Mitigation of SARS-CoV-2 transmission at a large public university

Abstract

In Fall 2020, universities saw extensive transmission of SARS-CoV-2 among their populations, threatening health of the university and surrounding communities, and viability of in-person instruction. Here we report a case study at the University of Illinois at Urbana-Champaign, where a multimodal "SHIELD: Target, Test, and Tell" program, with other non-pharmaceutical interventions, was employed to keep classrooms and laboratories open. The program included epidemiological modeling and surveillance, fast/frequent testing using a novel low-cost and scalable saliva-based RT-qPCR assay for SARS-CoV-2 that bypasses RNA extraction, called covidSHIELD, and digital tools for communication and compliance. In Fall 2020, we performed >1,000,000 covidSHIELD tests, positivity rates remained low, we had zero COVID-19-related hospitalizations or deaths amongst our university community, and mortality in the surrounding Champaign County was reduced more than 4-fold relative to expected. This case study shows that fast/frequent testing and other interventions mitigated transmission of SARS-CoV-2 at a large public university.

Fig. 1. Target.

a Sensitive testing can reveal a positive case early in the infection, and thus isolation of the index case reduces the number of people infected by this index case. Frequent testing and rapid isolation reduce the time period during which a person is infectious but not isolated (Area A). As a result, the R₀ multiplier for testing is the ratio between the truncated area under the curve (Area A) and the untruncated area under the curve (Area A + Area B). The dashed vertical line between Area A and Area B represents the moment an infected individual is isolated; as this line moves to the left, M is decreased and viral spread is reduced. b Effect of different mitigation interventions on the basic reproduction number R₀ as computed in our agent-based model. Mean R₀ values (n = 10) are indicated for each conditions tested. Error bars represent SEM. If R₀ is >1 (orange dashed line at R₀ = 1), the epidemic grows exponentially. If R₀ is <1, any outbreak diminishes exponentially. Without any mitigation, R₀ is close to 3 and a runaway epidemic will occur. Masking and social distancing help reduce transmission but can’t suppress growth of cases on their own as R₀ is still greater than one. However, when these measures are combined with frequent testing (2 tests a week), R₀ drops to 0.35 and containment of epidemic becomes possible. Adding extra mitigation interventions such as manual contact tracing and risk based exposure notification being R₀ further down to 0.19 suggesting the potential for strong control of the epidemic on campus. The details of the agent-based model are given in the “Methods” section. The results shown here are computed assuming that 100% of the students are compliant with twice a week testing, isolation, and quarantine. We also ran the same simulation assuming 60% compliance, and the same general trends were observed with R₀ for the full SHIELD program predicted to still be manageable (0.5, see Supplementary Fig. 1). Simulated effects of delays and imperfect contact tracing on the final size of epidemic and peak quarantined population, using the agent-based model, are shown in Supplementary Fig. 2.

Fig. 2. Test and tell.

a The effect of heat on SARS-CoV-2 nucleic acid detection in saliva. γ-irradiated SARS-CoV-2 (1.0 × 10⁴ viral copies/mL) was spiked into fresh human saliva (SARS-CoV-2 negative). Samples diluted 1:1 with 2× Tris–borate–EDTA (TBE) buffer were incubated at 25 °C, or in a hot water bath at the indicated temperature and incubation time. All saliva samples were spiked with purified MS2 bacteriophage as an internal control and directly analyzed by RT-qPCR, in triplicate, for SARS-CoV-2 nucleic acid corresponding with ORF1ab gene, N-gene, and S-gene. Undetermined Ct values are plotted as ND. This experiment was repeated at least three times. b 25 clinical saliva samples were split into two aliquots upon receipt, one set was processed using our covidSHIELD assay and the other set was subjected to RNA extraction. 5 μL of processed saliva were subsequently used as templates for RT-qPCR. A positive result is called when two out of three viral target genes is detected. c Qualitative outcome of parallel testing of paired mid-turbinate swabs and saliva with the Abbott RealTime SARS-CoV-2 assay and covidSHIELD. 120 participants were enrolled in a clinical study comparing results from contemporaneously collected nasopharyngeal or mid-turbinate nasal swabs analyzed using both protocols. Overall concordance was 98.3% (95% CI, 94.1–99.8%), positive percent agreement was 96.8% (95% CI, 83.2–99.9%) and negative percent agreement was 98.9% (95% CI, 93.9–99.9%). All clinical trials were reviewed by the Western Institutional Review Board. All participants gave written and informed consent. d Additional clinical study outcome of 17 individuals confirmed to be positive for COVID-19 and to have low viral loads (Ct = 32–42, average 37) by mid-turbinate nasal swabs analyzed at the Johns Hopkins University School of Medicine using Abbott Alinity compared with contemporaneously collected saliva samples that were analyzed using the covidSHIELD assay at the University of Illinois Urbana-Champaign CLIA-registered laboratory. p-value = 0.0004 was calculated using 2-tailed, unpaired t-test. e Mock representative images from the Safer Illinois app. The screen on the left appears when a user is in compliance with the campus testing protocol and has received a recent negative test for SARS-CoV-2. The screen on the right appears when the user of the app is out of compliance, when they have had a recent exposure notification, or when they have tested positive for the virus.

Fig. 3. Deployment of frequent repeat testing at the University of Illinois in fall 2020.

a Timeline of detected cases during surveillance testing from July 6 to December 23, 2020. The top panel displays the daily new cases (blue) and the daily case positivity (orange). The daily case positivity is computed as the (number of new cases)/(unique number of individuals tested during the day). The lower panel shows the number of daily tests performed (green), which to an excellent approximation is the same as the number of unique individuals tested in a day. b Ct values of the first positive test as a function of time elapsed since the last negative test. The difference between the first (1 day) and the second (2 days) bins is highly statistically significant (p-value 2.2e−7), and that between the second (2 days) and the third (3 days) bins is statistically significant (p-value 3.7e−4). p-values shown are for the two-sided hypothesis of non-zero Pearson correlation between the number of days since the last negative test (x-axis) and the Ct value of the first positive test (y-axis). The exact sample sizes are: 341 patients who test positive 1 day since the last negative, 616 patients—2 days since the last negative, 716 patients—3 days since the last negative, and 1230 patients—between 4 and 7 days since the last negative. The box plots were generated using the command boxplot in Matlab 9.10.0.1684407(R2021a). The bottom and top of each box are the 25th and 75th percentiles of the sample, respectively. The distance between the bottom and top of each box is the interquartile range. The red line in the middle of each box is the median. The whiskers go from the end of the interquartile range to the furthest observation within the whisker length which is 1.5× the interquartile range. The outliers are marked with red + sign and are defined as the observations that go beyond the whisker length. c Head-to-head daily testing with covidSHIELD and antigen-based lateral flow assays in a subgroup of participants (n = 190) from October 1, 2020 to Dec 23, 2020. A total of 13,299 contemporaneous tests were performed. Of the 190 individuals, 6 tested positive for SARS-CoV-2 on Day 0 using the covidSHIELD test but all six tested negative using the antigen test. Blue and orange bars represent the percentage of participants that tested positive for SARS-CoV-2 on day 0 using covidSHIELD and the antigen test, respectively. d Box-and-whiskers plot of mean Ct values of the 6 individuals who tested positive for SARS-CoV-2 using the covidSHIELD assay on Day 0. Data was plotted using GraphPad Prism v9.3.1, where the lower and upper box extends from 25th to 75th percentiles, respectively, the line in the middle of the box is the median, the lower and upper error lines are the minimum and maximum value, respectively; each individual values are plotted as circles superimposed on the graph.

Fig. 4. Epidemiological analysis for COVID-19 cases on campus.

a Cases estimated by date of infection. b Estimated growth rate. c Estimated effective reproduction number R_eff by the date of infection as a function of time using the method of Cori et al. assuming gamma-distributed generation time distribution and a distribution between consecutive tests shown in Supplementary Fig. 7. The shaded periods (I–IV) correspond to periods of the semester when R_eff > 1 suggesting transient growth of cases. The unshaded time periods correspond to R_eff < 1 suggesting that the epidemic is controlled, and cases are decreasing. In each subplot, the credible intervals for the calculation of that output are shown with three shades. These three shades correspond to credible intervals of 20% (darkest green), 50% (lighter green), and 90% (lightest green). Analysis was done using Epidemiological toolbox EpiNow2. Please see the “Methods” section for more details.

Fig. 5. Mitigation of SARS-CoV-2 spread in the context of the larger Champaign-Urbana Community.

The daily number of 7-day averaged daily new cases between faculty/staff and residents in Champaign County a, undergraduates and faculty/staff b, and undergraduate students and residents in Champaign County c, for the period between August 15 and December 23. All points in three plots are colored according to their categories (orange: undergraduate students, blue: faculty/staff, green: residents in Champaign County). Pearson correlation coefficient, 95% confidence interval, and p-values for two-tailed test were calculated using GraphPad Prism software.

Fig. 6. Other communities with SHIELD.

SHIELD was deployed at other locations in Winter 2020/Spring 2021. The results from four representative examples are shown: a a university (1×/week), b a pair of high schools (2×/week), c a courthouse (2×/week), and d a large private corporate campus (2×/week).

Fig. 7. Other communities without SHIELD.

a and b Relationship between observed and predicted COVID-19 cases and mortality among counties with large university enrollments (student enrollment > 15,000 at start of Fall 2020 semester, n = 251). Predicted COVID-19 case and mortality rates were analyzed with a 3-week lag (to adjust for delays between exposure and outcome) using each county’s social vulnerability index (SVI) and age-adjusted COVID-19 mortality, accounting for state (due to policy differences in COVID-19 management). COVID-19 human case data^, and SVI were provided from CDC (2021) and population data was provided from the U.S. Census Bureau (2019). The lines indicate the fit lines for the linear regressions, with shaded areas indicating the 95% confidence intervals around the fits and circles indicating observations with Champaign County shown in orange and other Big 10 Conference Universities (n = 16*) in black. *Hennepin and Ramsey Counties (MN) were both used for University of Minnesota.