Genomic surveillance of SARS-CoV-2 Omicron variants on a university campus

Universities have been important settings for COVID-19 surveillance and outbreaks due to congregate living and high contact rates. With the designation of Omicron (B.1.1.529) as a variant of concern in late November 2021 and its rapid global spread, there was a need to characterize its emergence and transmission dynamics in highly vaccinated communities. Prior work suggested Omicron’s enhanced transmissibility, partial immune escape, and high rates of vaccine breakthrough. Evidence on comparative viral loads and household transmission metrics versus Delta was mixed, and most serial interval studies did not leverage genomic confirmation within households. This study aimed to use a university-wide testing program to track the rapid emergence of Omicron, compare clinical characteristics, viral kinetics, and transmission dynamics to Delta, and integrate genomic epidemiology to assess within-residence clustering and serial intervals.

Background literature cited indicates Omicron’s rapid epidemic expansion and enhanced transmissibility relative to prior variants; decreased vaccine effectiveness against Omicron with improvements after booster doses; mixed findings on whether Omicron viral loads are lower or comparable to Delta; and higher household attack rates with shorter serial intervals for Omicron versus Delta. Many earlier household and cluster studies lacked genomic confirmation to rule out multiple introductions. The study builds on university-based surveillance literature demonstrating the value of campus testing programs for real-time epidemiology and public health decision-making.

Setting and participants: The Husky Coronavirus Testing research study at the University of Washington provided free SARS-CoV-2 testing to students, staff, and faculty from September 2021 to February 2022 under consistent campus mitigation (indoor masking, improved ventilation, gathering size limits, and mandatory vaccination with near-universal primary series completion by January 2022). Eligibility included English-speaking university affiliates. Participants completed daily attestations via SMS and were offered tests for new symptoms, known exposure, recent travel, or on request. Data were collected via REDCap. Specimen collection: Self-collected anterior nasal swabs were obtained via three modalities: observed at kiosks, unobserved dropbox returns, or courier returns. Two swab types were used due to supply issues: US Cotton #3 swabs (for unobserved courier returns and some observed collections) and RHINostics RH-5000 swabs (for observed kiosk and unobserved dropbox collections). Laboratory testing: Dry swabs were eluted and processed using Swab-Express extraction-free RT-qPCR targeting SARS-CoV-2 Orf1b and S genes multiplexed with human RNase P as internal control. Positive results required detection of viral targets in 3–4 reactions plus RNase P in ≥3 reactions. The Delta-associated S gene 157–158 deletion caused S-gene target failure/delay in this assay. Viral sequencing: Sequencing was attempted for positives with Orf1b Ct ≤30. Nucleic acids were extracted (MagNA Pure 96), libraries prepared with Illumina COVIDSeq, and ARTIC V4 primers were adopted Nov 18, 2021, switching to V4.1 on Jan 12, 2022, to better cover Omicron. Sequencing used Illumina NextSeq 2000, with consensus genomes assembled against MN908947 using a modified iVar-based pipeline. Consensus sequences were deposited to GISAID; BA.1 and BA.2 designations included parental lineages and sublineages. Epidemiologic and statistical analysis: The first positive swab date per person defined the infection date. Symptom proportions were compared by variant using Pearson’s chi-squared tests. For symptomatic individuals in shared-residence clusters with identical genomes, serial intervals (index symptom onset to subsequent case onset) were compared by variant via Mann–Whitney U tests. Ct analyses used multiple linear regression on Orf1b Ct values (to avoid S-gene assay differences), adjusting for age, symptom status, days since symptom onset (for symptomatic subset), vaccination status and timing, and average RNase P value; analyses were restricted to RHINostics swabs (excluding US Cotton #3) where specified. BA.1 vs BA.2 Ct differences were evaluated similarly. Phylogenetics and introductions: Consensus genomes were combined with Washington State genomes from GISAID (screened with Nextclade; poor-quality sequences excluded). Pairwise nucleotide distances ignored positions with Ns. Maximum likelihood phylogenies were built via Nextstrain Augur pipeline: masking ends and specific positions, alignment with MAFFT, tree inference with IQ-TREE, rooting on Wuhan/Hu-1, clock rate estimation, and visualization with Auspice. To estimate introductions to campus, campus genomes were combined with all contemporaneous WA State genomes (Delta: N≈15,406; Omicron: N≈14,359), and Augur ‘traits’ was used to infer state transitions (campus vs community) at internal nodes; introduction counts were validated by subsampling sensitivity analyses. Transmission dynamics modeling: Using all sequenced samples and observed daily case counts, the team reconstructed variant-specific incidence (multinomial likelihood for variant assignment and negative binomial likelihood for total cases) and estimated time-varying effective reproduction numbers (R) for Delta and Omicron, incorporating observed shorter serial intervals for Omicron.

- Testing volume and positivity: 74,955 samples from 24,393 participants yielded 3,630 positives (4.8%) representing 3,038 individuals (Sep 10, 2021–Feb 14, 2022). Sequencing of 2,101 samples from 1,939 individuals detected only Delta (N=209) and Omicron (N=1,730). - Rapid displacement: Omicron rapidly displaced Delta on campus over ~two weeks in December 2021, becoming the predominant strain and driving a larger outbreak. - Vaccination context: Among cases with known vaccination status, primary series completion was high in both variants (Delta ~97%, Omicron ~97%). Booster receipt ≥2 weeks before infection was 2.1% for Delta versus 28.5% for Omicron. Time since primary series was longer for Omicron cases (median 217 days, IQR 125–292) than Delta (median 194 days, IQR 169–224). - Symptoms: Loss of taste/smell was more frequent with Delta (11.1%) than Omicron (2.8%, P<0.001). Fever, myalgia, and chills were more common among Omicron cases; some gastrointestinal and upper respiratory symptoms varied by variant (e.g., rhinorrhea, diarrhea), though detailed rates were heterogeneous. - Viral load (Ct) comparisons: Adjusted analyses showed higher mean Orf1b Ct values for Omicron than Delta, indicating lower semiquantitative viral loads in Omicron by approximately 1.07–1.30 Ct on average (e.g., Variant [Omicron vs Delta] coefficient ~1.08–1.30 across models; P≤0.001). Higher Ct (lower viral load) was associated with longer duration since symptom onset and higher RNase P values. No significant Ct difference was found between BA.1 and BA.2. - Household clustering and serial interval: In 78 within-residence clusters with identical genomes (N=173 individuals; Delta N=25, Omicron N=148), the median serial interval between symptomatic index and first subsequent case was shorter for Omicron (2 days) than Delta (6 days; P=0.021). All cases within a household were detected within ≤15 days. - Effective reproduction number (R): Estimated R peaked at 1.85 (95% CrI 1.3–2.4) during the Delta outbreak (Sept–Oct) and at 2.4 (95% CrI 1.9–2.8) during the Omicron outbreak (Dec–Jan), with Omicron remaining above 1 for a longer period. - Introductions: Phylogenetic trait inference estimated 83 separate Delta introductions into campus represented by sequenced samples (≈2.5 sequenced cases per introduction). A similar approach was applied to Omicron using 14,359 WA Omicron genomes to estimate introductions. - Phylogenetics: Delta genomes formed several monophyletic clusters with high bootstrap support (100%), whereas Omicron genomes were more evenly distributed across the tree and had lower average bootstrap support, reflecting rapid diversification and uncertainty in specific transmission links.

Within a highly vaccinated university community with consistent mitigation measures, Omicron rapidly supplanted Delta and drove a substantial surge, driven by higher transmissibility and shorter serial intervals. Genomic confirmation within-residence clusters reduced confounding by multiple introductions and supported a true reduction in serial interval for Omicron relative to Delta. Symptom profiles differed by variant—loss of taste/smell was more characteristic of Delta, while systemic symptoms (fever, myalgia, chills) were more frequent with Omicron—although clinical overlap limits differential diagnosis based on symptoms alone. Higher peak effective reproduction numbers and rapid spread of Omicron were observed despite high vaccine coverage, emphasizing immune escape and shorter generation times as key factors. The study underscores the value of integrating genomic surveillance with campus testing programs to monitor variant dynamics, inform isolation and quarantine guidance, and detect changes in transmission patterns in near real time.

This study documents the rapid replacement of Delta by Omicron in a university population and quantifies Omicron’s higher transmissibility (higher R) and shorter serial intervals compared to Delta, within a highly vaccinated setting. Omicron infections displayed modestly higher Ct values (lower semiquantitative viral loads) relative to Delta, and symptom patterns differed between variants. These findings highlight the importance of university-based genomic surveillance to guide timely public health responses. Future work should expand multi-institutional surveillance, refine serial interval and generation time estimates by setting and vaccination/booster status, assess durability of protection against evolving sublineages, and integrate wastewater and environmental sampling with clinical genomics to better resolve introduction events and transmission networks.

- No routine surveillance testing of the entire campus population; participation-based sampling may bias case detection. - Missing follow-up symptom data for some individuals may have led to misclassification of asymptomatic status. - Vaccination status relied on self-report, with limited ability to validate against potentially incomplete state registries. - Change in swab type during the study could affect Ct/viral load measurements; Ct analyses were restricted to a single swab type to mitigate this. - Repeat infections were not included. - Single-campus study population that is younger and generally healthier than the general population, limiting generalizability. - Considerable phylogenetic uncertainty for Omicron limited direct comparison of fine-scale transmission patterns to Delta.