COVID-19 Vaccine Effectiveness Against Omicron Infection and Hospitalization

We included community-dwelling children aged 5 to 11 years who had COVID-19-relevant symptoms or a severe outcome attributable to COVID-19 and who were tested for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) by reverse-transcription polymerase chain reaction (RT-PCR) between January 2 and August 27, 2022, as recorded in the Ontario Laboratories Information System. Children who tested positive at least once during the study period were considered cases. For those with multiple positive tests, the first was used. Once an individual became a case, they could not re-enter the study. The index date was the date of specimen collection. Children who were asymptomatic at the time of testing, who had symptoms that were unrelated to COVID-19 (incidental SARS-CoV-2 infection), or who had no information on their symptoms at the time of testing were excluded. The list of COVID-19-related symptoms has been described elsewhere.^10,11 

Severe outcomes attributable to Omicron infection were defined as either hospitalization or death from COVID-19, identified from the Public Health Case and Contact Management Solution (CCM).^10–14  CCM records hospitalizations only if cases are admitted because of COVID-19, and not if admitted for other reasons with incidental detection of SARS-CoV-2. The index date for severe cases was the earliest of specimen collection, hospitalization, or death. Cases of multisystem inflammatory syndrome in children were not included because they are inadequately captured by CCM.

Symptomatic children with a negative SARS-CoV-2 RT-PCR test were considered controls. For children with multiple negative tests during the same calendar week, a randomly selected negative test was used. Children with multiple negative tests in different calendar weeks during the study period could be considered controls multiple times. Controls could later be considered cases if they subsequently tested positive.

Because of overwhelming volumes, diagnostic RT-PCR testing was restricted to certain populations, including hospitalized patients, patients seen in emergency departments, outpatients for whom COVID-19 treatment was being considered, household members of patient-facing healthcare workers, and symptomatic students who received a RT-PCR self-collection kit through their school.¹⁵  At-home SARS-CoV-2 rapid antigen detection tests could be accessed by those not among these groups. By January 2, 2022, >95% of circulating SARS-CoV-2 cases in Ontario were determined by representative surveillance to be B.1.1.529 (Omicron) sublineage BA.1/BA.2, and starting on July 2, 2022 the majority were BA.4/BA.5.⁶  Thus, cases were assumed to be Omicron throughout the study period, unless they were determined by whole genome sequencing or S-gene target failure screening to be B.1.617 (Delta), in which case they were excluded.

We excluded immunocompromised children (defined as solid organ or hematopoietic stem cell transplant recipients, taking any immunosuppressive medication, or those with HIV, cancer, sickle cell disease, or any other immunocompromising conditions);^16,17  those who received Moderna’s mRNA-1273 vaccine (because of limited use in this age group), a non-Health Canada-authorized COVID-19 vaccine, or three doses of COVID-19 vaccine within a span of 12 weeks (suggesting an immunocompromising condition); those who were tested after their third dose; and those tested <14 days after their first dose, or tested positive <90 days before testing (for cases).

COVID-19 vaccination status, vaccine product, dose administered, date of administration, dose number, and dosing interval were obtained from COVaxON, a centralized information system that includes comprehensive documentation of all COVID-19 vaccination events in Ontario.

Age, sex, and postal code were obtained from the Ontario Registered Persons Database. Comorbidities were determined from various databases using validated algorithms and commonly accepted diagnostic codes, as described elsewhere.¹⁶  Receipt of influenza vaccination (as a measure of health behaviors) during the 2019 to 2020 or 2020 to 2021 influenza seasons was obtained using physician and pharmacist billing claims from the Ontario Health Insurance Plan and Ontario Drug Benefit databases, respectively. COVID-19 vaccination of the child’s mother was retrieved using COVaxON and the MOMBABY database, an ICES-derived dataset linking children born in Ontario hospitals to their mothers using birth hospitalization records.¹⁸  Maternal COVID-19 vaccination before April 1, 2021 was used as a proxy to identify children of healthcare workers, given that only individuals aged ≥80 years, long-term care home residents and staff, Indigenous adults, and healthcare workers were eligible for COVID-19 vaccines in Ontario before that date.¹⁹  The public health unit of residence was determined using the postal code and Statistics Canada Postal Code Conversion File Plus (version 7B). Neighborhood-level information on household income quintile, household density quintile, and essential workers quintile was obtained from 2016 Census data using residential postal codes.

All datasets were linked using unique encoded identifiers and analyzed at ICES (formerly the Institute for Clinical Evaluative Sciences).^10–14 

Means (continuous variables) and frequencies (categorical variables) of participants’ baseline characteristics were calculated. Standardized differences were used to compare test-positive cases and test-negative controls and vaccinated (≥1 dose) and unvaccinated children.

Multivariable logistic regression models were employed to calculate adjusted odds ratios of vaccination in test-positive cases versus test-negative controls after adjusting for potential confounders. Confounders included week of test, age, sex, medical comorbidities, prior influenza vaccination, being a child of a healthcare worker, positive SARS-CoV-2 RT-PCR test ≥90 days before specimen collection date, household income quintile, household density quintile, essential workers quintile, and public health region.^20–22 

We calculated vaccine effectiveness (VE) using the formula: VE = (1–adjusted odds ratio) x 100%. We estimated VE against infection and severe outcomes caused by Omicron by the number of vaccine doses received; for children who had received only 1 dose by the index date, we estimated VE ≥14 days after the first dose, whereas for children who had received 2 doses, we estimated VE ≥7 days after the second dose.^1,23  VE was also estimated by varying time periods after vaccination. For children who had received 2 doses, we estimated VE by dosing interval and by both dosing interval and time since second dose receipt. Characteristics of children with different dosing intervals were compared using analysis of variance for continuous variables and χ² tests for categorical and dichotomous variables.

All analyses were conducted using SAS Version 9.4 (SAS Institute). All tests were 2-sided and used P < .05 as the level of statistical significance.

Use of data were authorized under section 45 of Ontario’s Personal Health Information Protection Act; research ethics board review was not required.

A total of 114 637 SARS-CoV-2 RT-PCR testing episodes occurred in children aged 5 to 11 years between January 2 and August 27, 2022. We excluded 9792 episodes based on predefined exclusion criteria (Fig 1). An additional 90 193 episodes were also excluded, with 10 261 testing episodes occurring in children who were asymptomatic, and 79 932 testing episodes with no information on symptoms or symptoms unrelated to COVID-19. We included 14 673 testing episodes, with 1004 children (7.5%) having multiple testing episodes eligible for inclusion during the study period. Among the included episodes, 6284 (45.4%) were considered symptomatic test-positive cases and 8389 (54.6%) were considered symptomatic test-negative controls.

Compared with test-negative controls, test-positive cases were older, more likely to reside in areas with more persons per dwelling, less likely to be vaccinated against COVID-19, less likely to have had a positive SARS-CoV-2 RT-PCR test >90 days before index date, less likely to have received prior influenza vaccination, and less likely to be the child of a healthcare worker (Table 1). Among children who had received 2 doses, the mean dosing interval was longer for test-negative controls than for test-positive cases (52.3 [standard difference (SD) 17.7] days versus 47.2 [SD 15.2] days).

Compared with unvaccinated children, vaccinated children were older, more likely to reside in the highest neighborhood income quintile, more likely to have received prior influenza vaccination or to be the child of a healthcare worker, and less likely to reside in lower income areas (Table 2).

Overall, VE against symptomatic infection was 9% (95% CI, 0% to 18%) ≥14 days after a first dose, and 49% (95% CI, 43% to 54%) ≥7 days after a second dose. VE declined over time after each dose, from a peak of 24% (95% CI, 8% to 36%) 14 to 29 days after a first dose to −1% (95% CI, −17% to 13%) ≥60 days after a first dose, and from a peak of 66% (95% CI, 60% to 71%) 7 to 29 days after a second dose to −4% (95% CI, −30% to 17%) ≥120 days after a second dose (Fig 2A, Supplemental Table 3).

VE was notably higher for children with longer dosing intervals. Children with a dosing interval of ≥56 days had a VE of 57% (95% CI, 51% to 62%), compared with children with a dosing interval of 15 to 27 days (12% [95% CI, −11% to 30%]) and 28 to 41 days (38% [95% CI, 28% to 47%]) (Fig 3A, Supplemental Table 3).

However, when stratifying VE by dosing interval and time since second dose receipt, we observed declining VE over time within each dosing interval group, although confidence intervals frequently overlapped (Fig 3B, Supplemental Table 4). We also noted higher VE for children with longer dosing intervals for some time-since-second-dose periods but not others. For example, 30 to 59 days after a second dose, VE was only 6% (95% CI, −47% to 40%) for children with a dosing interval of 15 to 27 days, compared with 60% (95% CI, 52% to 67%) for children with a dosing interval of ≥56 days. However, VE estimates became comparable between all dosing intervals at 90 to 119 days after vaccination, with confidence intervals overlapping 0% ≥120 days after vaccination.

Children with a shorter dosing interval were more likely to have received prior influenza vaccination (P < .001), be the child of a healthcare worker (P < .001), reside in Toronto (P < .001), and reside in higher income areas (P < .001) or more persons per dwelling (P < .001) compared with children with an extended dosing interval (Supplemental Table 5).

We identified 138 cases of severe outcomes (COVID-19-related hospitalization or death), with ≤5 deaths. Compared with test-negative controls, cases of severe COVID-19 were older, more likely to reside in lower income areas, more likely to have comorbidities, less likely to be vaccinated against COVID-19, less likely to have received prior influenza vaccination, and less likely to be the child of a healthcare worker.

The majority of severe COVID-19 cases occurred in unvaccinated children (n = 84, 60.9%), with few cases occurring in children who had received 1 dose (n = 30, 21.7%) or 2 doses (n = 24, 17.4%). VE against severe outcomes was 79% (95% CI, 63% to 88%) ≥7 days after a second dose. VE against severe outcomes also showed suggestions of waning, declining from 94% (95% CI, 57% to 99%) 7 to 29 days after a second dose to 57% (95% CI, −20% to 85%) ≥120 days after a second dose, but confidence intervals overlapped (Fig 2B, Supplemental Table 3).

We found that for children aged 5 to 11 years, VE against symptomatic Omicron infection was low after 1 dose of BNT162b2, and moderate after 2 doses, with waning VE over time after each dose. Against severe outcomes, VE was very high initially after a second dose (94%) but may also wane over time. These findings have important implications for recommendations on the use of COVID-19 vaccines in children.

Data on COVID-19 VE in children aged 5 to 11 years are still emerging. A test-negative design study employing data from a US pharmacy-based network found that VE against symptomatic infection during Omicron predominance decreased from 60.1% (95% CI, 54.7% to 64.8%) 2 to 4 weeks after a second dose to 28.9% (95% CI, 24.5% to 33.1%) 2 months after a second dose.²⁴  Similar decreasing trends in VE were observed in cohort studies, including in New York state, where the incidence rate ratio of Omicron infection in unvaccinated versus vaccinated children decreased from 3.1 (95% CI, 2.7 to 3.6) within 13 days after vaccination to 1.1 (95% CI, 1.1 to 1.2) 42 to 48 days after vaccination, and in Italy, where VE against infection peaked at 38.7% (95% CI, 37.7% to 39.7%) 0 to 14 days after vaccination and decreased to 21.2% (95% CI, 19.7% to 22.7%) 43 to 84 days after vaccination.^25,26  Another cohort study from Israel reported VE against symptomatic infection to be 48% (95% CI, 29% to 63%) at 7 to 21 days after the second dose, but longer-term trends were not provided.²⁷  The overall VE seen in our study seems comparable to these previously published estimates, yet the VE observed at 60 to 89 days after a second dose (51% [95% CI, 43% to 58%]) was higher than what has been previously described, with significant decreases occurring only starting 90 days after vaccination (35% [95% CI, 21% to 46%]). It is possible that this slower decline in VE against Omicron infection relates to the extended dosing interval that was employed for a significant proportion of our study population. Although the previously referenced studies did not provide specific information on the dosing intervals that were employed, a 3-week dosing interval was preferentially recommended in Israel, Italy, and the United States.^28–30  Our study revealed low VE at ≥120 days after vaccination with 95% confidence intervals overlapping 0% (−4% [95% CI, −29% to 17%]), which may relate to waning protection with time and lower VE against BA.4/BA.5 compared with BA.1/BA.2. Although VE against severe outcomes also showed signs of waning at ≥90 days after vaccination, confidence intervals overlapped.

Direct comparisons with other jurisdictions are challenging because of differences in study methodology, outcome definitions, sensitivity of administrative and surveillance data, and criteria for hospitalization. Compared with studies from Ontario that used the same databases to assess VE against Omicron infection in other age groups, we found that VE in children aged 5 to 11 years 7 to 59 days after a second dose was comparable to estimates for adolescents aged 12 to 17 years (51% [95% CI, 38% to 61%]), and higher than estimates reported for adults (36% [95% CI, 24% to 45%]).^12,31 

In our study, VE was overall higher for children with longer dosing intervals, with slower waning after a second dose following an interval of ≥56 days, although VE became comparable ≥90 days after vaccination. Multiple immunogenicity studies have demonstrated that extended dosing intervals result in higher spike antibodies, receptor-binding domain antibodies, neutralizing antibodies, and specific T-cell responses compared with shorter intervals, likely reflecting immune maturation between prime-boost events.^32,33  However, very few studies have directly compared the clinical effectiveness of varying dosing intervals; moreover, they are limited to adults and are from a limited number of jurisdictions that have allowed varying dosing intervals. A study from the Canadian provinces of Québec and British Columbia revealed higher VE for mRNA vaccines against infection with wild-type SARS-CoV-2 or the Alpha, Beta, or Gamma variants in adults with a dosing interval of 7 to 8 weeks (90% in British Columbia and 87% in Québec) compared with an interval of 3 to 4 weeks (84% in British Columbia and 79% in Québec).³⁴  A study from England also revealed higher VE for BNT162b2 against COVID-19 in adults aged 50 to 89 years with a dosing interval of >6 weeks compared with 3 weeks.³⁵  These previously described trends are consistent with our study’s findings for children aged 5 to 11 years. Ultimately, decisions about the optimal dosing interval for vaccine programs should balance achieving more rapid versus more sustained protection from vaccination and be guided by jurisdictions’ current SARS-CoV-2 epidemiology. Additional considerations on dosing interval include the reduced risk of myocarditis and pericarditis that has been described with longer dosing intervals.³⁶ 

Our study has some limitations. First, there were important restrictions to SARS-CoV-2 RT-PCR testing in Ontario during the study period, and our study did not capture rapid antigen detection tests results, which could have led to residual confounding from undetected SARS-CoV-2 infection, especially in children with longer dosing intervals. Limitations to testing may also restrict the external validity of our findings, as symptomatic and tested children differed from untested children regarding the presence of comorbidities, COVID-19 vaccination, prior influenza vaccination, and being a child of a healthcare worker. We were also unable to measure VE in immunocompromised children and those who received the Moderna vaccine, given their exclusion from the study. Further, because of the small number of testing episodes occurring after BA.4/BA.5 circulation and the limited use of whole genome sequencing, we could not estimate VE against Omicron sublineages. Lastly, we did not assess VE against other COVID-19-related outcomes, such as multisystem inflammatory syndrome in children or long COVID-19. Further studies assessing the impact of vaccination on these outcomes, as well as the duration of protection for longer intervals after vaccination in this age group, are needed.

Our results suggest that 2 doses of the BNT162b2 COVID-19 vaccine were initially effective against symptomatic infection from Omicron in children aged 5 to 11 years, but VE decreased steadily with time, especially >120 days after vaccination. Although extended dosing intervals were associated with higher effectiveness against symptomatic infection in the first 3 months after vaccination, VE estimates became similar to those of shorter intervals starting 90 days after the second dose. Cases of severe COVID-19 were rare among vaccinated children in this age group, and VE was high against severe outcomes, but with potential waning over time. Further studies are warranted, especially with bivalent COVID-19 vaccines and the potential emergence of other Omicron sublineages and SARS-CoV-2 variants.

Peter C. Austin

Sarah A. Buchan

Hannah Chung

Gaston De Serres

Deshayne B. Fell

Cindy Fong

Jonathan B. Gubbay

Naveed Z. Janjua

Sarah C. J. Jorgensen

Jeffrey C. Kwong

Shannon E. MacDonald

Salaheddin M. Mahmud

Sharifa Nasreen

Pierre-Philippe Piché-Renaud

Christiaan H. Righolt

Kevin L. Schwartz

Maria E. Sundaram

Sarah Swayze

Mina Tadrous

Nisha Thampi,

Kumanan Wilson

Sarah E. Wilson

We would like to acknowledge Public Health Ontario for access to vaccination data from COVaxON, case-level data from CCM, and COVID-19 laboratory data, as well as assistance with data interpretation. We 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. We thank IQVIA Solutions Canada Inc for use of their Drug Information File.

CCM

Public Health Case and Contact Management Solution

COVID-19

coronavirus disease 2019

RT-PCR

reverse-transcription polymerase chain reaction

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2

VE

vaccine effectiveness