key: cord-0684347-a5h88q5f authors: Hamer, D. H.; White, L.; Jenkins, H. E.; gill, c. J.; Landsberg, H. N.; Klapperich, C.; Bulekova, K.; Platt, J.; Decarie, L.; Gilmore, W.; Pilkington, M.; McDowell, T. L.; Fari, M. A.; Densmore, D. M.; Landaverde, L.; Li, W.; Rose, T.; Burgay, S. P.; Miller, C.; Doucette-Stamm, L.; Lockard, K.; Elmore, K.; Schroeder, T.; Zaia, A. M.; Kolaczyk, E. D.; Waters, G.; Brown, R. A. title: Control of COVID-19 transmission on an urban university campus during a second wave of the pandemic date: 2021-02-26 journal: nan DOI: 10.1101/2021.02.23.21252319 sha: 33c6fdeeb52c6f96d1fee2f154f949235098a19d doc_id: 684347 cord_uid: a5h88q5f Importance: The COVID-19 pandemic had a wide-ranging impact on educational institutions across the United States. Given potential financial challenges and adverse psychosocial effects of campus closure, as done in the spring of 2020 in response to the first wave, many institutions of higher education developed strategies to allow campuses to reopen and operate in the fall despite the ongoing threat of COVID-19. Many however opted to have limited campus re-opening in order to minimize potential risk of spread of SARS-CoV-2. Objective: To analyze how Boston University (BU) fully reopened its campus in the fall of 2020 and controlled COVID-19 transmission despite worsening transmission in the city of Boston. Design: Multi-faceted intervention case study. Setting: Large urban university campus. Interventions: The BU response included a high-throughput SARS-CoV-2 PCR testing facility with capacity to delivery results in less than 24 hours; routine asymptomatic screening for COVID-19; daily health attestations; compliance monitoring and feedback; robust contact tracing, quarantine and isolation in on campus facilities; face mask use; enhanced hand hygiene; social distancing recommendations; de-densification of classrooms and public places; and enhancement of all building air systems. Main Outcomes and Measures: Between August and December 2020, BU conducted >500,000 COVID-19 tests and identified 719 individuals with COVID-19: 627 (87.2%) students, 11 (1.5%) faculty, and 212 (25.5%) staff. Overall, about 1.8% of the BU community tested positive. Infections among faculty and staff were mostly acquired off campus, while undergraduate infections were more likely acquired in non-classroom campus settings. Of 837 close contacts traced, 86 (10.3%) tested positive for COVID-19. BU contact tracers identified a source of transmission for 51.5% of cases with 55.7% identifying a source outside of BU. Among infected faculty and staff with a known source of infection, the majority reported a transmission source outside of BU (100% for faculty and 79.8% for staff). Conclusions and Relevance: BU was successful in containing COVID-19 transmission on campus while minimizing off campus acquisition of COVID-19 from the greater Boston area. A coordinated strategy of testing, contact tracing, isolation and quarantine, with robust management and oversight, can control COVID-19 transmission, even in an urban university setting. 4 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 February 26, 2021. ; https://doi.org/10.1101/2021.02.23.21252319 doi: medRxiv preprint The SARS-CoV-2 global pandemic resulted in nearly 1.8 million deaths worldwide in 2020. 1 After a major surge in cases in spring 2020 in the northeast United States, other parts of the country experienced continued transmission through the summer with a second wave arising in the northeast in fall 2020. 2 This mandated several mitigation actions, including, in spring 2020, the closure of in-person learning at public and private higher education institutions. Faced with potentially dire financial challenges and adverse social impacts associated with continued closure, some universities and colleges developed strategies in spring 2020 to allow campuses to reopen and operate in the fall despite the ongoing threat of COVID-19. 3, 4 Boston University (BU) is a private university with a student/staff/faculty population of around 40,000 individuals. BU is located in the center of a large US city, thus creating a scenario for potential widespread COVID-19 transmission. Despite these challenges, the BU administration pursued an aggressive risk mitigation strategy involving widespread asymptomatic screening for COVID-19, environmental modifications, de-densification strategies, and contact tracing, isolation and quarantine, all with the goal of providing its students the opportunity to return to inperson learning in the fall semester of 2020. Here we describe the BU experience as a case study that we believe offers important lessons that may be broadly applicable to other higher education institutions. Initial planning, starting in March 2020, determined an approach that centered on active surveillance for asymptomatic and symptomatic cases via on-site molecular testing for SARS-CoV-2 (Supplement). This involved setting up de novo systems for high throughput laboratory testing, timely communication of results, rapid contact tracing, isolation of infected individuals, and quarantine of close contacts ( Figure 1 ). Surveillance was complemented by additional control measures, including mask use, enhanced hygiene practices, social distancing recommendations, daily health attestations, de-densification of classrooms and public places, and enhancement of all building air systems. This process was 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 February 26, 2021. ; https://doi.org/10.1101/2021.02.23.21252319 doi: medRxiv preprint 6 aided through mathematical modeling, and the use of multiple data systems and 24 hour/7 day per week monitoring and data-driven adaptation throughout the semester. This required a coordinated leadership and management structure ( Figure S1 ). Efforts were coordinated through interlocking management committees and leadership groups across campus, as well a series of systematic and sustained communication strategies. Here, we briefly summarize each of the BU COVID-19 response components. Starting in spring 2020, several different groups were organized to coordinate all aspects of COVID-19 control on campus, including, but not limited to, monitoring incoming data, modifying campus operations, implementing best public health and medical practices, surveillance, and budgeting (Supplement 1.1, Figure S1 ). BU implemented hybrid teaching in which all undergraduate and graduate students could attend classes in-person or remotely. This blended modality allowed classes to accommodate on-line only students using Zoom software and in-person students simultaneously. 5 Accommodations for on-line only teaching were made for educators who were elderly or had medical conditions that increased their COVID-19 risk. 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. Information Technology. A system was developed by the University's Information Services and Technology experts that linked the electronic medical record (EMR) systems for students, faculty, and staff, and the laboratory information system in our testing facility with the webbased system for daily symptom attestation of all community members on campus, reservations for tests and negative test reporting. SARS-CoV-2 testing categories. Based on guidance from public health authorities, the University developed four SARS-CoV-2 testing categories, related to an individual's risk of becoming infected on campus. These categories determined frequency of testing, ranging from twice weekly for category 1 (e.g., on-campus undergraduates) to no testing for category 4 (e.g., students, faculty, or staff entirely off-campus) (Supplement 1.4). Managing and responding to test results. All individuals with a negative test were automatically notified by email to access their test results through a secure, online portal when their results were available. Anyone with a positive test result was called by a healthcare professional. The contact tracing protocol was based on CDC and Massachusetts Community Tracing Collaborative processes 6,7 , with adaptation from BU academic programs and student input. Contact tracers followed a detailed script to identify all close contacts. A close contact was defined as someone within six feet of an infected person for 15 minutes or more over a 24 hour period (Supplement 1.5). Students who tested positive for SARS-CoV-2 had to isolate for 10 days after symptom onset and resolution of fever for at least 24 hours, and with improvement of other symptoms, or for 10 days from the positive test date if asymptomatic. 8 Students identified as a close contact had to 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 February 26, 2021. ; https://doi.org/10.1101/2021.02.23.21252319 doi: medRxiv preprint 8 quarantine for 14 days from the exposure date. Based on evolving data from the CDC, this period was reduced to 10 days of quarantine on November 20 th . 9 Students living on campus were moved to units allocated by BU (650 units for quarantine and 342 for isolation) (Supplement 1.6-1.8). Additional measures including face mask use, enhanced hand hygiene, social distancing recommendations, daily health attestations, de-densification of classrooms and public places, and enhancement of all building air systems including optimization of filtration units were implemented (Supplement 1.9-1.10). We used probabilistic SEIR (susceptible-exposed-infectious-recovered) transmission modeling during summer 2020 to provide insight into the expected relative efficacy of potential interventions for reducing COVID-19 transmission and burden in the BU community with the goal of achieving only linear increases of cases from transmission outside BU (Supplement 1.11). We used a stochastic agent-based model, implemented using the COVID agent-based simulator (covasim) framework. 10 BU developed a dedicated COVID-19 external communications platform called Back2BU. 11 The pre-existing ecosystem of data warehousing and analysis systems that traditionally supported the university was augmented to support the data storage, management, and analysis requirements necessary to allow for near-real time surveillance of outcomes associated with BU's COVID-19 response. Surveillance efforts focused on monitoring not only standard metrics around incidence, isolation, and quarantine, but also metrics around testing, contact tracing, and compliance with the collection of campus control measures. An external dashboard was created and updated daily to allow the broader BU community and beyond to track now-standard, fundamental metrics (Supplement 1.12-1.15). In addition, an augmented, internal dashboard was created to aid the various groups working with BU leadership to not only monitor but adapt the BU COVID-19 response. 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. A comprehensive phased move-in program was adopted to reduce crowding and lines, giving students the time to test and the university the time to respond to any positive cases on arrival (Supplement 1.18). The plan for this analysis was reviewed by the Boston University CRC Institutional Review Board (IRB) and was classified as non-human subjects research. The BUMC IRB reviewed the safe behavior quality improvement project (Supplement 1.17) and determined it to be exempt. We structure our results to provide information on the: 1) operational aspects of the systems BU put in place to manage the epidemic, and 2) resulting epidemiological features of COVID-19 in BU and the surrounding community. In general, we describe results during the semester: September 3-December 19; however, we include some results from initialization of systems during the summer. Overall, the systems designed in the summer to mitigate the pandemic performed well throughout fall semester, as detailed below. Housing. While 99% (7266) of graduate students lived off-campus, most undergraduates lived in on-campus housing, including a total of 7,131 students as of October 13, 2020, representing 67% of the fall 2020 capacity. Due to de-densification efforts, 3453 (48%) of students lived alone and 3678 (52%) of students lived with one roommate (Supplement 1.5). 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. Testing. Initial operations, staffing and supplies were built for a maximum of 6200 tests per day (~42,000 tests per week) with the aim of reliably processing samples within 24 hours of collection. After an expected initial period of learning and improvement, the laboratory reached a stable run rate by October, averaging 5-6,000 tests per day, with lower volume on weekends. Testing turnaround times decreased dramatically over the semester, leveling off to around 12-15 hours between sample collection to receipt of results ( Figure 2A) . Ultimately, the BU testing laboratory conducted 467,382 tests for the BU population during the fall semester (517,357 including the pre-semester move-in). Figure 2B ). If an individual did not meet criteria to be a close contact, but was part of a social group (e.g., Greek-life membership) or affiliation (e.g., sports team, music group) of recent cases, they were notified of their potential risk of COVID-19 exposure, reminded to follow public health guidelines, and required to increase testing frequency to 3 times per week. Despite having quarantine and isolation capacities of 650 and 342 respectively, occupancy only reached a maximum of 13.7% and 12.9% respectively at any one time ( Figure S3 ). Compliance. On average 12% of on-campus students did not comply with testing or attestation protocols in October and November which was lower than off campus students during the same time (51.5% for off campus students in categories 1, 2 and 3 or 20% if including students in category 4) ( Figure 3A ). Reported violations of other control measures had an initial early semester burst, but gradually reduced throughout the semester ( Figure 3B ). Early semester 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. Classroom Density. Class attendance data were not collected systematically. An upper bound on attendance levels is given by the percentage of registered students indicating through LfA that they would attend in person. For medium, large, and "very large" classes, this number was on average approximately 45% in October (with medians similar) but dropped to roughly 25-30% in November. For small classes, while the mean was similar to the other groups, the variability was substantially greater, with the distribution as a whole being closer to a uniform (especially in October) (Figure 4) . Limited survey data suggest that actual attendance figures tended to vary somewhat by class (undergraduate versus graduate), program, and campus. In addition, attendance changed (generally decreasing) throughout the semester. Changing epidemiology of COVID-19 in Boston. Following the use of statewide control measures in spring 2020, case numbers in Suffolk County (including Boston) remained low throughout the summer, averaging around 0.1 reported case per 1000 population up to mid-October. As the state moved through the reopening phases 12 and the weather became colder (and people spent more time indoors), reported cases began to increase to 0.40/1000 by early November and rose to 0.75/1000 in early December, following Thanksgiving gatherings ( Figure 2C ). These increasing trends were observed statewide and also in collection of SARS-CoV-2 from Boston area wastewater, indicating that they were not just due to increased testing in the Boston area. 13,14 During the fall semester, 719 individuals tested positive for COVID-19 in the BU community; including 627 (87.2%) students, 11 (1.5%) faculty, and 212 (25.5%) staff (Table 1) . Cases increased following holidays, particularly Thanksgiving, and that was more pronounced for undergraduate students and non-faculty staff. While the incidence rate among students and staff tracked that of Suffolk County ( Figure 2C ), there were two distinctions. First, BU surveillance testing resulted in detection of many asymptomatic cases (37.7% of total cases) likely more than the passive testing regime in the 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. Of the students who were close contacts and entered quarantine, 10.3% (86/837) tested positive for SARS-CoV-2 during their time in quarantine with lower overall test positivity rate of 2.1% due to multiple testing regimen in quarantine. The test positivity rate for those who self-attested having symptoms was notably higher (4.9%) than that of asymptomatic individuals (0.10%) indicating utility of this feature. Sources and locations of transmission. BU contact tracers identified a source of transmission for 51.5% of cases with 55.7% identifying a source outside of BU (Table 1) . Among infected faculty and staff with a known source of infection, the overwhelming majority reported a transmission source outside of BU (100% for faculty and 79.8% for staff). Students identified more BU contacts as sources of infection (39.8% for graduate students and 59.2% undergraduate students, Table 1 ). Notably, BU household contacts were identified as a source of infection less than 1% of the time (Table 1) and anecdotally, there were no sustained transmission events in on-campus housing, indicative of the efficacy of efforts to control spread in campus housing. When asked to identify their close contacts, students rarely identified close contacts in the classroom setting and most close contacts tended to be friends and, after Thanksgiving, family ( Figure 2D ). This indicates that while BU students tended to be more likely to identify another BU affiliate as a source of infection, the contacts leading to infection occurred in places that were not directly targeted by BU interventions. Although many institutions of higher education across the US reopened in fall 2020, some experienced large COVID-19 outbreaks compelling them to revert to on-line only or remote education. [15] [16] [17] The BU experience is important because it has an urban campus situated in a community that experienced high and increasing COVID-19 incidence from August to December 2020 with no option for the university to isolate from the wider community. Despite this challenge, the university benefited from having substantial resources including funding to establish and run a SARS-CoV-2 testing laboratory and sampling centers, a hybrid learning 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 February 26, 2021. ; https://doi.org/10.1101/2021.02.23.21252319 doi: medRxiv preprint 13 approach (LfA), and diverse expertise among university faculty in staff in the fields of medical epidemiology, modeling, biostatistics, and public health control measures. In addition, strong central leadership; internal communication to students, staff, and faculty; frequent and adaptive testing of students 18, 19 ; short turnaround time on testing; highly effective contact tracing coupled with isolation and quarantine; and vigorous enforcement all combined to prevent widespread campus outbreaks of COVID-19 despite the worsening situation in the greater Boston area. The interventions that were designed over the summer were supported by a strong leadership structure with multiple subcommittees targeting important aspects of the response. There was frequent communication and coordination between these groups to ensure that, if a cluster of cases was emerging, all parties were aware. This meant that testing cadences could be adapted, compliance efforts could be modified, and messaging adapted to blunt any potential risk of an outbreak. This coordinated effort was key to ensuring a high level of compliance and the success of planned interventions. Short turnaround time of results followed by rapid isolation of infected individuals, contact tracing and quarantine of close contacts resulted in limited transmission in the BU community. Faculty and staff were almost always infected outside of the university campus. While the majority of students with a known source of infection reported another BU affiliate as their contact, these infectious events appeared to occur outside of BU housing and instructional settings, where interventions were targeted. When clusters of cases did appear in settings the university could not directly target, e.g., due to social gatherings or other off campus gatherings, these were quickly controlled, due to our vigorous testing regime, rapid contact tracing, and strict enforcement measures, and consequently no major outbreaks were observed. This meant that the resulting numbers of cases throughout the semester were consistent with our goal of maintaining a linear, rather than exponential, increase in cases, which was manageable with our intervention strategies. Surveillance testing allowed us to identify and isolate many close contacts in advance of contact tracing efforts. Importantly, we note that due to the surveillance testing system, BU tended to detect increases in cases in advance of the surrounding community where people were mostly 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. 14 only tested once symptomatic. This is supported by the fact that 37.7% of positive individuals were asymptomatic at the time of isolation indicating that BU was detecting cases early in the disease course. This finding is in direct contrast to recently published research suggesting that BU cases led the surrounding area in a causal manner. 20 There were several limitations to the latter study including the selection criteria for inclusion of universities that resulted in only one Boston-area institution of higher education in the analysis and assuming that transmission in the community arose from the university campus rather than the converse, which did not account for evidence from the BU contact tracing efforts, as described above, that suggested 55.7% of the transmission was exogenous (i.e. arising in the community). BU's success is consistent with current understanding of best practices for COVID-19 control. In fact, these strategies of aggressive testing, contact tracing, and quarantine and isolation have been successfully implemented in many countries, including Singapore, South Korea, and Taiwan. [21] [22] [23] However, unlike these countries, the BU setting did not allow the restriction of travel between the campus and nearby community, making this a strong demonstration of the utility of these approaches even despite substantial importation of cases from the surrounding community. This can potentially serve as a model for other institutions nested within a broader community. The semester was not without challenges. There were times when the contact tracing team was unable to identify contacts due to some students' reluctance to divulge information regarding where they had been or who they had been with. In these cases, coordination between the dean of students and the contact tracing team was critical in identifying other students who were associated with the infected student(s) through team or club membership so increased frequency of testing (adaptive testing) could be performed. This approach to infection control on a university campus carries a high financial cost 24 ; BU had to implement budget adjustments, including hiring freezes, salary freezes, and several other costcutting measures, to meet the cost of these services and respond to declining revenue because of pandemic-related changes to operations. While we do not attempt a cost effectiveness analysis here, we note that the university was able to meet its financial obligations and avoid large layoffs 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. or other consequential financial impacts in the fall. Also, providing students an in-person option was beneficial for varying learning styles and meeting immigration requirements. The remote option also benefitted students who were unable to attend in person, either due to health considerations or travel restrictions. We recognize that BU benefited from being a large research university with much of the required expertise for our strategy available within the university, saving money and facilitating substantial control over the operations. This is clearly not feasible for all institutions of higher education. This implies that broader efforts in the community, supported by government public health agencies, are required to control spread. We demonstrate the efficacy of a multipronged response in controlling the spread of SARS-CoV-2 on an urban university campus, despite rising community burden of disease. This approach relied on the main axioms of infection control: frequent testing, vigorous contact tracing, and rapid isolation and quarantine, and a strong leadership structure to ensure nimble decision-making and rapid adaption to emerging data. (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 February 26, 2021. (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 February 26, 2021. ; https://doi.org/10.1101/2021.02.23.21252319 doi: medRxiv preprint COVID-19 Weekly Epidemiological Update Sauber-Schatz EK. COVID-19 trends among persons aged 0 -24 years -United States ?CDC_AA_refVal=https%3A%2F%2Fwww.cdc.gov%2F coronavirus%2F2019-ncov%2Fcommunity%2Fguidance-ihe-response.html 4. Mass.gov. Reopening: Higher Education. Accessed Centers for Disease Control and Prevention. Contact tracing for COVID-19 Duration of Isolation and Precautions for Adults with COVID-19. Accessed Covasim: An agent-based model of COVID-19 dynamics and interventions. medRxiv 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 Wastewater COVID-19 tracking Notre Dame and Michigan State Shifting Online as Campus Outbreaks Grow Response to a COVID-19 Outbreak on a University Campus -Indiana Multiple COVID-19 clusters on a University Campus -North Carolina Repeat SARS-CoV-2 testing models for residential college populations Assessment of SARS-CoV-2 screening strategies to permit the safe reopening of college campuses in the United States Are college campuses superspreaders? A data-driven modeling study Evaluation of the effectiveness of surveillance and containment measures for the first 100 patients with COVID-19 in Singapore Contact tracing during coronavirus disease outbreak, South Korea Contact tracing assessment of COVID-19 transmission dynamics in Taiwan and risk at different exposure periods before and after symptom onset 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 College campuses and COVID-19 mitigation: clinical and economic value 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 We would like to thank the members of the Augmented Budget Committee, Medical Advisory Group, Community Health Oversight Group, Healthway team, and all Boston University faculty, staff, and students who played important roles in keeping the campus safe and healthy during the 2020 fall semester. 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 February 26, 2021. week. C) BU COVID and Suffolk county cases per 1000 from August 25, 2020 to December 24, 2020 with a seven day smoother applied. Cases are shown overall, and by key subpopulations, as noted in the legend. D) For students who tested positive, the affiliation of their close contacts: classmate, faculty, coworker, friend, family, and other. 21 4, as 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. (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (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 February 26, 2021. ; https://doi.org/10.1101/2021.02.23.21252319 doi: medRxiv preprint