key: cord-0789608-gl266pvt
authors: Gunawardana, M.; Breslin, J.; Cortez, J. M.; Rivera, S.; Webster, S.; Ibarrondo, F. J.; Yang, O. O.; Pyles, R. B.; Ramirez, C. M.; Adler, A. P.; Anton, P. A.; Baum, M. M.
title: Longitudinal COVID-19 Surveillance and Characterization in the Workplace with Public Health and Diagnostic Endpoints
date: 2020-07-28
journal: nan
DOI: 10.1101/2020.07.25.20160812
sha: 31ffdfa40b42a254bdefa8a7d513d879c54a4a11
doc_id: 789608
cord_uid: gl266pvt

Background The rapid spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and the associated coronavirus disease 2019 (COVID-19) have precipitated a global pandemic heavily challenging our social behavior, economy, and healthcare infrastructure. Public health practices currently represent the primary interventions for managing the spread of the pandemic. We hypothesized that frequent, longitudinal workplace disease surveillance would represent an effective approach to controlling SARS-CoV-2 transmission among employees and their household members, reducing potential economic consequences and loss of productivity of standard isolation methods, while providing new insights into viral-host dynamics. Methodology and Findings On March 23, 2020 a clinical study (OCIS-05) was initiated at a small Southern California organization. Results from the first 3 months of the ongoing study are presented here. Study participants (27 employees and 27 household members) consented to provide frequent nasal or oral swab samples that were analyzed by RT-qPCR for SARS-CoV-2 RNA using CDC protocols. Only participants testing negative were allowed to enter the "safe zone" workplace facility. Optional blood samples were collected at baseline and throughout the 3-month study. Serum virus-specific antibody concentrations (IgG, IgM, and IgA) were measured using a selective, sensitive, and quantitative ELISA assay developed in house. A COVID-19 infection model, based on traditional SEIR compartmental models combined with Bayesian non-linear mixed models and modern machine learning, was used to predict the number of employees and household members who would have become infected in the absence of workplace surveillance. Two study participants were found to be infected by SARS-CoV-2 during the study. One subject, a household member, tested positive clinically by RT-qPCR prior to enrollment and experienced typical COVID-19 symptoms that did not require hospitalization. While on study, the participant was SARS-CoV-2 RNA positive for at least 71 days and had elevated virus-specific antibody concentrations (medians: IgM, 9.83 ug mL-1; IgG, 11.5 ug mL-1; IgA, 1.29 ug mL-1) in serum samples collected at three timepoints. A single, unrelated employee became positive for SARS-CoV-2 RNA over the course of the study, but remained asymptomatic with low associated viral RNA copy numbers. The participant did not have detectable serum IgM and IgG concentrations, and IgA concentrations decayed rapidly (half-life: 1.3 d). The employee was not allowed entry to the safe zone workplace until testing negative three consecutive times over 7 d. No other employees or household members contracted COVID-19 over the course of the study. Our model predicted that under the current prevalence in Los Angeles County without surveillance intervention, up to 7 employees (95% CI = 3-10) would have become infected with at most 1 of them requiring hospitalizations and 0 deaths. Conclusions Our clinical study met its primary objectives by using intense longitudinal testing to provide a safe work environment during the COVID-19 pandemic, and elucidating SARS-CoV-2 dynamics in recovering and asymptomatic participants. The surveillance plan outlined here is scalable and transferrable. The study represents a powerful example on how an innovative public health initiative can be dovetailed with scientific discovery.

All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07. 25.20160812 doi: medRxiv preprint Introduction 68 The rapid spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and the 7 protocol. At baseline, participants also completed a survey that included basic demographic 132 information (name, address, phone number, date of birth, gender, race, and ethnicity), relevant 133 co-morbid conditions, contact with possible COVID-19 positive individuals, and recent travel 134 history ( Table 1) . A symptom diary also was completed ( Table 1) . Where relevant, entries were 135 updated weekly. The above items were the extent of personal history and demographics 136 collected. No medical records were obtained/reviewed. 137 The clinical study was unblinded and there were no study-specific interventions. Descriptive 141 and summary statistics are provided for this small observational study. 142

Test results for nasal or oral swabs collected in the morning typically were available early in 143 the afternoon of the same day. When a participant tested positive or the results were 144 inconclusive, testing was repeated for confirmation or collected at increased frequency (daily 145 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07.25.20160812 doi: medRxiv preprint 8 when feasible) until the participant had repeated negative results. Participants were advised in the 146 informed consent and, again, when informed of research test results that these results were 147 research-based and not intended for clinical diagnosis. Participants received referrals for testing 148 and recommendations (e.g., self-quarantine) based on the current, preferred public health 149 requirements at the time. 150

Nasal sample collection was overseen by two researchers equipped with personal protective 152 equipment (PPE) in the Oak Crest parking lot, starting at 8:30 AM. Study participants were 153 required to wear masks, waited in their parked vehicles, and received a sterile nasopharyngeal 154 swab (synthetic applicator tip, synthetic handle) from one of the two study researchers (Fig 1A) . (which was not certified by peer review) 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 July 28, 2020. Optional saliva samples were self-collected in Falcon tubes (50 mL) at the participant's home 179 or in their sealed vehicle and stool swabs were collected at the participant's home. Specific 180 written instructions were provided to participants opting to provide these specimens. Blood (5-8 181 mL, ×2) using Vacutainer® (Becton, Dickinson and Company, Franklin Lakes, NJ) tubes for 182 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. (which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07.25.20160812 doi: medRxiv preprint a positive control (1×3 = 3 wells), a no template control (1×3 = 3 wells), leaving three wells that 205 were used for additional validation (e.g., naïve swab). This panel configuration meets or exceeds 206 the CDC guidelines [9] . 207 96-well PCR plates were prepared using reagents on ice and centrifuged at 1,000 RPM for 1 208 min at room temperature. The plates were run on a CFX96 Touch Real-Time The normalized SARS-CoV-2 RNA copy number, ΔCt was calculated according to eq. 1: 214

where Ct RP is the Ct value for the RP gene transcript in the sample and Ct N is the mean Ct for 215 corresponding two SARS-CoV-2 N gene transcripts (N1 and N2). 216

Calibration plots of Ct versus RNA copy number were generated for every batch of TaqPath™ 1-218

Step RT-qPCR Master Mix kits. Predetermined amounts of RNA standard for SARS-CoV-2 N 219 gene transcript fragments (vide supra) were serially diluted in nuclease-free water (Promega 220 Corporation, Madison, WI). In a typical experiment, eight calibration standards were used, 221 spanning the 0.15-1.5×10 6 RNA copy number range. The samples (20 µL) were analyzed (N1 222 and N2 probes) as described above. 223 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07.25.20160812 doi: medRxiv preprint

The diagnostic panel result interpretation followed CDC guidelines [9] . Specimens with a cycle 225 threshold (Ct) less than 40 for N1, N2, and RP were considered positive. Samples with a Ct 226 greater than 40 for N1 and N2, but less than 40 for RP were considered negative. Samples with a 227

Ct for either N1 or N2 less than 40 (i.e., only one of the two probes resulted in a Ct less than 40) 228 and a Ct less than 40 for RP were considered inconclusive. Samples with a Ct for RP greater than 229 40 were considered invalid. 230

A range of collection swabs were used in the clinical study due to supply shortages (see Results 232 section). The two most common swab types employed were model 25-2000-H and 25-806-2PD 233 (Puritan, Guilford, MA). All swabs were evaluated for fungal and bacterial contamination by 234 qPCR using published methods [11, 12] . Swab batches that were found to be free of microbial 235 contamination were evaluated for nasal sample collection efficiency. Typically, 3-6 volunteers 236 from the clinical study self-swabbed nasally according to the study protocol with the test swabs. 237

The swabs were analyzed as described above using the RP probe, and Ct values below 25 238 typically were indicative of efficient sample collection. Swabs from that batch then were used in 239 the study. Gene RBD with C-Terminal Hexa-Histidine Tag (generously provided by Dr. F. Krammer). 245 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. Each plate also contained wells with the control anti-RBD monoclonal IgG antibody CR3022 265 (Creative Biolabs, New York, NY) plated in serial dilutions to establish a standard curve, or an 266

IgM or IgA monoclonal antibody with the same variable region as CR3022 produced in-house. 267 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. Bayesian case velocity and machine learning was used for model analysis [15] . This model fuses 278 three methods to provide accurate predictions of case counts and hospitalizations. Briefly, a 279 Bayesian nonlinear mixed model was fitted to the velocity (first-derivative) of the cumulative 280 case counts for each location, such as California, or a specific county, zip code, or even 281 workplace and incorporated prior information such as local interventions to obtain the posterior 282 distribution of the trajectory of the cases. This was then fed into the compartmental model to 283 predict deaths using the random forest algorithm trained on COVID-19 data and population-level 284 characteristics such as age, expected proportion of co-morbidities, gender, and race/ethnicity. 285

This yielded daily projections and interval estimates for cases and deaths for each location. The 286 approach combined the strengths of traditional epidemiologic compartmental models with curve-287 fitting statistical models and modern machine learning. 288 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. The probability of multiple, sequential false negative SARS-CoV-2 results was estimated using 293 traditional probability theory. Each serial test was assumed to be independent as the probability of 294 a test result outcome was not dependent on the result of the previous test. The sensitivity of the 295 test was assumed to be 0.99 and the specificity 0.95. The probability of being COVD-19 positive 296

given a negative test result was estimated according to Bayes' theorem, eq. 2: 297

where P (D|N) is a conditional probability, the likelihood of event D occurring given that N is 298 true; P(N|D) is a conditional probability, the likelihood of event N occurring given that D is true; 299 P(D) and P(N) are the probabilities of observing D and N, respectively. This probability was 300 estimated for prevalence ranges from 0.05 to 0.5 and the greatest upper bound was taken as the 301 probability bound. 302 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. Asian 2 (7) 2 (7)

Other 2 (7) 0

The majority of clinical specimens consisted of nasal swab samples. Only 3 of the 54 participants 310 (5.6%) could not tolerate the nasal swab procedure and self-collected oral samples instead. A 311 total of 942 samples were analyzed in the three-month study. Figure 2A compares the 312 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. (which was not certified by peer review) 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 July 28, 2020. Due to a shortage of appropriate swabs at the onset of the study, 5 different swab types -each 330 tested prior to use as described in the Methods and Materials section-were used. Figure 2B  331 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. There were 5 invalid measurements in the first three study days attributed to lack of 344 familiarity with the sampling protocol and therefore these were not included in the analysis. An 345 additional 6 invalid samples were obtained during the remaining study period, accounting for 346 0.63% [6/(6+948)] of total measurements. 347

There were a total of 3 inconclusive measurements (0.32%) over the course of the study, with 348 the following caveats: the inconclusive results did not include those obtained for the 2 349 participants who tested positive for SARS-CoV-2 RNA (vide infra), nor did they include 9 350 inconclusive results obtained on 2 study days (June 3 and 5) for participants who repeatedly 351 tested negative for SARS-CoV-2 RNA. The high number of inconclusive measurements obtained 352 on those 2 consecutive sampling days were attributed to reagent trace contaminants that resulted 353 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. Two participants were found to be positive for SARS-CoV-2 RNA over the course of the study 357 (Fig. 3) . (which was not certified by peer review) 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 July 28, 2020. Table 3 and Fig. 4 . Other than subject 39, only one participant had detectable serum IgG 387 concentrations (collected on May 15, 0.12 µg mL -1 ). 388 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07.25.20160812 doi: medRxiv preprint Table 3 (which was not certified by peer review) 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 July 28, 2020. (which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07.25.20160812 doi: medRxiv preprint asymptomatic individual who maintained low SARS-CoV-2 RNA copy numbers and did not 498 appear to trigger a traditional host immune response. This result is not entirely unexpected as 499 antibody responses are not detectable in all COVID-19 patients, especially those who experience 500 low grade symptoms [17, 21] . 501

One household member (subject 39) was symptomatic and clinically diagnosed with 502 COVID-19 by RT-qPCR and began participating in the study during convalescence after 503 symptoms resolved (Fig. 3A) . The subject developed a robust immune response in terms of anti-504 SARS-CoV-2 IgM, IgG, and IgA serum concentrations (Table 3, Fig. 4C ) that persisted for over 505 2.5 months from the onset of symptoms. While the IgM and IgA concentrations decreased ca. fold between April 14 and May 26, the IgG concentrations remained stable over the studied 2.5 507 months (Fig. 4C) . This individual continues to be followed to measure the serum IgG kinetics, 508 Little is known about this potential subpopulation of individuals with persistent, long-term viral 513 recalcitrance. 514

Intermittent news reports suggest that some convalescing COVID-19 patients who have 515 tested negative for SARS-CoV-2 RNA by RT-qPCR may retest positive later in recovery, 516 prompting speculation that some individuals are vulnerable to reinfection. Others hypothesize 517 that positive retests are the result of non-infectious, residual viral fragments. Unfortunately, 518 many of these reports are anecdotal and not scientifically controlled. Scientific studies on the 519 alleged SARS-CoV-2 reinfection phenomenon are scarce. There are isolated case reports where a 520 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. All Oak Crest employees with the exception of subject 31 remained negative for SARS-CoV-536 2 RNA over the course of the 3-month study. In our hands, the RT-qPCR test appeared to 537 provide reliable results, with no known false positives or negatives. With the exception of 538 subjects 31 and 39, inconclusive measurements (i.e., only one of the two oligonucleotide probes 539 targeting the viral nucleocapsid protein gene transcript fragment met the assay threshold for 540 positivity) made up a small fraction (0.32%) of the test results. We contend that in the context of 541 a person developing or recovering from COVID-19 infection, inconclusive SARS-CoV-2 RNA 542 measurements can define the transition phase between positive and negative, as supported by 543 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. IgG, which has the longest half-life of the three measured virus-specific antibodies (Fig. 4) [33]. 556

The findings remain unexplained and could be due to cross-reactivity in the IgM assay. Subject 557 18, a self-quarantined employee who had just recovered from suspected COVID-19 (based on 558 symptomology) at the start of the study, repeatedly tested negative for SARS-CoV-2 RNA, but 559 tested positive for IgM antibodies that rapidly declined (τ1/2 = 8.8 d, Fig. 4A) . No IgG or IgA 560 antibodies were detected in serum samples from this participant. 561

There are some caveats to interpreting our antibody data. We measured only IgM, IgG, and 562

IgA against RBD and not the whole spike or E proteins, as discussed in more detail elsewhere 563

[33]. The IgM assay is subject to greater uncertainty compared to IgA and IgG at low 564 concentrations (ca. 0.25 µg mL -1 , or lower) need to be interpreted with caution. However, all 565 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07.25.20160812 doi: medRxiv preprint 31 serum samples were analyzed at three levels of dilution (1:100, 1:40, and 1:20) and only 566 detectable concentrations in at least 2 of the 3 diluted samples are reported. 567

The clinical study described here has met its goals to date by providing a safe work 568 environment for employees of a small academic institute. The ability to perform routine, frequent 569 COVID-19 testing not only has allowed us to responsibly maintain a productive work 570 environment during the pandemic, but also provided continuous reassuring data for the 571 participating households, many including vulnerable individuals. This had the important mental 572 health benefits of reducing anxiety and providing a sense of normalcy in the workplace (i.e., safe 573 zone) during an acutely stressful period. 574

Modeling local study impacts predicted that without our intervention up to 7 employees or 575 household contacts could have become infected with SARS-CoV-2 by July 7, 2020. Our finding 576 that only 2 subjects contracted COVID-19, one prior to the start of the study, suggests that 577 workplace disease surveillance based on frequent, longitudinal testing combined with 578 recommended public health practices (social distancing, frequent hand sanitizing, and mask 579 wearing on select days), when feasible, can be effective at creating a safe zone or bubble 580 preventing COVID-19 from entering into the workplace. 581

Our simulations predicted an important, local public health benefit from our scalable 582 surveillance plan. Similar approaches are being adopted by professional sports leagues (e.g., the 583 National Hockey League), the entertainment industry, and others to support responsible 584 resumption of their activities. The study also generated important, new scientific data on the 585 SARS-CoV-2 host dynamics enabled by its longitudinal, intense sampling design. Our clinical 586 study represents a powerful example on how an innovative public health initiative can be 587 dovetailed with scientific discovery. 588 All rights reserved. No reuse allowed without permission.

(which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07.25.20160812 doi: medRxiv preprint

Prevent Getting Sick

Six Asian Countries/Areas: A Follow-up Study

Coronavirus Disease 2019 in China

Challenges of Managing the Asymptomatic Carriers of

Presumed Asymptomatic Carrier 611 Transmission of COVID-19

Transmission 615 of 2019-nCoV Infection from an Asymptomatic Contact in Germany

SARS-CoV-2 Viral Load in

Upper Respiratory Specimens of Infected Patients

The Rachel Maddow Show

2019-nCoV) Real-time RT-PCR Diagnostic Panel

Detection of 628 2019 Novel Coronavirus (2019-nCoV) by Real-time RT-PCR

No reuse allowed without permission. (which was not certified by peer review) 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

Broad-coverage Fungal Quantitative Real-time PCR Assay

Broad-coverage Bacterial Quantitative Real-time PCR Assay

CoV-2 Seroconversion in Humans

Antigen Production, and Test Setup

Bayesian Case Velocity Model with Random Forest for Predicting COVID-19 in the

2020:under review

Testing Data in the

No reuse allowed without permission. (which was not certified by peer review) 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

Clinical and Immunological 653 Assessment of Asymptomatic SARS-CoV-2 Infections

Estimating the Asymptomatic Proportion 656 of Coronavirus Disease 2019 (COVID-19) Cases on Board the Diamond Princess Cruise 657

Estimation of 661 the Asymptomatic Ratio of Novel Coronavirus Infections (COVID-19)

COVID-19: Asymptomatic Carrier Transmission Is 665 an Underestimated Problem

RNA in COVID-19: A Case Report

RNA in a COVID-19 Patient. Research Square. 2020

A Case of a Readmitted Patient Who Recovered from

Positive SARS-CoV-2 Test in a Woman with COVID-19 at 22 Days After Hospital 681 Discharge: A Case Report

A Special Case of COVID-19 with

Long Duration of Viral Shedding for 49 Days

Case Report: Viral Shedding for 60 Days in a Woman with

Novel Coronavirus Disease (COVID-19)

Positive Virus Detection in Two Medical Staff Recovered from Coronavirus Disease

Persistence and Clearance of Viral 698 RNA in 2019 Novel Coronavirus Disease Rehabilitation Patients

Recovered COVID-19 Patients with Re-detectable Positive RNA Test

Positivity in COVID-19 Patients: A Single Center Experience and Potential Implications

Rapid 708 Decay of Anti-SARS-CoV-2 Antibodies in Persons with Mild COVID-19

The study was funded entirely through discretional, institutional funds. The authors gratefully 590

The model was run for California with prevalence levels seen on March 23, 2020 and on July 7, 411 2020 in Los Angeles County. Given 27 employees and 27 household contacts with the 412 characteristics of participants from the current study and without surveillance interventions, the 413 model produced an expectation that up to 7 employees (95% CI = 3-10) would have become 414 infected with at most 1 of them requiring hospitalizations and 0 deaths. The lack of deaths stems 415 from the population demographic tending towards younger persons and improved in-hospital 416 treatment procedures at the time of writing. 417

A number of policies and procedures were instituted during the study, as described below. Some 419 were part of the original protocol, and others were added based on practical considerations. 420While working on site at Oak Crest, all individuals practiced social distancing, hand hygiene, 421 and wore face coverings when in close proximity to other employees on Mondays (vide infra); 422 testing negated the need for mandatory face masks on other days. Outside of work, employees 423 and their household members were encouraged to strictly adhere to current public health 424 advisories regarding gathering in groups, wearing face coverings, and social distancing. 425Participants who tested positive for SARS-CoV-2 RNA received referrals for testing and 426 recommendations (e.g., self-quarantine) based on the current, preferred public health 427 requirements. Symptomatic participants, even if testing negative for SARS-CoV-2 RNA by RT-428 qPCR, would have been asked to return home (none were reported in the study), self-quarantine 429 at a minimum until symptoms resolved, and contact their health care provider about confirmatory 430 clinical testing and symptom management. Participants testing positive for SARS-CoV-2, even if 431 asymptomatic, were informed as soon as possible and asked to go home, self-quarantine, and 432 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) 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 July 28, 2020. testing, treatment, and quarantine recommendations. Participants also were given educational 434 information to guide other household members of potential risk and some of their options. Any 435 employee who was asked to self-quarantine, or did not choose to participate in the study, worked 436 remotely as much as possible and continued to receive full pay. 437Participants who tested negative and obtained a subsequent invalid test result were asked to 438 wear protective face covering (surgical masks were provided by Oak Crest if needed) until they 439 could be re-tested. Initially, participants who tested negative and obtained a subsequent 440 inconclusive test result were sent home immediately and told to self-quarantine until they could 441 be re-tested (i.e., new swab sample collected and analyzed). As the study progressed, that 442 protocol was modified to be the same as an invalid result (i.e., wear protective face covering 443 rather than send the employee home for self-quarantine), assuming that previous test results were 444 negative. The longest period between tests was Friday to the following Monday (3 d). Starting in 445May, all employees were required to wear protective masks on Monday mornings until the test 446 results were finalized and communicated. The same procedure was instituted for 3-day week-447 ends when there was an additional day between tests. Employees who missed one day of testing 448 were required to wear a protective mask until they were re-tested. Employees who missed more 449 than one day of testing only were permitted to re-enter the facility after a new sample tested 450 negative. 451 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) 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 July 28, 2020. 

The current, ongoing clinical study has two primary aims: (1) characterize the rate of SARS-453CoV-2 acquisition in a small cohort of participants interacting on a daily basis in the workplace; 454(2) determine the ability of these data to manage workplace SARS-CoV-2 exposure and 455 consequences, minimizing further spread as per public health advisories. Workplace study 456 participants include staff, students, and volunteer researchers, referred to as "employees". Family 457 members and housemates also were recruited leading to a sub-cohort referred to as "household 458 members". Related exploratory study goals include: characterizing the rate of SARS-CoV-2 459 acquisition in employee and household members; quantifying antibody-specific responses in 460 blood at baseline (previously exposed) and while on study (to capture asymptomatic/pre-461 symptomatic, newly infected); characterizing viral shedding parameters in saliva and stool 462

Clinical SARS-CoV-2 RNA test kits that have received Emergency Use Authorization 464 (EUA) by the FDA and are processed in a laboratory certified according to the Clinical 465 Laboratory Improvement Amendments (CLIA). Such kits have been a rare resource during this 466 clinical study and, consequently, were not available for investigative purposes. Our clinical study 467 used research reagents to achieve outcomes, but did not interfere with the availability of reagents 468 in short supply for clinical diagnosis. The RT and amplification kit (TaqPath™ 1-Step RT-qPCR 469Master Mix) and primers have not been in short supply. The employed RNA extraction kit 470 (Quick RNA Viral Kit) did not have FDA EUA in March 2020 and has been readily available 471 since then. Nasopharyngeal swabs were in short supply early in the study requiring the use of 472 existing, in-house stocks until they became more readily available starting in May 2020 (Fig.  473   2B) . The various swab types performed similarly and met the study criteria. To date, we have 474 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) 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 July 28, 2020. The study was designed to be efficient in terms of manpower commitment and protecting the 478 study team from exposure to SARS-CoV-2 during sampling. Employees were trained in self-479 sample collection while isolated in their vehicles 3 times per week (Mon, Wed, Fri) between 480 8:30 and 9:15 AM with support from the study team. The samples (typically 23-28 per day) were 481 processed the same day with results were available by early afternoon. A team of 4 researchers 482(2 handling samples and 2 observing for quality assurance, Fig. 1B ) performed the RNA 483 extraction and loading of the 96-well plate for RT and cDNA gene fragment amplification, 484 typically completed within 90 min. Additional labor included quality assurance and subsequent 485 data analysis. These resource commitments made the study feasible to continue indefinitely until 486 recovery from the COVID-19 pandemic allows for a safe workplace. 487During the course of the study, one employee (subject 31) became SARS-CoV-2 RNA 488 positive for ca. 2 weeks (Fig. 3B) . The viral RNA copy number (median, 13.1 copies per swab; 489 IQR, 10.9-18.8 copies per swab) remained low throughout and the subject did not report any 490 COVID-19 symptoms. Blood draws on May 14, 15, and 26 from the subject produced serum 491 samples that were below the analytical LLD for 2 of the measured anti-SARS-CoV-2 492 monoclonal antibodies (IgM and IgG). However, anti-RBD IgA concentrations were detectable 493 in all samples and decayed rapidly (τ1/2 = 1.3 d, Fig. 4B) . 494It is becoming increasingly evident that a significant proportion of individuals with positive 495 SARS-CoV-2 RT-qPCR test results remain asymptomatic or minimally symptomatic [5, [17] [18] [19] [20] ] 496 yet still contagious [5, 20] . However, ours is the first report on longitudinal analysis of an 497 All rights reserved. No reuse allowed without permission.(which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07.25.20160812 doi: medRxiv preprint All rights reserved. No reuse allowed without permission.(which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07.25.20160812 doi: medRxiv preprint All rights reserved. No reuse allowed without permission.(which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07.25.20160812 doi: medRxiv preprint All rights reserved. No reuse allowed without permission.(which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07.25.20160812 doi: medRxiv preprint All rights reserved. No reuse allowed without permission.(which was not certified by peer review) 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 July 28, 2020. . https://doi.org/10.1101/2020.07.25.20160812 doi: medRxiv preprint