n \n1. California Institute of Technology, 1200 E. California Blvd., Pasadena, CA, USA 91125 \n2. City of Pasadena Public Health Department, 1845 N. Fair Oaks Ave., Pasadena, CA, USA 91103 \n \n+ These authors contributed equally \n*Correspondence to: rustem.admin@caltech.edu \n \n  \nOne sentence summary \nIn some human infections, SARS-CoV-2 viral load rises slowly (over days) and remains near the limit of \ndetection of rapid, low-sensitivity tests. \n \nAbstract  \nTransmission of SARS-CoV-2 in community settings often occurs before symptom onset, therefore testing \nstrategies that can reliably detect people in the early phase of infection are urgently needed. Early detection \nof SARS-CoV-2 infection is especially critical to protect vulne rable populations who require frequent \ninteractions with caretakers. Rapid COVID-19 tests have been pr oposed as an attractive strategy for \nsurveillance, however a limitation of most rapid tests is their low sensitivity. Low-sensitivity tests are \ncomparable to high sensitivity tests in detecting early infections when two assumptions are met: (1) viral \nload rises quickly (within hours) after infection and (2) viral  load reaches and sustains high levels (>10\n5–\n106 RNA copies/mL). However, there are no human data testing these  assumptions. In this study, we \ndocument a case of presymptomatic household transmission from a  healthy young adult to a sibling and a \nparent. Participants prospectiv ely provided twice-daily saliva samples. Samples were analyzed by RT-\nqPCR and RT-ddPCR and we measured the complete viral load profiles throughout the course of infection \nof the sibling and parent. This study provides evidence that in at least some human cases of SARS-CoV-2, \nviral load rises slowly (over days, not hours) and not to such high levels to be detectable reliably by any \nlow-sensitivity test. Additional viral load profiles from different samples types across a broad demographic \nmust be obtained to describe the early phase of infection and d etermine which testing strategies will be \nmost effective for identifying SARS-CoV-2 infection before transmission can occur.   \n \nIntroduction \nAs of early December 2020, nearly one year after the first COVID-19 outbreak in Wuhan, China, there \nhave been more than 65 million cases and 1.5 million deaths glo bally.\n1 Transmission of SARS-CoV-2 in \ncommunity settings often occurs before symptom onset,2,3 putting at great risk people who require frequent \ninteractions with caregivers, such as residents of nursing home s. Better strategies for using the available \nCOVID-19 diagnostic tests are cr itically needed t o decrease overall transmission, thereby reducing \ntransmission to these vulnerable populations.4  \n \nTransmission from asymptomatic or presymptomatic individuals is  c o n s i d e r e d  t h e  A c h i l l e s ’  H e e l  o f  \nCOVID-19 infection control.\n3 In a recent epidemiologic inves tigation of 183 confirmed COVID -19 cases \nin Wanzhou, China, about 76% of transmissions occurred from ind ividuals without symptoms (either \nasymptomatic or presymptomatic). 5 Numerous transmission events originating from individuals with out \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20239467doi: medRxiv preprint \nNOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.\n\n2 \n \nsymptoms have been documented in a variety of locations, such a s dinner parties, 6 skilled nursing \nfacilities,2,7 correctional facilities,8 sporting events,9,10 religious ceremonies,6 and spring break trips.11 \nThe importance of effective testing strategies to quell transmi ssion from individuals without symptoms is \nunderscored by an outbreak at an overnight camp in Wisconsin 12 w h e r e  a  9th grade student developed \nsymptoms the day after arrival, prompting quarantine of 11 clos e contacts. All 11 contacts were \nasymptomatic and released from quarantine after receiving negat ive rapid antigen test results. However, 6 \nof the 11 went on to develop symptoms, and an outbreak ensued, with more than 100 additional individuals \n(a total of 3/4 of camp attendees) infected. In contrast, a mor e successful containment of an outbreak was \ndocumented in a skilled nursing facility in Los Angeles, where serial surveillance PCR testing was initiated \nimmediately after three residents became symptomatic and tested positive.13 PCR results prompted isolation \nof 14 infected individuals without symptoms, which limited the outbreak to a total of only 19 out of 99 \nresidents over the course of two weeks. These cases demonstrate  the value of testing strategies that can \ndetect and isolate infected individuals in the early phase in t he infection, reducing the potential for \ntransmission to others during the elicitation window (the perio d when a person is infectious but not \nisolating).14  \n \nMore than 200 in vitro diagnostics have received Emergency Use Authorization (EUA) from the U.S. Food \nand Drug Administration for identification of acute SARS-CoV-2 infection.15 These tests have a wide range \nof sensitivities. The most sensitive tests, with limits of detection (LOD) of 102-103 RNA copies/mL, include \nthe RT-qPCR assays. These tests typically involve more intensiv e sample-preparation methods to extract \nand purify RNA and most are run in centralized laboratories (wi th a few exceptions of point-of-care tests \nthat integrate rigorous sample preparation and RNA detection 15,16). At the other extreme are the low-\nsensitivity tests (LODs of ~10 5–107 RNA copies/mL), such as antigen tests or molecular tests that do not \nperform rigorous sample preparation. These tests offer tangible advantages, such as being fast (rapid antigen \ntests yield results in minutes), less expensive to manufacture, and can be deployable outside of laboratories.  \n \nRapid, low-sensitivity tests are clearly a valuable part of the overall infection control strategy; however, the \nuse of such tests as a strategy for diagnosing infected persons at the early phase of infection is controversial. \nThe U.S. Food & Drug Administration (FDA) has authorized such tests for use in symptomatic populations.  \nData in several reports suggest that such tests may miss presymptomatic and asymptomatic individuals \nearly in the infection.\n12,17 However, logical arguments have also been made,18,19 in favor of widely deploying \nsurveillance tests with “analytic sensitivities vastly inferior to those of benchmark tests.”18   \n \nLow-sensitivity tests will be equally effective to high-sensitivity tests at minimizing transmission if the \nfollowing two assumptions about the early phase of SARS-CoV-2 i nfection hold true: (i) viral load \nincreases rapidly, by orders of magnitude within hours, and (ii ) viral load reaches and sustains high levels \nduring the infectious window, such that a rapid low-sensitivity  test would have a similar ability to detect \nearly-phase infections compared with high-sensitivity tests. These two assumptions have not been tested in \nhumans. Viral load at the early phases of SARS-CoV-2 infection remains a knowledge gap necessary to \ninform the use of testing resources to effectively minimize tra nsmission. To fill this knowledge gap, and \ninform selection of diagnostic tests appropriate for identifyin g infections in the earliest phases, requires \nstudies that monitor SARS-CoV-2 viral load with high temporal r esolution (beginning at the incidence of \ninfection) and in a large, diverse cohort of individuals.   \n \nWe are conducting a case-ascertained observational study in whi ch community members recently \ndiagnosed with COVID-19 and their SARS-CoV-2-presumed-negative household contacts prospectively \nprovide twice-daily saliva samples. We are quantifying absolute  SARS-CoV-2 RNA viral load from these \nsaliva samples using RT-qPCR and RT digital droplet PCR (RT-ddP CR) assays. This article documents \npreliminary results from the study, with the complete SARS-Cov- 2 viral load profiles from two cases of \nobserved household transmission. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20239467doi: medRxiv preprint \n\n3 \n \nMethods \n \nParticipant Population \nThis study was reviewed and approved by the Institutional Revie w Board of the California Institute of \nTechnology, protocol #20-1026. All participants provided written informed consent prior to participation. \nIndividuals ages 6 and older were eligible for participation if  they lived within the jurisdiction of a \npartnering public health departm ent and were recently (within 7  days) diagnosed with COVID-19 by a \nCLIA laboratory test or were currently living in a shared residence with at least one person who was recently \n(within 7 days) diagnosed with COVID-19 by a CLIA laboratory te st. The exclusion criteria for the study \nincluded physical or cognitive impairments that would affect the ability to provide informed consent, or to \nsafely self-collect and return samples. In addition, participan ts must not have been hospitalized, and they \nmust be fluent in either Spanish or English. Individuals withou t laboratory confirmed COVID-19 but with \nsymptoms of respiratory illness in the 14 days preceding screening for enrollment were not eligible. Study \ndata were collected and managed using REDCap (Research Electron ic Data Capture) hosted at the \nCalifornia Institute of Technology. \n \nSymptom Monitoring  \nParticipants in the study completed a questionnaire upon enroll ment to provide information on \ndemographics, health factors, COVID-19 diagnosis history, COVID -19-like symptoms since February \n2020, household infection-control practices and perceptions of COVID-19 risk. Additionally, participants \nrecorded any COVID-19-like symptoms (as defined by the U.S. Cen ters for Disease Control 20) that they \nwere experiencing on a symptom-t racking card at least once per day. Participants also filled out an \nadditional questionnaire at the c onclusion of the study to docu ment behaviors and interactions with \nhousehold members during their enrollment.  \n \nCollection of Respiratory Specimens \nParticipants self-collected saliva samples using the Spectrum SDNA-1000 Saliva Collection Kit (Spectrum \nSolutions LLC, Draper, UT, USA) at home twice per day (after wa king up and before going to bed), \nfollowing the manufacturer's guidelines. Participants were instructed not to eat, drink, smoke, brush their \nteeth, use mouthwash, or chew gum  for at least 30 min prior to donating. These tubes were labelled and \npackaged by the participants and transported at room temperature by a medical courier to the California \nInstitute of Technology daily for analysis.   \n \nNucleic Acid Extraction \n \nAn aliquot of 400 µL from each saliva sample in Spectrum buffer  was manually extracted using the \nMagMAX Viral/Pathogen Nucleic Acid Isolation Extraction Kit (Ca t. A42352, Thermo Fisher Scientific) \nand eluted in 100 µL. Positive extraction controls and negative  extraction controls were included in every \nextraction batch: Positive extraction controls were prepared by  combining 200 µL commercial pooled \nhuman saliva (Cat. 991-05-P, Lee Bioscience) with 200 µL SARS-C oV-2 heat inactivated particles (Cat. \nNR-52286, BEI Resources) at a concentration equivalent to 7500 genomic equivalent units/mL, mixed with \nthe buffer from the Spectrum SDNA-1000 Saliva Collection Device . Negative extraction controls were \nprepared by combining 200 µL commercial pooled human saliva and  200 µL Spectrum buffer. Spectrum \nbuffer contained components that inactivate the virus and stabilize the RNA, facilitating the study.  \n \n \nQuantification of Viral Load by RT-qPCR  \nAn aliquot of 5 µL of eluent was input into duplicate 20 µL RT-qPCR reactions (Cat. A15299, TaqPath 1-\nStep RT-qPCR Master Mix, CG) with multiplex primers and probes from Integrated DNA Technologies \n(Coralville, IA, USA) targeting SARS-CoV-2 N1 (Cat. 10006821, 1 0006822, 10006823) and N2 (Cat. \n10006824, 10006825, 10007050), and human RNase P (Cat. 10006827, 10006828, 10007061). Positive \nand negative reaction controls were included on every plate: Te mplates for positive control reactions \ncontained 4 copies per µL of SARS-CoV-2 genomic RNA from nCoV 2 019-nCoV/USA-WA1/2020 (Cat. \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20239467doi: medRxiv preprint \n\n4 \n \nNR-52281, BEI) in 5 µL of nuclease-free water (Cat. AM9932, The rmoFisher Scientific), and negative \nreaction controls contained only nuclease-free water. Reactions were run on a CFX96 Real-time PCR \nSystem (Bio-Rad Laboratories) according to the amplification protocol defined in the CDC 2019-Novel \nCoronavirus (2019-nCoV) Real-Time RT-PCR Diagnostic Panel, with  Ct determination by auto-\nthresholding each target channel. Ct values were converted to viral load relative to the resulting value of \nknown input of SARS-CoV-2 heat-inactivated particles in the pos itive extraction control. The conversion \nwas done across all qPCR samples, using the average Ct values of the positive control for the two SARS-\nCoV-2 gene targets (N1 and N2; each 32.50 and N=11) after auto-thresholding on cycles 10-45.\n \n \nQuantification of Viral Load by RT-ddPCR  \nAn aliquot of 5.5 µL from a dilution of the eluent (samples were diluted to be within the range required for \nddPCR) was input into 22 µL reactions of the Bio-Rad SARS-CoV-2 ddPCR Kit (Cat. 12013743, BioRad \nLaboratories) for multiplex quantification of SARS-CoV-2 N1 and N2 targets, and human RNase P targets. \nDroplets were generated on a QX200 droplet generator (#1864002,  Bio-Rad Laboratories) and measured \nusing a QX200 Droplet Digital PCR System (#1864001, Bio-Rad Lab oratories), with analysis using \nQuantaSoft Analysis Software.  \n \nConversion of Ct Values to Viral Load \nRT-qPCR Ct values from our assay were converted to viral load (copies/mL) using the following equation: \n \n𝑉𝑖𝑟𝑎𝑙 𝐿𝑜𝑎𝑑 ൌ  7500 ∗ ሺ2ଷଶ.ହ଴ି஼௧ሻ \n \nBoth the CLIA Laboratory and the laboratory analysis in Kissler et al.21 utilized similar assays to ours, and \ntherefore we assumed this same equation could be used to estima te viral load from Ct values from those \nsources. \n \nSequencing  \nExtracted RNA from samples taken from the early, peak, and late infection stages of each of the three \nindividuals infected with SARS-CoV-2 was sequenced by the Chan Zuckerberg Biohub (San Francisco, \nCA, USA). All sequences, throughout infection, and across household contacts, were found to be identical.   \n \nResults and Discussion  \n \nWe report a case of SARS-CoV-2 transmission in a household of f our individuals, who we refer to as \nParent-1, Parent-2, Sibling-1, and Sibling-2 (Table S1). Siblin g-1 (who reported recent close contact with \nsomeone infected with SARS-CoV-2) and Sibling-2 returned home t ogether from out-of-state and were \nCLIA-lab RT-qPCR tested for COVID-19 the next day. The followin g day, Sibling-2’s specimen resulted \nnegative, and Sibling-1’s specimen resulted positive, prompting  Sibling-1 to isolate and all household \nmembers to quarantine. Within hours of receiving Sibling-1’s positive-test result, they were enrolled in the \nstudy. Parent-2 remained SARS-CoV-2-negative in all samples. Pa rent-1 and Sibling-2 were SARS-CoV-\n2-negative upon enrollment and b ecame continuously positive sta rting ~36 hours after enrollment. Viral \nsequencing determined that the SARS-CoV-2 in samples from Sibli ng-1, Parent-1 and Sibling-2 shared \nidentical sequences to each other, highly supportive of household transmission.  \n \nAll nucleic acid measurements from saliva samples included human RNase P target measurements (Figure \nS2) as an indicator of sample quality to confirm that viral loa d dynamics were not an artifact of sample \ncollection. RNase P Ct values were consistent across samples from each participant: during the early phase \nof infection (from enrollment to just beyond peak viral load, u p to day 8 of enrollment), the 15 samples \nfrom Parent-1 had an average RNas e P Ct value of 27.28 (±1.12 S D), and the 15 samples from Sibling-2 \nhad an average Ct value of 24.51 (±1.31 SD). Also, a pattern of  lower RNase P Ct values (more human \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20239467doi: medRxiv preprint \n\n5 \n \nmaterial) in morning samples than evening samples is occasionally discernable, although this pattern does \nnot appear to dominate viral load signal.  \n \nRT-qPCR and RT-ddPCR measurements of viral load from saliva sam ples provided by the three infected \nindividuals (Figure 1) offer three insights.  (i) Presymptomatic viral loads in some humans can rise over the \ncourse of days not hours, which is slower than expected.\n18,22  This slow rise increases the utility of sensitive \ntests (such as PCR) to enable earlier detection and isolation o f infected individuals before their viral load \nincreases to a presumably more infectious level. Sibling-2’s viral load rose slowly—within PCR detection \nrange—for 3 days until viral load reached the limit of detection (LOD) for rapid tests. This slow rise also \nmakes it more dangerous to assume that most persons with low viral load will not become infectious\n19 and \ntherefore do not need to self-isolate: Sibling-2 produced 6 positive samples in the 103-105 copies/mL range \npresymptomatically before peaking at ~10 7 copies/mL. (ii) Peak viral load does not always rise above the  \nLOD of rapid, low-sensitivity tests, as expected.18,22  Of 88 positive SARS-CoV-2 samples, only one sample \nwas well above the LOD range of rapid tests. Furthermore, in Si bling-1, the logical source of the two \ninfections, neither the CLIA-lab test nor our testing detected viral loads above 10 7 copies/mL during the \npresumed period of household transmission.  (iii) The LOD of a test can affect how early in infection we \ncan diagnose an infected person, and how consistently we can de tect early-phase infections. Of the 52 \npositive samples from the first 10 days of the study, 33 were near or above the LOD of the more sensitive \nrapid test (ID NOW, as determined by the FDA) whereas only 3 were near or above the least sensitive LOD \nof 9.3*106 copies/mL (Table S2). Importantly, there were several days during the presymptomatic period \nthat Parent-1 and Sibling-2 were detected by RT-qPCR, but may n ot have been reliably detected by many \nlow-sensitivity tests.   \n \nThe insights from this study are also supported by analysis of data by Kissler et al.,21 reporting longitudinal \n(but less frequent) testing of anterior nares and oropharyngeal swabs from individuals associated with the \nNational Basketball Association (Figure S1). In Kissler et al.\n21, viral loads rose slowly (for up to 5 days in \nsome individuals) between the first PCR positive test to the LOD of rapid tests. Few of these samples ever \nreached viral loads well above the LODs of most rapid antigen tests (Table S2).  Low-sensitivity tests have \na role in the COVID-19 testing strategies, but our limited data from this study and data from Kissler et al.21 \ndemonstrate clearly that for at least some individuals, low-sensitivity tests will likely be unable to reliably \ndiagnose SARS-CoV-2 infection during the early phase of infection.  Our limited data are consistent with \nthe use of low-sensitivity tests for point-of-care confirmation of suspected COVID-19 in symptomatic \nindividuals, as authorized by the FDA, but not for universal su rveillance testing of asymptomatic \nindividuals, as has been proposed.\n18,19,22  \n \nAdditional studies are urgently needed to address several limit ations of this work.  High-frequency viral \nload measurements from the incidence of infection must be observed in a larger, diverse pool of participants \nto infer the distribution of viral load profiles in human SARS-CoV-2 infection. Both saliva and nasal swabs \nhave been proposed as sample types for rapid, low-sensitivity t ests; however, the LOD of these tests is \nbetter defined in nasal swabs. Our study only analyzed saliva; although saliva has been demonstrated to be \na more sensitive sample type than nasopharyngeal swab by some studies,\n23 other studies have arrived at the \nopposite conclusion. 24,25  The details of saliva collection, sample stabilization, preanalytical handling, \nsample-preparation protocols, and timing of sampling may play a role in the apparent sensitivity measured \nin different sample types. No previous study has directly compa red saliva with other sample types during \nthe early phase of SARS-CoV-2 infection. Quantitative compariso ns of multiple respiratory sample types \n(including nasal, oropharyngeal, and nasopharyngeal swabs) at t he same time points are needed to clarify \nviral load profiles in different respiratory specimen types. La stly, to understand the relationship between \nviral load and infectiousness, direct comparisons of RNA viral load to culturable virus titer across the entire \ncourse of infection are needed. Data to address these limitatio ns are needed to inform optimal testing \nstrategies to reduce SARS-CoV-2 transmission.   \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20239467doi: medRxiv preprint \n\n6 \n \n \nUnderstanding the kinetics of viral load from the incidence of infection and throughout the infectious period \nfor a broad demographic will have implications on SARS-CoV-2 testing policies, including policies around \ntravel. For example, several states require recent negative test results for out-of-state visitors prior to arrival. \nAlthough some states acknowledge the risk of false negative results by low-sensitivity tests and require \nPCR confirmation,\n19,26 others do not specify the type of negative test result require d for arrival.27  If rapid \nlow-sensitivity tests are used, they would risk missing presymptomatic individuals before they rise to their \nhighest (and presumably most infectious) viral load.  Such fals e negative tests have the potential to create \na costly false sense of security; individuals who may have been  recently exposed and receive a negative \nresult from a low-sensitivity test may be more likely—compared to individuals who did not get a test—to \ncome into contact with other members of the community, including the most vulnerable populations.    \n \nAll tests have value when used properly and within the right st rategy, and we anticipate that these \npreliminary results will stimulate studies that will provide a better understanding of SARS-CoV-2 viral load \nin the early stages of infection and lead to the development of effective testing strategies. \n   \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20239467doi: medRxiv preprint \n\n7 \n \n \nFIGURE 1. Quantified SARS-CoV-2 saliva viral load after transmission betw een household contacts relative \nto detection limits of rapid tests.  SARS-CoV-2 viral load over time for “Sibling-1” (the household  index case), as \nwell as “Parent-1” and “Sibling-2.” All three viral sequences w ere identical. Star indicates the viral load estimated \nfrom the cycle threshold (Ct) result from the commercial CLIA laboratory test used to diagnose Sibling-1. Diamonds \nindicate conversion from N1 target cycle threshold values obtained by RT-qPCR to SARS-CoV-2 viral load. Bullseyes \nindicate viral load obtained by single-molecule RT droplet digi tal PCR (RT-ddPCR). Black lines represent periods \nwhen participants reported no symptoms; orange lines indicate p eriods when participants reported at least one \nsymptom. Vertical bars indicate the before noon (white) and aft er noon (grey) periods of each day. Pink shading \nindicates the presumed period of the household transmission events. Horizontal blue lines depict the limit of detection \n(LOD) of the Abbott ID NOW (3 x 10\n5 copies/mL) for upper respiratory specimens from U.S. FDA SARS- CoV-2 \nReference Panel Comparative testing data. Horizontal grey bars depict the range of LODs estimated for commercial \nantigen tests for upper respiratory specimens (1.90 x 105 copies/mL to 9.33 x 106 copies/mL; see Table S2).  \n   \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20239467doi: medRxiv preprint \n\n8 \n \nDATA AVAILABILITY \nRaw Ct values and calculated viral loads are available at CaltechDATA, \nhttps://data.caltech.edu/records/1702 \n \nCOMPETING INTERESTS STATEMENT \nThe authors report grants from the Bill & Melinda Gates Foundat ion (to RFI) and the Jacobs Institute for \nMolecular Engineering for Medicine (to RFI), as well as a Calte ch graduate fellowship (to MMC), two \nNational Institutes of Health Biotechnology Leadership Pre-doctoral Training Program (BLP) fellowships \n(to MKP and JTB) from Caltech’s Donna and Benjamin M. Rosen Bioengineering Center (T32GM112592), \nand a National Institutes of Health NIGMS Predoctoral Training Grant (GM008042) (to AW) during the \ncourse of the study. Dr. Ismagilov is a co-founder, consultant,  and a director and has stock ownership and \nreceived personal fees from Talis Biomedical Corp., outside the submitted work. In addition, Dr. Ismagilov \nis an inventor on a series of patents licensed to Bio-Rad Labor atories Inc. in the context of ddPCR, with \nroyalties paid.  \n \nFUNDING \nThis study is based on research funded in part by the Bill & Me linda Gates Foundation. The findings and \nconclusions contained within are those of the authors and do not necessarily reflect positions or policies of \nthe Bill & Melinda Gates Foundation. This work was also funded by the Jacobs Institute for Molecular \nEngineering for Medicine. AW is supported by an NIH NIGMS Predo ctoral Training Grant (GM008042); \nMMC is supported by a Caltech Graduate Student Fellowship; and MP and JTB are each partially supported \nby National Institutes of Health Biotechnology Leadership Pre-d octoral Training Program (BLP) \nfellowships from Caltech’s Donna and Benjamin M. Rosen Bioengineering Center (T32GM112592). \n \nACKNOWLEDGEMENTS \nWe thank Lauriane Quenee, Junie Hildebrandt, Grace Fisher-Adams, RuthAnne Bevier, Chantal D’Apuzzo, \nRalph Adolphs, Victor Rivera, Steve Chapman, Gary Waters, Leona rd Edwards, and Shannon Yamashita \nfor their assistance and advice on study implementation or administration. We thank Jessica Leong, Jessica \nSlagle, and Angel Navarro for volunteering their time to help w ith this study. We thank Kissler et al. for \nmaking their data publicly available to the community. We thank Maira Phelps, Lienna Chan, Lucy Li and \nAmy Kistler at the Chan Zuckerberg Biohub for performing SARS-CoV-2 sequencing. We thank Jennifer \nFulcher, Debika Bhattacharya and Matthew Bidwell Goetz for their ideas on potential study populations \nand early study design. We thank Martin Hill, Alma Sanchez, Sco tt Kim, Debbie Noble, Nina Paddock, \nWhitney Harrison, and Stu Miller for their support with recruitment. We thank Allison Rhines, Karen \nHeichman, and Dan Wattendorf for valuable discussion and guidan ce. Finally, we thank all the Pasadena \nPublic Health Department case investigators and contact tracers  for their efforts in study recruitment and \ntheir work in the pandemic response.  \n \nSUPPLEMENTARY MATERIALS \nSupplemental Tables 1-2 \nSupplemental Figures 1-2 \nSupplemental References \nContributions of Non-Corresponding Authors \n \n \n   \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20239467doi: medRxiv preprint \n\n9 \n \nREFERENCES \n \n1 CSSE. COVID-19 Dashboard by the Center for Systems Science and Engineering (CSSE) at \nJohns Hopkins University (JHU), https://gisanddata.maps.arcgis.com/apps/opsdashboard/ \nindex.html#/bda7594740fd40299423467b48e9ecf6 (2020). \n2 Arons, M. M.  et al. Presymptomatic SARS-CoV-2 Infections and Transmission in a Skilled \nNursing Facility. New England Journal of Medicine 382, 2081-2090, \ndoi:10.1056/NEJMoa2008457 (2020). \n3 Gandhi, M., Yokoe, D. S. & Hav lir, D. V. Asymptomatic Transmission, the Achilles’ Heel of \nCurrent Strategies to Control Covid-19. New England Journal of Medicine 382, 2158-2160, \ndoi:10.1056/NEJMe2009758 (2020). \n4 Richmond, C. S., Sabin, A. P ., Jobe, D. A., Lovrich, S. D. & Kenny, P. A. SARS-CoV-2 \nsequencing reveals rapid transmission from college student clusters resulting in morbidity and \ndeaths in vulnerable populations. medRxiv, doi:10.1101/2020.10.12.20210294 (2020). \n5 Shi, Q.  et al. Effective control of SARS-CoV-2 transmission in Wanzhou, China. Nature \nMedicine, doi:10.1038/s41591-020-01178-5 (2020). \n6 Wei WE  et al. Presymptomatic Transmission of SARS-CoV-2 — Singapore, January 23–March \n16, 2020. MMWR. Morbidity and mortality weekly report 69, 411–415, \ndoi:http://dx.doi.org/10.15585/mmwr.mm6914e1 (2020). \n7 Kimball A, Hatfield KM, Arons M & et al. Asymptomatic and Presymptomatic SARS-CoV-2 \nInfections in Residents of a Long-Term Care Skilled Nursing Facility — King County, \nWashington, March 2020. MMWR. Morbidity and mortality weekly report 69, \ndoi:http://dx.doi.org/10.15585/mmwr.mm6913e1 (2020). \n8 Pringle JC, Leikauskas J, Ransom-Kelley S & et al. COVID-19 in a Correctional Facility \nEmployee Following Multiple Brief Exposures to Persons with COVID-19 — Vermont, July–\nAugust 2020. MMWR. Morbidity and mortality weekly report 69, 1569–1570, \ndoi:http://dx.doi.org/10.15585/mmwr.mm6943e1 (2020). \n9 Atrubin D, Wiese M & Bohinc B. An Outbreak of COVID-19 Associ ated with a Recreational \nHockey Game — Florida, June 2020. MMWR. Morbidity and mortality weekly report 69, 1492–\n1493, doi:http://dx.doi.org/10.15585/mmwr.mm6941a4 (2020). \n10 Teran RA, Ghinai I, Gretsch  S & et al. COVID-19 Outbreak Among a University’s Men’s and \nWomen’s Soccer Teams — Chicago, Illinois, July–August 2020. MMWR. Morbidity and \nmortality weekly report 69, 1591–1594, doi:http://dx.doi.org/10.15585/mmwr.mm6943e5 (2020). \n11 Lewis M, Sanchez R, Auerbach S & et al. COVID-19 Outbreak Among College Students After a \nSpring Break Trip to Mexico — Austin, Texas, March 26–April 5, 2020. MMWR. Morbidity and \nmortality weekly report 69, 830-835, \ndoi:http://dx.doi.org/10.15585/mmwr.mm6926e1 (2020). \n12 Pray IW, Gibbons-Burgener SN, Rosenberg AZ & et al. COVID-19  Outbreak at an Overnight \nSummer School Retreat ― Wisconsin, July–August 2020. MMR Morbidity and mortality weekly \nreport 69, 1600-1604, doi:http://dx.doi.org/10.15585/mmwr.mm6943a4 (2020). \n13 Dora AV, Winnett A, Jatt LP & et al. Universal and Serial La boratory Testing for SARS-CoV-2 \nat a Long-Term Care Skilled Nursing Facility for Veterans — Los Angeles, California, 2020. \nMMWR. Morbidity and mortality weekly report 69, 651–655, \ndoi:http://dx.doi.org/10.15585/mmwr.mm6921e1 (2020). \n14 CDC. Investigating a COVID-19 Case. (2020). https://www.cdc. gov/coronavirus/2019-\nncov/php/contact-tracing/contact-tracing-plan/investigating-covid-19-case.html  \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20239467doi: medRxiv preprint \n\n10 \n \n15 FDA. Emergency Use Authorization (EUA) information, and list of all current EUAs, \nhttps://www.fda.gov/emergency-preparedness-and-response/mcm-legal-regulatory-and-policy-\nframework/emergency-use-authorization#2019-ncov (2020). \n16 Smithgall, M. C., Scherber kova, I., Whittier, S. & Green, D. A. Comparison of Cepheid Xpert \nXpress and Abbott ID Now to Roche cobas for the Rapid Detection of SARS-CoV-2. Journal of \nClinical Virology 128, 104428, doi:https://doi.org/10.1016/j.jcv.2020.104428 (2020). \n17 Torres, I., Poujois, S., Alber t, E., Colomina, J. & Navarro, D. Real-life evaluation of a rapid \nantigen test (Panbio COVID-19 Ag Rapid Test Device) for SARS-CoV-2 detection in \nasymptomatic close contacts of COVID-19 patients. medRxiv, doi:10.1101/2020.12.01.20241562 \n(2020). \n18 Mina, M. J., Parker, R. & La rremore, D. B. Rethinking Covid-19 Test Sensitivity — A Strategy \nfor Containment. New England Journal of Medicine, doi:10.1056/NEJMp2025631 (2020). \n19 New York Times. Thinking of Traveling in the U.S.? Check Which States Have Travel \nRestrictions, https://www.nytimes.com/2020/07/10/travel/state-travel-restrictions.html (2020). \n20 CDC. Symptoms of Coronavirus. https://www.cdc.gov/coronaviru s/2019-ncov/symptoms-\ntesting/symptoms.html?CDC_AA_refVal (2020).  \n21 Kissler, S. M.  et al. Viral dynamics of SARS-CoV-2 infection and the predictive value of repeat \ntesting. medRxiv, 2020.2010.2021.20217042, doi:10.1101/2020.10.21.20217042 (2020). \n22 Larremore, D. B.  et al. Test sensitivity is secondary to frequency and turnaround time for \nCOVID-19 screening. Science Advances, eabd5393, doi:10.1126/sciadv.abd5393 (2020). \n23 Wyllie, A. L.  et al. Saliva or Nasopharyngeal Swab Specimens for Detection of SARS-CoV-2. \nNew England Journal of Medicine 383, 1283-1286, doi:10.1056/NEJMc2016359 (2020). \n24 Schwob, J. M.  et al. Antigen rapid tests, nasopharyngeal PCR and saliva PCR to detect SARS-\nCoV-2: a prospective comparative clinical trial. medRxiv, 2020.2011.2023.20237057, \ndoi:10.1101/2020.11.23.20237057 (2020). \n25 Becker, D.  et al. Saliva is less sensitive than nasopharyngeal swabs for COVID-19 detection in \nthe community setting. medRxiv, doi:10.1101/2020.05.11.20092338 (2020). \n26 Mass.gov. Information on the Outbreak of Coronavirus Disease 2019 (COVID-19) : COVID-19 \nTravel Order, https://www.mass.gov/info-details/covid-19-travel-order (2020). \n27 DC Health. Coronavirus (COVID-19) Situational Update: Thursday, Nov. 5, 2020. \nhttps://coronavirus.dc.gov/sites/default/files/dc/sites/coronavirus/release_content/attachments/ \nSituational-Update-Presentation_11-05-20.pdf (2020). \n \n . CC-BY 4.0 International licenseIt is made available under a \n is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint \nThe copyright holder for thisthis version posted December 11, 2020. ; https://doi.org/10.1101/2020.12.09.20239467doi: medRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}