SARS-CoV-2 Vaccine Breakthrough by Omicron and Delta Variants, New York, USA

Alexander C. Keyel; Alexis Russell; Jonathan Plitnick; Jemma V. Rowlands; Daryl M. Lamson; Eli Rosenberg; Kirsten St. George


Emerging Infectious Diseases. 2022;28(10):1990-1998. 

In This Article


Omicron Emergence

In analysis 1, >80% of 272 case-patient/control pairs were 18–69 years of age; most were from the Capital and Mid-Hudson regions (Table 1; Figure 1). Among controls, 8% had received a booster, and among case-patients, 22% had received a booster. Among controls, 56.6% were unvaccinated; among case-patients, 30% were unvaccinated (Table 1). Sample sizes were 177 for Pfizer, 109 for Moderna, and 22 for Janssen vaccine recipients. The variables most associated with an Omicron lineage identity were vaccination (OR 3.1, 95% CI 2.0–4.9; p<0.001) and booster status (OR 6.7, 95% CI 3.4–13.0; p <0.001) (Table 2; Figure 2).

Figure 2.

Visualization of the fixed effects from the second Omicron emergence analysis on a log-odds scale (without age matching) in a study of SARS-CoV-2 vaccine breakthrough by Omicron and Delta variants, New York, USA. Odds scale in Appendix ( Stratum-specific effects were often strong but were excluded for visual clarity. Increased values indicate an increased probability of infection with Omicron instead of Delta. Lines show ± 1 SE.

In analysis 2 (309 pairs), when patient age was removed as a matching criterion, younger age was also predictive of an Omicron infection; log-odds of infection with Omicron generally decreased as age increased (OR 0.962, 95% CI 0.950–0.974) (Table 2). Significant patterns beyond this log-linear age effect were found for persons in 2 age groups. log-odds of infection with Omicron were lower for persons 0–4 years of age than predicted by a log-linear age effect alone (Figure 2) and higher for persons 18–29 years of age than predicted by a log-linear age term alone; risk was highest for those 18–29 years of age (Figure 2). OR estimates for vaccination status (OR 4.8, 95% CI 2.8–8.1) and booster status (OR 38.5, 95% CI 15.9–93.2) were higher than in the analysis that used age as a matching criterion (Table 2).

In analysis 3 (vaccinated-only persons, 129 pairs), the probability of infection with Omicron decreased with an increased number of days after the last vaccine dose (OR 0.996, 95% CI 0.993–0.999) (Table 2). Vaccine type was also included in the top statistical models (Appendix Table 1) and the trend toward reduced odds of Omicron infection after vaccination with the Janssen vaccine was borderline significant (OR 0.351, 95% CI 0.132–0.935, relative to Pfizer vaccine; OR 0.388, 95% CI 0.149–1.009, relative to any mRNA vaccine) (Table 2).

Delta Emergence

In analysis 1 (55 pairs), 75% were 18–69 years of age; 89% of case-patients/controls were from the Finger Lakes, Long Island, and the Mid-Hudson regions (Table 3). A total of 74.5% of controls and 61.8% of case-patients were unvaccinated (Table 3). Vaccine type, time from last vaccination, and an interaction of the 2 were not significantly associated with an increased likelihood of infection with Delta than any other virus lineage in the fully matched conditional logistic regression (Table 4). Vaccination status was the top model (OR 2.4, 95% CI 0.8–6.8; p = 0.08). Vaccine type had no significant effect (p = 0.12), but estimated ORs were 2.9 (95% CI 0.9–8.9) for Pfizer, 0.38 (95% CI 0.04–4.2) for Moderna, and 2.0 (95% CI 0.17–23.6) for Janssen.

The power analysis showed that a sample size of 110 (55 pairs) would have a 15%–45% chance of obtaining a significant result for an OR of 2 under the simulated probability distributions. A sample size of ≥255 would be needed to have ≥80% power for an OR of 2. A sample size of 110 could have ≤78% power to detect an OR of 3 and 93% power to detect an OR of 4. A sample size of 24 could detect an OR of 22 with 80% power but would only have 36% power to detect an OR of 4.

When case-patients and controls were no longer matched on the basis of age (66 pairs), vaccine type was the top model (Appendix Table 2), suggesting that odds of being infected with Delta rather than any other virus lineage increased by a factor of 7.3 (2.0–26.7) for those receiving the Pfizer vaccine relative to unvaccinated persons. Effects for Moderna (2.0, 95% CI 0.25–17.1) and Janssen (0.46, 95% CI 0.04–4.76) vaccines were substantial but not individually significant.