Protection Conferred by COVID-19 Vaccination, Prior SARS-CoV-2 Infection, or Hybrid Immunity Against Omicron-Associated Severe Outcomes Among Community-Dwelling Adults

Abstract

Introduction

We assessed protection from coronavirus disease 2019 (COVID-19) vaccines and/or prior severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection against Omicron-associated severe outcomes during successive sublineage-predominant periods.

Methods

We used a test-negative design to estimate protection by vaccines and/or prior infection against hospitalization/death among community-dwelling, polymerase chain reaction (PCR)-tested adults aged ≥50 years in Ontario, Canada, between 2 January 2022 and 30 June 2023. Multivariable logistic regression was used to estimate the relative change in the odds of hospitalization/death with each vaccine dose (2–5) and/or prior PCR-confirmed SARS-CoV-2 infection (compared with unvaccinated, uninfected subjects) up to 15 months since the last vaccination or infection.

Results

We included 18 526 cases with Omicron-associated severe outcomes and 90 778 test-negative controls. Vaccine protection was high during BA.1/BA.2 predominance but was generally <50% during periods of BA.4/BA.5 and BQ/XBB predominance without boosters. A third/fourth dose transiently increased protection during BA.4/BA.5 predominance (third-dose, 6-month: 68%, 95% confidence interval [CI] 63%–72%; fourth-dose, 6-month: 80%, 95% CI 77%–83%) but was lower and waned quickly during BQ/XBB predominance (third-dose, 6-month: 59%, 95% CI 48%–67%; 12-month: 49%, 95% CI 41%–56%; fourth-dose, 6-month: 62%, 95% CI 56%–68%, 12-months: 51%, 95% CI 41%–56%). Hybrid immunity conferred nearly 90% protection throughout BA.1/BA.2 and BA.4/BA.5 predominance but was reduced during BQ/XBB predominance (third-dose, 6-month: 60%, 95% CI 36%–75%; fourth-dose, 6-month: 63%, 95% CI 42%–76%). Protection was restored with a fifth dose (bivalent; 6-month: 91%, 95% CI 79%–96%). Prior infection alone did not confer lasting protection.

Conclusions

Protection from COVID-19 vaccines and/or prior SARS-CoV-2 infections against severe outcomes is reduced when immune-evasive variants/subvariants emerge and may also wane over time. Our findings support a variant-adapted booster vaccination strategy with periodic review.

Omicron became the globally predominant circulating variant of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in early 2022 and since has evolved into ever more transmissible and immune-evasive sublineages [1, 2]. Since December 2021, Canada's National Advisory Committee on Immunization (NACI) has recommended at least one vaccine booster dose for all adults aged ≥18 years. In the province of Ontario, the first booster (third dose) was offered to all community-dwelling adults by December 2021, and the second booster (fourth dose) to those aged ≥60 years in April 2022, and to all adults by July 2022 [3, 4]. On 1 September 2022, the Moderna BA.1 bivalent vaccine was authorized by Health Canada for adults aged ≥18 years, and on 7 October 2022, the Pfizer-BioNTech BA.4/BA.5 bivalent vaccine was authorized for individuals aged ≥12 years. Moderna BA.1 and Pfizer-BioNTech BA.4/BA.5 were introduced in Ontario through a population-based booster program shortly after authorization, in September and October 2022, respectively [5, 6]. Following their introduction, bivalent vaccines were the preferred products; however, NACI recommended receiving any available messenger RNA (mRNA) booster dose product to ensure timely protection [7]. Therefore, original monovalent mRNA vaccines were still available and accessible as booster doses. As of May 2023, over 70% of Ontario adults aged ≥50 years had received at least 1 booster dose [8].

Although booster doses restore some protection against Omicron [5, 9–11], it is unclear if such protection is sustained and to what extent emergent sublineages will impact vaccine effectiveness. Vaccine effectiveness estimates of booster doses during periods when more immune-evasive sublineages, such as BA.4/BA.5, BQ, and XBB, have become predominant are lacking. Furthermore, several studies suggest that prior infection alone or “hybrid immunity” (from a combination of vaccination and a recovered SARS-CoV-2 infection) may offer sustained protection against Omicron-associated severe outcomes [12–14]. To inform future vaccination strategies, we investigated possible waning of protection from booster vaccine doses and/or the influence of prior SARS-CoV-2 infections against Omicron-related severe outcomes, from when Omicron first emerged and the following 18 months as the variant evolved.

METHODS

Study Design, Population, and Data Sources

We conducted a population-based, test-negative design study using linked databases in Ontario to investigate protection conferred by coronavirus disease 2019 (COVID-19) mRNA vaccination and/or prior SARS-CoV-2 infection against Omicron-associated severe outcomes (hospitalization or death) over time.

The study population included community-dwelling adults aged ≥50 years who underwent ≥1 polymerase chain reaction (PCR) test for SARS-CoV-2 from 2 January 2022 to 30 June 2023. In Ontario, the Omicron variant and its sublineages represented nearly 100% of all positive samples since late January 2022. We excluded individuals who had received only 1 dose of COVID-19 vaccine, did not receive 1 dose of mRNA vaccine in the first 2 doses, had Delta variant infections (for the index PCR test), were immunocompromised, or received a third dose earlier than 3 November 2021 (Supplementary Figure 1). Among vaccinated subjects, 96% had received exclusively mRNA vaccines (Moderna Spikevax monovalent, Pfizer-BioNTech Comirnaty monovalent, Moderna Spikevax BA.1 bivalent, or Pfizer-BioNTech Comirnaty BA.4/BA.5 bivalent) [9].

We used provincial data from ICES (formerly the Institute for Clinical Evaluative Sciences), an independent, non-profit research institute whose legal status under Ontario's health information privacy law allows it to collect and analyze health care and demographic data for health system evaluation and improvement. As described in previous studies [3–5, 9], SARS-CoV-2 laboratory testing, COVID-19 surveillance, COVID-19 vaccination, and health administrative data sets were linked using unique encoded identifiers and analyzed at ICES. The use of the data in this study is authorized under section 45 of Ontario's Personal Health Information Protection Act and does not require review by a research ethics board.

Definitions and Sampling

We defined cases as COVID-19-associated hospitalization or death, due to, or partially due to SARS-CoV-2 infection as confirmed by PCR. Under Ontario's provincial COVID-19 surveillance program, hospitalization data were entered only for patients who received in-hospital treatment and/or their duration of hospital stay was extended due to COVID-19 [5, 9]. We excluded hospitalizations when the diagnostic specimen was collected >3 days post-admission. We defined controls as individuals who were symptomatic and tested negative for SARS-CoV-2 by PCR at healthcare settings, irrespective of outcome [3, 5]. Cases and controls were sampled by week of test; therefore, controls could be included in the study repeatedly. However, a control could not re-enter the study once classified as a case. This sampling strategy was used to help control for calendar time in the analysis.

We defined the index date as the date of specimen collection, hospitalization, or death, whichever was earliest. We classified subjects according to their vaccination status (vaccinated with ≥2 doses, or unvaccinated) and the number of doses received as of the index date. The primary series was defined as 2 monovalent vaccine doses, and boosters as third to seventh doses, either monovalent or bivalent [6, 9]. Prior infection was defined as PCR-confirmed SARS-CoV-2 infection (by any pre-Omicron and Omicron virus strains) ≥ 60 days before the index date [15]. Results of rapid antigen tests were unavailable for study. We defined “time since the last immunogenic event” with respect to the index date as the time since the most recent vaccine dose or prior infection, whichever occurred later. Protection against hospitalization or death was reported at 3 monthly increments (to a maximum of 15 months), since the last immunogenic event, and according to Omicron sublineage predominant periods. A predominant period was defined as ≥50% of the sequenced samples being confirmed with a specific Omicron sublineage. In Ontario, the predominant sublineages were identified as BA.1/BA.2 (2 January to 18 June 2022), BA.4/BA.5 (19 June to 26 November 2022), and BQ/XBB (27 November 2022 to 30 June 2023) (Figure 1). The grouping was based on available evidence indicating differences in vaccine effectiveness between these sublineages due to immune evasion and other factors [1, 2, 5, 7, 10].

Statistical Analyses

We used means and proportions to describe the study population; cases and controls were compared, and the differences quantified using standardized differences (SD) during the BA.1/BA.2, BA.4/BA.5, and BQ/XBB-predominant periods (SD ≥0.1 was considered potentially clinically relevant).

We used multivariable logistic regression to estimate the relative change in the odds of severe outcomes (defined as the proportionate reduction in the odds of hospitalization or death), through vaccination or hybrid immunity (vaccination and documented prior infection), compared to the reference group of subjects with neither vaccination nor prior infection. The estimated protection was calculated using the formula: (1-adjusted odds ratio) × 100%, and reported at 3, 6, 9, 12, and 15 months since the last immunogenic event. The models were adjusted for sex, age category (50–59, 60–69, 70–79, ≥80 years), public health unit region, neighborhood-level sociodemographic variables (income quintile, essential worker quintile, persons per dwelling quintile, self-identified visible minority quintile), influenza vaccination status (proxy for health behaviors), number of SARS-CoV-2 tests within 3 months prior to 14 December 2020 (proxy for healthcare workers), comorbidities, receipt of home care services, and week of test (modeled using restricted cubic splines with knots at weeks 9, 28, 37, and 49) as previously described [3–5, 9]. A 3-way interaction term between the number of vaccine doses (coded as a categorical variable), prior infection status (yes or no), and the log-transformed time since last immunogenic event was used. Because the timing of vaccine dose administration with respect to a sublineage-predominant period varied, the available periods of observation for boosters could be shorter than 15 months. In the analyses where immunogenic events were restricted to vaccination, the results were interpreted as vaccine effectiveness. We further estimated the marginal protection from hybrid immunity in the same manner, comparing vaccinated individuals with/without prior infection at the corresponding number of vaccine doses and time point. We conducted a post-hoc sensitivity analysis using individuals who received their last COVID-19 vaccine more than 12 months ago, regardless of the cumulative number of doses, as the reference group to estimate the marginal protection from booster doses among those without documented prior infections.

We used SAS version 9.1 (SAS Institute Inc., Cary, North Carolina, USA) for all analyses. All tests were 2-sided, and a P value <.05 indicates statistical significance.

RESULTS

We included 18 526 cases with Omicron-associated severe outcomes and 90 778 test-negative controls (among a total of 91 113 unique individuals), with 54 994, 29 362, and 24 948 subjects in the BA.1/BA.2-, BA.4/BA.5-, and BQ/XBB-predominant periods, respectively. In all 3 periods, cases were older than controls, and higher proportions were male or had comorbid conditions; fewer had prior PCR-confirmed SARS-CoV-2 infections (0.9% vs 4.8%, 1.7% vs 6.5%, and 3.7% vs 6.3%, respectively), and fewer had received COVID-19 vaccines (≥2 doses, 65.9% vs 95.6%, 86.1% vs 95.2%, and 88.8% vs 94.4%, respectively) [Table 1]. In the BA.1/BA.2-predominant period, booster vaccines (3–5 doses) had been received by 2665 (34.3%) and 35 091 (74.3%) of cases and controls, respectively. In the BA.4/BA.5-predominant period, booster vaccines (3–6 doses) had been received by 4302 (70.1%) and 19 068 (82.1%) of cases and controls, respectively; 7.9% of these were bivalent vaccines, given as the fourth (8.1%) or fifth (97.9%) dose. In the BQ/XBB-predominant period, booster vaccines (3–7 doses) had been received by 3459 (74.8%) and 16 571 (81.5%), respectively; 48.5% of these were bivalent vaccines (BA.1-containing, 24.1%; BA.4/BA.5-containing, 24.3%), given mostly as the fourth (51.9%) and fifth (99.2%) dose. Few subjects had received sixth or seventh doses. Information on bivalent booster vaccine use by period and dose count is provided in Supplementary Table 1.

Protection Conferred by Vaccination

Among subjects with no PCR-confirmed prior infection (>94% of this cohort), we estimated vaccine effectiveness at 3, 6, 9, 12, and 15 months since the last immunogenic event (ie, last vaccine dose), where applicable (Figure 2, panel left [gray]; Supplementary Table 2). With the primary 2-dose series, protection against severe outcomes throughout the BA.1/BA.2-predominant period was approximately 80% (eg, at 6 months: 80% [95% CI, 78%–82%]). During the BA.4/BA.5-predominant period, protection declined markedly to below 50% (eg, at 6 months: 46% [95% CI, 33%–57%]), and further reduced to below 30% during the BQ/XBB-predominant period. Reduced protection was also observed for boosters across these periods. For example, protection at 6 months with a third-dose decreased from 94% (95% CI, 93%–95%) to 68% (95% CI, 63%–72%) to 59% (95% CI, 48%–67%) in the BA.1/BA.2-, BA.4/BA.5-, and BQ/XBB-predominant periods, respectively; whereas with a fourth-dose, protection decreased from 80% (95% CI, 77%–83%) to 62% (95%CI, 56%–68%) in the BA.4/BA.5- and BQ/XBB-predominant periods, respectively. During the BA.4/BA.5-predominant period, there was a trend toward increased protection with more booster doses. During the BQ/XBB-predominant period this was less apparent (fifth-dose at 6 months: 69%; 95% CI, 63%–74%); and we also observed waning protection as time elapsed. For example, protection at 12 months was 49% (95% CI, 41%–56%) with a third-dose and 51% (95% CI, 41%–56%) with a fourth-dose.

Protection by Prior SARS-CoV-2 Infection Among the Unvaccinated

Among the unvaccinated, prior infection alone was associated with protection generally >90% throughout the BA.1/BA.2-predominant period. However, during the BA.4/BA.5-predominant period, protection waned from 85% at 6 months to 76% at 12 months and was only 32% at 12 months during the BQ/XBB-predominant period (Figure 2, Supplementary Table 2).

Protection Conferred by Hybrid Immunity

We estimated protection conferred by hybrid immunity against severe outcomes at 3, 6, 9, 12, and 15 months since the last immunogenic event, where applicable (Figure 2, panel right [black]; Supplementary Table 2). We observed high levels of protection (>90%) during the BA.1/BA.2-predominant period, lasting for at least 12 months. Protection was slightly lower during the BA.4/BA.5-predominant period but increased with booster doses (eg, third-dose at 6 months 88% [95% CI, 82%–92%]; fourth-dose at 6 months 91% [95% CI, 81%–95%]), and remained high for 9–12 months. However, protection decreased during the BQ/XBB-predominant period for the primary series, and third, and fourth dose recipients (eg, third-dose at 6 months: 60% [95% CI, 36%–75%], 12 months 63% [95% CI, 48%–73%]; fourth-dose at 6 months: 63% [95% CI, 42%–76%], 12 months 49%; [95% CI, 1%–47%]) and also tended to wane as time elapsed. Protection was restored for those who had received a fifth dose (at 6 months: 91% [95% CI, 79%–96%]).

Marginal Protection Conferred by Hybrid Immunity

Among vaccinated subjects, prior infection (compared with no prior infection) was associated with significantly greater protection against severe outcomes for the corresponding number of vaccine doses and time points, during the BA.1/BA.2-predominant (for second-dose, at 3–12 months; third-dose, at 3–6 months) and BA.4/BA.5-predominant (for second-dose, at 6–15 months; third-dose, at 3–12 months; fourth-dose, at 3–6 months) periods (Figure 3, Supplementary Table 3). It was generally not significant at most time points during the BQ/XBB-predominant period, except for those who received a fifth dose, at 6–9 months.

Marginal Protection Conferred by Booster Doses

The marginal vaccine effectiveness conferred by booster doses using individuals who received the last vaccine dose more than 12 months ago as the reference group exhibited similar trends but were consistently lower than the corresponding estimates of vaccine effectiveness using unvaccinated individuals as the reference (Figure 4; Supplementary Table 4).

DISCUSSION

In this population-based study spanning 3 Omicron sublineage-predominant periods over 1.5 years, we found that vaccine effectiveness against severe outcomes of the two-dose primary vaccination series alone was generally below 50% during the BA.4/BA.5- and BQ/XBB-predominant periods. Booster vaccine doses transiently increased protection during the BA.4/BA.5-predominant period, but protection was lower and waned relatively quickly over months during the BQ/XBB-predominant period. Prior infection alone also did not provide lasting protection. Hybrid immunity was associated with higher levels of protection in earlier periods but was reduced and waned more quickly during the BQ/XBB-predominant period. Our results suggest that a variant-adapted booster vaccination strategy is necessary as the virus continues to evolve.

Our findings are in line with a recent meta-analysis that found marked reduction of vaccine effectiveness with the primary series against hospitalizations and deaths in the Omicron era; booster vaccine doses partially restored protection but it waned over months [11]. This study provides new data on Omicron sublineages and extended observation up to 15 months since the last vaccine dose. Our point estimates of protection conferred by the primary series among subjects without documented prior infection were all <50% during the BA.4/BA.5-predominant and <30% during the BQ/XBB-predominant periods; even with a booster (third or fourth) dose, protection was generally <80% during the former and <62% after 6 months during the latter period, and declined progressively as time elapsed. We found that repeated booster dosing, alongside with the introduction of bivalent vaccines, given mostly as a fifth dose, restored protection to some degree (about 70% at 6 months), similar to other studies [16–22]. Vaccine effectiveness of different bivalent vaccine products among adults in Ontario has been explored separately [9]. Our analysis on marginal protection using subjects vaccinated longer than 12 months prior as the reference group showed consistent results.

On hybrid immunity, a recent meta-regression analysis of 15 studies showed high levels of protection (>90%) against severe outcomes with either primary series or booster vaccine doses, which lasted for at least 6 months [12]. However, the analysis only included studies before June 2022, thus the impact of Omicron sublineages that emerged later (ie, BA.4, BA.5, BQ, and XBB) were not examined. Our study found similar point estimates at around 90%, lasting for >6 months, during both BA.1/BA.2- and BA.4/BA.5-predominant periods; and we showed significantly greater (marginal) protection with hybrid immunity compared with vaccination alone at almost all time points. Notably, such protection dropped significantly to generally <70% when the more immune-evasive BQ/XBB sublineages became predominant, only to be restored with a fifth dose (nearly all being bivalent vaccines, Supplementary Table 1) [23, 24]. Although earlier studies had reported nearly 80% protection against severe outcomes with prior infection alone (ie, unvaccinated), lasting for about 6 months in the BA.1/BA.2- and BA.4/BA.5-predominant periods [12–14], we observed progressive decline from 6 to 15 months during the latter period. Protection was very low during the BQ/XBB-predominant period. Taken together, we found that neither vaccine-induced, infection-induced, nor hybrid immunity provide long-lasting protection against severe COVID-19 outcomes as virus sublineages continue to evolve over time. Our data support the approach to review vaccine composition based on continual surveillance and offer booster doses periodically, especially for high-risk individuals, even in immune-experienced populations where vaccine coverage is high and prior SARS-CoV-2 infection is prevalent [25].

The strengths of our study include being population-based, the large sample size, evaluation of multiple Omicron sublineage-predominant periods, stratification by the latest vaccine dose and longer observation durations, and adjustment for available confounding factors. Our study has several limitations. We could only capture prior infections confirmed by PCR tests performed in healthcare settings in Ontario; results of rapid antigen tests were unavailable. Canada-wide data indicate that infection-acquired seropositivity has risen steadily during the Omicron waves among older adults (from <10% to about 62% by February 2023) [26], thus undocumented infections are probable. Despite some level of infection-induced immunity existing—likely among both vaccinated subjects and the unvaccinated reference group—we still found greater risk reduction with the combination of vaccination and prior PCR-confirmed disease, suggesting that the contemporary observed vaccine effectiveness has been modified by the collective immune experience in the population. It is unclear if pauci-symptomatic or asymptomatic infections (untested or diagnosed by antigen tests in community settings) generate less durable protection, which deserve further study [27, 28]. Whether there was differential impact on the vaccinated/unvaccinated comparators as a growing proportion of population was infected over time, leading to some degree of underestimation of vaccine effectiveness is uncertain, and cannot be ruled out [29]. Second, we were unable to evaluate protection against specific Omicron sublineage due to limited sequencing data, but our categorization according to sublineage-predominant periods allowed study of the temporal changes as the virus evolved. Finally, although the restoration of protection with bivalent boosters (fifth dose) in the BQ/XBB-predominant period was reassuring, we acknowledge that at the time of writing, the observation periods for the fifth doses were relatively short, and few subjects had received sixth/seventh doses.

In conclusion, among community-dwelling adults, protection from COVID-19 vaccines and/or prior SARS-CoV-2 infections against severe outcomes is reduced when immune-evasive variants/subvariants emerge and may also wane over time. Protection was overall enhanced with hybrid immunity, whereas neither prior infection, prior vaccination alone, nor hybrid immunity provides lasting protection. Our results indicate that a variant-adapted booster vaccination strategy with periodic review is likely necessary even in immune-experienced populations with high vaccine coverage and/or prior infection prevalence.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments and disclaimers. The authors acknowledge the Canadian Immunization Research Network (CIRN) Provincial Collaborative Network (PCN) Investigators, Public Health Ontario for access to vaccination data from COVaxON, case-level data from the Public Health Case and Contact Management Solution (CCM) and coronavirus disease 2019 (COVID-19) laboratory data, as well as assistance with data interpretation. They also thank the staff of Ontario's public health units who are responsible for COVID-19 case and contact management and data collection within CCM. They thank IQVIA Solutions Canada Inc. for use of their Drug Information File. The authors are grateful to the Ontario residents without whom this research would be impossible.

This document used data adapted from the Statistics Canada Postal Code^OM Conversion File, which is based on data licensed from Canada Post Corporation, and/or data adapted from the Ontario Ministry of Health Postal Code Conversion File, which contains data copied under license from ^©Canada Post Corporation and Statistics Canada. Parts of this material are based on data and/or information compiled and provided by: Ministry of Health (MOH), Ontario Health, the Canadian Institute for Health Information, Statistics Canada, and IQVIA Solutions Canada Inc. The analyses, conclusions, opinions, and statements expressed herein are solely those of the authors and do not reflect those of the funding or data sources; no endorsement is intended or should be inferred. Adapted from Statistics Canada, Canadian Census 2016. This does not constitute an endorsement by Statistics Canada of this product.

Financial support. This work was supported by funding from CIRN through a grant from the Public Health Agency of Canada and the Canadian Institutes of Health Research (CNF 151944) and also by funding from the Public Health Agency of Canada, through the Vaccine Surveillance Working Party and the COVID-19 Immunity Task Force. This study was supported by Public Health Ontario and by ICES, which is funded by an annual grant from the Ontario MOH and Ministry of Long-Term Care (MLTC). This work was also supported by the Ontario Health Data Platform (OHDP), a Province of Ontario initiative to support Ontario's ongoing response to COVID-19 and its related impacts. J. C. K. is supported by a Clinician-Scientist Award from the University of Toronto Department of Family and Community Medicine. The study sponsors did not participate in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review or approval of the manuscript; or the decision to submit the manuscript for publication.

References

Author notes