Health Reports
COVID-19 infection in the Canadian household population

by Tracey Bushnik, Steven Earl, Jonathan Cabot and Janine Clarke

DOI: https://www.doi.org/10.25318/82-003-x202200400003-eng

Abstract

Background

Certain population groups face a disproportionate burden of exposure to COVID-19. This study examined characteristics of Canadians living in private households in fall 2020 and winter 2021 who had been infected with COVID-19.

Data and methods

With an online questionnaire and an at-home finger-prick blood test, the Canadian COVID‑19 Antibody and Health Survey was designed to estimate the prevalence of COVID‑19 infection among people in private households in Canada. Data were collected from respondents aged 1 or older in the 10 provinces and the three territorial capitals, from November 2020 to April 2021. Descriptive statistics and logistic regression were used to identify characteristics that were associated with being seropositive for a past COVID-19 infection. Gender differences in observed associations were examined.

Results

After covariate adjustment, younger age and visible minority status were associated with an increased likelihood of being seropositive for a past COVID-19 infection. For males, having a visible minority status, having less education and living in a multi-unit dwelling increased the likelihood of being seropositive. Females were more likely to have been seropositive if they worked in health care in direct contact with others.

Interpretation

As Canada navigates the fifth wave of the pandemic, understanding who was more likely to be infected in earlier waves can help ongoing public health efforts to stop the transmission of COVID-19.

Keywords

burden, COVID-19, exposure, pandemic, SARS-CoV-2, seroprevalence, gender

Authors

Tracey Bushnik is with the Health Analysis Division, Analytical Studies and Modelling Branch, Statistics Canada. Steven Earl and Janine Clarke are with the Centre for Population Health Data, Statistics Canada, and Jonathan Cabot is with the Centre for Social Data Integration and Development, Statistics Canada.

End of text box

Introduction

Since the onset of the pandemic, studies worldwide have found that certain population groups face a disproportionate burden of exposure to COVID-19. These include people of working age, people in racially marginalized groups or who are at a social disadvantage and health care workers.^{Note 1Note 2Note 3} Not only are these groups at increased risk of infection, but some may also experience barriers to COVID-19 testing^Note 4 and vaccination.^Note 5 Gender and sex differences in exposure and susceptibility have also been reported, but results have varied across countries.^Note 6

Exposure risk varies by these characteristics for a number of reasons. It may vary by gender because of occupational, risk-taking and health-seeking behaviour differences, and sex can influence a person’s immune response or ability to resist or withstand infection.^Note 6 Younger age is a factor in the household population because children, youth and working-age people have an increased likelihood of exposure owing to more encounters with non-household members through active social behaviour, school, work or caring for others.^Note 7 Being part of a racialized or visible minority group is associated with an increased exposure risk because of the increased likelihood of living in higher-risk residential settings and working in higher-risk occupations.^{Note 8Note 9} Housing can contribute to increased exposure through the amount of shared living space or amenities in the same household,^Note 10 in multi-unit buildings or in more densely populated communities.^Note 8 Working in health care or other essential occupations can contribute to increased exposure through more frequent and close contact with others who may be infected.^Note 11

In Canada, the first wave of the pandemic began in March 2020 and peaked in mid-April 2020^Note 12 (Figure 1). At that time, outbreaks in long-term care facilities accounted for the majority of COVID-19-related deaths.^Note 13 The second wave emerged in August 2020 and peaked in early January 2021, followed closely by the third wave that peaked in April 2021.^Note 12 Provisional weekly estimates of excess mortality—that is, mortality above the expected level that is directly or indirectly related to the pandemic—suggested that more than 9,000 excess deaths occurred in Canada during the second wave, with two-thirds of these excess deaths among adults aged 65 or older.^Note 14Vaccination efforts began near the end of December 2020, with priority given to high-risk people owing to a limited initial vaccine supply. From December 2020 through the beginning of April 2021, about 15% of Canadians (about 5.8 million people)—mainly adults aged 80 or older, residents and staff of congregate living settings that provided care for seniors, health care workers, and adults living in Indigenous communities^Note 15—received at least one dose of a COVID-19 vaccine.^Note 16

The first of its kind at the national level, the Canadian COVID-19 Antibody and Health Survey (CCAHS) provides information about the prevalence of a past COVID-19 infection among people aged 1 or older living in private households in Canada. The CCAHS collected questionnaire and dried blood spot (DBS) data during the second and third waves of the pandemic, from November 2020 to April 2021 (Figure 1). Using these data, this study examines associations between a past COVID-19 infection and the characteristics of Canadians living in private households. It also examines gender differences in these associations. With the pandemic in its fifth and perhaps entering a sixth wave, understanding past patterns in COVID-19 infection and identifying who was at increased risk can help inform ongoing and future policy.

Methods

Data source

The Canadian COVID-19 Antibody and Health Survey (CCAHS) is cross-sectional. It used a direct sampling frame for people aged 1 to 24 years and a multi-stage sampling frame for those aged 25 or older to select a sample of 47,900 people living in private households across the 10 provinces and the three territorial capitals. Selected respondents were asked to complete an online electronic questionnaire (EQ) and an at-home finger-prick blood test, called a dried blood spot (DBS) test. After filling in the EQ, respondents were asked to complete the DBS test and mail it using the provided pre-addressed envelope. Data were collected from November 2020 to April 2021, with the bulk of collection completed in January and February 2021 (86%). This yielded 11,026 questionnaire and DBS test responses for a 23% combined response rate. A survey weight to account for the complex sampling design and survey nonresponse was produced. Thus, weighted estimates from the CCAHS are representative of the target population. The CCAHS does not include people living on reserves and other Indigenous settlements in the provinces, individuals living on Canadian Forces bases, the institutionalized population or residents of certain remote regions—these groups account for approximately 3% of the Canadian population aged 1 or older. Details about the CCAHS design and weighting can be found in the user guide.^Note 17

The present analysis excluded 206 respondents who were seropositive for vaccination or who reported being vaccinated. The final analytical sample included 10,820 respondents of whom 207 had tested positive for antibodies indicating a past COVID-19 infection.

Measures

Seropositive for past COVID-19 infection

Completed DBS tests were sent to two reference laboratories that used a 384-wells format chemiluminescent enzyme-linked immunosorbent assay (ELISA) to test the DBS samples for detection of IgG antibodies against three antigens: the trimer of SARS-CoV-2 protein spike (S), the receptor binding domain (RBD) of spike and nucleocapsid (N). Laboratory staff followed standard operating procedures and uniform assay protocols to ensure similar results and consistent performance. Details about protocols and procedures can be found in the CCAHS user guide.^Note 17

A respondent whose DBS result was positive for two or more antigen tests and who did not report receiving a COVID-19 vaccine was deemed as seropositive for antibodies against SARS-CoV-2 because of a past COVID-19 infection.

Covariates

Respondents were asked sociodemographic and COVID-19 health-related questions in the EQ. Sex at birth and gender were equal for 99.5% of the analytical sample, and 100% of those who were seropositive. This study examined gender (male, female), age groups (1 to 19, 20 to 59 and 60 or older), highest level of household education (postsecondary school graduate versus lower level) household size (one, two, three or four or more people), currently smoke tobacco (no, yes) and number of chronic conditions (none, one or two or more). Visible minority status was assigned if the respondent reported a non-white cultural or racial background.

Adjusted household income quintiles were derived by using a modified version of the equivalence score method where total household income is adjusted by a weight factor based on the number of people in the household.^Note 18

Multi-unit dwelling type included duplex, low-rise apartment (fewer than five storeys) or high-rise apartment (five storeys or more) versus single detached, double, row or terrace.

Respondents who reported working in direct contact with others (not virtually or over the phone) were asked their occupation and were categorized as being in health care if their first level National Occupational Classification (NOC) 2016 code was “3.”  If the NOC 2016 was not “3,” they were categorized as being in an occupation other than health care. Respondents younger than 15, respondents who did not work in direct contact with others or respondents who did not work were all combined into the same work status category.

Respondents were categorized as having a positive nose or throat swab test result if they reported receiving a positive result versus having no positive test result if they reported either a negative result or that they never had a nose or throat swab test.

Respondents were categorized as not reporting COVID-19 symptoms in the past six months if they answered “no” to each of the following: cough, fever, chills, sore throat, shortness of breath or difficulty breathing, pain, runny nose, nausea, vomiting or diarrhea, headache, general weakness or new loss of taste or smell.

Analytical techniques

Proportions, differences and 95% confidence intervals (CIs) were estimated to examine the characteristics of people with a past COVID-19 infection versus those without. Multivariable logistic regression models with listwise deletion (n=429 of 10,820 were excluded) were used to estimate the association between population characteristics (gender, age group, visible minority status, education, income, household size, dwelling type and work status) and a past COVID-19 infection. Adjusted seroprevalence estimates, differences, ratios and their 95% CIs were estimated from the full model by calculating average marginal effects, adjusting for all covariates. Additional models—each with an interaction term between gender and one covariate, adjusting for all remaining covariates—assessed whether the association between a specific characteristic and the outcome differed by gender. Supplementary analyses examined whether smoking or the number of chronic conditions played a role in gender differences.

All estimates were weighted by using the survey weight, and the sampling variance was calculated by using the bootstrap weights. Statistical significance of t-tests and model parameters was assessed at two levels: p < 0.05 and 0.05 ≤ p < 0.10. All analyses were conducted in SAS 9.4 and SAS-callable SUDAAN 11.0.3.

Results

People in private households who were seropositive for a past COVID-19 infection were similar to those who were seronegative according to education, income, type of dwelling, smoking and number of chronic conditions (Table 1). However, a lower proportion of those with a past infection were aged 60 or older compared with those who had not been infected (14% versus 27%) or lived alone (8% versus 13%), and a higher proportion were in a visible minority group (35% versus 20%) or in a household of four or more people (52% versus 42%). Moreover, a higher proportion of those with a past infection reported experiencing symptoms in the previous six months (61% versus 35%) or a positive nose or throat swab test result (54% versus 0.4%). Figure 2 shows that among people who were seropositive, 76% of those without symptoms and 27% of those with symptoms reported no previous positive swab test result. Among those who were seronegative, almost 100% reported no positive swab test result, regardless of the presence of symptoms.

Table 2 presents unadjusted (column A) and adjusted seroprevalence estimates, differences and ratios (columns B to D). After controlling for gender, visible minority status, education, income, household size, dwelling type and work status, people younger than 60 had a higher average adjusted seroprevalence of a past COVID-19 infection. The adjusted prevalence ratios (column D) indicate that people younger than 20 were 1.9 times (95% CI: 0.9, 4.1) and those between 20 and 59 were 1.7 times (95% CI: 1.0, 3.0) more likely to have been infected than people aged 60 or older. Moreover, people who reported being in a visible minority group were 1.7 times (95% CI: 1.1, 2.5) more likely to have been infected with COVID-19 than those who were not in a visible minority group, regardless of other characteristics.

Table 3 presents the effect of gender on the observed associations. After covariate adjustment, the higher average adjusted seroprevalence of a past infection among those who reported being in a visible minority group was predominantly among males and not females. Visible minority males were 2.5 times more likely (95% CI: 1.4, 4.2) to have been infected than non–visible minority males, but there was little difference for females. Males who lived in a multi-unit dwelling were twice as likely (95% CI: 1.2, 3.6) as males who did not live in a multi-unit dwelling to have been infected, but there was no association between dwelling type and a past infection for females. The results also suggest that males with less than a postsecondary school graduation may have been twice as likely (95% CI: 1.0, 3.9) as males with more education to have been infected and that females who worked in a health occupation in direct contact with others may have been 2.5 times more likely (95% CI: 1.0, 6.2) to have been infected compared with females who did not work in direct contact with others.

Discussion

The CCAHS reported that about 2.6% of Canadians living in private households between November 2020 and April 2021 had SARS-CoV-2 antibodies indicating a past infection in their blood.^Note 19 The present analysis found that younger age and visible minority status were associated with an increased likelihood of a past infection regardless of other characteristics. It also found that gender modified the associations between being part of a visible minority group, living in a multi-unit dwelling, having less than a postsecondary education or working in a health occupation in direct contact with others and the likelihood of a past infection, after covariate adjustment.

The finding of increased infection risk among younger people is in line with results from other studies. While susceptibility to COVID-19 is greater for older people, younger people have increased exposure through school, working or caring for others.^{Note 7Note 20}

A higher prevalence of COVID-19 has been observed among people in racialized groups in Canada and in other countries such as the United States and the United Kingdom.^{Note 2Note 21Note 22Note 23} The present study found an increased likelihood of a past infection among those who reported being in a visible minority, but mainly for males and not females. Gender differences in health behaviours and health status may partly explain this finding. A meta-analysis of 57 studies predominantly from China found a higher prevalence of smoking and comorbidities among males, and that it was associated with increased prevalence of COVID-19.^Note 24 In supplementary analyses, the present study also found that a higher proportion of visible minority males smoked (8%) and reported at least one chronic condition (22%) compared with visible minority females (1.5% and 15% respectively).

Interaction in common areas such as elevators, stairways and laundry rooms in multi-unit residential buildings represents opportunity for infection transmission.^Note 25 The present study found an increased likelihood of a past infection for males, but not for females, who lived in a multi-unit dwelling. The multi-unit dwellings of males may have differed in some way that increased the risk of exposure (e.g., number of elevators, number of floors, number of shared spaces), or there may have been other underlying differences. For example, a higher proportion of males than females in multi-unit dwellings smoked (13% versus 8%), a behaviour that has been linked to higher COVID-19 prevalence among males.^Note 24 The present study also found that males who had not graduated from postsecondary school were more likely to have been infected, but not females. One study based on repeated cross-sectional surveys found that a lower education level for males, but not females, was associated with less factual knowledge about COVID-19. Less factual knowledge was then associated with reduced adherence to some protective behaviours,^Note 26 which in turn could increase the risk of exposure to COVID-19.^Note 27

Increased infection risk among health care workers has been reported worldwide.^{Note 1Note 28} The present study found that females, but not males, working in health care in direct contact with others had a higher likelihood of past infection compared with those who did not work in direct contact with others. This gender-specific finding may have been related to occupation type. In the present study, the most common occupation among females working in health care was registered nurse (29%) or nurse’s aide, orderly or patient services (16%), roles associated with inpatient settings and higher COVID-19 exposure risk.^Note 29 On the other hand, the most common occupation for males in health care in direct contact with others was pharmacist (15%) or general practitioner or family physician (14%).

This study also found that four out of 10 people who were seropositive because of a past infection reported not experiencing any symptoms in the previous six months. Furthermore, three-quarters of these people reported they had not previously received a positive nose or throat swab test result. Together, these findings speak to the issue of asymptomatic COVID-19 infection and the challenge of preventing transmission when people may be unaware of carrying the virus.

The COVID-19 landscape continues to evolve in Canada. COVID-19 vaccines are now widely available, and to date about 85% of Canadians aged 5 or older are fully vaccinated.^Note 16 Nevertheless, certain groups who were at greater risk of infection according to the CCAHS may continue to be at greater risk owing to emerging variants of concern, vaccine hesitancy or lower rates of vaccination.^{Note 16Note 30}

Strengths and limitations

A key strength of this study is that it is based on a nationally representative survey of Canadians—past COVID-19 infection status was ascertained through a combination of questionnaire data and a blood test. The main limitation of this study is its small number of seropositive respondents. It resulted in less statistical power and wider CIs around estimates, and prevented further disaggregation of results (e.g., by occupation type, by visible minority group). Also, the CCAHS recommends that the additional variance component originating from the sensitivity and specificity of the laboratory tests be incorporated into estimates of seroprevalence.^Note 17 However, the suggested method could not be followed because it does not pertain to model-adjusted seroprevalence. Therefore, even though the sensitivity and specificity of the tests used in CCAHS were estimated at 100%,^Note 17 the results presented here may not reflect the total variance associated with estimates of seroprevalence. Moreover, despite adjustments and calibrations reflected in the survey weights, there is still a risk of non-response bias in the survey estimates owing to the high degree of non-response to the CCAHS.

Conclusion

Being of younger ages, being male in a visible minority group or with less than a postsecondary education or living in a multi-unit dwelling and being female in a health occupation in direct contact with others were characteristics associated with an increased likelihood of a past COVID-19 infection in fall 2020 to winter 2021 in Canada. As the country navigates the fifth and possibly a sixth wave of the pandemic, understanding who was more likely to be infected in earlier waves can help ongoing public health efforts to stop the transmission of COVID-19.