Understanding national trends in COVID-19 vaccine hesitancy in Canada: results from five sequential cross-sectional representative surveys spanning April 2020-March 2021.

OBJECTIVE: To examine rates of vaccine hesitancy and their correlates among Canadian adults between April 2020 and March 2021.
DESIGN: Five sequential cross-sectional age, sex and province-weighted population-based samples who completed online surveys.
SETTING: Canada. PARTICIPANTS: A total of 15 019 Canadians aged 18 years and over were recruited through a recognised polling firm (Leger Opinion). Respondents were 51.5% female with a mean age of 48.1 (SD 17.2) years (range 18-95 years) and predominantly white (80.8%). PRIMARY AND SECONDARY OUTCOME MEASURES: Rates of vaccine hesitancy over the five surveys (time points) and their sociodemographic, clinical and psychological correlates.
RESULTS: A total of 42.2% of respondents reported some degree of vaccine hesitancy, which was lowest during surveys 1 (April 2020) and 5 (March 2021) and highest during survey 3 (November 2020). Fully adjusted multivariate logistic regression analyses revealed that women, those aged 50 and younger, non-white, those with high school education or less, and those with annual household incomes below the poverty line in Canada were significantly more likely to report vaccine hesitancy, as were essential and healthcare workers, parents of children under the age of 18 and those who do not get regular influenza vaccines. Endorsing prevention behaviours as important for reducing virus transmission and high COVID-19 health concerns were associated with 77% and 54% reduction in vaccine hesitancy, respectively. Having high personal financial concerns was associated with 1.33 times increased odds of vaccine hesitancy.
CONCLUSIONS: Results highlight the importance of targeting vaccine efforts to specific groups by emphasising the outsized health benefits compared with risks of vaccination. Future research should monitor changes in vaccine intentions and behaviour to better understand underlying factors. © Author(s) (or their employer(s)) 2022. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

Assesses changes in vaccine intentions over time across three critical waves of the pandemic in Canada through five survey waves from April 2020 to March 2021.

Large sample size with good distribution across provincial regions, age groups, gender, employment status and income compared with census data available through Statistics Canada.

Sample under-represents people of colour and perhaps non-native English and French speakers, as the survey was only available in these two languages.

Data reflect trends in vaccine intentions over time but not in the same individuals.

Results reflect a subanalysis of Canadian representative data from the International COVID-19 Awareness and Responses Evaluation (iCARE) study (100 000 participants from 190 countries) alongside ongoing efforts to collect similarly representative samples in eight other countries (see www.icarestudy.com), which will enable comparisons with international datasets.

Introduction

The SARS-CoV-2 virus causing COVID-19 has caused a global pandemic, resulting in significant morbidity, mortality, and economic and social disruption in Canada and around the world.1 Key to reducing disease morbidity and mortality and reducing the need for future lockdowns is widespread acceptance of COVID-19 vaccines, several of which have been approved for those aged 5 and older by Health Canada.2 3. High rates of vaccine acceptance were thought to be necessary for achieving target levels of herd immunity,4 but it has proven difficult to estimate the minimum threshold of immunisation needed to achieve this due to the emergence of highly virulent strains like Delta, whose R0 has been estimated to be five to six times greater than the original Wuhan SARS-CoV-2 strain.5 6 This has led experts to recommend vaccinating as much of the population as possible and exploring the need for additional ‘booster’ or yearly doses.7 Regardless of how COVID-19 vaccination schedules unfold over the short and longer terms, the ultimate success of vaccination programmes depends on people’s willingness to get vaccinated. However, several reports from nations where vaccines have been widely available indicate that intentions to get a COVID-19 vaccine have been steadily declining (and rates of vaccine hesitancy steadily increasing)8 since the first pandemic wave. For example, a longitudinal study in the USA reported significant declines in the likelihood of getting vaccinated (somewhat or very likely to get vaccinated), from a high of 74% in early April 2020 to a low of 56% by early December 2020.9 These declines were observed for both men and women and in all age, racial/ethnic and education subgroups. Similar trends were also observed in Australia, where 31.9% of Australians reported being less willing to get vaccinated between August 2020 and January 2021, and were particularly prevalent among Indigenous populations and those who did not complete high school.10 Since then, 175 studies worldwide have been published on vaccine hesitancy through to the end of August 2021, including 21 reporting data from Canada. According to a living systematic review by Crawshaw et al,11 the IQR of vaccine hesitancy was 12%–24%, with a mean of 17%. Overall, these results raise important questions about vaccine attitudes and intentions among Canadians, whose willingness to get vaccinated now and in the future will be critical for optimising the success of Canada’s vaccine strategy and our successful transition out of the pandemic.

Vaccine hesitancy has been defined as ‘…a delay in acceptance or refusal of vaccination despite availability of vaccination services’. 8 Key to optimising vaccination rates is understanding patterns and correlates of hesitancy over time. This will allow us to improve vaccine policy planning, develop targeted interventions and enhance tailoring of vaccine messaging to vulnerable groups. To this end, we examined rates of vaccine hesitancy and their correlates among Canadians by analysing data from five cross-sectional age, sex and province-weighted population-based samples who completed online surveys between April 2020 and March 2021. In order to explore the factors associated with vaccine hesitancy over time, data across all surveys were examined as a function of key sociodemographics, clinical characteristics and psychological factors known to be important for vaccine behaviour.12

Methods

Study design

The International COVID-19 Awareness and Responses Evaluation (iCARE) Study (www.icarestudy.com)13 is an ongoing international, multiwave, cross-sectional observational survey study of public awareness, attitudes and responses to COVID-19 public health policies. The study is led by researchers from the Montreal Behavioural Medicine Centre (www.mbmc-cmcm.ca) in collaboration with a team of over 200 international collaborators from more than 40 countries. The survey was designed with international experts to assess constructs from the capability, opportunity, motivation–behaviour model of the behaviour change wheel14 and from the health belief model.15 16 The survey also includes questions on sociodemographics, physical and mental health conditions, general health behaviours, previous COVID-19 infection, awareness of local government prevention policies, perceptions and attitudes about these policies, adherence to prevention behaviours, COVID-19-related concerns and impacts, and vaccine attitudes and intentions. The surveys include approximately 75 questions (approximately one per page), take 15–20 min to complete, and can all be found online (www.osf.io/nswcm). Questions were presented in the same order, but the response set order was randomised for questions with multiple subitems to reduce bias. Some questions were conditionally displayed based on responses to previous items to reduce the number and complexity of the items. Completing all questions is mandatory to move forward, but many questions included the option ‘I don’t know/prefer not to answer’. Full details about survey development and general methodology have been published elsewhere,13 and the Checklist for Reporting Results of Internet E-Surveys (CHERRIES) can be found in online supplemental table S1.

Participants

For this study, we report data from five nationally representative online surveys of Canadians aged 18 years and over using a recognised polling firm which recruits participants through their closed, proprietary online panel (Leger Opinion). This panel includes over 400 000 Canadians, the majority of which (61%) were recruited within the past 10 years. Two-thirds of the panel were recruited randomly by telephone, with the remainder recruited via publicity and social media. Respondents are invited to complete the survey via email and did so voluntarily. Leger Opinion sends panellists a unique link to complete the survey so they cannot complete it more than once (receive a message: ‘you have already completed this survey’). Using data from Statistics Canada, results were weighted within each province according to the sex and age of the respondents. Then, the weight of each province was further adjusted to represent their actual weight within the Canadian federation. Data were collected between 9 and 20 April 2020 (survey 1), 5 and 17 June 2020 (survey 2), 29 October and 11 November 2020 (survey 3), 27 January and 7 February 2021 (survey 4), and 11 and 29 March 2021 (survey 5), respectively, using a self-administered computer-assisted web interface. Online consent is provided by participants prior to completing the survey. No personal identifying information is collected from any participant. Participants are offered nominal compensation through the polling firm (participants collect points that can be traded in for gift cards); no direct compensation is provided by the research team.

Assessment of vaccine intentions and hesitancy

To assess vaccine hesitancy, we asked: ‘If a vaccine for COVID-19 were available today, what is the likelihood that you would get vaccinated?’ Response options (very unlikely, unlikely, somewhat likely, extremely likely, I don’t know/prefer not to answer) were dichotomised into ‘very unlikely, unlikely, somewhat likely’ to describe those indicating at least some degree of hesitancy, versus ‘very likely’ to describe those with very high intentions to get vaccinated. A dichotomous outcome was chosen to identify all those who could benefit from intervention, with those responding ‘very likely’ to get vaccinated treated as the comparator/reference.

Assessment of psychological factors

We assessed two psychological factors that are often important motivators of engaging in protective health behaviours: perceived importance of engaging in infection prevention behaviours, and the nature and extent of people’s COVID-19-related concerns.15–17 Perceived importance of engaging in COVID-19 prevention behaviours (including getting vaccinated) was assessed using a single question: ‘To what extent do you believe that the measures asked of you by your government or local health authority are important to prevent and/or reduce the spread of COVID-19?’ Response options (not at all important, not very important, somewhat important, very important and I don’t know/prefer not to answer) were dichotomised into ‘very important’ versus all others.

To assess the concerns people have about the COVID-19 virus and its impacts, individuals were presented with the following prompt: ‘Because of COVID-19, I am concerned about…’. Respondents then had to indicate the extent which they had 10 specific concerns, choosing among ‘not at all’, ‘very little’, ‘somewhat’, ‘to a great extent’ and ‘I don’t know/prefer not to answer’. To cluster COVID-19-related concerns, we performed a principal component analysis on a polychoric correlation matrix of the 10 variables in the concerns module (ordinal scale, as detailed earlier), details of which can be found elsewhere.18 We observed a three-component structure that included ‘health concerns’, ‘personal financial concerns’ and ‘social and economy concerns’. Mean values (M) and SD for each of the three components are reported as a score out of 4, from 1=not at all to 4=to a great extent. Internal consistency for the components ranged from satisfactory (social/economy concerns α=0.69) to excellent (personal financial concerns α=0.82, health concerns α=0.91) for the individual components.18

Statistical analysis

Several survey questions included an answer I don’t know/I prefer not to answer, which was recoded as a missing value, and analyses were based on complete case records. Descriptive statistics (weighted means, SDs, and proportions) were calculated to describe the sample in terms of demographic characteristics, across all surveys. Univariate analyses were conducted to examine differences in sociodemographic characteristics (weighted proportions) as a function of vaccine hesitancy across the five time points. Three separate multivariable logistic regression models were performed to assess associations between vaccine hesitancy (dependent variable) and participant sociodemographic (ie, age, sex, ethnicity, education, employment status, annual household income, parental status, worker status and provincial region) and clinical characteristics (ie, health risk conditions, history of influenza vaccine and previous COVID-19 infection) (independent variables: model 1), vaccine hesitancy (dependant variable) and perceived importance of prevention behaviours (independent variable: model 2), and vaccine hesitancy (dependent variable) and the nature and extent of the three types of COVID-19-related concerns (independent variables: model 3). Analyses were conducted across all surveys combined and models were partially (covariates included age, sex, ethnicity and survey wave) and fully adjusted (covariates included age, sex, ethnicity, survey wave, education, employment status, annual household income, health risk condition, essential worker, healthcare worker, parental status, history of influenza vaccine and COVID-19 test result). All variables were selected a priori based on pre-existing data.12 Analyses were also conducted as a function of time point/survey to examine trends over time, assessed using the Welch test. All statistical tests were two-sided and a p value of <0.05 was considered as statistically significant. Statistical analysis was performed in SAS V.9.4.

Patient and public involvement

This study was designed in collaboration with over 200 international collaborators, many of which are from the general public (https://mbmc-cmcm.ca/covid19/research/icare-collaborators/). As such, we were able to use both input from the community and behavioural theory to inform the construction of our surveys. The use of a series of survey waves also enabled us to adapt each questionnaire to the changing nature of the pandemic and of its impact on the population. Finally, the public has been called on to contribute to the dissemination of study results through sharable infographics made available on the study website.

Results

Participant characteristics

Our sample included a total of 15 019 respondents (survey 1, n=3003; survey 2, n=3005; survey 3, n=3005, survey 4, n=3000; and survey 5, n=3006) who completed a survey between 9 April 2020 and 29 March 2021. Response rates (total number of completed surveys divided by total number of invitations) ranged between 16% (survey 4) and 25% (survey 5), which is average for online panels.19 However, participation and completion rates as defined by CHERRIES ranged between 86.6% (survey 1) and 95.4% (surveys 3 and 4) and 90.5% (survey 2) and 94.7% (survey 5), respectively. Participant characteristics collapsed across all surveys and then as a function of survey round can be found in table 1 and online supplemental table S2, respectively. Respondents across all five surveys were 51.6% female (range 18–95 years) with a mean age of 48.1 (SD 17.2) years. The majority of the sample were white (81.8%), had a high school or less education (72.3%) and reported total family annual incomes over $60 000 (51.7%). Nearly half (49.7%) reported being currently employed. Just over 44% reported having at least one physician-diagnosed health risk condition (eg, cardiovascular or lung disease, cancer, diabetes and obesity), and just over a quarter (26%) reported having a physician-diagnosed psychiatric disorder (eg, depressive or anxiety disorder). About 16.0% identified as being an essential service worker; just over 4.0% identified as being a healthcare worker; and 21.5% identified as being parents of children under 18. Approximately 17% of respondents had gotten tested for COVID-19, with nearly 1% reporting testing positive. Only 43% of respondents reported getting an influenza vaccine at least three times or more over the past 5 years. In general, compared with census data available through Statistics Canada, participants across all five surveys were well distributed across provincial regions, age groups, employment status and income, and there were equal proportions of men and women. However, those with a graduate or postgraduate degree and people of colour were less represented.

Table 1

Participant characteristics (weighted proportions)

All surveys (N=15 019)
Variable	N (%)
Sex
Male	7239 (48.4)
Female	7724 (51.6)
Age (numerical)	48.1±17.2
Age (years) (categorical)
≤25	1808 (12.2)
26–50	6138 (41.4)
≥51	6897 (46.5)
Race/ethnicity
Non-white	2687 (18.2)
White	12 047 (81.8)
Education level
High school or lower	10 642 (72.3)
Graduate or postgraduate degree	4085 (27.7)
Current employment status
Unemployed	7412 (50.3)
Employed	7338 (49.7)
Annual household income
<$60 000/year	6405 (48.3)
≥$60 000/year	6853 (51.7)
Provincial region
Western*	4702 (31.3)
Ontario	5762 (38.4)
Quebec	3523 (23.5)
Atlantic†	1032 (6.9)
Health-risk condition‡
No	8192 (55.4)
Yes	6596 (44.6)
Psychiatric disorder§
No	10 680 (74.0)
Yes	3747 (26.0)
Essential service worker
No	12 192 (84.1)
Yes	2307 (15.9)
Healthcare worker
No	13 867 (95.6)
Yes	632 (4.4)
Parent of children <18 years
No	11 490 (78.5)
Yes	3145 (21.5)
Results of COVID-19 test
Others	14 702 (99.0)
COVID-19 positive	144 (1.0)
History of getting influenza vaccine
<3 times in the past 5 years	8348 (57.0)
≥3–5 times in the past 5 years	6304 (43.0)

*Western provinces: British Columbia, Alberta, Saskatchewan and Manitoba.

†Atlantic provinces: Nova Scotia, New Brunswick, Prince Edward Island and Newfoundland/Labrador.

‡Health risk conditions: cardiovascular disease, chronic respiratory disease, diabetes, obesity, cancer and other autoimmune diseases.

§Psychiatric disorders: any mood and/or anxiety disorder and dementia.

Estimates of vaccine hesitancy and changes over time

Rates of vaccine hesitancy across time/survey round are presented in figure 1. Overall, 42.2% of respondents reported vaccine hesitancy over the course of the study period, though we observed significant variations in vaccine hesitancy rates over time (survey 1: 36.8%, survey 2: 44.6%; survey 3:52.9%, survey 4: 39.6%, survey 5: 36.9%). As can be seen in figure 1, vaccine hesitancy was lowest during surveys 1 (April 2020) and 5 (March 2021), and highest during survey 3 (November 2020).

Figure 1

Rates of vaccine hesitancy across the five surveys/time points.

Participant characteristics presented as a function of vaccine hesitancy status across all surveys/time points are presented in figure 2 (individual survey data can be found in online supplemental table S3). Across all surveys, rates of vaccine hesitancy were significantly higher among younger age groups (<25 years and 26–50 years compared with those aged 50+), non-white, those currently employed, those reporting less than $60 000 in annual family income, and those living in Western provinces (British Columbia, Alberta, Saskatchewan and Manitoba) and Ontario compared with Quebec and the Atlantic provinces. In addition, rates of vaccine hesitancy were significantly higher among those without a health risk condition, those identifying as essential workers, those identifying as healthcare workers, and parents of children under 18. Finally, rates of vaccine hesitancy were significantly higher among those reporting getting the influenza vaccine less than three times in the past 5 years (all p<0.05).

Figure 2

Participant characteristics presented as a function of being hesitant* versus extremely likely to get a COVID-19 vaccine across the three surveys: univariate analyses. Western provinces: British Columbia, Alberta, Saskatchewan. Manitoba Atlantic provinces: Nova Scotia, New Brunswick, Prince Edward Island, and Newfoundland/Labrador. * High-risk health conditions: cardiovascular disease, chronic respiratory disease, diabetes, obesity, cancer, autoimmune disease. Psychiatric disorders: any mood and/or anxiety disorder and dementia. *Hesitant: those reporting being ‘somewhat likely’, ‘unlikely’ or ‘extremely unlikely’ to seek out the COVID-19 vaccine.

Sociodemographic predictors of vaccine hesitancy

Multivariable logistic regression analyses examining associations between vaccine hesitancy and sociodemographic and clinical variables across all surveys/time points are presented in table 2. The partially adjusted model revealed that women were 19% more likely to be vaccine hesitant (ORpadj 1.19, 95% CI 1.08 to 1.32), those aged less than 25 years (ORpadj 2.07, 95% CI 1.74 to 2.46) and 26–50 years (ORpadj 2.41 95% CI 2.16 to 2.69) were 2.07 times and 2.41 times more likely to be hesitant compared with those aged 51 and over, and those who identified as non-white were 1.3 times more likely to be vaccine hesitant compared with those who identified as white (ORadj 1.30, 95% CI 1.14 to 1.49). The fully adjusted model revealed that in addition to women, younger age groups and non-white, those with high school or less education were 1.15 times more likely to be vaccine hesitant compared with those with graduate or postgraduate degrees (ORadj 1.15, 95% CI 1.041 to 1.28); those earning less than $60 000 per year in household income were 1.42 times more likely to be vaccine hesitant that those earning $60 000 or more (ORadj 1.42, 95% CI 1.26 to 1.61); essential and healthcare workers were 1.44 (ORadj 1.44, 95% CI 1.21 to 1.71) and 1.35 (ORadj 1.35, 95% CI 1.04 to 1.75) times more likely to be vaccine hesitant, respectively, compared with those not in those fields. Finally, parents of children under 18 were 1.51 times more likely to be vaccine hesitant compared with non-parents (ORadj 1.51, 95% CI 1.30 to 1.75); and those reporting getting the influenza vaccine three times or more in the past 5 years were 73% less likely to be vaccine hesitant compared with those reporting getting the influenza vaccine less than three times in the past 5 years (ORadj 0.27, 95% CI 0.23 to 0.30).

Table 2

Multivariate associations between sociodemographic characteristics and COVID-19 vaccine hesitancy

Variable	Estimate	SE	P value	OR	95% CI
					Lower	Upper
Partially adjusted
Intercept	−1.163	0.070	<0.0001
≤25 years vs ≥51 years	0.729	0.087	<0.0001	2.073	1.749	2.457
26–50 years vs ≥51 years	0.880	0.056	<0.0001	2.411	2.161	2.690
Woman versus man	0.173	0.052	0.0008	1.189	1.075	1.316
Fully adjusted model*
Intercept	−0.910	0.101	<0.0001
≤25 years (vs ≥51 years)	0.371	0.107	0.006	1.449	1.174	1.788
26–50 years (vs ≥51 years)	0.403	0.075	<0.0001	1.496	1.292	1.732
Woman (vs man)	0.198	0.059	0.0008	1.218	1.085	1.368
White (vs non-white)	0.388	0.082	<0.0001	1.474	1.254	1.733
Graduate/postgraduate degree (vs high school or lower)	0.143	0.053	0.007	1.154	1.041	1.279
Employed (vs unemployed)	0.020	0.074	0.791	1.02	0.882	1.178
Annual household income ≥$60 000 vs <$60 000	0.354	0.062	<0.0001	1.424	1.26	1.609
Parent (vs not)	0.411	0.077	<0.0001	1.508	1.297	1.753
Essential worker (vs not)	0.364	0.087	<0.0001	1.439	1.214	1.705
At-risk health condition (vs none)	−0.077	0.062	0.217	0.926	0.82	1.046
Influenza vaccine ≥3–5 times in last 5 years (vs <3 times in the last 5 years)	−1.329	0.063	<0.0001	0.265	0.234	0.299
Positive COVID-19 test result (vs negative)	0.282	0.307	0.358	1.326	0.726	2.421

Psychological predictors of vaccine hesitancy

Perceptions of the importance of engaging in infection prevention behaviours across the five surveys/time points is presented in figure 3. Overall, 76% of respondents reported believing that engaging in infection prevention behaviours was extremely important, though we observed significant variations in perceived importance over time. Perceived importance was highest at survey 1 (87%), which then dropped to 71.3% by survey 2 and remained generally stable across survey 3 (74.5%), survey 4 (75.7%) and survey 5 (71.3%). Concern trends generally followed a similar pattern: M for each concern type were highest at survey 1 and dropped significantly by survey 2 and remained generally stable across surveys 3–5 (p<0.0001 for trend, see figure 3). Across all five surveys/time points, respondents reported having the greatest concerns about the social and economic impacts of the pandemic (M=3.18, SD 0.76), followed by health concerns (M=2.98, SD=0.86) and personal financial concerns (M=2.43, SD=1.08).

Figure 3

Perceptions of the importance of engaging in infection prevention behaviours (percentage of respondents reporting ‘extremely important’, dashed line) and mean COVID-19 concern levels (solid lines) across the five surveys/time points.

Partially and fully adjusted multivariate logistic regression analyses examining associations between vaccine hesitancy and perceived importance of engaging in infection prevention behaviours and COVID-19-related concern types across all surveys/time points are presented in tables 3 and 4. Respondents who perceived engaging in infection prevention behaviours to be extremely important were 78% (partially adjusted) and 77% (fully adjusted) less likely to be vaccine hesitant than those who believed engaging in these behaviours was only somewhat, not very or not at all important (ORpadj 0.22, 95% CI 0.19 to 0.25, and ORadj 0.23, 95% CI 0.20 to 0.27, respectively). Although social and economy concerns were the most endorsed by respondents, they were not predictive of vaccine hesitancy in partially or fully adjusted analyses. However, health concerns were associated with a 58% (partially adjusted) and 54% (fully adjusted) reduced odds of vaccine hesitancy (ORpadj 0.42, 95% CI 0.39 to 0.46, and ORadj 0.46, 95% CI 0.42 to 0.50, respectively), while having high personal financial concerns was associated with a 1.41 and 1.34 times greater odds of vaccine hesitancy in partially (ORpadj 1.41, 95% CI 1.32 to 1.49) and fully adjusted (ORadj 1.34, 95% CI 1.25 to 1.43) models.

Discussion

The present study analysed Canadian survey data from five age, sex and province-weighted population-based samples to describe vaccine intentions between April 2020 and March 2021 and their correlates. Over 40% of Canadians reported some degree of vaccine hesitancy over the course of the study period. Vaccine hesitancy was lowest during pandemic waves 1 and 3, and highest during pandemic wave 2, just prior to vaccine approval in Canada (December 2020). These results are consistent with data from the USA covering the same time period, which also demonstrated significant increases in vaccine hesitancy between April and December 2020 among 8167 online respondents in the Understanding America Study.9 These results are also aligned with those of a study conducted by the World Economic Forum, which reported a decline in positive vaccine intentions between August (77%) and October 2020 (73%) among 18 526 respondents from 15 countries (including 1000 from Canada).20

Profile of Canadians who are vaccine hesitant

We examined the profile of Canadians who were more likely to report being vaccine hesitant and found that in fully adjusted analyses (including survey/time point), women, younger individuals (aged 50 and younger), non-white individuals, those with lower levels of education (high school or less), and those reporting lower annual household incomes (less than $60 000/year) were significantly more likely to report being vaccine hesitant over the study period. Overall, this profile is consistent with the results of similar studies in Canada and other Western nations (eg, USA, UK, France, Italy, Germany and Australia),21–31 suggesting a robust phenomenon of higher vaccine hesitancy among women, younger individuals, non-white individuals and those of lower socioeconomic status.

The reasons for the lower vaccine intentions among women remains poorly understood and seems paradoxical, given evidence that women are more adherent to COVID-19 prevention measures in general.18 32 Some speculate it might be related to their tendency to have greater health risk perceptions in general,33 which may lead to heightened fears of experiencing vaccine side effects compared with men, resulting in less willingness to get vaccinated. These fears may not be completely unfounded, in light of evidence showing that women tend to have stronger immune reactions to vaccines than men, which may lead to more adverse events following vaccination.34 35 More recent data suggest that women may be more reluctant to get vaccinated due to reproductive factors, as women who are pregnant or planning to get pregnant appear to be delaying vaccination due to safety concerns affecting the fetus.36 37 Given evidence to suggest that pregnancy in the presence of COVID-19 may confer increased risk for severe illness, hospitalisation and intensive care unit admission, and preliminary findings of no obvious safety concerns among pregnant women who received mRNA vaccines,38 addressing vaccine hesitancy in this group will be important for protecting this vulnerable population.

Contrary to women, younger adults may be less willing to get vaccinated due to lower COVID-19 risk perception compared with older adults.39 These perceptions may have been fuelled by early reports of lower risks of COVID-19 hospitalisation and complications among younger age groups.40 While overall COVID-19-related mortality among those under age 20 remains low (proportion of all-cause deaths attributed to COVID-19 has been estimated to be 0.48%41), those aged 2–59 have accounted for 63% of all infections and 30% of all hospitalisations in Canada since the start of the pandemic.42 This suggests that this age group remains an important vector of community virus transmission, and a need to optimise vaccination uptake in this age cohort.

Our results also revealed lower vaccine intentions among non-white individuals, those with high school or less education and those with annual household incomes of less than $C60 000/year (below the poverty line in Canada).43 These results are consistent with those from previous studies in the USA,21 23 24 44 Australia28 and across Europe.29 30 45 46 Results of greater vaccine hesitancy among people of colour are a cause for concern, given that these individuals are more likely to work in industries worst affected by the COVID-19 pandemic, such as food and beverage, hospitality and long-term care services.47 Reasons for higher rates of hesitancy among these groups may include lower health literacy48 and lack of trust in vaccines and the healthcare system,49 the latter of which may be exacerbated by low representation of people of colour in vaccine trials and experience with discrimination and systemic racism.50 Clearly, greater efforts need to be made to motivate and enable those from racial and ethnic minority groups to get vaccinated.

We also identified two important groups of individuals at greater risk of being vaccine hesitant: essential and healthcare workers. Evidence of greater hesitancy among essential and healthcare workers was both surprising and a cause of concern, given that they are the individuals most likely to be exposed, and expose others to COVID-19. However, our results do seem to be in line with US data from a survey of 16 970 employed adults in the USA showing that those working in essential service sectors (ie, leisure and hospitality, manufacturing, construction, retail, transportation, and food and beverage) had the highest rates of vaccine hesitancy (45%–54%) compared with non-essential sectors like technology (25%), financial services (26%), public administration (36%) and entertainment (37%).51 Our finding of high vaccine hesitancy among healthcare workers is also consistent with other studies both within52 and outside53–55 of Canada. Though we were not able to determine what types of healthcare workers are more likely to be vaccine hesitant, data from previous reports suggests this is more common among female healthcare workers,52 53 56 as well as nurses and paramedical professionals rather than physicians or health administrators.54–56 While the reasons for vaccine hesitancy among healthcare workers remain poorly understood, available evidence suggest their hesitancy is linked to vaccine novelty and concerns about safety.52 55 Further research is needed to identify barriers to vaccination among essential and healthcare workers due to their high risk of virus exposure and transmission.

There were two additional findings from our analyses that warrant discussion. The first is that vaccine hesitancy was higher among those with an inconsistent history of getting the influenza vaccine. This is consistent with previous reports22 44 46 57 58 and suggests that having favourable vaccine attitudes and behaviours in general is associated with greater likelihood of getting vaccinated against COVID-19. The other finding is that parents of children under age 18 were 1.5 times more likely to be vaccine hesitant compared with non-parents. Given the recent approval of vaccines among children 5–11 years of age in Canada, this finding is a cause for concern and consistent with at least one study out of the UK that also found that parents of young children were more likely to report vaccine hesitancy or refusal.59 The reasons for this are remain poorly understood but may reflect more general trends of parental hesitancy to vaccinate their children against common infectious diseases (eg, mumps, measles and pertussis).60 Given that COVID-19 infection rates are currently highest among school-aged children in Canada,61 parents represent an important target for vaccination. Further research is needed to understand the reasons for vaccine hesitancy in this group and the impact of personal vaccine hesitancy on their willingness to get their children vaccinated against COVID-19, in order to optimise vaccination rates in this vulnerable group.

In addition to sociodemographic predictors, we also assessed psychological predictors of vaccine hesitancy. One of the strongest predictors of positive vaccine intentions was the extent to which Canadians believed engaging in preventive health behaviours (eg, vaccination) was important for reducing virus transmission. Those who believed that engaging in preventive health behaviours (like getting vaccinated) was ‘extremely important’ were 77% less likely to be vaccine hesitant after adjustment for covariates including sociodemographics and survey period/time point. This finding is consistent with previous reports linking high perceived benefits (of getting vaccinated) to positive vaccine intentions,62 highlighting the need for vaccination campaigns to clearly and consistently emphasise how the benefits of getting vaccinated far outweigh any risks. We also found that different types of COVID-19-related concerns were important determinants of vaccine hesitancy. Interestingly, even though social and economy concerns were the most highly endorsed at each survey/time point, only high health-related concerns and personal financial concerns were significant predictors of vaccine hesitancy—but not in the same direction. In fact, we found that those with high health concerns (ie, concerned about becoming infected and/or infecting others) were 54% less likely to be vaccine hesitant, while those with high concerns about their personal financial situation (eg, were worried about job and income loss or not having enough money to feed their family) were 1.33 times more likely to report being vaccine hesitant. Results linking high health concerns to lower vaccine hesitancy are consistent with those of other studies in Canada, the USA, Australia and Europe,21 23 24 26 29 58 63–66 and provide further support of the need for vaccination campaigns to highlight how getting vaccinated is going to be health protective. However, to our knowledge, this is the first study to date to observe a link between high personal financial concerns and increased vaccine hesitancy, and suggests that those whose livelihoods were negatively impacted by the virus may be less willing or able to get vaccinated. Further research is needed to determine the extent to which this reflects a lack of motivation or desire to get vaccinated, or a perceived inability to get vaccinated due to practical barriers or limitations (eg, lack of access to paid leave to get vaccinated).

Limitations and strengths

This study should be interpreted in light of some methodological limitations. First, although we included large, national samples of Canadians with representation across age, sex and province, the absolute number of participants in certain provinces (eg, Atlantic) was lower, making interprovincial comparisons difficult. Second, the survey was available only in English and French, which may have led to an under-representation of certain non-native English or French speaking groups. Further, our surveys included fewer people of colour, which may reflect participation on online panels, so results might not generalise as well to non-white participants. Third, since the surveys were voluntary and participants were drawn from a polling firm’s subject pool, participation may have been subject to some degree of selection bias. Fourth, though this study presents data depicting vaccine intentions over time, it was drawn from three separate cohorts of online panels, so data reflect trends in vaccine intentions over time but not in the same individuals. Finally, data were self-reported, which may have been subject to social desirability bias.67 However, the fact that the surveys were anonymous likely mitigated this limitation.

Despite some limitations, this study also had a number of important strengths. The study included a large sample size; respondents were well distributed across provincial regions, age groups, employment status and income compared with census data available through Statistics Canada; and there were equal proportions of men and women. This study also collected data during peak lockdown of the first wave (April 2020) through to the end of the third wave (end of March 2021) when vaccines started becoming available in Canada. This allowed for the assessment of changes in vaccine intentions over time across three critical waves of the pandemic in Canada. We conducted principal component analysis to determine the structure of our concerns module, which was found to have excellent internal consistency, which is important for ensuring the validity of our results linking concern types to vaccine hesitancy. Finally, results reflect a subanalysis of Canadian representative data from the iCARE Study, which has collected data from more than 100 000 people from 190 countries to date alongside ongoing efforts to collect similarly representative samples in eight other countries (see: www.icarestudy.com). This will facilitate comparisons with international datasets to contribute important evidence to support the development and implementation of COVID-19 vaccine policy strategies worldwide.

Conclusions

Over 40% of Canadians reported some degree of vaccine hesitancy betweenApril 2020 and March 2021. Vaccine hesitancy was lowest during pandemic waves 1 and 3, and highest during pandemic wave 2, just prior to vaccine approval in Canada. Women, individuals aged 50 and younger, non-white individuals, those with high school education or less, and those with annual household incomes below the poverty line in Canada (ie, $60 000) were significantly more likely to report being vaccine hesitant over the study period. Three important groups of Canadians were identified as being vaccine hesitant: essential and healthcare workers, parents of children under the age of 18 and those without a previous history of influenza vaccination. Finally, perceived importance of engaging in infection prevention behaviours (like vaccination) and having high COVID-19-related health concerns were predictive of lower levels of vaccine hesitancy, whereas having high COVID-19-related personal financial concerns was predictive of higher levels of vaccine hesitancy. Overall, results point to the importance of targeting vaccine efforts to subgroups who may be socioeconomically disadvantaged, who also happen to be disproportionately represented in essential service occupations including healthcare. Finally, vaccine messaging should emphasise how the benefits of getting vaccinated (particularly to health) far outweigh the risks, particularly those associated with personal financial losses. Future research is needed to monitor ongoing changes in vaccine intentions and behaviour, as well as to better understand motivators and facilitators of vaccine acceptance, particularly among vulnerable groups.

Table 3

Multivariate logistic regression model estimating the association between COVID-19-related concerns and vaccine hesitancy

Variable	Estimate	SE	P value	OR	95% CI
					Lower	Upper
Partially adjusted models*
Intercept	0.701	0.154	<0.0001
Health concerns	−0.861	0.040	<0.0001	0.423	0.391	0.458
Personal financial concerns	0.341	0.031	<0.0001	1.406	1.324	1.493
Social and economy concerns	−0.035	0.042	0.394	0.966	0.891	1.047
Fully adjusted models†
Intercept	0.914	0.184	<0.0001
Health concerns	−0.780	0.043	<0.0001	0.458	0.421	0.499
Personal financial concerns	0.290	0.035	<0.0001	1.336	1.248	1.429
Social and economy concerns	−0.064	0.046	0.1633	0.938	0.858	1.026

*Partially adjusted for sex, age, ethnicity/race and survey/time point.

†Fully adjusted for sex, age, ethnicity/race, and survey/time point, education, employment status, annual household income, health risk condition, essential worker, healthcare worker, parental status, history of influenza vaccine and COVID-19 test result.

Table 4

Multivariate logistic regression model estimating the association between perceived importance of COVID-19 prevention measures and vaccine hesitancy

Variable	Estimate	SE	P value	OR	95% CI
					Lower	Upper
Partially adjusted models*
Intercept	0.168	0.092	0.0665
Perceived importance	−1.536	0.065	<0.0001	0.215	0.19	0.245
Very important versus others†
Fully adjusted models‡
Intercept	0.371	0.123	0.0026
Perceived importance	−1.462	0.073	<0.0001	0.232	0.201	0.267
Very important versus others†

*Partially adjusted for sex, age, ethnicity/race and survey/time point.

†Others: somewhat important, not very important, not at all important.

‡Fully adjusted for sex, age, ethnicity/race and survey/time point, education, employment status, annual household income, at-risk health condition, essential worker, healthcare worker, parental status, history of flu vaccine and COVID-19 test result.