Rapid emergence of SARS-CoV-2 Omicron variant is associated with an infection advantage over Delta in vaccinated persons.

Background: The SARS-CoV-2 Omicron variant became a global concern due to its rapid spread and displacement of the dominant Delta variant. We hypothesized that part of Omicron's rapid rise was based on its increased ability to cause infections in persons that are vaccinated compared to Delta.
Methods: We analyzed nasal swab PCR tests for samples collected between December 12 and 16, 2021, in Connecticut when the proportion of Delta and Omicron variants was relatively equal. We used the spike gene target failure (SGTF) to classify probable Delta and Omicron infections. We fitted an exponential curve to the estimated infections to determine the doubling times for each variant. We compared the test positivity rates for each variant by vaccination status, number of doses, and vaccine manufacturer. Generalized linear models were used to assess factors associated with odds of infection with each variant among persons testing positive for SARS-CoV-2. Findings: For infections with high virus copies (Ct < 30) among vaccinated persons, we found higher odds that they were infected with Omicron compared to Delta, and that the odds increased with increased number of vaccine doses. Compared to unvaccinated persons, we found significant reduction in Delta positivity rates after two (43.4%-49.1%) and three vaccine doses (81.1%), while we only found a significant reduction in Omicron positivity rates after three doses (62.3%).
Conclusion: The rapid rise in Omicron infections was likely driven by Omicron's escape from vaccine-induced immunity. Funding: This work was supported by the Centers for Disease Control and Prevention (CDC).

Introduction

The emergence of SARS-CoV-2 variants continues to shape the COVID-19 pandemic. The success of the Alpha (lineage B.1.1.7) and Delta (B.1.617.2) variants that dominated the pandemic for most of 2021 was primarily driven by successive increases to their intrinsic transmissibility. As population immunity to SARS-CoV-2 increases through infections and vaccination, selection for variants that are partially resistant to the immune response, in particular neutralizing antibodies, should also increase. Mathematical modeling suggests that SARS-CoV-2 variants with increased transmissibility and partial immune escape may significantly increase infections even in a well-immunized population. A variant with these properties could significantly limit vaccine effectiveness against infections and lead to a new “wave” of COVID-19 cases.

The detection and rapid spread of the SARS-CoV-2 Omicron variant (B.1.1.529) in Botswana and South Africa grew as a global concern because it contained 15 mutations in the spike protein immunogenic receptor binding domain. , Subsequent in vitro assays showed that antibody-mediated neutralization using sera derived from vaccinees was significantly lower for Omicron than the previously dominant Delta variant.6, 7, 8, 9, 10, 11 For example, serum antibody neutralization from mRNA-1273 vaccinees within 3 months of the second vaccine dose was diminished 43x with Omicron compared to Delta and from BNT162b was diminished 122x. However, neutralization against Omicron was significantly enhanced after a booster vaccine dose, including for Ad26.COV2.S. , While these data suggest that Omicron may have an infection advantage over Delta in vaccinated persons, in vitro neutralization is not a direct correlate for human protection from infection.

The emergence of Omicron led to record-setting levels of COVID-19 cases in many parts of the world, even in well-vaccinated regions. , 14, 15, 16, 17 Using a population in southern Connecticut, USA, in which 48.5% have received at least one vaccine dose (including children and adults), we tested the hypothesis that the rapid increase in Omicron infections was at least partially influenced by its ability to cause infections in persons that are vaccinated compared with Delta. We established a surveillance system to differentiate Delta and Omicron cases using PCR and genome sequencing, and we selected a period in mid-December 2021 for the study when Delta and Omicron were relatively equal. From this period, we analyzed 37,877 nasal swab PCR test results and compared the Delta and Omicron positivity rates by the number of vaccine doses received. We confirmed our results using a logistic regression model to calculate the odds of detecting Omicron relative to Delta among infected persons and further assessed the effect of the number of COVID-19 vaccine doses and vaccine manufacturer (Ad26.COV2.S, mRNA-1273, or BNT162b2). We found that three vaccine doses were required to reduce Omicron positivity rates in our population, and that Omicron has an infection advantage in vaccinated persons relative to Delta that is proportional to the number of vaccine doses.

Results

Rapid emergence of Omicron

In late November 2021, we established a surveillance program in southern Connecticut, USA, to investigate the emergence of Omicron. At that time, BA.1 (also known as B.1.1.529.1) was the primary Omicron lineage spreading globally. Similar to the Alpha variant, Omicron BA.1 has a spike gene deletion (Δ69/70 HV) that causes “spike gene target failure” (SGTF) when using the ThermoFisher TaqPath COVID-19 Combo Kit qRT-PCR assay, allowing us to quickly identify potential Omicron infections. Yale New Haven Health (YNHH) uses TaqPath for testing mid-turbinate nasal swabs from symptomatic and asymptomatic outpatients for SARS-CoV-2 at collection sites in southern Connecticut. Our SGTF case definition included having an ORF1ab gene target PCR cycle threshold (Ct) of <30 and spike gene target “not detected.” We retrospectively applied the SGTF case definition to a cross-sectional study of samples collected since November 15, 2021, and prospectively to January 10, 2022 (Figure 1 A).

Figure 1

Variant case counts, test positivity, and odds of infection by vaccination status

(A) Number of persons infected with Delta and Omicron SARS-CoV-2 variants and the proportion of Omicron cases in southern Connecticut. Overlaid on the plot showing the number of positive cases is the proportion of Omicron variants (dots) with a fitted smoothed curve. The growth rate of Omicron compared to Delta during their respective emergence periods is shown in Figure S1.

(B) The proportion of positive SARS-CoV-2 PCR tests (Ct ≤ 30) for Delta and Omicron variants (using SGTF to differentiate) by vaccination status. Data shown as means with 95% confidence intervals. The positivity rate values are listed in Table 2.

(C) Odds of infection with Omicron relative to Delta variants by age, sex, and vaccination status among individuals who tested positive for SARS-CoV-2. We regressed the binary outcome for the SARS-CoV-2 variant (Delta as the reference group) and specified females and unvaccinated persons as the reference categories for the sex and vaccination status predictor variables in the model. ORs >1 indicate higher odds of detecting Omicron relative to Delta in persons testing positive for SARS-CoV-2 infection. Data shown as means with 95% confidence intervals. The OR values are listed in Table S1. The positivity rates and ORs stratified by vaccine manufacturers are shown in Figure S2 and Table S2.

Table 2

Positivity rates for Omicron and Delta among PCR tests performed between December 12 and 26, 2021. Rates shown in parentheses as means with 95% confidence intervals.

Variant	All (vaccinated & unvaccinated, n = 33,416)	0 dose: unvaccinated (n = 17914)	Vaccinated, 1 dose (n = 1,405)	Vaccinated, 2 doses ≥5 months before test (n = 1,186)	Vaccinated, 2 doses <5 months before test (n = 10,322)	Vaccinated, 3 doses (n = 2,589)
Delta (Ct ≤ 30)	1,374 (0.041: 0.039, 0.043)	954 (0.053: 0.05, 0.057)	57 (0.041: 0.03, 0.051)	32 (0.027: 0.018, 0.036)	306 (0.03: 0.026, 0.033)	25 (0.01: 0.006, 0.013)
Omicron (Ct ≤ 30)	1,387 (0.042: 0.039, 0.044)	793 (0.044: 0.041, 0.047)	52 (0.037: 0.027, 0.047)	58 (0.049: 0.037, 0.061)	433 (0.042: 0.038, 0.046)	51 (0.02: 0.014, 0.025)
Negative or Ct > 30	30,655 (0.917: 0.914, 0.92)	16,167 (0.902: 0.898, 0.907)	1,296 (0.922: 0.908, 0.936)	1,096 (0.924: 0.909, 0.939)	9,583 (0.928: 0.923, 0.933)	2,513 (0.971: 0.964, 0.977)
Combined (Delta and Omicron)	2,761 (0.083: 0.08, 0.086)	1747 (0.098: 0.093, 0.102)	109 (0.078: 0.064, 0.092)	90 (0.076: 0.061, 0.091)	739 (0.072: 0.067, 0.077)	76 (0.029: 0.023, 0.036)

We detected the first sample meeting our SGTF case definition on December 4, 2021, which we sequence-confirmed as Omicron lineage BA.1. We sequenced a subset of samples collected from November 22 to December 27 (n = 695), and 100% (216/216) of the SGTF samples were confirmed as Omicron (BA.1), and 100% (479/479) of samples without SGTF (i.e., the spike gene was detected) were confirmed as Delta (B.1.617.2 or AY.x; Data S1). This established our case definitions as adequate proxies for Omicron (BA.1) and Delta infections during our study period.

We found that Omicron became the dominant variant in our population 16 days after its first detection (December 20, 2021; Figure 1A). Fitting an exponential curve to cumulative cases, we estimated that Omicron cases doubled every 3.1 days (95% confidence interval [CI]: 2.8–3.4), 4.3x shorter than the initial doubling time for Delta during its emergence period from April 18 to July 29, 2021 (13.4 days [95% CI: 12.5–14.5]; Figures S1A–S1C). The rapid emergence of Omicron in southern Connecticut was also associated with a rapid rise in COVID-19 cases (Figure 1A), as seen in many places around the world. When we first detected Omicron, the US Centers for Disease Control and Prevention estimated that 71%–74% of the population in southern Connecticut had completed a primary COVID-19 vaccine series (1 dose of Ad26.COV2.S or two doses of mRNA-1273 or BNT162b2). Therefore, we hypothesized that part of the rapid increase in Omicron infections stemmed from its increased ability to cause infections in persons that are vaccinated compared with Delta.

Omicron versus Delta in vaccinated persons

To investigate if Omicron is more likely than Delta to cause infections in vaccinated persons, we analyzed 37,877 nasal swab PCR tests conducted from December 12 to 26 when the total number of probable Delta and Omicron infections were relatively equal (Delta = 1,374/2,761, 49.8%; Omicron = 1,387/2,761, 50.2%; Figures 1A and S2). We conducted a medical records review to identify that the 37,877 tests during that period were from 33,416 unique persons with known vaccination status. Since some individuals tested multiple times during the study period, only the first test was included. For each PCR test, we collected information on age and sex of the person tested, test date, test outcome (negative, positive >30 Ct, positive ≤30 Ct Delta, and positive ≤30 Ct Omicron; Figure S2), and date and manufacturer of each COVID-19 vaccine administered (Ad26.COV2.S, mRNA-1273, and/or BNT162b2). We excluded persons who indicated in their records a preference to opt out of research, and the number of doses was regarded as those taken at least 14 days before the SARS-CoV-2 test. In our population (including children and adults), 53.6% were unvaccinated, 46.4% received at least one vaccine dose, 42.2% received at least two vaccine doses, and 7.5% received three vaccine doses. Additional details regarding the characteristics of the population are provided in Table 1 .

Table 1

Demographic characteristics among PCR tests performed between December 12 and 26, 2021

	Number of vaccine doses
Age group, y	0 (n = 18,072)	1 (n = 1,594)	2 (n = 11,537)	3 (n = 2,212)
0–5	2,859 (15.8%)	66 (4.1%)	37 (0.3%)	0 (0%)
6–15	3,921 (21.7%)	424 (26.6%)	930 (8.1%)	1 (0.1%)
16–30	3,570 (19.8%)	192 (12.1%)	2,033 (17.6%)	162 (7.3%)
31–45	3,855 (21.3%)	377 (23.7%)	2,953 (25.6%)	508 (23.0%)
46–60	2,416 (13.4%)	307 (19.3%)	3,009 (26.1%)	626 (28.3%)
>60	1,451 (8.0%)	228 (14.3%)	2,575 (22.3%)	915 (41.4%)
Sex
Female	9,842 (54.5%)	847 (53.1%)	6,879 (59.6%)	1,398 (63.2%)
Male	8,219 (45.5%)	747 (46.9%)	4,656 (40.4%)	814 (36.8%)
Unknown	11 (0.1%)	0 (0%)	2 (0.0%)	0 (0%)

We then calculated the ≤30 Ct test positivity rates for each variant stratified by vaccination status (Figure 1B, Table 2 ). We found that the positivity rate among unvaccinated persons was higher for Delta than Omicron (5.3% [95% CI: 5.0%–5.7%] versus 4.4% [95% CI: 4.1%–4.7%], p < 0.0001). We found similar results in persons who received a single vaccine dose. Conversely, our results show that Omicron had higher positivity rates than Delta among those who received two doses within 5 months (Omicron: 4.2% [95% CI: 3.8%–4.6%] versus Delta: 3% [95% CI: 2.6%–3.3%], p < 0.0001), two doses more than 5 months ago (Omicron: 4.2% [95% CI: 3.8%–4.6%] versus Delta: 2.7% [95% CI: 1.8%–3.6%], p=0.007), and three vaccine doses (Omicron: 2% [95% CI: 1.4%–2.5%] versus Delta: 1.0% [95% CI: 0.6%–1.3%], p=0.04). Our estimates of Omicron positivity rates in persons receiving one or two vaccine doses were not significantly lower than unvaccinated persons but were 49.7% lower after three doses. In comparison, the reduction in Delta positivity rates from unvaccinated to two vaccine doses was 45.6%–49.6% and to three vaccine doses was 83.2% (Table S2). Despite the higher positivity rates for Omicron in vaccinated persons, we still found that 57.2% (793/1,387) of the Omicron infections in our population occurred in persons who were unvaccinated, and 96.3% (1,336/1,387) were eligible for one or more vaccine doses at the time of PCR testing.

We confirmed our ≤30 Ct test positivity analysis by calculating the odds of detecting Omicron relative to Delta using a logistic regression model (Figure 1C; Tables S1 and S2). We used the first SARS-CoV-2 test in the logistic regression model as some persons were tested multiple times. For infections among persons who were vaccinated, we found higher odds that they were infected with Omicron (versus Delta), and that the odds appeared to increase with increased number of vaccine doses (1 dose odds ratio [OR] = 1.3 [95% CI: 1.8–2.0]; two doses ≥5 months OR = 2.3 [95% CI: 1.5–3.7]; two doses <5 months OR = 1.9 [95% CI: 1.5–2.2]; three doses OR = 3.0 [95% CI: 1.8–4.9]). The odds of infection did not vary by sex or age, and our results were similar when we stratified the data by Ad26.COV2.S, mRNA-1273, or BNT162b2 (Figure S2). These findings support our hypothesis that Omicron has an infection advantage in vaccinated persons relative to Delta.

PCR cycle thresholds by variant and vaccination status

Next, we sought to determine if infection advantage for Omicron relative to Delta in vaccinated persons (Figure 1) was related to virus copies in the nasal passage. We compared the mean nasal swab PCR Ct values by variant category (Omicron or Delta) and stratified by the number of vaccine doses received (Figure 2 A). Lower PCR Ct values correspond to higher virus copies. Combining positive tests from unvaccinated and vaccinated persons, we found that the overall mean PCR Ct values were higher for infections with Omicron than Delta (Omicron = 20.96 [95% CI: 12.70–29.21] versus Delta = 20.68 [95% CI: 11.02–30.34], p < 0.001, Kruskal-Wallis test; Figures 2A and 2B). Similarly, the PCR Ct values were consistently lower for Omicron compared to Delta across all vaccination categories in our population, although we only found statistically significant differences in persons vaccinated with two doses (Omicron = 21.19 [95% CI: 12.67–29.71] versus Delta = 20.62 [95% CI: 11.10–30.14], p = 0.049; Figures 2A and 2B).

Figure 2

Effect of sex, age, variant, and vaccination status on the nasal swab PCR cycle threshold

(A) Nasal swab PCR cycle threshold (Ct) values for the Delta and Omicron SARS-CoV-2 variants by vaccination status. Data shown as means with 95% confidence intervals.

(B) Association of age, sex, and vaccination status with PCR Ct values. The effect sizes >1 indicate a higher CT value (lower virus RNA) for Omicron compared to Delta, males relative to females, and vaccinated relative unvaccinated persons who received different doses. Data shown as means with 95% confidence intervals. The OR values are shown in Table S3.

(C) Association of age, sex, vaccination status, and vaccine manufacturer with PCR Ct values. Data shown as means with 95% confidence intervals. The OR values are shown in Table S4.

To adjust for age, sex, vaccine doses, and vaccine manufacturers, we compared nasal swab PCR Ct values of Omicron relative to Delta by fitting a regression model with a Gaussian family distribution. After adjusting for covariates, we found that the PCR Ct values were consistent across vaccine doses, but confirming our analysis above, Omicron infections had higher Ct values (i.e., lower virus copies) than those infected with Delta (OR = 1.55, 95% CI: 1.10–2.17; Figure 2B; Table S3). We found similar trends for the different vaccine manufacturers (Figure 2C; Table S4). Our results suggest that the enhanced transmissibility of Omicron, and its ability to cause infections in vaccinated persons compared to Delta, is not from higher nasal passage virus copies.

Discussion

We hypothesized that the rapid emergence and spread of the SARS-CoV-2 Omicron variant was partly due to its increased ability to evade immunity from prior infection and/or vaccination. Using a study population seeking outpatient testing when Omicron and Delta were overall relatively equal among infections, we found that Omicron has a relatively higher propensity to cause infections in COVID-19-vaccinated persons. Furthermore, our results show that the advantage of Omicron compared to Delta increases with the number of vaccine doses. Although we were not able to study the impact of prior infections, a recent study from South Africa estimated that Omicron had an increased risk of causing SARS-CoV-2 reinfections compared with the Beta (B.1.351) or Delta variants, consistent with our hypothesis. Considering the high vaccination rates and the recent “wave” of Delta infections, the large increase in COVID-19 cases caused by Omicron is likely due in part to a larger population of persons susceptible to Omicron infection that were protected from Delta.

Our findings should not be interpreted as implying that vaccination increases the risk for Omicron infections. On the contrary, vaccination decreased the positivity rates for Omicron, and most (57.2%) Omicron infections in our population occurred in persons that were unvaccinated or eligible for a booster dose. Thus, further vaccination would have likely decreased the number of Omicron infections. What our findings imply is that the reductions in infections from vaccination is greater for Delta than Omicron. Compared to unvaccinated persons, we found that three vaccine doses were required to significantly reduce the Omicron positive rate (∼55% reduction), which was similar to the reduction in Delta positivity rates from two doses and significantly lower than the Delta reduction from three doses (∼81%). While we did not design this study to directly measure vaccine effectiveness, our results are consistent with vaccine effectiveness studies indicating that a third/booster vaccine dose is needed to significantly reduce Omicron infections.19, 20, 21, 22, 23, 24 To maintain effectiveness against new divergent SARS-CoV-2 variants, the administration schedule for COVID-19 vaccines designed to the original (“Wuhan-Hu-1”) SARS-CoV-2 spike gene sequence needs to be continuously evaluated. Overall, this further highlights the need for variant-specific or broad-acting coronavirus vaccines as a long-term solution.

We demonstrate that the ability to cause infections in vaccinated persons and increased transmissibility of Omicron compared to Delta is not associated with higher virus copies in the nasal passage. First, our data add further evidence supporting that although vaccination reduces the likelihood of SARS-CoV-2 variants to establish infection, once infected, vaccination does not significantly reduce virus copies in diagnostic samples. We show this for both Omicron and Delta, though we previously reported that vaccination can shorten the duration of infection. Second, the increased transmissibility of some previous variants may have been driven by increased viral loads, causing persons to be more infectious. For example, the displacement of Alpha by Delta in mid-2021 was associated with increased virus copies for Delta in diagnostic samples. , In contrast, the rapid growth rate of Omicron, as shown by our estimates of ∼4.3x shorter doubling time compared to Delta, was associated with lower virus copies in nasal swabs (as also shown with anterior nares/oropharyngeal combined swabs). Thus, the increased transmissibility of Omicron relative to Delta may stem from a combination of immune evasion, lower infectious dose, and/or a change in infection tropism to the upper respiratory tract that potentially shortens the generation time and serial interval between infections. , 31, 32, 33, 34, 35, 36

In conclusion, escape from vaccine-induced immunity likely contributed to the rapid rise in Omicron infections. Our findings may also explain why Omicron has been associated with more reinfections. While Omicron was more likely to cause infections in vaccinated persons than Delta, vaccination remains effective in reducing severe disease, even for Omicron. Together with the rebound of vaccine effectiveness after administering a booster dose, measures to expand the uptake of the primary vaccine series and additional booster doses remain an important strategy for controlling the COVID-19 pandemic.

Limitations of the study

Our study had several limitations. First, probable Delta and Omicron infections were inferred based on the SGTF PCR data. Although we validated the SGTF results by sequencing a representative number of samples, we could not sequence every positive sample. Moreover, we classified SARS-CoV-2 infections as probable Delta or Omicron only from samples with high virus copies (Ct < 30), which may be biased against Omicron as Omicron infections tend to have higher PCR Ct values than Delta. Second, although our vaccination history data is extensive, our records may not have captured some administrations. We excluded persons with incomplete vaccine information from analysis, but this did not significantly decrease the sample size. Third, we did not have access to data on previous positive test results, serology, or household attack rates, which would have allowed us to study reinfections and variant-specific transmissibility. Fourth, while our data were from outpatients testing for a variety of reasons, including COVID-19 symptoms, or pre-travel, -event, or -procedure, we did not have access to this level of information for each person. Asymptomatic testing for travel or parties increased during the holidays, which can decrease the test positivity rates. However, such changes would not likely introduce a significant bias against either variant. Fourth, the demand for SARS-CoV-2 tests was high during the study period, causing many people to conduct at-home tests or forego testing altogether. Vaccinated persons, especially those who received a booster dose, may have been less likely to seek a PCR test if they were asymptomatic. Since we compare Omicron to Delta by vaccine dose, this change in healthcare-seeking behavior would not likely impact our findings. Fifth, we did not directly assess the effectiveness of COVID-19 vaccines against the Delta and Omicron variants; therefore, any potential implied conclusions regarding the vaccine effectiveness and immunity against these variants should be interpreted with caution. Finally, our study compared the odds of detecting Omicron relative to Delta among infected persons by vaccine administration; our findings should not be erroneously interpreted as vaccination increases the risk for infection with Omicron.

STAR★Methods

Key resources table

Resource availability

Lead contact

Further information and requests for data, resources, and reagents should be directed to and will be fulfilled by the Lead Contact, Nathan D. Grubaugh (nathan.grubaugh@yale.edu).

Materials availability

This study did not generate new unique reagents.

Experimental model and subject details

Ethics statement

The Institutional Review Board from the Yale University Human Research Protection Program determined that obtaining de-identified test results linked to vaccination status and sequencing of de-identified remnant COVID-19 clinical samples obtained from clinical partners conducted in this study is not research involving human subjects (IRB Protocol ID: 2000031374).

Study participants

Our study consisted of 34,980 unique persons that tested for SARS-CoV-2 (37,877 tests) from outpatient sites, including mass testing locations, in New London, New Haven, and Fairfield Counties, Connecticut, through Yale New Haven Health (YNHH). Provided indications for testing were being symptomatic for COVID-19, exposure to a known case of COVID-19, required testing (e.g. for work, school, or travel), and testing prior to undergoing an aerosol generating procedure. The participants included a diversity of ages from 0 to 5 to >60, and 55% were female. We did not obtain information about race or ethnicity. We obtained COVID-19 vaccination information from each person by combining information from the YNHH system’s electronic medical records and the Connecticut immunization registry (CT-WiZ), the latter to capture possible out-of-system vaccinations. However, it is possible that some out-of-state vaccinations were missed. The vaccinated persons received Ad26.COV2.S, mRNA-1273, and/or BNT162b2. Details regarding the characteristics of the population are provided in Table 1.

Study outcomes

We quantified the positivity rates for the Omicron and Delta SARS-CoV-2 variants in our cross-sectional study, and estimated the ORs of detecting Delta in persons testing positive by sex, age, and vaccination status category. We also calculated the doubling times (in days) for the Omicron and Delta variants to understand their transmissibility. Finally, we assessed the association between the nasal swab PCR Ct value and sex, age, variant, and vaccination status category stratified by vaccine manufacturer.

Method details

PCR testing for variant differentiation

Mid-turbinate nasal swabs from outpatient collection sites were tested for SARS-CoV-2 by the YNHH COVID-19 and Clinical Virology Laboratories using the MagMAX viral/pathogen nucleic acid isolation kit and TaqPath COVID-19 Combo Kit. The TaqPath qRT-PCR assay reports Ct values from three SARS-CoV-2 gene targets: ORF1ab, spike, and nucleocapsid. ORF1ab with Ct values <30 were investigated for spike gene detection. If the spike gene was detected, the sample was categorized as “probable Delta” and if the spike gene was not detected (i.e., SGTF), the sample was categorized as “probable Omicron”.

Sequence confirmation of variants

Mid-turbinate nasal swabs in viral transport media were received from SARS-CoV-2 infections from YNHH. Nucleic acid was extracted from 300 μL of the original sample using the MagMAX viral/pathogen nucleic acid isolation kit, eluting in 75 μL of the elution buffer. The extracted nucleic acid was again tested for SARS-CoV-2 RNA using a “research use only” (RUO) RT-qPCR assay, which generates an SGTF result similar to the TaqPath assay. For rapid confirmation of the initial suspected Omicron samples with SGTF, we used the NEBNext ARTIC SARS-CoV-2 Companion Kit and sequenced pooled libraries on the Oxford Nanopore Technologies (ONT) MinION. The standard NEB protocol with PCR Bead Cleanup was slightly modified by using V4 or V4.1 primer pools for amplicon generation, by including an additional bead cleanup step (1:1 beads:sample) after the NEBNext end prep reaction, and by scaling up the barcode ligation reaction by using 16 μL of end-prepped DNA. Final pooled libraries were quantified using the Qubit High Sensitivity dsDNA kit, and the ONT SQK-LSK109 protocol was followed to prime and load the ONT MinION for sequencing. Samples were processed in sets of 14–46 samples with two negative controls. The RAMPART application developed by the ARTIC Network was used to monitor the sequencing run until sufficient coverage was reached (https://artic.network/ncov-2019/ncov2019-using-rampart.html). The ARTIC bioinformatics pipeline was used to generate consensus genomes with fast basecalling done by MinKNOW (https://artic.network/ncov-2019/ncov2019-bioinformatics-sop.html). A threshold of 20x coverage was used to call consensus genomes, and negative controls were confirmed to completely consist of Ns.

For routine sequencing of samples with nucleocapsid gene target Ct values ≤35, we used the Illumina COVIDSeq Test RUO version. The protocol was slightly modified by using V4 primers for amplicon generation, by lowering the annealing temperature of the amplicon generation step to 63 °C, and by shortening the tagmentation step to 3 min. Final libraries were pooled and cleaned before quantification with the Qubit High Sensitivity dsDNA kit. The resulting libraries were sequenced using a 2x150 approach on an Illumina NovaSeq at the Yale Center for Genome Analysis. Each sequenced sample had at least one million reads. Samples were typically processed in sets of 93 or 94 with negative controls incorporated during the RNA extraction, cDNA synthesis, and amplicon generation steps. The reads were aligned to the Wuhan-Hu-1 reference genomes (GenBank: MN908937.3) using BWA-MEM v.0.7.15. Adaptor sequences were trimmed, primer sequences were masked, and consensus bases were called with simple majority >60% frequency using iVar v1.3.1 and SAMtools v1.7. An ambiguous ‘N’ was used when fewer than 20 reads were present at a site. In all cases, negative controls were analyzed and confirmed to consist of at least 99% Ns. For both rapid and routine sequencing, Pangolin v.3.1.17 was used to assign lineages. Consensus genomes were submitted to GISAID and included in weekly updates on our website (https://covidtrackerct.com/).

Quantification and statistical analysis

Variant growth rates

We calculated daily variant proportions using SGTF samples as a proxy for Omicron and sequence-confirmed lineages for Delta from samples obtained by YNHH. We smoothed these daily variant proportions using a 7-day rolling average. We defined the emergence period for Omicron and Delta as the time since its first sequence-confirmed detection to when the variant reached 95% of total samples in our dataset We defined Delta’s emergence period as April 18, 2021 to July 29, 2021 (102 days), and Omicron’s emergence period as December 4, 2021 to January 7, 2022 (34 days).We multiplied the daily variant proportions by the daily fitted cases from Covidestim for the three counties in our study to determine the number of variant cases during the emergence periods. Using these data, we ran a logistic regression analysis for each variant separately, with a sample corresponding to a specific variant category as the binary outcome and the number of days since the first detection of the variant as the predictor. We plotted the smoothed fitted curves for the emergence periods with their 95% CIs (Figure S1A), which shows the probability of a given case belonging to a specific variant category over time. We estimated the doubling time by fitting an exponential curve to cumulative cases over time for each variant and dividing log(2) by the resulting coefficient. We show the total fitted cases for each emergence period in Figure S1B.

Positivity rates

The PCR positivity rates for each variant were estimated using the ORF1ab Ct values ≤30 and SGTF signatures to define as Omicron or Delta. For this analysis, ORF1ab Ct values from 30 to 40 were included as “negatives” as we could not assign a variant category, and thus the variant-specific positivity rates that we show are not the true overall test positivity rates. We estimated the positivity rates for different SARS-CoV-2 variants as the proportion of persons testing positive during the study period with PCR Ct value < 30 for the ORF1ab and S gene targets. To estimate the test positivity by vaccination status, we counted the number of doses received >14 days before the SARS-CoV-2 test. We calculated the CIs for the proportion based on the standard errors for the binomial distribution. We show each rate with the 95% CI.

Odds of infection with omicron relative to delta

To assess the odds of detecting Omicron relative to Delta variant in infected persons, we fitted a logistic regression model to determine the effect of the covariates, namely, sex, age, and vaccination status stratified by the vaccine manufacturer. Similarly, we fitted a generalized linear regression with Gaussian distribution to assess the association between the ORF1ab PCR Ct value with covariates, namely, sex, age, and vaccination status stratified by the vaccine manufacturer. We specified females and unvaccinated persons as the reference categories for the sex and vaccination status covariates in the model. To estimate the odds of infection with Omicron relative to Delta by vaccination status, we counted the number of doses received >14 days before the SARS-CoV-2 test.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Deposited data
Confirmed COVID-19 cases	Yale New Haven Hospital	https://www.ynhh.org/
Estimated COVID-19 infections	Covidestim	https://covidestim.org/
Data and software availability
R	RStudio	R version 4.0.3 https://cran.r-project.org/; https://www.rstudio.com/
Validation of spike gene target failure (SGTF) data	Mendeley Data	https://doi.org/10.17632/t3d3nd5wb9.1
Other
SARS-CoV-2 variant frequencies	Yale University, Yale New Haven Hospital	https://github.com/grubaughlab/2022_paper_omicron-v-delta