COVID-19 vaccination intention during early vaccine rollout in Canada: a nationwide online survey.

BACKGROUND: Understanding vaccination intention during early vaccination rollout in Canada can help the government's efforts in vaccination education and outreach.
METHOD: Panel members age 18 and over from the nationally representative Angus Reid Forum were invited to complete an online survey about their experience with COVID-19, including their intention to get vaccinated. Respondents were asked "When a vaccine against the coronavirus becomes available to you, will you get vaccinated or not?" Having no intention to vaccinate was defined as choosing "No - I will not get a coronavirus vaccination" as a response. Odds ratios and predicted probabilities are reported for no vaccine intentionality in demographic groups.
FINDINGS: 14,621 panel members completed the survey. Having no intention to vaccinate against COVID-19 is relatively low overall (9%) with substantial variation among demographic groups. Being a resident of Alberta (predicted probability = 15%; OR 0.58 [95%CI 0.14-2.24]), aged 40-59 (predicted probability = 12%; OR 0.87 [0.78-0.97]), identifying as a visible minority (predicted probability = 15%; OR 0.56 [0.37-0.84]), having some college level education or lower (predicted probability = 14%) and living in households of at least five members (predicted probability = 13%; OR 0.82 [0.76-0.88]) are related to lower vaccination intention.
INTERPRETATION: The study identifies population groups with greater and lesser intention to vaccinate in Canada. As the Canadian COVID-19 vaccination effort continues, policymakers may use this information to focus outreach, education, and other efforts on the latter groups, which also have had higher risks for contracting and dying from COVID-19. FUNDING: Pfizer Global Medical, Unity Health Foundation, Canadian COVID-19 Immunity Task Force.

Evidence before this study

Several surveys assessing national-level vaccine hesitancy or acceptance have been published in final or in pre-print form for other countries (France, England, Italy, US). Only two other studies have surveyed vaccination acceptance among the general population in Canada, but their sample sizes are small compared to the Ab-C study. Other Canadian surveys focused on specific populations, such as healthcare workers and ethnic minorities, rather than the general population. We searched PubMed, medRxiv, bioRxiv, arXiv, and medRxiv up to May 5, 2021 for preprints and peer-reviewed articles reporting on COVID-19 vaccination hesitancy or intention, using search terms, such as “vaccine hesitancy COVID,” “covid vaccine,” “covid vaccine acceptance.”

Added value

The Ab-C study is the first nationwide and reasonably representative study of SARS-CoV-2 in Canada, including intention to be vaccinated against COVID during the early phase of vaccination rollout, beginning in late 2020. The study produced estimated probabilities for having no vaccination intention by specific demographic groups. Our findings confirm that most Canadians intend to be vaccinated, with 9% of adult Canadians having no intention to vaccinate. However, vaccination intention is comparatively lower in key demographic groups, including residents of Alberta, individuals without a university degree, and visible minorities.

Implications of all the available evidence

Consistent with earlier understanding, Ab-C documents a relatively high vaccination intention in Canada. However, because children do not currently have priority for vaccination and key demographic groups with higher proportions with no intention to vaccinate tend to cluster in networks, the combination may fuel subsequent local waves . Policymakers may wish to focus their resources on vaccination campaigns targeting the most reluctant groups and to begin vaccinating children as soon as possible.

Alt-text: Unlabelled box

Introduction

Although the Canadian COVID-19 mortality rate is low compared to the United States and some European countries, many parts of Canada experienced a significant third wave in March-May 2021, driven by more infectious variants and lifting of public health measures [[1], [2], [3]].

Canada started vaccinating healthcare workers and nursing home residents in December 2020, and has prioritized the rest of the population by mortality risk factors (e.g., age, comorbidities), and some provinces further prioritized based on risk of disease acquisition (e.g., essential workers, living in highly impacted regions) [4,5]. Although Canada's COVID-19 vaccination efforts had a slow start with less than 4% of the population vaccinated with one dose by the end of February 2021 (compared to 15% in the United States) [6,7], two-thirds of Canadians had received at least 1 dose of a COVID-19 vaccine compared just over half in the United States by mid-June [6,7].

As a part of the ongoing Action to Beat Coronavirus (Ab-C) study [8], we report on levels and predictors of self-reported intention to vaccinate against COVID-19 and identify key groups with low intention to vaccinate at the start of the vaccine rollout in Canada. in a reasonably representative sample of Canadians.

Intentionality toward vaccination is related to, but not synonymous with, vaccine hesitancy. The World Health Organization defines vaccine hesitancy as “delay in acceptance or refusal of vaccines despite availability of vaccine services” [9] that is influenced by complacency, convenience, and confidence [9,10], which applies in situations where vaccines are already on offer. There are other reasons for refusing or delaying vaccination (e.g., not meeting age eligibility requirements, in quarantine after natural SARS-CoV-2 infection, contraindication to the vaccine), which also affect both intentions and vaccination acceptance when available [11].

Methods

Study design and survey timeline

The Ab-C study is a collaboration between the University of Toronto, Unity Health Toronto, the Centre for Global Health Research, and the Angus Reid Forum (ARF) [12], an established panel of Canadian adults used for online opinion polling. The bilingual (English and French) Ab-C study started in 2020, with an initial online survey in May, including questions about COVID-19 symptoms and SARS-CoV-2 testing history. Respondents were then asked to self-collect a dried blood sample using a mailed kit so that seroprevalence during the first viral wave could be determined [[13], [14], [15]]. A second online survey, coinciding with Canada's vaccination rollout (December 2020 through February 2021) asked about COVID-19 symptoms and SARS-CoV-2 testing since the first survey, and about their intention to be vaccinated, which is the basis of this report [16]. A second dried blood sample was also requested. Ab-C participants receive no monetary compensation, but do receive modest redeemable points from the ARF.

Representativeness of the survey and sampling

44,270 members were invited to participate in the initial Ab-C survey (May to September 2020), stratified by age, sex, education, region, and census metropolitan area to match national demographic profiles. Of the 19,994 (45%) who responded, all of them were invited again in early 2021 to complete a second 20-item survey, which included a question about vaccination intention. Up to four reminder emails were sent after the initial invitation. 14,621 (73%) respondents completed the second survey.

Ab-C respondents were similar demographically to the 2016 Canadian Census profile, with some exceptions [14,17,18]. First, people 60 years and older were intentionally oversampled to capture a greater share of the population at highest risk of SARS-CoV-2 [[13], [14], [15]]. Second, panel members with higher education levels were overrepresented (23% with a Bachelor's degree or higher in Canada and 43% in the second survey sample). Hence, all subsequent analyses were adjusted for education levels to achieve the same distribution as in the Canadian census [19]. The prevalences of smoking, diabetes, and hypertension were similar to national surveys [[20], [21], [22], [23]], suggesting that the health profile of the cohort was similar to national population distributions.

Survey questions

Participants were asked “When a vaccine against the coronavirus becomes available to you, will you get vaccinated or not?” The five possible responses were: 1. “No – I will not get a coronavirus vaccination,” 2. “Not sure,” 3. “Yes – I will eventually get a vaccination, but I will wait awhile first,” 4. “Yes – I will get a vaccination as soon as one becomes available to me,” 5. “I have already been vaccinated.” In the analysis, we combined responses 4 and 5 as indicating positive intent, leaving four ordered categories. Having no vaccination intention was defined as choosing “No – I will not get a coronavirus vaccination” as a response. No questions were asked about why respondents chose their responses.

Analytical method

We calculated odd ratios (ORs) of intention to vaccinate by province, age, sex, education, visible minority, Indigenous ancestry, household size, body mass index (BMI), smoking status, and history of diabetes and hypertension using a survey weighted ordered logit model. The variables were selected to address the inherent differences in disease outcome in demographic groups (e.g., age), healthcare access (e.g., Indigenous ancestry), transmission of diseases in close quarters (e.g., household size), provincial policies, and other common factors that may affect individuals’ health (e.g., smoking status, BMI, history of diabetes and hypertension). To identify high-risk groups, we modeled predicted probabilities (prevalences) of each vaccination intention level by demographic group after obtaining the ORs. These probabilities have been adjusted for educational attainment levels by weighting.

Prior to calculating ORs, parallel line assumptions (also known as the proportional odds assumption, which assumes that the relationship between demographics and different thresholds of the outcome is proportional) [24] were tested for each demographic factor at the 0.05 level within the survey setting, and all variables met this assumption. The overall – multivariable - model also meets the parallel line assumption with an adjusted Wald test. All analyses were done in Stata 16 [25].

Ethics

The Ab-C study was approved by the IRB of Unity Health Toronto. Informed consent was obtained from all participants through an online form at the beginning of the survey, which could not proceed without consent.

Role of the funding source

The funders had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Results

Between January and March 2021, 14,621 participants out of 20,359 polled responded to the survey. The largest number of respondents were from Ontario (39.4%), followed by British Columbia and Yukon (18.9%), Quebec (15.7%), Alberta (12.6%), Manitoba and Saskatchewan (7%), and the Atlantic provinces (7%). Most participants are below age 70 (30% age 18-39, 35% age 40-59, 24% age 60-69), and 12% age 70 or above. 53% were women and 46% were men, and 34% had some (incomplete) university education and 43% were university graduates. About 14% identified as visible minorities and 9% had Indigenous ancestry. Almost half (43%) of the respondents lived with another person, while others lived alone (18%) or in a household of three (17%), four (15%) or at least five (7%). 32% and 27% had BMI classified as overweight [[25], [26], [27], [28], [29], [30]] or obese (>30), respectively. 48% were former or current smokers. Almost 10% of the participants were diagnosed with diabetes and 27% had hypertension.

Among Ab-C participants, 9% had no intention to vaccinate (Table 1 ). Alberta (16%) and other Prairie provinces (14%) had higher proportions not intending to vaccinate compared to less than 10% in all other provinces. Participants aged 40-59 had the lowest vaccination intention, with 11.6% reporting no intention to vaccinate, compared to less than 10% in all other age groups. Men had greater proportion with no intention to vaccinate (11%) than women (8%), and those with less than college level education (e.g., associate or technical degree) or less had more respondents with no intention to vaccinate (14%) than participants with other levels of education. Other groups with lower intention to vaccinate than average included those identifying themselves as visible minority (12%), Indigenous (15%), and those living in households of five or more people (16%).

Table 1

Unweighted survey sample characteristics (n = 14,621)

	Canada	Ab-C		Ab-C respondents who "Will not vaccinate" by demographics		Test of difference in vaccination intention in groups
	%	n	%	n	row %	p value
Total		14,621	100%	1355	9.3%
Province
Ontario	38%	5,766	39.4%	448	7.8%	<0.0001
British Columbia	13%	2,761	18.9%	199	7.2%
Quebec	23%	2,299	15.7%	191	8.3%
Alberta	12%	1,839	12.6%	301	16.4%
Manitoba & Saskatchewan	7%	1,027	7%	142	13.8%
Atlantic provinces	7%	929	6.4%	74	8%
Age groups
18 to 39 years	49%	4,270	29.2%	396	9.3%	<0.0001
40 to 59 years	28%	5,106	34.9%	590	11.6%
60 to 69 years	12%	3,472	23.7%	257	7.4%
70 years or older	11%	1,773	12.1%	112	6.3%
Sex
Male	49%	6,742	46.1%	736	10.9%	<0.0001
Female	51%	7,777	53.2%	602	7.7%
Prefer to self-describe	0%	102	0.7%	17	16.7%
Education
Some college or lower	45%	3,402	23.3%	468	13.8%	<0.0001
Some university	32%	4,935	33.8%	564	11.4%
Bachelor or higher	23%	6,284	43%	323	5.1%
Visible minority
No	78%	12,526	85.7%	1,098	8.8%	<0.0001
Yes	22%	2,095	14.3%	257	12.3%
Indigenous ancestry
Not Indigenous	95%	13,317	91.1%	1,164	8.7%	<0.0001
Indigenous	5%	1,304	8.9%	191	14.6%
Household size
One person	28%	2,678	18.3%	262	9.8%	<0.0001
Two people	34%	6,315	43.2%	501	7.9%
Three people	15%	2,415	16.5%	219	9.1%
Four people	14%	2,163	14.8%	204	9.4%
Five or more people	8%	1,050	7.2%	169	16.1%
Body Mass Index
Under or normal weight (< 25 kg/m2)	37%	4,318	29.5%	396	9.2%	0.162
Overweight (≥ 25 kg/m2 & < 30 kg/m2)	37%	4,716	32.3%	447	9.5%
Obese (≥ 30 kg/m2)	27%	3,914	26.8%	337	8.6%
Unknown		1,673	11.4%	175	10.5%
Smoking status
Never	46%	7,310	50%	632	8.6%	0.016
Former or current	54%	7,063	48.3%	693	9.8%
Unknown	0%	248	1.7%	30	12.1%
Diabetic
No	91%	13,064	89.4%	1,222	9.4%	0.122
Yes	9%	1,424	9.7%	116	8.1%
Unknown	0%	133	0.9%	17	12.8%
Hypertensive
No	77%	10,419	71.3%	1,026	9.8%	<0.0001
Yes	23%	3,980	27.2%	299	7.5%
Unknown	0%	222	1.5%	30	13.5%

Notes:

British Columbia includes Yukon. Manitoba & Saskatchewan include Northwest Territories.

The category "Unknown" represents non-responses in the survey.

The category "Prefer to self-describe" is a choice of sex in Angus Reid Forum's member profile.

n – unweighted counts

% - survey weighted percentages

Sources of % of characteristics at the Canadian national-level:

aCensus Canada 2016 [[17], [18], [19],30]

bHealth Fact Sheets [20,21]

cHealth Reports [22]

dPublic Health Agency of Canada [23]

Table 2 presents participants’ weighted ORs of vaccination intention, and Table 3 represents the predicted probabilities to not vaccinate for each demographic group. Compared to the reference groups, the demographics with lower odds of intending to vaccinate included age 40-59 years (OR = 0.87, 0.78 – 0.97, predicted probability = 12%), being a visible minority (OR = 0.56, 0.37 – 0.84, predicted probability = 15%), and belonging to a household of five or more people (OR = 0.82, 0.76 – 0.88, predicted probability = 13%). Being from Alberta and having some college level education or lower also had higher predicted probability of having no intention to vaccinate at 15% and 14%, respectively. Having BMI <25 was also associated with a slightly higher probability of having no intention to vaccinate at 11%, compared to those being overweight (10%) or obese (9%). The predicted probabilities of unwillingness for key demographic groups are displayed in Figure 1 .

Table 2

Results of the weighted ordered logit for intention to vaccinate

	n	Univariate results		Multivariate results
		Weighted OR	Weighted 95% CI	Weighted OR	Weighted 95% CI
Province
Ontario	5,766	Ref		Ref
British Columbia	2,761	1.00	(0.66, 1.53)	1.09	(0.67, 1.77)
Quebec	2,299	1.04	(0.68, 1.57)	1.01	(0.72, 1.43)
Alberta	1,839	0.57	(0.14, 2.24)	0.58	(0.17, 2.01)
Manitoba & Saskatchewan	1,027	0.63	(0.17, 2.36)	0.70	(0.21, 2.28)
Atlantic provinces	929	0.93	(0.64, 1.35)	0.95	(0.52, 1.74)
Age groups
18 to 39 years	4,270	Ref		Ref
40 to 59 years	5,106	0.85*	(0.78, 0.92)	0.87*	(0.78, 0.97)
60 to 69 years	3,472	1.45*	(1.21, 1.73)	1.41*	(1.03, 1.92)
70 years or older	1,773	1.94*	(1.86, 2.03)	1.75*	(1.50, 2.04)
Sex
Male	6,742	Ref		Ref
Female	7,777	1.23	(1.00, 1.51)	1.23	(0.97, 1.56)
Prefer to self-describe	102	0.50	(0.21, 1.21)	0.76	(0.38, 1.53)
Education
Some college or lower	3,402	Ref		Ref
Some university	4,935	1.32*	(1.10, 1.58)	1.33*	(1.10, 1.62)
Bachelor or higher	6,284	2.57*	(1.54, 4.29)	2.83*	(1.96, 4.09)
Visible minority
No	12,526	Ref		Ref
Yes	2,095	0.56*	(0.37, 0.84)	0.56*	(0.37, 0.84)
Indigenous ancestry
Not Indigenous	13,317	Ref		Ref
Indigenous	1,304	0.71*	(0.54, 0.94)	0.95	(0.86, 1.05)
Household size
One person	2,678	Ref		Ref
Two people	6,315	1.22*	(1.14, 1.31)	1.28*	(1.09, 1.50)
Three people	2,415	1.01	(0.82, 1.24)	1.19	(0.87, 1.62)
Four people	2,163	0.96	(0.79, 1.16)	1.20	(0.82, 1.75)
Five or more people	1,050	0.62*	(0.49, 0.78)	0.82*	(0.76, 0.88)
Body Mass Index
Under or normal weight (< 25 kg/m2)	4,318	Ref		Ref
Overweight (≥ 25 kg/m2 & < 30 kg/m2)	4,716	1.04	(0.75, 1.43)	1.11	(0.75, 1.63)
Obese (≥ 30 kg/m2)	3,914	1.04	(0.85, 1.26)	1.18*	(1.05, 1.34)
Unknown	1,673	0.79	(0.58, 1.07)	0.93	(0.59, 1.48)
Smoking status
Never	7,310	Ref		Ref
Former or current	7,063	0.90	(0.77, 1.07)	0.99	(0.85, 1.15)
Unknown	248	0.57*	(0.36, 0.90)	0.83	(0.57, 1.22)
Diabetic
No	13,064	Ref		Ref
Yes	1,424	1.23*	(1.08, 1.39)	1.11	(0.85, 1.44)
Unknown	133	0.45*	(0.38, 0.53)	0.77	(0.32, 1.85)
Hypertensive
No	10,419	Ref		Ref
Yes	3,980	1.32*	(1.25, 1.38)	1.21*	(1.09, 1.33)
Unknown	222	0.50*	(0.26, 0.95)	0.72	(0.47, 1.10)

Notes:

British Columbia includes Yukon. Manitoba and Saskatchewan include Northwest Territories.

The category "Unknown" represents non-responses in the survey.

The category "Prefer to self-describe" is a choice of sex in Angus Reid Forum's member profile.

* p value <0.05

OR – odds ratio

95% CI – confidence interval

Table 3

Predicted probabilities for vaccination hesitancy from multivariate ordered logit

	Will not vaccinate
	Prob	SE
Province
Ontario	9.2%*	(1.2%)
British Columbia	8.5%*	(1.7%)
Quebec	9.1%*	(1.4%)
Alberta	14.6%	(4.7%)
Manitoba & Saskatchewan	12.6%	(3.1%)
Atlantic provinces	9.6%*	(1.4%)
Age groups
18 to 39 years	10.7%*	(2.2%)
40 to 59 years	12%*	(2.2%)
60 to 69 years	7.9%*	(1.2%)
70 years or older	6.5%*	(1.2%)
Sex
Male	11%*	(2%)
Female	9.1%*	(1.7%)
Prefer to self-describe	13.8%	(3.7%)
Education
Some college or lower	14%*	(2.7%)
Some university	11%*	(2%)
Bachelor or higher	5.6%*	(0.8%)
Visible minority
No	9.2%*	(1.8%)
Yes	15%*	(1.6%)
Indigenous ancestry
Not Indigenous	10%*	(1.8%)
Indigenous	10.4%*	(1.7%)
Household size
One person	11.3%*	(2.4%)
Two people	9.1%*	(1.7%)
Three people	9.7%*	(1.5%)
Four people	9.7%*	(1.3%)
Five or more people	13.4%*	(2.8%)
Body Mass Index
Under or normal weight (< 25 kg/m2)	10.7%*	(1.6%)
Overweight (≥ 25 kg/m2 & < 30 kg/m2)	9.8%*	(2.1%)
Obese (≥ 30 kg/m2)	9.2%*	(1.6%)
Unknown	11.3%*	(2.3%)
Smoking status
Never	10%*	(1.9%)
Former or current	10.1%*	(1.7%)
Unknown	11.7%*	(2.5%)
Diabetic
No	10.1%*	(1.8%)
Yes	9.2%*	(2.1%)
Unknown	12.6%	(3.2%)
Hypertensive
No	10.4%*	(1.8%)
Yes	8.8%*	(1.8%)
Unknown	13.7%*	(1.6%)

Notes:

British Columbia includes Yukon. Manitoba & Saskatchewan include Northwest Territories.

The category "Unknown" represents non-responses in the survey.

The category "Prefer to self-describe" is a choice of sex in Angus Reid Forum's member profile.

* p value <0.05

Prob – predicted probabilities

SE – standard error

Figure 1

Probabilities of unwillingness to vaccinate for key demographic groups among participants in the Ab-C study

Discussion

This study assessed intention to be vaccinated in the early phase of vaccination rollout in a nationally representative sample of Canadians, identifying differences by age, education, ethnicity, and provincial residence.

The 9% with no intention to vaccinate in our survey of Canadian adults is significantly lower than the 15-35% reported in Latin American countries (64.6% vaccination intention in Paraguay to 85% vaccination intention in Puerto Rico U.S.) in a recent cross-sectional study [26] and the 20% respondents who are unlikely to receive a COVID-19 vaccine in Canada from a population-based study that surveyed 4,948 respondents in British Columbia [27]. Our result that visible minorities are more likely not to vaccinate than others is consistent with the findings of a recent cohort study from the United States and the United Kingdom [28]. Surprisingly, a longitudinal study that examined vaccination attitudes in the United States between March and August 2020 has shown a decline in favorable attitudes, in contrast to an initial assumption that increasing experience with the vaccines would favor increasing acceptance [28]. This finding is a reminder to the public health community that although most people in Canada intend to be vaccinated at some point, 9% of people without intention to vaccinate may still impose a notable hurdle to reaching herd immunity (which potentially has a threshold of at least 70% of the population for SARS-CoV-2 if all restrictions on activities were lifted [29]), especially given that the virus and its variants are expected to become endemic [30] and that people under age 12 are still not eligible for vaccination [31].

Applying the 9% without vaccination intention to Canada as a whole, the approximately 3 million adult Canadians who might not get vaccinated could play an important role in perpetuating viral transmission. This group may interact with the approximate 5 million Canadians below age 12 [32] who are not likely to be vaccinated soon. Moreover, vaccine effectiveness in reducing infection (and perhaps reducing transmission) is about 90% for most of the vaccines available in Canada [33], which suggests that some vaccinated people may also contribute to continuing viral circulation. Within particular sub-groups, such as visible minorities, more people wishing not to vaccinate paired with greater exposure to children might lead to fueling transmission within local networks [34,35]. Thus, encouraging vaccine intention overall (and eventually vaccinating children) is important.

This study has three limitations. First, the online survey did not ask about participants’ attitudes and beliefs towards vaccination generally prior to asking them about their intention to vaccinate against COVID-19, nor did it ask the reasons for their response to the vaccination question. People who have preferences regarding vaccine types may be more willing to wait for a particular vaccine which could affect vaccination intention (36). The next phases of the Ab-C study will include questions to elicit information about the reasons for declining vaccination and to determine whether people followed through on their intentions. Second, this cross-sectional survey cannot detect changes in attitudes toward COVID-19 vaccination. Third, although the oversampling of older age groups was only relevant for the initial Ab-C survey, the higher proportions of older age groups persisted in this survey. The survey weights did not adjust for the oversampling of older age groups. The limitation on sampling may affect the generalizability of the results among younger age groups, while the lack of information about attitudes and beliefs about vaccines limits the understanding about why vaccination hesitancy exists. The survey may have underestimated the proportion of the population with no intention to get vaccinated if people felt that the “socially responsible” response was to say they would take a COVID vaccine.

As the Canadian COVID-19 vaccination effort continues, policymakers may want to focus outreach, education, and other efforts on the groups identified with the lowest inclination to get vaccinated, which are also the groups with higher risks for contracting and dying from COVID-19 [36]. Influence from family physicians or Provincial Health Officers may address intention to receive a COVID-19 vaccine [27]. As Canada's vaccination campaign continues, with a goal of vaccinating the majority of the population by September 2021 [4], we will periodically re-survey the Ab-C cohort and report on actual vaccination history, which will shed light on the relationship between intentionality and hesitancy.

Contributors

Conceived the idea and developed the study design: PJ. Data collection and analyses: XT, EM, TL. XT, NN, HG and PJ wrote the initial draft, and all authors were involved in commenting on subsequent revisions. PJ is the guarantor.

Declaration of interests

We declare no competing interests.

Data sharing

The Ab-C Data will be publicly available through the Canadian Immunity Task Force data release.