Health Reports
Sociodemographic disparities in COVID-19 vaccine uptake and vaccination intent in Canada

by Mireille Guay, Aubrey Maquiling, Ruoke Chen, Valérie Lavergne, Donalyne-Joy Baysac, Jackie Kokaua, Catherine Dufour, Eve Dubé, Shannon E. MacDonald and Nicolas L. Gilbert

DOI: https://www.doi.org/10.25318/82-003-x202201200004-eng

Abstract

Background

This study’s objective was to examine sociodemographic disparities in COVID-19 vaccine uptake and vaccination intent in the Canadian provinces by identifying factors associated with vaccine uptake in seniors prioritized for vaccination at the time of the survey and vaccination intent in all adults.

Data and methods

A cross-sectional survey of Canadian adults was conducted in all provinces from mid-April to mid-May 2021. In addition to sociodemographic characteristics, respondents (n=10,678) provided information on their COVID-19 vaccination status or their intent to get vaccinated. Logistic regression models were fitted using sociodemographic factors as explanatory variables and vaccination status (unvaccinated versus at least one dose) or vaccination intent (unlikely versus likely or already vaccinated) as outcomes. To account for vaccine prioritization groups, multiple regression models were adjusted for province of residence, age, Indigenous identity and health care worker status.

Results

Seniors with a lower household income (less than $60,000) and those living in smaller communities (fewer than 100,000 inhabitants) had higher odds of being unvaccinated. Among Canadian adults, the odds of being unlikely to get vaccinated were higher for males (adjusted odds ratio [AOR] 1.3), individuals younger than 60 (AOR between 3.3 and 5.1), non-health care workers (AOR 3.3), those with less than a high school education (AOR 3.4) or a household income of less than $30,000 (AOR 2.7) and individuals who do not identify as South Asian, Chinese, Black, Filipino, Arab, Latin American, Southeast Asian, West Asian, Korean or Japanese (AOR 1.7).

Interpretation

COVID-19 vaccine uptake (80%) and vaccination intent (95%) were high among Canadians; however, relative disparities were observed among specific groups. Continued efforts targeted toward these groups are essential in reducing potential inequity in access or service provision.

Keywords

COVID-19; vaccine; vaccination coverage; intention; health equity; Canada.

Authors

Mireille Guay, Aubrey Maquiling, Ruoke Chen, Valérie Lavergne, Donalyne-Joy Baysac and Nicolas L. Gilbert are with the Vaccine Rollout Task Force, Public Health Agency of Canada, Ontario, Canada. Nicolas L. Gilbert is also with the École de santé publique de l’Université de Montréal, Quebec, Canada. Jackie Kokaua and Catherine Dufour are with the Centre for Social Data Integration and Development; Social, Health and Labour Statistics; Statistics Canada; Ontario; Canada. Eve Dubé is with the Institut national de santé publique du Québec, Quebec, Canada, and the Département d’anthropologie de l’Université Laval, Quebec, Canada. Shannon E. MacDonald is with the Faculty of Nursing, University of Alberta, Alberta, Canada.

Introduction

Canada’s COVID-19 vaccination campaign began on December 14, 2020, with the arrival of a limited number of doses that were prioritized for high-risk populations identified by Canada’s National Advisory Committee on Immunization (NACI), such as the elderly, residents and staff of congregate living settings, frontline health care workers, and Indigenous populations.^Note 1 As vaccines became more widely available, provinces and territories considered NACI advice and assessed their local situation to determine their vaccine rollout plans. COVID-19 vaccine prioritization therefore varied by province or territory, with the elderly being among the first to be vaccinated, followed by priority age groups identified by decreasing 5- or 10-year age bands. By mid-April 2021, most provinces had started vaccinating adults aged 60 years and older, except for New Brunswick (which began on April 27, 2021), Nova Scotia (April 19, 2021), and Newfoundland and Labrador (May 3, 2021). Following the establishment of the vaccination campaigns, evidence of COVID-19 vaccine effectiveness has started to emerge worldwide.^{Note 2Note 3Note 4} In Canada, studies have shown up to 95% protection against symptomatic infection and severe outcomes with various vaccine products and dose intervals.^{Note 5Note 6Note 7Note 8} Therefore, ensuring equitable distribution and high uptake of COVID-19 vaccines is critical to protecting the population from infection, thereby reducing transmission and the risks associated with COVID-19 morbidity and mortality.^{Note 9Note 10}

Vaccine hesitancy, defined by the World Health Organization as “a delay in acceptance or refusal of safe vaccines despite availability of vaccine services,” has been on the rise over the last decade and is recognized as a global threat.^Note 11 As with all vaccines, there is a range of factors that contribute to COVID-19 vaccine hesitancy, such as doubts about the safety and effectiveness of approved and available vaccines.^Note 12 Access barriers are also major challenges to vaccine uptake. Despite believing in the value of getting vaccinated, some people do not do so because of a lack of time and scheduling limitations, a lack of access and capacity to navigate the technologies and online service platforms, and a lack of transportation to reach vaccination services.^{Note 13Note 14}

By May 1, 2021, 41% of individuals aged 18 and older and 77% of those aged 60 and older had received at least one dose. While the majority of Canadian seniors had already received at least one dose, the rollout of COVID-19 vaccines may not have sufficiently addressed the needs of the population, as social inequities contribute to vaccine accessibility and uptake.^Note 15 Previous Canadian studies showed that females, Indigenous people, racialized populations, individuals with a lower level of education and essential non-health care workers had lower intentions of getting vaccinated,^{Note 16Note 17} while being male, being older (older than 60 years), having a university-level education, having or living with someone with chronic medical conditions (CMCs) and working with patients infected with COVID-19 were significantly associated with acceptance of COVID-19 vaccines.^{Note 12Note 17Note 18} In other countries, being Black or from a one-parent household were factors associated with lower COVID-19 vaccine uptake.^{Note 19Note 20}

Given the disproportionate health and economic impacts COVID-19 has had on racialized populations and those with a lower socioeconomic status^Note 21, understanding the differences in COVID-19 vaccine uptake among different population groups and the drivers of COVID-19 vaccine hesitancy is crucial for improving vaccine acceptance and ensuring equitable vaccine uptake to avoid worsening inequalities in COVID-19-related burdens.^{Note 22Note 23}

The COVID-19 Vaccination Coverage Survey (CVCS) was developed to measure COVID-19 vaccination coverage in the Canadian provinces and territories and to compare vaccination coverage levels for various sociodemographic and economic subgroups during the early stages of the vaccination campaign (March to May 2021). Using CVCS data, this study’s objective was to examine sociodemographic disparities in COVID-19 vaccine uptake and vaccination intent by identifying the social and economic determinants associated with vaccine uptake among seniors aged 60 years and older (i.e., those eligible for vaccination at the time of the survey) and the determinants of vaccination intent in the entire adult population.

Data and methods

Study population

The CVCS was designed by the Public Health Agency of Canada (PHAC) and Statistics Canada and was funded solely by PHAC.^Note 24 It was a cross-sectional, voluntary survey consisting of two cycles. The first cycle of the CVCS took place between March 15 and April 12, 2021, and targeted Canadians aged 18 years and older living in the three territorial capitals; the second cycle was conducted in the 10 provinces between April 12 and May 12, 2021. According to the information available on provincial websites, individuals 60 years and older were eligible to receive their first dose by May 2021, and individuals 18 years and older were eligible by June 2021. The CVCS target population excluded people living on reserves or in other Indigenous settlements, in institutions and in collective dwellings. The present study focused on analyzing the data collected from the second cycle of the CVCS.

Survey description

The CVCS is a sample survey with a cross-sectional design. The sample was stratified by province of residence, and a simple random sample of dwellings was independently selected from the Dwelling Universe File^Note 25 within each province. Invitation letters were sent out by mail, and one person per household was randomly selected to participate in the survey. Respondents could complete the survey questionnaire in either official language (English or French), and the primary method of collection was self-response via electronic questionnaire. In case of non-response, computer-assisted telephone interviewing was used to follow up with the selected individuals to encourage participation and maximize response rate. Sampling weights based on gender, age and province of residence were produced for each respondent to create nationally representative estimates.^Note 24 A total of 10,678 respondents completed the Cycle 2 survey questionnaire, yielding a response rate of 59.2%.

The CVCS questionnaire was reviewed and tested by Statistics Canada. It collected information on COVID-19 vaccine uptake by asking respondents whether they have been vaccinated against COVID-19. For those who were unvaccinated, a subsequent question on their intention to be vaccinated was asked using a four-point ordinal response scale: very likely, somewhat likely, somewhat unlikely or very unlikely. Because of the very small proportion of respondents in the last two categories, they were collapsed and considered to reflect those who were “unlikely to get vaccinated.” Additionally, sociodemographic characteristics of the participants, such as gender, age, education, pre-pandemic household income, marital status, country of birth, Indigenous identity and population groups, were collected using Statistics Canada’s multiple-choice standardized questions. Belonging to a population group is defined as individuals who self-identify using one of the response categories set out in the Employment Equity Act, specifically, South Asian, Chinese, Black, Filipino, Arab, Latin American, Southeast Asian, West Asian, Korean and Japanese. In addition, information on community size was obtained by linking the respondent’s postal code with Statistics Canada’s Postal CodeOM Conversion File Plus (PCCF+ version 7D based on 2016 Census data). 

Main objectives

Two separate logistic regression analyses were conducted to examine (1) the sociodemographic determinants associated with self-reported COVID-19 non-vaccination among seniors and (2) the sociodemographic determinants associated with the intent to not get vaccinated against COVID-19 in all adults aged 18 years and older. For the first objective, only adults aged 60 and older were included in the analyses, as they were prioritized for COVID-19 vaccination and were the only group eligible to be vaccinated across all provinces at the time of the survey.

Quantitative variables

Outcome variables

The two outcome variables were COVID-19 vaccination status and COVID-19 vaccination intent. The former was categorized as “not vaccinated” and “received at least one dose,” and the latter was categorized as “unlikely to get vaccinated” (combining those who reported themselves to be “somewhat” or “very” unlikely to get vaccinated) and “vaccinated or likely to get vaccinated.” Vaccination intent was measured using a four-point likelihood scale.

Control variables

To control for differences in provincial vaccine rollout plans and vaccination eligibility at the time of the survey, province of residence, age, occupation in the health care sector and Indigenous identity—factors used to define eligibility criteria—were considered as covariates in the multiple regression models.

Independent variables

Sociodemographic factors such as gender, level of education, household income before tax, CMCs, marital status, country of birth, population groups and community size were included as independent variables. Variables included in the models have been previously demonstrated to be related to the modelled outcome and are considered by the authors to conceptually have a potential association with the outcome. For variables that had categories with largely unequal sample sizes, the largest category was used as a reference to avoid any problems associated with using a small group as a reference group. In other cases, the reference category was selected as the normative group, that is, the group from which the most logical or important comparisons can be drawn. When there was no obvious norm and samples sizes were similar, the highest or lowest category (e.g., education, income, age or community size) or the category with the lowest odds to provide odds ratios (ORs) greater than 1 was selected for easier interpretation. (See Table 1 for the definition of the categories for each variable.)

Statistical analysis

To describe the study sample, categorical independent variables were summarized using unweighted frequencies and proportions. Additionally, weighted proportions of unvaccinated seniors were estimated and stratified by the sociodemographic characteristics listed above. Weighted prevalence estimates (percentages) of vaccination intent among all adults aged 18 years and older were also calculated. Confidence intervals (CIs) for prevalence proportions were adjusted using the Wilson score interval.^Note 26 Associations between sociodemographic factors and COVID-19 vaccine uptake or vaccination intent were identified using simple and multiple logistic regression models. From these models, ORs were calculated and multiple comparison tests with Tukey‒Kramer adjustment were performed to identify statistically significant differences in vaccine uptake or vaccination intent. Simple and multiple models were fitted using the complete case analysis method as the proportion of missing cases was very low (less than 4%) and these cases had the same distribution of all the variables as the complete cases (data not shown). For the multiple logistic regression models, multicollinearity was assessed by performing chi-square associations between all independent variables (data not shown). While some associations were present between some independent variables, there was no evidence of inflated standard errors or severe multicollinearity in the models. Given the complex survey design, adequate design-adjusted variances were estimated using the bootstrap resampling method (1,000 samples). The data were analyzed in SAS Enterprise Guide 7.1 (SAS version 9.4), and the SURVEVYFREQ and SURVEYLOGISTIC procedures were exclusively used to ensure the complex survey design was accounted for in the analysis.

Results

Sample characteristics

More than half (56.1%) of the full sample (N=10,678) of respondents were female, and 43.2% of them were 60 years of age or older. The vast majority (79.4%) were born in Canada and more than half (58.9%) were married or living in common law. Close to half (45.2%) of the respondents had a pre-pandemic before-tax household income between $30,000 and $90,000 and most (67.8%) of them had a postsecondary education or higher. Finally, 15.5% self-identified as being part of a population group (Table 1).

Vaccination status among seniors aged 60 years and older

Overall, a large majority (80%) of adults aged 60 years and older had received at least one dose of a COVID-19 vaccine at the time of the survey (Table 2).

In the adjusted multiple logistic regression model, some groups had greater odds of being unvaccinated. At the time of the survey, seniors in all provinces except Quebec had higher odds of being unvaccinated (compared with seniors in Saskatchewan). Seniors aged 60 to 69 years had 5.3 (95% CI: 2.7 to 10.3) to 11.2 (95% CI: 5.8 to 21.7) times higher odds of being unvaccinated than seniors aged 80 years and older. The observed increasing trend with decreasing age group was expected since the oldest age groups were the first to be vaccinated during the initial vaccine rollout. Moreover, an increasing trend was observed with decreasing income. Compared with seniors with household incomes of $150,000 or greater, the odds of being unvaccinated were greater for those with a household income of less than $60,000 (AOR: 2.3, 95% CI: 1.0 to 5.3 and AOR: 3.9, 95% CI: 1.6 to 9.3 for those with less than $60,000 and less than $30,000, respectively). Furthermore, seniors in smaller communities (fewer than 100,000 inhabitants) had 2.4 (95% CI: 1.3 to 4.2) to 2.5 (95% CI: 1.4 to 4.6) times higher odds of being unvaccinated than those living in larger communities of 1,500,000 inhabitants or more. Results for Indigenous identity and population groups could not be broken down further because of limited sample sizes (Table 2).

Associations observed between vaccination coverage and the four covariates listed at the top of Table 2 should be interpreted with caution, as the differences could be explained by differences in provincial vaccine rollout plans and vaccination eligibility at the time of the survey and not by sociodemographic inequality.

Vaccination intent among all adults 18 years and older

At the time of the survey, 95% of adults aged 18 years and older stated that they had been vaccinated (45%) or were likely to be vaccinated in the future (49%) (Appendix Table A).

In the adjusted multiple logistic regression analysis, the odds of being unlikely to get vaccinated for individuals younger than 60 years were 3.3 (95% CI: 1.6 to 6.9) to 5.1 (95% CI: 2.5 to 10.7) times greater than for individuals who were 70 years or older. An increasing trend was observed with decreasing age (Table 3). In addition, non-health care workers had almost 3.3 (95% CI: 1.7 to 5.0) times greater odds of being unlikely to get vaccinated compared with health care workers. The odds for individuals who completed less than a secondary education were 3.4 (95% CI: 1.6 to 7.2) times greater than for those with a university education. Individuals with a household income of less than $30,000 had 2.7 (95% CI: 1.0 to 7.0) times greater odds of being unlikely to get vaccinated than those with a household income greater than or equal to $150,000. An increasing trend was observed with decreasing household income. Finally, the odds for males were 1.3 (95% CI: 1.0 to 1.7) times greater than for females, and the odds for people not belonging to a population group were 1.7 (95% CI: 1.0 to 3.3) times greater than for those part of a population group.

Sensitivity analyses

Several sensitivity analyses were conducted to test the robustness of the models. First, seniors residing in New Brunswick, Nova Scotia, and Newfoundland and Labrador became eligible for vaccination during, rather than before, the study collection period. A sensitivity analysis showed that the results for vaccination status among seniors were not affected by the removal of these three provinces (data not shown). Second, as seniors (60 years and older) and non-seniors (aged 18 to 59 years) had different vaccine uptake during the study period, determinants of non-vaccination were also examined in non-seniors as the original model included only seniors. The same models were applied to non-seniors, and the results showed that all independent variables were significantly associated with non-vaccination except for marital status, population groups and country of birth (Appendix Table B). Likewise, the sociodemographic determinants of being unlikely to get vaccinated were compared between seniors and non-seniors (Appendix Table C). In seniors, the sociodemographic determinant of being unlikely to get vaccinated was community size. For non-seniors, the sociodemographic determinants of being unlikely to get vaccinated were gender, health care worker status, education and self-identification in a population group. The lack of association between sociodemographic factors and vaccination intent observed among seniors could be attributable to low sample sizes and decreased statistical power (Appendix Table C). Lastly, because exact income was not available, income categories were used and income adjusted for household size could not be computed. However, as a sensitivity analysis, household size was included as an additional variable in the models, and no major changes in the findings were observed (data not shown).

Discussion

Vaccination status among seniors 60 years and older

According to the CVCS, seniors in all provinces except Quebec had higher odds of being unvaccinated (compared with seniors in Saskatchewan). This was expected as provinces rolled out vaccines for this age group at different times. Saskatchewan and Quebec lowered the vaccination eligibility age to 60 years and older in late March to early April 2021, a few weeks earlier than other provinces.^{Note 27Note 28Note 29}

Significant differences in COVID-19 non-vaccination among seniors were found. Those with a household income of less than $60,000 had lower coverage than those earning at least $150,000. The association between lower household income and being unvaccinated suggests a lower vaccine uptake in people of lower socioeconomic status. However, data from this survey provide no insight on the causal mechanisms underlying this association. Few Canadian studies have yet to cover income disparities in COVID-19 vaccine uptake. However, in Canada, some parallels were observed between COVID-19 vaccine uptake and vaccination against influenza, where lower income was a significant predictor of non-vaccination.^{Note 30Note 31}

A cross-sectional study in Wales, United Kingdom, that targeted individuals aged 50 years and older found lower rates of vaccination among ethnic minority groups (e.g., Black ethnic group).^Note 30 In contrast, among individuals aged 60 years and older in Canada, this study (when adjusted for other sociodemographic factors) found no significant difference in the odds of being unvaccinated when comparing those who were part of population groups with those who were not.^Note 30 While some studies have shown that specific ethnic minority groups had higher or lower vaccine uptake compared with White-,^{Note 13Note 30} the small sample sizes in the CVCS of seniors who self-identified as part of a population group prevented further breakdown, thereby obscuring any potential differences. Further research on COVID-19 vaccination coverage with larger samples of racialized populations in the Canadian context would be needed to ascertain the current situation.

The study in Wales, United Kingdom also found that the odds of being vaccinated against COVID-19 were lower for males compared with females.^Note 30 Although not statistically significant, the current study showed that senior males had higher odds of being unvaccinated, aligning with this previous finding. In Canada, no gender differences were observed in influenza vaccination coverage.^Note 31 The association between community size and vaccination coverage showed a clear dose-response relationship in the CVCS, with a lower proportion of vaccinated seniors living in smaller communities. This aligns with previous findings in Canada, where health care workers living in rural British Columbia had lower vaccination rates compared with health care workers in urban centres.^Note 32 Similarly, lower COVID-19 vaccination rates in rural areas were observed in the United States.^Note 33 The level of education, health care infrastructure and political views were the reasons put forward by the authors to explain this difference in the United States.

Vaccination intent among all adults 18 years and older

In the adjusted intent model, the odds of being unlikely to get vaccinated were significantly greater in individuals who were male, younger (younger than 60 years of age), less educated, not part of a population group, not a health care worker and in lower-income households. Similar evidence was found for age and education in recent Canadian studies.^{Note 12Note 18Note 34Note 35} People part of a population group, Indigenous people and essential non-health care workers also had lower adjusted odds of intending to receive a COVID-19 vaccine.^Note 17 However, in the present study, Indigenous identity did not present a statistically significant association with vaccination intent, and, in contrast, the odds of being unlikely to get vaccinated were greater in people not part of a population group than in people who were. Recent studies conducted in the United States found that ethnic groups, particularly Asian and Hispanic people, were more likely to accept COVID-19 vaccines than White people, with Asian people having the lowest levels of hesitancy.^{Note 36Note 37} With NACI recommending that health care workers be prioritized to receive a vaccine in Canada,^Note 38 this may in part explain the present study’s findings of higher vaccine uptake and vaccination intent among health care workers. Conversely, studies found higher vaccine hesitancy in non White ethnic populations;^{Note 39Note 40} therefore, further investigation is warranted to examine the association between COVID-19 vaccination intent and racialized populations, especially as it may vary across these groups.

According to the CVCS results, males had higher odds of being unlikely to get vaccinated against COVID-19. This contrasts with some findings observed in other studies conducted in Canada and other countries that females had a lower vaccination intent.^{Note 12Note 16Note 17Note 20Note 36Note 40} In a study conducted in Saskatchewan, being a recent immigrant (born outside Canada and living in the country for less than 20 years) was associated with a greater likelihood of vaccine hesitancy.^Note 16 Conversely, in the present study, no significant association was observed with country of birth at the national level. Moreover, there was no strong association between being unlikely to get vaccinated and having a CMC in the simple or the multiple regression models. Nevertheless, a recent study from Quebec demonstrated that having, or living with someone with, CMCs and increased risk perceptions of COVID-19 remained one of the strongest predictors of COVID-19 vaccination intent.^Note 12 This potential difference between Canada and Quebec might be in part because of the high mortality rate seen in Quebec during the first wave of the pandemic,^Note 41 increasing the perceived risks of COVID-19 among Quebecers.^Note 12

Strengths and limitations

As with any large-scale survey, the CVCS has several strengths and limitations that must be considered when interpreting the results. The major strengths of the survey were the sufficiently large sample size and the high response rate of 59.2%. These allowed for analysis by various sociodemographic factors. Analyses were conducted using calibrated sampling weights to ensure the estimates were representative of all Canadians living in the 10 provinces.

Some study limitations need to be acknowledged. According to the CVCS sampling frame, approximately 95.4% of the dwellings had a mailing address. The small proportion of the population without a mailing address, which may differ from the rest of the target population, was therefore excluded from the sampling frame.

Like many other Statistics Canada surveys, the CVCS excluded First Nations communities on reserve and institutionalized individuals. Exclusion of the latter group could impact the ability to generalize the findings on determinants of COVID-19 non-vaccination in the entire population aged 60 years and older. The territories were also excluded from the analysis, and this could impact coverage of Indigenous people. Interviews were conducted only in English or French, excluding respondents who are not fluent in either official language. This limitation could affect the representation of some population groups in the CVCS results because the language barrier could prevent them from participating. These populations may have differences in access to, or utilization of, health care services from that of other Canadian adults.

The measures in the CVCS are self-reported and may also be subject to recall bias and social desirability bias. However, recall bias is less likely to occur in the present study given that data collection ended less than five months after the beginning of the COVID-19 vaccine rollout plans. Despite a satisfactory response rate, response bias cannot be ruled out. Indeed, those with greater interest in the topic would be more likely to respond to the survey. The survey estimates are adjusted to account for non-response through the survey weights. However, to the extent that non-responding households and people differ from the rest of the sample, residual selection bias cannot be completely eliminated.

It is conceivable that other factors not measured in this survey could also be associated with the modelled outcomes. Further research is warranted to explore other potential determinants.

Conclusion

The present study highlighted Canadians’ positive attitudes toward vaccination. By May 2021, the majority of the adult population was either vaccinated with at least one dose (45%) or stated to be likely to receive a COVID-19 vaccine in the future (49%). In the CVCS, those who were male, were younger, were not part of a population group, not a health care worker and had a lower level of education or household income were associated with being unlikely to be vaccinated against COVID-19. Continued efforts to reach vaccine hesitant groups and promote vaccine uptake are essential to protecting all Canadians against the disease.

Acknowledgements

The authors wish to thank Statistics Canada for its collaboration and all of the study participants for taking the time to complete the survey.

Appendix