key: cord-0740177-ogsicraa authors: Morgan, S. C.; Aigner, S.; Anderson, C.; Belda-Ferre, P.; De Hoff, P.; Marotz, C. A.; Sathe, S.; Zeller, M.; Ahmed, N.; Audhya, X.; Baer, N. A.; Barber, T.; Barrick, B.; Batachari, L.; Betty, M.; Blue, S. M.; Brainard, B.; Buckley, T.; Case, J.; Castro-Martinez, A.; Chacon, M.; Cheung, W.; Chong, L.; Coufal, N. G.; Crescini, E. S.; DeGrand, S.; Dimmock, D. P.; Donofrio-Odmann, J. J.; Eisner, E. R.; Estaki, M.; Franco Vargas, L.; Freddock, M.; Gallant, R. M.; Galmozzi, A.; Gao, N. J.; Gilmer, S.; Grzelak, E. M.; Hakim, A.; Hart, J.; Hobbs, C.; Humphrey, G.; Ilkenhans, N.; Jacobs, M.; Kahn, C. A title: Automated, miniaturized, and scalable screening of healthcare workers, first responders, and students for SARS-CoV-2 in San Diego County date: 2021-06-28 journal: nan DOI: 10.1101/2021.06.25.21257885 sha: f7ef3022d8d6eb05daae48cd5664c71b5a89b5b7 doc_id: 740177 cord_uid: ogsicraa Background: Successful containment strategies for SARS-CoV-2, the causative virus of the COVID-19 pandemic, have involved widespread population testing that identifies infections early and enables rapid contact tracing. In this study, we developed a rapid and inexpensive RT-qPCR testing pipeline for population-level SARS-CoV-2 detection, and used this pipeline to establish a clinical laboratory dedicated to COVID-19 testing at the University of California San Diego (UCSD) with a processing capacity of 6,000 samples per day and next-day result turnaround times. Methods and findings: Using this pipeline, we screened 6,786 healthcare workers and first responders, and 21,220 students, faculty, and staff from UCSD. Additionally, we screened 6,031 preschool-grade 12 students and staff from public and private schools across San Diego County that remained fully or partially open for in-person teaching during the pandemic. Between April 17, 2020 and February 5, 2021, participants provided 161,582 nasal swabs that were tested for the presence of SARS-CoV-2. Overall, 752 positive tests were obtained, yielding a test positivity rate of 0.47%. While the presence of symptoms was significantly correlated with higher viral load, most of the COVID-19 positive participants who participated in symptom surveys were asymptomatic at the time of testing. The positivity rate among preschool-grade 12 schools that remained open for in-person teaching was similar to the positivity rate at UCSD and lower than that of San Diego County, with the children in private schools being less likely to test positive than the adults at these schools. Conclusions: Most schools across the United States have been closed for in-person learning for much of the 2020-2021 school year, and their safe reopening is a national priority. However, as there are no vaccines against SARS-CoV-2 currently available to the majority of school-aged children, the traditional strategies of mandatory masking, physical distancing, and repeated viral testing of students and staff remain key components of risk mitigation in these settings. The data presented here suggest that the safety measures and repeated testing actions taken by participating healthcare and educational facilities were effective in preventing outbreaks, and that a similar combination of risk-mitigation strategies and repeated testing may be successfully adopted by other healthcare and educational systems. limit the spread of COVID-19 in schools (https://www.cdc.gov/coronavirus/2019-138 ncov/community/schools-childcare/operation-strategy.html), which include universal and correct 139 use of masks, physical distancing, handwashing, and contact tracing, but, so far, few studies 140 have investigated how these measures may work in practice in this environment (5) (6) (7) (8) . 141 The countries that have demonstrated the greatest success in controlling the COVID-19 142 pandemic, including New Zealand, Australia, and South Korea, have combined risk-mitigating 143 strategies with large-scale and widespread testing, as well as aggressive contact tracing, to 144 both identify outbreaks and curb community transmission. Importantly, this broad testing must 145 be implemented with a minimal turnaround time to allow for effective contact tracing and 146 isolation of affected persons. Even now, nearly 1.5 years after the initial reports of infections, it can be difficult for many people in the US and other countries to obtain a test for 148 SARS-CoV-2 if they do not exhibit symptoms or have not been in known contact with a person 149 with a confirmed case of COVID-19. In the context of overall limited testing capacity, these 150 measures are intended to allow medical doctors to triage patients into appropriate treatment 151 pipelines. However, in the context of a pandemic, these measures exclude key demographics, 152 namely asymptomatic and pre-symptomatic carriers, who contribute to the "silent spread" of the 153 virus (9,10). Therefore, the current symptom-and contact-based testing strategies employed in 154 much of the world are both unlikely to accurately capture the full extent to which this novel 155 coronavirus can spread throughout our communities. 156 Some regions have screened large, representative proportions of their target populations The objective of this study was to develop and establish an accurate, high-throughput, 163 rapid, and inexpensive SARS-CoV-2 screening pipeline for use at the population level. To this 164 end, we performed SARS-CoV-2 screening on over 5,000 healthcare workers from two large 165 healthcare systems in San Diego County and 1,162 first responders from San Diego Fire and 166 Rescue. Additionally, we screened 21,220 students, faculty, and staff from the University of 167 California, San Diego (UCSD), and 6,031 students and staff from preschool-grade 12 schools 168 across San Diego County that had remained fully or partially open for in-person teaching 169 throughout the COVID-19 pandemic (4,750 from 11 private schools and 1,281 from 13 public 170 schools). We developed a RT-qPCR testing pipeline that included the miniaturization of a 171 testing kit that was granted Emergency Use Authorization by the U.S. Food and Drug 172 Administration. This pipeline was then used to establish a Clinical Laboratory Improvement 173 Amendments (CLIA) certified laboratory at UCSD, and further refinements were made to 174 increase throughput capacity to 6,000 samples per day with next-day results, and to enable 175 accurate detection from self-collected anterior-nares swab samples. A secondary objective of 176 this study was to identify asymptomatic and pre-symptomatic infections, and to evaluate the 177 effectiveness of health and safety measures implemented by healthcare and educational 178 facilities in response to the COVID-19 pandemic, particularly those developed at educational 179 facilities to safely re-open schools for in-person learning. 180 181 The SEARCH study (San Diego Epidemiology and Research Study for Health) was a multi-site study whose aim was to evaluate the presence of COVID-19 infection in 185 potentially exposed healthcare worker and first responder populations in San Diego County (SD 186 County). In particular, employees from Rady Children's Hospital San Diego (RCHSD), Rady 187 Potential healthcare worker participants were informed of the study through an 237 organizational all-user email communication as well as a flyer that was sent to Rady and 238 Scripps Health employees and which contained a QR code that allowed them to fill out an 239 electronic consent form and symptom questionnaire prior to arrival at a study site using 240 REDCap (Research Electronic Data Capture) software hosted at RCHSD. REDCap is a secure, 241 web-based software platform designed to support data capture for research studies (18, 19) . 242 Paper versions of these forms were also available at the study site if an eligible participant was 243 not able to complete them electronically. Potential firefighter and lifeguard participants were 244 informed of the study through an assignment that was posted to every SDFD worker on their 245 Target Solutions (Vector Solutions, Tampa, FL, USA) accounts. Target Solutions is an online 246 training management system which requires assignments to be opened, read, and confirmed 247 prior to the assignment deadline, to ensure all personnel receive and acknowledge their 248 contents. The assignment contained the details of the study as well as instructions on how to 249 participate. In this way, all firefighters and lifeguards in the SDFD were informed of this study. 250 Potential participants were then asked to complete a three-question quiz: (1) Is 251 participation in this study voluntary? (Answer: yes); (2) Will participants be told about their study 252 findings? (Answer: not necessarily); (3) Is participating in this study the same as receiving a 253 clinical test? (Answer: no). Those who answered any questions incorrectly were re-educated by 254 study staff prior to final enrollment and the collection of samples. The Enrollment Screening 255 Form -completed either electronically using a secure REDCap survey, or using a paper version 256 -collected identifiable information including name, address, contact phone number, date of 257 birth, workplace information, and date and site of testing, in addition to information regarding 258 comorbidities and any active symptoms on the day of testing. A barcode was entered into an 259 electronic data capture file for each participant with matching labels placed on a participant 260 information sheet as well as on the NP swab collection tube, in order to link each sample with 261 the participant. A barcode-associated "synthetic name" was also generated, and was cross-262 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 checked with each participant at registration and again at the swabbing station to prevent 263 sample mix-up. Protected health information (PHI) was not provided to the research testing 264 laboratories. Instead, names and identifiable information were stored on REDCap at RCHSD, as 265 specified in the IRB protocol (IRB approval #200470). 266 An NP swab sample was collected from each participant by sampling the posterior 267 nasopharynx through both nares, according to instructions for NP swab collection from the CDC 268 (https://www.cdc.gov/flu/pdf/professionals/flu-specimen-collection-poster.pdf). A qualified team 269 member with appropriate personal protective equipment (PPE), consisting of gown, gloves, N95 270 respirator, and face shield, obtained the sample and placed the NP swab into the barcode-271 labeled collection tube containing 3 mL viral transport medium (VTM), which was prepared 272 according to CDC guidelines (https://www.cdc.gov/coronavirus/2019-ncov/downloads/Viral-273 Transport-Medium.pdf). At the end of each sampling day, samples were transported on wet ice 274 in a cooler to a Biosafety Laboratory 2 Plus (BSL-2+) laboratory at Scripps Research (La Jolla, 275 CA, USA) for sample accessioning and plating, as described in the Sample processing, RNA 276 extraction, and RT-qPCR section below. 277 At RCHSD, NP sample collection of healthcare workers and firefighters for this study 278 involved 192 screening hours over 33 sampling days, between April 17 and June 30, 2020, with 279 two registered nurses (RNs) involved in consenting participants for the study, two respiratory 280 therapists involved in swabbing participants, and two additional RNs acting as support staff 281 involved in participant information data entry and specimen handling support, for a total of six 282 healthcare workers at each screening site on each day. Across all the Scripps Health locations, 283 NP sample collection for this study involved 76 screening hours over 20 sampling days, 284 between April 24 and June 30, 2020. The mobile sample collection team increased from six to 285 ten staff (four additional support staff for data entry and specimen handling) for locations with 286 high demand. For the SDFD employees, NP samples were collected at the permanent sampling 287 location at RCHSD between June 1 and June 30, 2020. 288 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. process used in the SEARCH study is available on protocols.io (21). Following RNA extraction 303 of samples, SARS-CoV-2 detection was performed using a miniaturized and automated RT-304 qPCR procedure. Viral RNA was detected using the TaqPath COVID-19 Combo Kit 305 (ThermoFisher Scientific, Waltham, MA, USA). This viral RNA detection kit is approved by the 306 Emergency Use Authorization (EUA) authority of the US FDA for the detection of SARS-CoV-2, 307 and targets three regions of the SARS-CoV-2 genome -N gene, S gene, and ORF1ab gene -as 308 well as an internal positive control, MS2. The miniaturization of the RT-qPCR process (from the 309 standard 25 µL to a 3 µL reaction volume) involved the use of two Mosquito robotic liquid 310 handlers (STP Labtech Ltd., formerly TTP Labtech Ltd., Boston, MA, USA): a 16-channel liquid 311 handler (HV Genomics), and an HV X1 hit/cherry picker single-channel liquid handler. RT-qPCR 312 was performed on a QuantStudio™ 5 Real-Time PCR System (Applied Biosystems). A detailed 313 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 protocol of the miniaturized and automated RT-qPCR process used in this study is available on 314 protocols.io (22). Using the TaqPath COVID-19 Multiplex Real-Time RT-PCR Assay, samples 315 are considered positive when at least two out of three viral target genes (S gene, N gene, and 316 ORF1ab) amplify using a threshold cycle (Ct) of 37 cycles, and samples are considered 317 negative when none of the three viral target genes amplify but the internal control (MS2) does 318 amplify. Inconclusive test results are those in which only one of three viral target genes 319 amplifies, while invalid tests are those in which no viral target genes amplify, and the internal 320 control also fails to amplify. Samples that produced inconclusive and invalid results from the 321 initial analysis were re-extracted and re-amplified, and the decision tree recommended by the 322 manufacturer (Thermo Fisher) was used to determine the final result; data presented here are 323 the final results obtained after re-extraction and re-amplification, in cases where this was 324 necessary. 325 A technical validation of the RT-qPCR miniaturization was performed by comparing the 327 miniaturization reactions to full-scale reactions. RT-qPCR performance of the miniaturized 328 reactions was compared to published results for full-scale reactions to evaluate equivalency (Fig 329 S1) . The limit of detection (LOD) was calculated for the miniaturized reactions using two 330 different methods: (1) SARS-CoV-2 viral RNA was spiked into the RT-qPCR reaction in different 331 concentrations (1-128 viral genome equivalents per reaction) along with RNA extracted from 332 negative NP control samples; (2) SARS-CoV-2 viral RNA was spiked into negative NP control 333 samples in lysis buffer, before RNA extraction, in different concentrations (1-100,000 viral 334 genome equivalents/mL input sample) (Fig S2) . RT-qPCR performance of the miniaturized 335 reactions were found to be equivalent to the full-scale reactions (Fig S1) . The limit of detection 336 for the miniaturized reactions (using the evaluation criteria recommended by the manufacturer, 337 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 and requiring all three replicates to show a positive result) was found to be 500 viral RNA 338 copies/mL input sample, with >4 viral RNA copies per RT-qPCR reaction (Fig S2 Mawi, and SDS are referred to here as "ANM", "ANW", and "AND", respectively. All three media 378 types inactivate the virus, allowing for self-collection and safer handling, but Mawi and SDS 379 have the additional benefit of being non-toxic substances, allowing for unsupervised self-380 collection of samples. 381 Third, to increase assay throughput, we centralized our operations from three research 382 laboratories located ~10 min drive apart into two adjacent rooms in the same building, and 383 obtained and integrated two Hamilton® Microlab STAR liquid handling systems for automated 384 accessioning and transfer of samples from collection tubes to 96-well plates prior to RNA 385 extraction (see "EXCITE sample accessioning, RNA extraction, and RT-qPCR" section below). 386 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 28, 2021. Validation of anterior nares samples 397 We clinically validated our clinical workflow using AN swabs collected in MTM ("ANM"), 398 Mawi ("ANW"), and 5% SDS ("AND"). First, we performed a technical validation by determining 399 the LOD of all three sample types using contrived samples with a range of SARS-CoV-2 viral 400 particle (VP) concentrations, from 250 VP/mL up to 32,000 VP/mL, with each test performed in 401 triplicate. We demonstrated that the sensitivity of our clinical workflow is excellent, with a 402 technical LOD of 500 VP/mL for MTM and 250 VP/mL for Mawi and SDS (Table S2 ). Next, a 403 second technical validation was performed using contrived samples containing 1,000 VP/mL; a 404 minimum of 20 replicate samples were run per sample type, and a minimum of 19/20 were 405 required to return a positive result via RT-qPCR (see "EXCITE sample accessioning, RNA 406 extraction, and RT-qPCR" section below). All ANM and AND replicates returned a positive 407 result, and 19/20 ANW replicates returned a positive result (Table S3) . Last, we performed 408 clinical validation on all three sample types, first validating ANM samples and then using the 409 validated ANM as a comparator to clinically validate ANW and AND. We used remnant AN 410 samples collected in MTM from 30 positive cases and 32 negative cases, kindly provided by 411 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. Collection-Nylon-Flocked-Nasopharyngeal-Nasal-Swabs.html), which were placed in tubes 431 containing 1 mL of either MTM, Mawi, or 5% SDS. All participants >12 years of age self-432 collected samples, and children 12 years of age or younger in participating schools had samples 433 collected by trained staff at each school. Sample collection occurred on-site at participating fire 434 departments and schools, and all samples were sent daily to the EXCITE lab for testing. For 435 UCSD participants, sample test kits were available in dedicated vending machines throughout 436 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. Scientific, Waltham, MA, USA). RT-qPCR reaction preparation was performed using a 449 Mosquito® HV Genomics robotic liquid handler (SPT LabTech), and miniaturized 3 µL RT-450 qPCR reactions were performed on a QuantStudio™ 7 Pro Real-Time PCR System (Applied 451 Biosystems). A detailed protocol of the miniaturized and automated RT-qPCR process used by 452 the EXCITE lab is available on protocols.io (22). Test results were determined as described 453 above in the "SEARCH sample processing, RNA extraction, and RT-qPCR" section. Samples 454 that tested positive through the EXCITE pipeline were re-extracted and re-amplified to confirm 455 the positive result, and then reported to SD County. EXCITE participants received the results of 456 every test they submitted, regardless of positivity. 457 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10. 1101 Chi-square tests and data summaries from Tables 5 and 6 Rescue Department (SDFD) were screened, along with a small number of employees from 486 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10. 1101 other institutions who heard of the study by word of mouth and were allowed to participate. The 487 EXCITE lab was designed for repeated screening of UCSD students and employees, but has 488 also partnered with a number of private and public preschool-grade 12 schools and SDFD to 489 provide repeated testing of their members. A recent study highlighted the potential positive 490 implications of institutions expanding their COVID-19 testing efforts to include members of 491 surrounding communities, which can benefit the institution itself as well as the community as a 492 whole (27) During the time period of this study, the EXCITE lab processed an average of 1,124 samples 497 per day. Additionally, the average total time from sample barcode scan to return of results was 498 15.2 ± 0.03 h, with an average of 6.6 ± 0.01 h of in-lab processing time (Fig S3) . Results were 499 returned within 24 h of receipt at the lab for 98.1% of samples, and between 24-60 h for 1.2% of 500 samples. The latter were largely samples that returned with Inconclusive or Invalid results on 501 the first analysis run, and were analyzed a second time before reporting a clinical result. We 502 note that there was a small proportion (0.7%) of samples with processing times >60 h; these 503 were primarily samples for which the lab did not initially receive the required patient-level 504 demographic information to report a clinical result, and therefore needed to request additional 505 information from the ordering healthcare provider prior to returning results. 506 The breakdown of test results by sample type is presented in Table 2 . In the SEARCH 507 study, 21 samples initially returned a positive result based on the RT-qPCR performed by the 508 collaborating SEARCH research labs. Samples from all positive tests based on the research 509 testing pipeline were sent to a CLIA-certified lab for validation using their clinical testing pipeline 510 before results were returned to participants. Of the 21 initial positive tests, 18 were validated as 511 true positives, and three were returned as false positives. Therefore, the false-positive rate for 512 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101/2021.06.25.21257885 doi: medRxiv preprint the SEARCH study was 0.037%. Because the EXCITE lab is itself a CLIA-certified lab, positive 513 results were reported directly without external validation. Of the 752 true positive results 514 identified through both SEARCH and EXCITE, 715 tests were obtained from unique individuals, 515 with 35 individuals testing positive on two different occasions and one individual testing positive 516 on three different occasions. The average time between positive tests for the same individual 517 was 5 days; no individuals from this study were identified to be re-infected after recovering from 518 COVID-19. 519 While the SEARCH study only collected NP swabs in VTM, the EXCITE lab used AN 520 swabs in three different preservation media; MTM, Mawi, and 5% SDS, corresponding to ANM, 521 ANW, and AND sample types, respectively. All three media inactivate the SARS-CoV-2 virus, 522 enabling self-collection. The ANM sample type was validated first, but was soon replaced by 523 ANW and AND, which are non-toxic, allowing for unobserved self-collection. AND samples 524 produced invalid results significantly more often than samples collected in other preservation 525 media (Table 2) ; AND samples comprised only 4.7% of all samples, but 67.8% of all invalid test 526 results. As a consequence, the EXCITE lab quickly discontinued the use of AND samples, 527 focusing instead on ANW samples. 528 Table 2 . SARS-CoV-2 RT-qPCR test results obtained from SEARCH and EXCITE, by sample 529 type: AND (anterior nares swab in 5% SDS); ANM (anterior nares swab in MTM); ANW (anterior 530 nares swab in Mawi); NPV (nasopharyngeal swab in VTM). All NPV samples were obtained 531 from SEARCH, and all AND, ANM, and ANW samples were obtained from EXCITE. Samples 532 where no viral target genes amplified but the internal control did amplify were considered 533 negative. Samples where at least two out of three viral target genes amplified were considered 534 positive. Samples where neither the viral target genes nor the internal control amplified were 535 considered invalid. Samples where only one out of three viral target genes amplified were 536 considered inconclusive. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. (100) 161582 (100) 538 The percentage of positive tests from the SEARCH study was very low: only 17 539 participants (0.27%) had a positive COVID-19 test throughout the 2.5 month duration of the 540 study (one of these participants submitted two positive samples, resulting in the count of 18 541 positive NPV test results in Table 2 ), and positive cases were spread relatively evenly 542 throughout the timeframe of the study, with the rolling average always below 1% positivity (Fig 543 1A) . Comparatively, SD County testing reported much higher positivity, with rolling average 544 positivity ranging from ~2-7% over the same time period (Fig 1C) . In the EXCITE lab, positivity 545 changed over time, starting low from mid-September through mid-November, and then rising 546 following the Thanksgiving holiday (Nov 26, 2020), and rising again to a peak in early January 547 2021, after the winter holidays ( Fig 1B) . The number of tests performed per day increased 548 throughout the fall, as the program was ramped up to test students residing in campus-owned 549 housing or coming to campus every two weeks, and then decreased in late December 2020 550 when many students left the UCSD campus for the holidays. Overall testing was highest in early 551 January 2021, when students residing in campus-owned housing or coming to campus were 552 required to participate in intensive testing upon their return to campus (at days 1, 5, and 10 after 553 return), and then went to a weekly testing cadence thereafter (see "Return to Learn at UCSD" 554 section below for more details). The EXCITE data show a similar trend in positive test rates to 555 the data from data from SD County (Fig 1C) , except that the increase in positive cases in the fall 556 started later (at the end of November 2020 for EXCITE, compared to the beginning of 557 November for SD County) and show a slightly earlier drop in positive cases in January 2021 at 558 EXCITE. The positivity rate of the SD County data was much higher than the rates reported by 559 both SEARCH and EXCITE over the same time frame; the rolling average positivity rate for 560 SEARCH never rose above 1%, and for EXCITE was less than 2% at its peak, as compared to 561 the SD County data, where test positivity ranged from 2-15% across the same time period (April 562 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101/2021.06.25.21257885 doi: medRxiv preprint 17, 2020 -February 5, 2021). We note that the populations being tested by SD County are 563 different from those tested in our study. Our data collection involved repeated screening of 564 asymptomatic populations, while samples collected by SD County are more often from 565 individuals who exhibit symptoms or have been exposed to a known COVID-19 case, likely 566 explaining the large discrepancy in positivity rates. In particular, students at UCSD who were 567 experiencing symptoms were referred to the Center for Advanced Laboratory Medicine (CALM), 568 a separate CLIA-certified laboratory on campus, to be tested. Similarly, students and staff at the 569 preschool-grade 12 partner schools were asked to stay home and obtain testing at SD County 570 testing sites if they experienced any symptoms. 571 The winter holidays are typically a very social time for people living in the US, and 572 despite the COVID-19 pandemic, millions of Americans traveled during the holidays in late 573 2020, contributing to a sharp increase in cases reflected both in our data and in the SD County 574 data. The decline in UCSD cases throughout January 2021 was due to early detection of cases 575 after return from winter break, minimal transmission on campus, and reduced community 576 transmission in SD County. The decline in SD County cases observed in January 2021 is likely 577 due to people returning to their homes following the winter holiday and the subsequent 578 decrease in travel and social gatherings, and to a lesser extent, an increase in vaccination 579 rates, which began in late 2020. Countries with higher vaccination rates in early 2021, including 580 the UK and Israel, saw sharp declines in new COVID-19 cases and hospital admissions in 581 February 2021, highlighting the remarkable efficacy of these vaccines, even after one dose 582 (28, 29) . 583 584 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. Almost all of the positive tests for this study came from the EXCITE lab dataset, and 593 most positive tests were obtained between mid-November, 2020, and mid-January, 2021. 594 UCSD participants made up most of the positive tests, which was expected because most of the 595 EXCITE testing was conducted on UCSD students and staff (Fig 2) . Only SDFD and UCSD 596 participants were tested by both SEARCH and EXCITE (Fig 2) . However, the UCSD population 597 tested during the SEARCH study was different than that tested by the EXCITE lab. During the 598 SEARCH study, some UCSD employees involved in the study chose to participate in testing. 599 While these employees likely also participated in testing through the EXCITE lab, they were not 600 the main targets of either testing effort; the SEARCH study was designed to screen healthcare 601 workers and first responders, and the EXCITE lab was established to screen students, faculty, 602 and staff at UCSD and some partner institutions, with a focus on testing schools. 603 604 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. COVID-19 case data from SD County show that the 20-39 age group represents the 613 highest number of cases (and case rates), followed by the 40-59 age group (Fig 3) . The age 614 groups represented by children (0-9 and 10-19) reported the lowest number of cases in the 615 County. Within the EXCITE data from this study, participants in the 20-39 age group also made 616 up the majority of cases, with all other age groups falling well below (Fig 3) . This is largely due 617 to the demographics of the EXCITE testing population, where the majority of participants were 618 university students. We note that the 10-19 age range represents a higher proportion of cases in 619 the EXCITE data than in the County data, also likely because of the large number of university 620 students in the 18-19 age range in our study. 621 622 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. No significant differences in viral load (estimated by Ct value) were observed for 628 participants from different age groups (Fig S4) . Previous studies have varied in their reports of 629 viral loads in patients of different age groups. Some, like this current study, have not found any 630 significant correlation between viral load and age groups (30-33), while others have (34-37). In 631 the studies that did observe a difference, the results are conflicting: Euser et al. (37) reported 632 that children <12 years had lower viral loads than adults, while Heald-Sargent et al. (36) 633 reported children <5 years had higher viral loads than older children and adults. Additionally, 634 Hasanoglu et al. (35) reported decreasing viral load with increasing age in adults, while To et al. 635 (34) reported the opposite trend. Consistent with previous reports (10,31), we did not find any 636 significant difference in viral load between male and female participants (t = 0.02, df = 645.7, p 637 = 0.98). 638 Only two of the three SARS-CoV-2 target genes were required to amplify for a test to be 639 considered positive. The S-gene was the most common 'dropout' (i.e. the gene that failed to 640 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101/2021.06.25.21257885 doi: medRxiv preprint amplify), occurring in approximately 10% of positive cases; N-gene and ORF1ab dropouts 641 occurred in approximately 1-2% of positive cases, respectively. With the exception of the cases 642 in this study infected with the B.1.1.7 SARS-CoV-2 variant of concern, which show a 643 characteristic S-gene dropout with low Ct values (<30) for the other two viral genes (38), 644 samples with viral gene dropouts had higher average Ct values (i.e. lower viral load) as 645 compared to samples without dropouts (Fig S5) ; samples containing gene dropouts had an 646 average Ct value of 32.3 ± 1.1, while samples without dropouts averaged 23.0 ± 5.3 (t = -29.7, 647 df = 2.2, p = 0.001). Seven cases were identified that were consistent with the B.1.1.7 variant 648 (9.2% of all cases with S-gene dropouts), and the average Ct value for these seven cases was 649 21.9 ± 5.5. 650 Symptom reporting differed between SEARCH and EXCITE (Fig 4) . In total, SEARCH 652 participants filled out 7,489 symptom questionnaires (Table 3) , and EXCITE participants filled 653 out 18,318 (Table 4) . Therefore, 92.8% of SEARCH samples were accompanied by symptom 654 reporting, compared with just 11.9% of EXCITE samples. This discrepancy is due to the 655 different nature of each screening system. The SEARCH study was designed as a prospective 656 research study, so occupation and symptom information was requested from all participants. 657 The EXCITE lab was designed for high-throughput screening of asymptomatic populations, with 658 no UCSD participants being asked to report symptoms at the time of testing, and participants 659 from other partners being asked an abbreviated set of questions. Additionally, EXCITE 660 participants were generally assumed to be asymptomatic, because any symptomatic individuals 661 were encouraged to stay home and seek other means of testing. Conversely, at the time when 662 SEARCH was implemented (April 2020), it was the only means of obtaining a COVID-19 test for 663 many participants, resulting in a number of participants seeking out testing through SEARCH 664 specifically because they were exhibiting symptoms they felt might be related to COVID-19. The 665 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101/2021.06.25.21257885 doi: medRxiv preprint timing of the SEARCH study also overlapped with the end of flu season, potentially explaining 666 the higher number of participants reporting symptoms at the beginning of the study as 667 compared to later in the study (Fig 4A) . Indeed, when SEARCH participants waiting in line to be 668 tested were asked why they chose to participate, their reasons mostly fell within four common 669 themes: (1) early in the study, many believed their symptoms could represent COVID-19 illness 670 but were unable to qualify for clinical testing because they were "not sick enough" per the 671 testing triage protocols in place at the time; (2) later in the study, others believed their symptoms 672 more likely represented seasonal allergies than COVID-19 but wanted to confirm, and 673 understood the importance of reporting these symptoms nonetheless; (3) some reported intense 674 stress regarding the pandemic and believed their symptoms of headache, fatigue, or muscle 675 aches might be due to worry or poor sleep, but again wanted to confirm; (4) finally, several said 676 they sought the privacy of testing outside their own healthcare workplaces. These unanticipated 677 explanations for study participation illuminate some unmet needs for frontline workers in 678 healthcare environments during the uncertainties of an unprecedented pandemic. 679 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. Among all SEARCH participants, the most common symptoms reported at the time of 693 testing were stuffy nose, runny nose, sore throat, fatigue, and cough (Table 3 ). The least 694 common symptoms reported were vomiting, trouble breathing, and fever. Among the 695 participants who tested positive and also completed a symptom questionnaire, the most 696 common symptoms at the time of testing were stuffy nose, runny nose, and myalgia (muscle 697 pain), and the least common symptoms were rash, vomiting, and trouble breathing (Table 3) . 698 More than 90% of participants who tested negative reported no symptoms, compared with 699 ~55% of participants who tested positive; nearly 45% of participants who tested positive 700 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. Among the EXCITE participants who completed symptom questionnaires, the most 708 common symptoms reported at the time of testing were headache and stuffy/runny nose (Table 709 4). The least common symptoms reported were trouble breathing and anosmia (loss of sense of 710 smell). Among the participants who tested positive and also completed a symptom 711 questionnaire, the most common symptoms at the time of testing were fever, headache, and 712 stuffy/runny nose, and the least common symptoms were trouble breathing, nausea, and 713 diarrhea (Table 4 ). More than 99% of participants who tested negative reported no symptoms, 714 compared with 89% of participants who tested positive; nearly 11% of participants who tested 715 positive reported at least one symptom at the time of their test, with two being the average 716 number of symptoms reported per person, out of a possible 10. The average age of EXCITE 717 participants who completed symptom questionnaires was 24.1 ± 16.4. 718 Different symptom questions were asked of SEARCH and EXCITE participants. Some 719 symptoms characteristic of COVID-19, such as anosmia (loss of sense of smell), were not 720 included in the SEARCH study because sample collection began before this symptom was 721 recognized as an indicator of COVID-19. Other symptoms, such as myalgia (muscle soreness) 722 and rash, were included in the SEARCH questionnaire but not the EXCITE questionnaire. This 723 is because only the most common cold, flu, and COVID-19 symptoms were included in the 724 EXCITE questionnaire. 725 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10. 1101 Of the 17 SEARCH patients who tested positive for SARS-CoV-2, 11 were available for 726 regular telephone follow-ups, undertaken for 2-3 weeks in order to gain an understanding of 727 symptom progression. Only three of these 11 people reported symptoms at the time of testing: 728 one reported fever, one reported fatigue, muscle aches, and chills, and one reported cough, 729 nasal congestion, and sore throat. Among the eight who did not report symptoms at the time of 730 testing, five developed symptoms within 24 h of participating in the study, ranging from ageusia 731 (loss of sense of taste) (n = 2) to sore throat with rhinorrhea (runny nose) (n = 1) to fever and 732 cough (n = 2). One patient developed anosmia within 72 h, followed by severe fatigue and later 733 a cough. Finally, two patients reported no symptoms of illness by two and three weeks, 734 respectively. None of the ill participants required hospitalization, though one had an extended 735 productive cough that began two days after testing positive and was placed on work leave for 736 three weeks. While more than half of the SEARCH participants who tested positive for SARS-737 CoV-2 in our study did not report any symptoms at the time of testing, most went on to develop 738 symptoms within 2-3 days of their first positive test, suggesting that within our study, being pre-739 symptomatic at the time of testing was more common than having an asymptomatic infection. 740 Interestingly, shortness of breath/trouble breathing, which is considered a hallmark symptom of 741 COVID-19, was reported by only one positive participant at the time of testing across both 742 SEARCH and EXCITE, suggesting that this symptom may typically develop later on in the 743 progression of the disease, as noted by some previous studies (39,40), and may not be a good 744 screening question to determine whether a person qualifies for a COVID-19 test. Reporting of 745 the prevalence of shortness of breath/trouble breathing as a major symptom is inconsistent: one 746 study also found shortness of breath to be a less common symptom (41), but other studies have 747 reported shortness of breath to be one of the most common symptoms (42-44). However, it 748 should be noted that these studies involved assessing symptoms in hospitalized patients, or 749 even patients who suffered fatal cases of COVID-19, which represents a small fraction of 750 infections. 751 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101/2021.06.25.21257885 doi: medRxiv preprint Viral load was significantly correlated with the presence of symptoms for all three target 752 genes (Fig 5) , and only one of the 19 positive tests that were accompanied by a positive 753 reporting of symptoms contained a gene dropout, suggesting that presentation of symptoms 754 may indicate higher viral load in the body. Some previous studies have identified a correlation 755 between viral load and symptom severity (45,46), but most of these studies were conducted on 756 hospitalized patients. Conversely, other studies have found viral loads to be unchanged (13) or 757 even higher (35) in asymptomatic patients as compared to symptomatic patients. Interestingly, 758 overall Ct values from SEARCH positives were higher than those from EXCITE positives, 759 regardless of symptom reporting (t = -2.7, df = 18.1, p = 0.01) (Fig 5 CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101/2021.06.25.21257885 doi: medRxiv preprint In order to keep the healthcare worker and first responder data together, we have 775 included the SDFD first responders who were tested through the EXCITE lab in this section. 776 Therefore, this section contains data from 6,786 individuals, while the SEARCH study itself only 777 included 6,376 individuals. Of 11,964 total swabs, 3,935 were collected at RCHSD, 4,138 were 778 collected at Scripps Health locations, and 3,891 were collected by the EXCITE lab. A proportion 779 of study participants (1,167) chose to be tested on multiple occasions, some as many as 22 780 times, although the majority of high-repeat participants were those from SDFD who were tested 781 routinely through the EXCITE lab (Fig S6a Genomic Medicine (RCIGM), and Rady Children's Specialists of San Diego (RCSSD), but on 789 the intake form, only RCHSD was one of the pre-specified options, meaning employees of 790 RCIGM and RCSSD likely would have selected the "Other" option as their employer, or simply 791 not specified. A conservative estimate of participation, therefore, suggests that out of 792 approximately 8,000 eligible Rady employees, 1,220 RCHSD employees (15.3%) participated. 793 A more liberal estimate of participation would include both RCHSD and "Other" employees, and 794 only those employees who opened their invitation email, resulting in 1,562 employees being 795 tested out of an estimated possible 6,000 (26.0%). However, a small number of employees from 796 Sharp HealthCare were informed of this study by word-of-mouth and were allowed to 797 participate; these employees would likely have chosen "Other" or "Unspecified" as their 798 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101/2021.06.25.21257885 doi: medRxiv preprint organization as well. The true proportion of the target population tested is therefore likely 799 somewhere between 15.3-26.0% for Rady employees. For simplicity, tables and Figs in this 800 section combine the employees who selected "RCHSD" and "Other/Unspecified" into a single 801 category, "Rady". A small number of UCSD Health employees (112) also opted to be tested one 802 or more times. These participants selected "UCSD" as their employer, but because UCSD 803 employees were not targeted by the SEARCH study, we cannot estimate the proportion of 804 employees tested for this population. 805 Participants were asked to add a job description on their intake forms. Because these 806 job descriptions were free-text and not pre-specified categories, a wide range of occupations 807 were reported. These occupations were manually combined into 14 major categories with the 808 remainder being assigned to "Other/Unspecified" (Table 5 ). Approximately one-third (34.0%) of 809 healthcare worker participants from Scripps and Rady were employed as nurses, with the next 810 most common job category being general healthcare worker (24.2%), followed by doctor 811 (11.9%). The proportions of nurses and doctors in this study is similar to their actual proportions 812 at the hospitals and clinics from this study. The most common occupation among COVID-19 813 positive participants was first responder (66.7%) -this is because SDFD members additionally 814 participated in repeated screening through the EXCITE lab in late 2020 and early 2021, and the 815 entirety of the positive first responder cases came from that later time period, when cases in SD 816 County were markedly higher. During the SEARCH study timeframe (April -June 2020), no 817 SDFD members tested positive through our pipeline, and the most common occupations testing 818 positive at that time were general healthcare workers and nurses (5 participants each). While 819 some of these participants were presumed to have contracted the virus from work, others were 820 able to trace their infection to other sources, such as a family gathering or a spouse with a 821 public-facing job. 822 823 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101/2021.06.25.21257885 doi: medRxiv preprint (100) Demographic questions were asked of the participants; these questions were not 825 mandatory for participation, but the majority of participants responded to most or all of the 826 questions. The median age of healthcare workers was 41 for all target populations (Scripps 827 Health, Rady, and SDFD) (Fig 6, Table 6 ). The median age of those participants who tested 828 positive was significantly lower overall (35 years) and for healthcare workers (32 and 28 years 829 for Scripps Health and Rady participants, respectively), but not significantly lower for SDFD 830 participants (38.5 years) ( Table 6 ). The overall data are consistent with CDC observations that 831 the median age of COVID-19 infections has declined over time, from >40 years to <36 years 832 between the months of March and July 2020 (47), and are in line with the median age (36 833 years) of confirmed COVID-19 cases among SD County residents at large 834 (https://www.sandiegocounty.gov/content/dam/sdc/hhsa/programs/phs/Epidemiology/COVID-835 19%20Watch.pdf). Greene et al. (47) suggest that this decrease in the median age of confirmed 836 COVID-19 cases is a result of changing testing patterns, not changes in the epidemiology of 837 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; Female participants made up 74.0% of healthcare workers, which is representative of 839 the makeup of the workforce at the hospitals and clinics involved in this study, which are 840 approximately 74% female across Scripps Health and approximately 83% female at RCHSD. 841 Conversely, male participants made up 84.9% of first responders from SDFD, which is slightly 842 lower than the national proportion of male firefighters (96%) (https://nfpa.org). Of the 843 participants who tested positive for SARS-CoV-2, 70.6% of healthcare employees were female, 844 while 85.7% of SDFD employees were male. Overall, males were significantly more likely to test 845 positive than females (Table 6) ; however, when separated by source (Scripps Health, Rady, 846 SDFD), no differences were observed. This is because the majority of the positive tests came 847 from SDFD workers, who were predominantly male. Previous studies have suggested that 848 males may be more susceptible to this virus (48), but our results are proportional to the 849 populations tested in this study. 850 The majority of participants (72.1%) did not report their ethnicity, and almost one third 851 (28.4%) did not report their race (Table 6 ). For the SEARCH study, race and ethnicity were 852 presented as a single "check all that apply" question, which resulted in many people choosing 853 only a race, only an ethnicity, or neither. Most participants did not select an ethnicity, and those 854 that did select an ethnicity overwhelmingly selected Hispanic and did not also select a race. For 855 the SDFD participants who were tested through the EXCITE lab, race and ethnicity were 856 included as separate questions, which we believe improved reporting compliance. The EXCITE 857 demographic data are presented for every SDFD participant who was tested through EXCITE, 858 regardless of whether they were also originally tested through SEARCH. As a result, fewer 859 SDFD participants did not specify their ethnicity or race, as compared to healthcare workers 860 from Scripps Health and Rady (Table 6 ). SD County is ethnically diverse, with 34.1% of the 861 population identifying as Hispanic (https://www.census.gov/). However, only 15.3% of the 862 participants in this study selected Hispanic as their ethnicity. It is possible that the population in 863 our study is not representative of the population of SD County, but it is also possible that 864 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101/2021.06.25.21257885 doi: medRxiv preprint Hispanic participants were less likely to report their ethnicity. There are many reasons why 865 people choose not to divulge this information in their medical or employment records, or when 866 participating in studies, including confusion over what category they fall into (especially when 867 there are limited and pre-specified categories from which to choose) as well as fear of 868 marginalization or receiving unequal quality of care (49). Additionally, both race and ethnicity 869 having been combined into a single "check all that apply" question for the SEARCH study likely 870 caused additional confusion. Most people who selected both a race and an ethnicity identified 871 as White and Hispanic. In the case of the SEARCH study, it may not have been clear to 872 participants that they were meant to select more than one option, and therefore most 873 participants selected the one answer they identified most strongly with. 874 More healthcare workers identified as Asian (16.9% for Scripps Health and 11.9% for 875 Rady, as compared to 2.7% for SDFD), while more SDFD participants identified as Other/Mixed 876 Race (9.0% for SDFD, compared to 2.0% for Scripps Health and 1.6% for Rady). In all three 877 populations, the majority of participants identified as White (44.8%, 47.7%, and 72.9% for 878 Scripps Health, Rady, and SDFD, respectively). However, direct comparisons between 879 healthcare workers and SDFD first responders may be inappropriate, since the methods used to 880 obtain demographic information were different for these populations. Overall, participants 881 identifying as White, Black/African American, and Other/Mixed Race were more likely to test 882 positive (p = 0.03), but no differences were observed when separated by source (Table 6 ). Non-883 Hispanic SDFD participants were more likely to test positive than Hispanic or Unspecified 884 participants (p = 0.02), but no similar differences were observed among healthcare workers. 885 886 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Participant-provided zip codes of residence were overlaid on a map of SD County (Fig 900 7) . Study participants were spread throughout the County, with the highest concentration in San 901 Diego and the communities slightly further north towards La Jolla, which was expected given 902 that many of the healthcare systems monitored in this study are located in or near these 903 regions. However, the broad distribution of participants across SD County suggests that the 904 mobile testing sites were successful in enabling a more representative population to participate. 905 The distribution of positive participants largely matched the overall distribution of participants, 906 with concentrations in more heavily tested/populated areas. We note that even when 907 participants chose to be tested multiple times, they are represented only once in . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. Data from UCSD, 11 private preschool-grade 12 (P-12) schools, and a group of 13 922 public P-12 schools located in SD County are included in this section. All participants were 923 tested via the EXCITE lab at UCSD, and this section contains data from 27,252 individuals, with 924 21,221 coming from UCSD, 1,281 coming from public P-12 schools, and 4,750 coming from 925 private P-12 schools. Because EXCITE was designed for repeated testing of asymptomatic 926 populations, the majority of participants (81.7%) were tested on multiple occasions, some over 927 40 times (Fig S6b) . The participants who tested most frequently were likely a combination of 928 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 EXCITE lab employees and UCSD student athletes, who were required to test frequently to 929 decrease the chance that their activities would be interrupted due to Demographic information was obtained from the participants. In this section, each 931 participant was counted once, regardless of how many times they participated in testing; for 932 positivity status, "positive" participants are those who tested positive through EXCITE at least 933 once, while "negative" participants are those who never tested positive. For UCSD participants, 934 demographic information was gathered from pre-existing student/employee records. For P-12 935 school participants, demographic information was gathered either as part of the consent 936 process for participation in this testing program, from student/employee records, or a 937 combination of the two. The median age was 22 for UCSD participants, 11 for public school 938 participants, and 16 for private school participants (Fig 8, Table 7 ). Most participants from each 939 group belonged to the 'student' age ranges appropriate for each educational facility. Overall, 940 participants were evenly split between male and female; however, within the P-12 schools, sex 941 was evenly split for student age ranges but skewed female for adult (teachers and staff) age 942 ranges (Fig 8) . We found no differences in positivity rate by sex, overall or within any of the 943 three populations studied. Our finding was similar to the overall case reporting by sex by SD 944 County (https://www.sandiegocounty.gov/). However, SD County did report a higher fatality rate 945 for males than for females, and previous research investigating sex disparities also found that 946 while males are not more likely to test positive, they are more likely to be hospitalized or die 947 from COVID-19 (50). We have no information on whether any of the participants testing positive 948 through the EXCITE lab were hospitalized. 949 Adults 19+ were overall more likely to test positive than younger age groups (p < 950 0.0001), partly because the majority of the participants were adults from UCSD (Table 7) . While 951 almost all UCSD participants were aged 19+, the average age of people testing positive at 952 UCSD was slightly lower than those testing negative (25.2 vs. 26.4, respectively; p = 0.002). For 953 private school participants, we found that students tested positive less frequently than adults, 954 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101/2021.06.25.21257885 doi: medRxiv preprint but this trend was not extended to public school participants. At private schools, adults were 955 more likely to test positive than children (2.3% of adults as compared to 1.0% of children, p = 956 0.002), but at public schools, children aged 11-13 were the most likely to test positive (p = 0.04). 957 This could be due to differences in testing uptake by the schools: for most private schools, 958 students and staff who attended in-person learning at school were required to participate in 959 regular testing, but for public schools, participation was voluntary. This testing program was 960 designed for asymptomatic screening, and participants who felt sick were discouraged from 961 coming to school and were encouraged to seek testing from SD County testing sites or primary 962 care providers. Therefore, these data do not show the complete picture, but they do show that 963 with proper protective measures, students attending in-person schooling are not testing positive 964 more frequently than the general population (Fig 3) , and in the case of the private schools in this 965 study, may be testing positive less frequently than adults in the same settings. Our private 966 school results are similar to other studies, which have also shown that transmission rates in 967 schools are low (5,6), and that students were less likely to test positive than staff in educational 968 settings (8). Also of note, there were no outbreaks in participating schools in this study that were 969 attributable to in-school transmission. The low positivity among school-aged children provides 970 an argument for opening P-12 schools for in-person learning; with teachers and some school-971 aged children eligible for vaccination across the US, and children representing the age group 972 that is currently least likely to test positive, our study and others suggest that schools can re-973 open with minimal risk of community transmission (5-8). However, it should be noted that the 974 low rate of positivity among children observed at the County level (Fig 2) could be a result of 975 school closures during this time, which would have drastically reduced the number of contacts 976 each child would have. 977 978 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. (Table 7) . Across all three populations studied, approximately 40% of participants did 985 not report their race, but the reporting of ethnicity varied by population: two-thirds (67%) of 986 UCSD participants did not report their ethnicity, compared to ~40% of private school participants 987 and just ~20% of public school participants. Because UCSD student/employee records default 988 to unspecified race and ethnicity unless a participant changes their status themselves, or their 989 status is changed by a healthcare worker during a medical visit, this could account for a 990 significant proportion of the participants who did not report a race or ethnicity. However, the 991 discrepancy between race reporting and ethnicity reporting at UCSD suggests that there is a 992 portion of this population reporting race, but declining to report ethnicity. Similar proportions of 993 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101/2021.06.25.21257885 doi: medRxiv preprint participants did not report their race at both public and private schools, but twice as many private school participants (proportionally) declined to report their ethnicity. As with the language barriers and lack of health insurance prevent access to testing or hospital care, while 1020 factors such as air pollution and comorbidities can increase the severity of an infection. 1021 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101/2021.06.25.21257885 doi: medRxiv preprint 9. Zip codes of residence of (A) all UCSD/P-12 school participants tested through the 1059 EXCITE lab, and (B) UCSD/P-12 school participants who tested positive for COVID-19 through 1060 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 the EXCITE lab. The deepness of color is proportionate to the number of participants who chose households over the Thanksgiving weekend. Students returning to campus from fall break were 1114 tested at days 1, 5, and 10 after return. Between Thanksgiving and the beginning of the winter 1115 quarter, weekly testing was encouraged but not mandatory, while bi-weekly testing remained 1116 mandatory. 1117 When the winter quarter began in early January 2021, student move-in was again 1118 staggered to allow for all students to get tested and to prevent crowding; students were also 1119 asked to test themselves on days 1, 3, and 5 after move-in, after which time weekly testing was 1120 mandatory for those residing in campus owned housing or coming to campus. During the winter 1121 quarter, UCSD housed 8700 on-campus students who were required to test weekly, in addition 1122 to 1850 off-campus students who came to campus for classes, not including student athletes 1123 who came to campus for training and were required to test more frequently. During the winter 1124 quarter, 2% of classes were taught in-person, all of which were taught in an outdoor classroom 1125 setting. In addition to asymptomatic screening by the EXCITE lab, UCSD also offered 1126 symptomatic and exposure testing via the CALM lab. Together, the EXCITE and CALM labs 1127 made up the entirety of COVID-19 testing conducted on-campus at UCSD. Testing data from 1128 the CALM lab were not included in this study. 1129 The EXCITE lab collaborated with 11 private schools and a group of 13 public schools in 1131 SD County to provide repeated screening of students, teachers, and staff. These schools 1132 offered different in-person learning schedules, with some offering in-person learning five days a 1133 week and others alternating between in-person and remote learning, with a portion of students 1134 attending in-person learning on different days. During the timeframe this study took place, the 1135 large majority of public schools in SD County were fully remote. However, some schools were 1136 allowed to remain open for in-person learning on a limited basis, based on the CDC's social 1137 vulnerability index and the California Department of Public Health's (CDPH) Healthy Places 1138 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 Index (HPI). Public schools that were in the lowest HPI quartile were selected, as they 1139 represented students from low socio-economic status communities, including many immigrant, 1140 refugee, and other socially vulnerable communities. Not all students from these communities 1141 participated in on-site learning; schools offered this choice to children who were struggling with 1142 remote learning due to their housing situation (homelessness, overcrowded housing, poor wifi) 1143 and/or those who were struggling academically. For public schools, testing was offered weekly 1144 but was not mandatory, and compliance was lower for students than for staff at these schools, 1145 with an average of 68% of students and 92% of staff who consented to testing. For most private 1146 schools, testing was mandatory in order to attend in-person learning, but remote learning was 1147 also available to students. The testing schedule varied for each private school, from weekly 1148 testing to exposure-based testing, but most schools required students and staff to provide a 1149 negative test before returning to school after breaks and holidays (Table S5) . Schools used a 1150 variety of safety and risk mitigation measures to attempt to ensure the health of those 1151 participating in in-person learning. We were provided general information for the public schools, 1152 and more specific information for most participating private schools, detailed in Table S5 and 1153 summarized below. 1154 At the private schools for which we obtained details of their risk mitigation strategies, 1155 cohort size ranged from 9-14 students up to the entire grade level. For at least one private 1156 school, a single positive test within a cohort meant that the entire cohort stayed at home for 1157 remote learning for two weeks, and contact tracing was used to determine any possible 1158 exposures outside the cohort. Because this method was quite disruptive for the students, 1159 administrative staff at this school recommend creating smaller cohort sizes. Desks in 1160 classrooms were placed at a distance of 4-6 ft (1.2-1.8 m), and while the established 1161 recommended distance is 6 ft (1.5 m), a recent report that has been adopted by the CDC 1162 suggests that a 3 ft (0.9 m) spacing may be just as effective at preventing the spread of SARS-1163 CoV-2, as long as everyone is masked (56). Masking was required at all times, with exceptions 1164 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 for very young children and when eating (physically distanced, outside) and napping (for young 1165 children). For most schools, indoor classrooms were either retrofitted with portable home-made 1166 HEPA air filtration systems or had upgraded HVAC systems installed with higher-quality air 1167 filters. Different schools employed different indoor/outdoor instruction methods, with some 1168 schools teaching exclusively indoors, some teaching almost exclusively in outdoor classrooms, 1169 and others using a combination, emphasizing mandatory time outdoors. We note that outdoor 1170 instruction or mandatory outdoor time may not be feasible everywhere, especially in the winter 1171 months. One private school described one-way walking traffic set up throughout the campus, as 1172 well as staggering student drop-off and pick-up to prevent crowding. Different schools used 1173 different methods of contact tracing and daily symptom checking, and all six private schools that 1174 provided details of their symptom screening policies mentioned using dedicated applications, 1175 including Emocha, SchoolPass, and ProCare. These schools also implemented temperature 1176 checks upon arrival on campus. At the presence of any symptoms of illness, students and staff 1177 were required to switch to remote teaching until symptoms resolved. 1178 At the public schools, the following extra measures were taken to ensure the health and 1179 safety of students and staff, following CDC and CDPH guidance: mandatory masking for 1180 students and staff; grouping students into cohorts to minimize interactions; 6 ft (1.5 m) distance 1181 between desks; scheduled drop-off and pick-up times; heightened ventilation in classrooms; 1182 increased sanitization and hand-washing protocols; daily symptom checks and mandatory two-1183 week at-home quarantines for students who present symptoms of illness; restricting 1184 parents/guardians and volunteers from campus; restricting the sharing of materials between 1185 students. For most of the public schools, a single positive test within a cohort resulted in the 1186 entire cohort returning to remote learning for two weeks. 1187 1188 1189 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. ; https://doi.org/10.1101 https://doi.org/10. /2021 In this study, we developed a high-throughput semi-automated pipeline for RT-qPCR 1191 detection of SARS-CoV-2, with scalable capacity and rapid turnaround times that was used for 1192 large-scale repeated asymptomatic screening on an individual basis. This pipeline was first 1193 used to test more than 6,000 healthcare workers and first responders in San Diego, California in 1194 the spring of 2020. The pipeline was then modified in the fall of 2020 and used to establish a 1195 dedicated CLIA-certified COVID-19 testing lab at the University of California, San Diego, 1196 allowing students and staff to return to campus safely by providing repeated asymptomatic 1197 screening. Testing was expanded to include firefighters and some preschool-grade 12 schools 1198 across San Diego County. Thus far, we have tested more than 150,000 nasal swabs from over 1199 28,000 individuals. More than half of participants who tested positive reported no symptoms at 1200 the time of testing, highlighting the importance of asymptomatic/pre-symptomatic screening. The 1201 presence of symptoms was significantly correlated with higher viral load. Hispanic and 1202 Black/African American participants from UCSD and partnering schools were more likely to test 1203 positive, highlighting the disproportionate toll the COVID-19 pandemic has had on already-1204 marginalized populations within the US. At UCSD and at public schools, students were more 1205 likely to test positive for COVID-19 than staff. However, in private schools, students were less 1206 likely to test positive than staff. No correlation between age/sex and viral load was observed. 1207 We note that the results reported here were obtained during a time period (April 17, 2020 to 1208 February 5, 2021) when COVID-19 vaccination rates in San Diego were less than 10%. Our 1209 results suggest that schools ranging from preschool to university may be opened safely, even 1210 without vaccination, when proper health and safety measures are implemented, such mandatory 1211 masking, increased desk spacing, reduced cohort size, and repeat testing. 1212 1213 1214 . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. . CC-BY-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 28, 2021. role in study design, data collection and analysis, decision to publish, or preparation of the 1238 manuscript Coronavirus Pandemic (COVID-19) 1246 [Internet]. Our World In Data Association of stay-at-home orders with COVID-19 1249 hospitalization in 4 states Evidence-based, cost-1251 effective interventions to suppress the COVID-19 pandemic: A systematic review. 1252 medRxiv The effect of stay-at-home orders on COVID-1254 19 cases and fatalities in the United States Pilot 1256 Investigation of SARS-CoV-2 Secondary Transmission in Kindergarten Through Grade 1257 12 Schools Implementing Mitigation Strategies -St Transmission of 1260 SARS-CoV-2 in Australian educational settings: a prospective cohort study 1263 Determining the optimal strategy for reopening schools, the impact of test and trace 1264 interventions, and the risk of occurrence of a second COVID-19 epidemic wave in the UK: 1265 a modelling study (sKIDs): an active, prospective surveillance study. Lancet Child Adolesc Heal Asymptomatic Transmission, the Achilles' heel of 1271 current strategies to control Covid-19 SARS-CoV-2 1273 viral load peaks prior to symptom onset: a systematic review and individual-pooled 1274 analysis of coronavirus viral load from 66 studies Prevalence of Asymptomatic SARS-CoV-2 Infection Spread of SARS-CoV-2 in the Icelandic population Suppression of COVID-19 outbreak in the municipality of Vo, Italy. medRxiv Presymptomatic SARS-CoV-2 Infections and Transmission in a Skilled Nursing Facility 1288 Transmission of 2019-nCoV Infection from an Asymptomatic Contact in Germany. N Engl 1289 Comparison of flocked and rayon swabs 1291 for collection of respiratory epithelial cells from uninfected volunteers and symptomatic 1292 patients 3D printed 1294 nasopharyngeal swabs with wrapped rayon fibers developed and validated by SCREEN 1295 Research electronic 1298 data capture (REDCap) -A metadata-driven methodology and workflow process for 1299 providing translational research informatics support The REDCap 1302 Consortium: building an international community of software platform partners INSPECT sample tracking 1305 system Automated Extraction of Viral RNA using the Omega Bio-tek Mag-Bind Viral DNA / RNA 1309 96 Kit for SARS-CoV-2 Detection Semi-Automated and Miniaturized SARS-CoV-2 1312 Detection using TaqPath COVID-19 Multiplex Real-Time RT-PCR Assay Available from: 1314 dx Semi-Automated Extraction 1316 of Viral RNA using the MagMax Viral Pathogen ( MVPII ) 96 Kit for SARS Available from: 1318 dx Welcome to 1320 the Tidyverse Dates and times made easy with lubridate Visualizations with statistical details: the "ggstatsplot" approach The Case 1326 for Altruism in Institutional Diagnostic Testing. medRxiv Covid-19 Vaccine in a Nationwide Mass Vaccination Setting Hospital admissions due to COVID-19 in Scotland after one dose of vaccine Severe Acute 1333 Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) Viral Load in the Upper Respiratory 1334 Tract of Children and Adults With Early Acute Coronavirus Disease Viral load of SARS-CoV-2 across patients and 1337 compared to other respiratory viruses Association of 1340 Viral Load in SARS-CoV-2 Patients With Age and Gender Temporal profiles 1346 of viral load in posterior oropharyngeal saliva samples and serum antibody responses 1347 during infection by SARS-CoV-2: an observational cohort study Higher viral 1350 loads in asymptomatic COVID-19 patients might be the invisible part of the iceberg Age-related 1354 differences in nasopharyngeal Severe Acute Respiratory Syndrome Coronavirus 2 1355 (SARS-CoV-2) levels in patients with mild to moderate Coronavirus Disease 2019 1356 (COVID-19) SARS-CoV-2 1358 viral load distribution reveals that viral loads increase with age: a retrospective cross-1359 sectional cohort study 1362 Emergence and rapid transmission of SARS-CoV-2 B.1.1.7 in the United States. med Clinical characteristics of 138 1365 hospitalized patients with 2019 novel coronavirus-infected pneumonia in Wuhan Clinical features of patients infected 1368 with 2019 novel coronavirus in Wuhan Clinical progression of patients with 1370 COVID-19 in Shanghai Characteristics and 1373 outcomes of 21 critically ill patients with COVID-19 in Washington state Dysregulation of Immune Response 1376 in Patients With Coronavirus 2019 (COVID-19) in Wuhan, China Clinical features of 85 fatal cases of 1379 COVID-19 from Wuhan: A retrospective observational study Viral dynamics in mild and severe 1382 cases of COVID-19 Kinetics of viral load and 1385 antibody response in relation to COVID-19 severity Decreasing median age 1387 of COVID-19 cases in the United States -changing epidemiology or changing 1388 surveillance? PLoS One Coronavirus: Why men are more vulnerable to COVID-19 than women? Race, ethnicity, culture, and disparities in health care Male sex identified by global COVID-19 meta-analysis as a risk factor for death and ITU 1396 admission COVID-19 1399 pandemic: Disparate health impact on the Hispanic/Latinx population in the United 1400 States 1402 Risk for COVID-19 infection and death among Latinos in the United States: examining 1403 heterogeneity in transmission dynamics African American Communities are at 1406 highest risk of COVID-19: spatial modeling of New York City ZIP Code-level testing 1407 results Assessing 1410 differential impacts of COVID-19 on black communities Effectiveness of 3 versus 6 ft of physical distancing for controlling spread of Coronavirus 1417 Disease 2019 among primary and secondary students and staff: A retrospective, 1418 statewide cohort study The number of times (A) SEARCH study participants and (B) EXCITE lab participants 1472 chose to be tested for SARS-CoV-2. Healthcare workers from Scripps Health and Rady systems 1473 were tested between April 17 and June 30, 2020 via the SEARCH study. Students, faculty, and 1474 staff from the University of California San Diego Fire-Rescue Department (SDFD) participants were tested via both the SEARCH 1477 study and the EXCITE lab during the same timeframes SEARCH study, legal status in the US was not a requirement for participation in the EXCITE 996 testing program, nor was it a question asked of potential participants. 997More UCSD participants identified as Asian (22.4%, as compared to 7.6% and 4.1% for 998 private and public schools, respectively), while more P-12 school participants identified as White 999 (44.9% and 46.7% for private and public schools, respectively, as compared to 22.9% for 1000 UCSD). While the overall number of participants identifying as Black/African American, 1001Indigenous/Native American, or Pacific Islander was comparatively low, more public school 1002 participants selected these options as their race identity (Table 7) . Additionally, approximately 1003 half of public school participants identified as Hispanic, compared to just 6.8% of UCSD and 1004 private school participants (Table 7) . Overall, participants identifying as Black/African American 1005 or Other/Mixed Race were more likely to have a positive test, as well as those with an 1006 unspecified race (p < 0.0001), although when separated by group the result was not statistically 1007 significant for P-12 schools. Additionally, participants identifying as Hispanic were also more 1008 likely to test positive, although this difference was only statistically significant for UCSD 1009 participants (p < 0.0001), not for P-12 schools. Even though the result was not statistically 1010 significant, there was a trend toward a larger proportion of Black/African American and Hispanic 1011 participants from the public schools testing positive for COVID-19 (Table 7) . Similar racial and 1012 ethnic disparities have been noted previously by other researchers, highlighting the 1013 disproportionate toll the COVID-19 pandemic has had on already-marginalized population in the 1014 US (51-55). In particular, Gil et al. (51) identified potential reasons why Hispanic people may 1015 test positive for COVID-19 more frequently, including: higher rates of coexisting medical 1016 conditions; lower rates of health insurance; immigration status; and language barriers. Similarly, 1017 Millett et al. (54) note that counties with a larger Black population were also more likely to have 1018 higher rates of air pollution, comorbidities, and lower rates of health insurance. Factors such as 1019 Participant-provided zip codes of residence were overlaid on a map of SD County (Fig 1028 9) . Participants involved in testing through the EXCITE lab came from all across SD County. For 1029 P-12 schools, testing was conducted on-site at each school, and for UCSD participants, testing 1030 was conducted on campus. Participant density was highest in regions surrounding UCSD in La 1031 Jolla, with densely tested regions extending north and south along the coast away from UCSD, 1032 and decreasing in density when moving inland and when approaching the US-Mexico border 1033 (Fig 9a) . Interestingly, participants who tested positive did not necessarily follow the same 1034 density pattern as overall testing (Fig 9b) . The highest density of positive participants was 1035 located around UCSD, which was expected given the large proportion of on-campus students 1036 participating in testing through the EXCITE lab. Apart from the UCSD campus, the areas with 1037 the highest proportion of positive participants included San Diego, National City, and Chula 1038 Vista, all located south of UCSD. Areas such as Encinitas and Solana Beach, to the north of 1039 UCSD, were heavily tested, but represented a disproportionately small portion of positive 1040 participants. Conversely, Chula Vista was not heavily tested, but represented a 1041 disproportionately large portion of positive participants. There are many possible explanations 1042 for this discrepancy. Median household income (2019) in Encinitas and Chula Vista was 1043 $116,022 and $81,272, respectively (https://www.census.gov/). A previous study found that a 1044 decrease of $10,000 median household income in New York City was correlated with a 1.6% 1045 increase in COVID-19 positivity rate (55). If this correlation is applied to San Diego County, we 1046 would expect positivity rates to be approximately 5.4% higher in Chula Vista as compared to 1047Encinitas. This same study found an increase in COVID-19 positivity in more densely populated 1048 areas (55); these areas are more likely to have multi-family housing such as apartment 1049 buildings, where physical distancing is more challenging. In Encinitas, population density in 1050 2010 was 3,164 people per square mile, while in Chula Vista, population density was 4,915/sq 1051 mi (https://www.census.gov/). Chula Vista also has significantly more Black/African American 1052 1061 that zip code as their area of residence. Zip codes from outside of San Diego County were 1062 assumed to belong to UCSD students living on-campus who provided their home address, and 1063were re-assigned as such. The icon on both images represents the location of the UCSD 1064 campus. 1065 1066Return to Learn at UCSD 1067 During the fall 2020 quarter, UCSD housed 9,129 on-campus students, and COVID-19 1068 testing on a bi-weekly basis was mandatory for any students who lived in campus owned 1069 housing or came to campus for classes. Student athletes were tested weekly when training, and 1070 more often when competing. During the fall quarter, 6% of classes were taught in-person, with a 1071 maximum in-person class size of 50, but this was reduced to 2% of classes in-person mid-1072 quarter as SD County restrictions on indoor teaching were implemented. Following County 1073 restrictions on indoor teaching, all in-person classes were moved to outdoor settings, with large 1074 tents acting as outdoor classrooms. Testing was provided free-of-charge to students, faculty, 1075 and staff, and while testing was not mandatory for faculty and staff, it was recommended. High-1076 volume testing sites for students moving on-campus were set up at the beginning of each 1077 quarter, and student move-in was staggered to allow for adequate testing and social distancing. 1078Testing at term start among students moving/returning to campus-owned housing was provided 1079 at days 1 and 10 in the fall quarter. Sequestration (masking in all areas including in residences 1080 with the exception of when inside a single bedroom or bathroom) was implemented during the 1081 move in testing period. 1082Students living on-campus who tested positive at term start or at any point thereafter 1083 were given options on how to self-isolate for the required 14 days before being allowed to move 1084 into/return to their residential suites. Many students availed themselves of the temporary 1085 isolation housing that was provided on campus, while others chose to return home to isolate, 1086 and some, mainly graduate students, chose to isolate in their campus residence with their 1087 families or alone (this option was not presented to students living in shared housing). Students 1088living off-campus who tested positive could remain in their homes if they could effectively isolate 1089 themselves from others, or were provided isolation housing on campus if desired. 1090Masking was required in all public spaces on campus, both indoors and outdoors, and 1091 beginning mid-way through the fall quarter, students were allowed to gather in groups of three 1092 for up to two hours at a time, provided everyone was outdoors, masked, and socially distanced. 1093At certain points throughout the quarter, cases rose and students were required to sequester 1094 temporarily. Bi-weekly asymptomatic testing was accomplished via the installation of vending 1095 machines throughout campus that supplied COVID-19 testing kits containing a collection tube 1096 pre-filled with media, a swab, and instructions on how to self-collect, as well as drop-off bins, 1097 from which samples were collected multiple times per day for processing by the EXCITE lab. If 1098 at any point a student developed symptoms characteristic of COVID-19, they were tested via 1099 the CALM lab on campus and asked to self-isolate until their test results were returned. 1100Integration of the testing program with two smartphone applications -the UCSD 1101 student/employee app and the UCSD Health EPIC Electronic Health Record MyChart app -1102 allowed for an easy method for linking test tube barcodes with student/employee ID numbers, 1103 and provided students and employees with their test results, whether they were negative or 1104 positive. Individuals with positive test results also received a telephone call from a clinical 1105 provider at UCSD Health, and were contacted daily during the isolation period. Most results 1106 were returned the day after sample collection. We believe this combination of quick turnaround 1107 time and the return of results for every sample increased compliance, because it provided 1108 continuous feedback to students and employees. 1109The number of students residing on campus decreased after the Thanksgiving 2020 1110 weekend, because students who left for the holiday were incentivized (with partial refunds of 1111 housing costs) to remain at home until the winter quarter started in January 2021. This measure 1112 aimed to reduce an influx of new COVID-19 cases from students who gathered with other 1113 Acknowledgements 1215 We would like to acknowledge the San Diego COVID Research Enterprise Network (SCREEN) 1216Initiative for engineering a 3D printable NP swab and scaling to meet the immediate diagnostic 1217 demands of the SEARCH study. We thank the RCHSD employees and leadership who 1218