Introduction
Acute gastrointestinal illness (AGI) has a substantial public health burden in Canada. An estimated 4 million Canadians experience AGI caused by enteric pathogens each year [Reference Thomas1]. Roughly 1.6 million of these cases are caused by four pathogens in Canada: norovirus, followed by Clostridium perfringens, Campylobacter spp., and nontyphoidal Salmonella spp. [Reference Thomas and Murray2]. Previous studies have highlighted the burden of AGI on the Canadian healthcare system. From 2000–2010, there were approximately 4,000 hospitalizations and 105 deaths across Canada associated with AGI caused by a group of 30 foodborne pathogens relevant to the Canadian context [Reference Thomas3]. Furthermore, there were an additional 7,600 hospitalizations and 133 deaths from unspecified agents each year [Reference Thomas3].
Prior work has identified several consistent risk factors for AGI. Children and older adults are at an increased risk due to immature and declining immune status, respectively [Reference McCabe-Sellers and Beattie4]. Similarly, living with one or more chronic medical conditions is a risk factor for developing AGI due to their direct or indirect impact on the immune system [Reference Barkley5]. A relationship between body mass index (BMI) and AGI has also been established in the literature with underweight and overweight or obese people being particularly susceptible [Reference Dobner and Kaser6, Reference De Heredia, Gómez-Martínez and Marcos7]. Consuming fresh fruits and vegetables has also been linked to outbreaks of AGI [Reference Bennett8]. Lastly, men who have sex with men (MSM) have been identified as a potentially high-risk group [Reference Simms and Gilbart9]. It has been reported that the increased risk of AGI observed in MSM is due to the specific sexual practices (i.e., anilingus) [Reference Simms and Gilbart9] but may also reflect different relationships of sexual minorities to the health care system [Reference Hatzenbuehler10].
Other risk factors that have been given less attention in the literature include education, income, and race/ethnicity [Reference Van Cauteren11, Reference Quinlan12]. Potential risk factors of interest may extend to living arrangements, including living alone and living with children, immigrant status, self-perceived health, and access to a regular healthcare provider. These factors have not yet been explored in the Canadian context. Therefore, this study aims to evaluate the effects of established and novel factors on the risk of incurring an ED visit or hospitalization for AGI through the use of linkage of Canada’s national population-based health survey (the Canadian Community Health Survey, CCHS) to records of hospital visits due to AGI to enable longitudinal analyses.
Methodology
Study design
A retrospective cohort study was conducted to examine associations between severe AGI and a range of factors queried in a Canadian population-based survey. Survey respondents formed a cohort followed for health outcomes via data linkage. Statistics Canada administered Canadian Community Health Survey (CCHS) and maintains linkages to Canadian Institute of Health Information (CIHI) records of hospitalization from the Discharge Abstract Database (DAD) and records of ED visits from the National Ambulatory Care Reporting System (NACRS) [13]. The CCHS is a national survey that collects information about respondents’ sociodemographic, health status, and behavioural characteristics [13]. The DAD is a database that captures inpatient hospitalization records across Canada excluding those in Quebec [14]. The NACRS captures ED visit records from hospitals across Canada [15], with reporting at a level where cause of ED visit is discernible varying throughout geography and time. Respondents were followed from their CCHS interview date to 31 December 2017 to assess for the incidence of an ED visit or inpatient hospitalization due to AGI.
Data linkage
Statistics Canada probabilistically linked the records of CCHS respondents who agreed to the linkage (84.6% of respondents) to a number of other administrative records using birthdate, sex, and postal code information [Reference Wallar and Rosella16]. During the linkage process, each CCHS record was assigned a unique merging key that allows researchers to identify individuals in these databases with multiple interactions in the same dataset, across datasets and within a fiscal year and across fiscal years [17]. Linked CCHS data were made available in Statistics Canada Research Data Centres in November 2017 [Reference Wallar and Rosella16]. For this study, the data were accessed through the Toronto Research Data Centre. Before outputs from analyses are released from the secure RDC environment to researchers, they are vetted by Statistics Canada analysts to ensure compliance with disclosure and privacy restrictions.
Cohort assembly
Eleven cycles of the CCHS were pooled together for this study to increase the sample size of the cohort (Figure 1). CCHS Cycles 1.1, 2016, and 2017 were excluded as they were missing key variables or had significant content changes. Respondents were excluded if they were pregnant, less than 18 years, or resided in Quebec. The decision to exclude respondents who were pregnant or less than 18 years was due to their behavioural risk factors likely being significantly different compared to adults who are not pregnant (i.e., alcohol consumption and smoking). Respondents residing in Quebec were excluded from this study, as records from this province are not captured by the DAD. A separate cohort of these CCHS respondents residing in Ontario was constructed to be linked to the NACRS. This separate cohort was limited to Ontario, due to the detailed diagnosis code reporting in Ontario ED records. Several cycles of the NACRS and DAD (2003–2017) were pooled together to identify CCHS respondents from the cohorts with ED visits or inpatient hospital admissions due to AGI. Only the first ED visit or hospitalization for each respondent was selected to be linked to the cohorts.
Figure 1. Assembly of analytic cohorts from Canadian Community Health Survey (CCHS) cycles 2003–2015. AGI: Acute Gastrointestinal Illness. CCHS: Canadian Community Health Survey. DAD: Discharge Abstract Database. ED: emergency department. NACRS: National Ambulatory Care Reporting System.
Table 1. Description of datasets used in study linking Canadian health survey data to health records for cohort study of acute gastrointestinal illness
Note: AGI, acute gastrointestinal illness; ED, emergency department.
Variable construction
The outcome variable for this study was an ED visit or inpatient hospitalization due to AGI. Cases were ascertained via AGI-relevant International Statistical Classification of Diseases and Related Health Problems (ICD)-9-CA (Supplementary Material 1) or ICD-10-CA codes (Supplementary Material 2). As an ED visit or inpatient hospitalization can have multiple diagnoses associated with the encounter, an encounter was considered to be a case if any of the 16 diagnosis codes attached to the record included the selected ICD-9/10-CA codes.
Explanatory variables selected from the CCHS included those identified by prior literature and novel potential risk factors that were identified following comprehensive deliberation by the research team, including sex, age, race/ethnicity, immigrant status, education level, income, living arrangements (living alone, living with children), and sexuality (MSM, non-MSM). Health status variables were BMI, cigarette smoking, alcohol consumption, self-perceived health, the presence of chronic conditions, access to a regular healthcare provider, and fruit and vegetable consumption. The organization of these variables and their construction for analyses is described in Supplementary Material 3.
Analytic strategy
Descriptive, unweighted proportions and frequency counts (rounded to the nearest 100 for disclosure compliance) of explanatory variables by outcome status were tabulated and summarized. Given the cohort study design, two series of time-to-event survival analyses were applied using Cox proportional hazards regression to estimate the association between explanatory variables and the outcomes of an ED visit or inpatient hospitalization due to AGI. The measures of association are hazard ratios (HRs) with 95% confidence intervals (CIs). In addition to evaluating the effects of the pre-selected factors, we aimed to assess how these estimates changed when adjusted for other factors. To achieve this, we constructed three models for each explanatory variable. The first series of models were the unadjusted, univariable models. The second series of models adjusted for age and sex. The third model was a multivariable model that adjusted for all other explanatory variables in the study. Statistical analysis was performed using the statistical software R 4.12, implemented using RStudio version 1.3.1073 [18]. Supplementary Material 4 describes the R packages used to conduct the analyses.
Ethics review
This project was deemed exempt from project specific ethics approval by the Research Ethics Board (REB) at Toronto Metropolitan University (REB 2023–454) given its exclusive use of secondary linked data per Article 2.2 of the Tri-Council Policy Statement 2 [19].
Results
Pooling of the CCHS cycles yielded a raw cohort of 673,900 respondents. Applying the study’s exclusion criteria yielded 470,700 respondents to be linked to the DAD, forming the hospitalization cohort (Figure 1). Respondents were then excluded from this cohort if they did not reside in Ontario, yielding 190,700 respondents to be linked to the NACRS to form the ED visit cohort (Figure 1). Six per cent of respondents from the ED visit cohort (n = 10,800) incurred an ED visit due to AGI, and 2% of the hospitalization cohort (n = 9,400) incurred a hospitalization due to AGI (Figure 1).
Table 2 summarizes the sociodemographic and health status characteristics of the ED visit and Hospitalization cohorts. In the ED visit cohort, most cases were female (64.8%) and aged 25–64 years (54.6%). In addition, the majority were White (88.9%) and born in Canada (82.4%). The sociodemographic make-up of the hospitalization cohort cases was very similar (Table 2), with most being female (64.9%), White (90.4%), and born in Canada (87.2%). However, the hospitalization cohort cases tended to be older with 52.1% being seniors (65+ years), compared to 33.3% in the ED visit cohort.
Table 2. Demographic and health status characteristics of the linked cohorts
*Suppressed due to low cell counts.
Table 3 presents the results of the series of Cox proportional hazards regression models constructed for the linked cohorts. Key unadjusted risk factors for AGI-related ED visits included being adults (18–24 years), having less than secondary school education, living alone, and being a cigarette smoker. Conversely, a decreased risk was observed among visible minorities, immigrants to Canada, MSM, people living with children, and people who consume any amount of alcohol. Adjusting for age and sex (Model Series 2) resulted in a decreased risk in people compliant with CFG recommendations. In the multivariable model (3), the effects of several variables were attenuated: being in the third income quintile, being a visible minority, living alone, being a low-risk alcohol drinker, and complying with CFG.
Table 3. Cox proportional hazards modelling results estimating unadjusted (Model 1), minimally adjusted (Model 2) and fully adjusted (Model 3) associations of hospitalization and ED visits for acute gastrointestinal illness in cohorts linking Canadian national survey data to health care records
a Model Series 1: unadjusted.
b Model Series 2: adjusted for age and sex.
c Model 3: multivariable model including all variables.
d These regressions were run excluding the binary sex variable (male, female).
In Model Series 1 for the hospitalization cohort, key unadjusted risk factors for hospitalization due to AGI were having less than secondary school education, being an MSM, and living alone. Key inverse associations were observed between young adults, visible minorities, immigrants to Canada, and people who consume any amount of alcohol. Adjusting for age and sex (Model Series 2) attenuated the effect of being in the second income quintile and being a visible minority. In the multivariable model (3), an increased risk was observed among those considered a visible minority, and the direction of association shifted to a reduced risk among people in the first three income quintiles and with secondary school education.
Discussion
To our knowledge, this is the first study in Canada to use the CCHS linked to the NACRS and DAD to explore risk factors associated with AGI-related ED visits and hospitalizations. It is also one of the first to comprehensively describe individual risk factors of requiring AGI-related hospital care in Canada in a longitudinal cohort. An estimated 1 in 8 Canadians (12.5%) will be affected by AGI related to foodborne causes each year [Reference Thomas1]. The rates of AGI-related ED visits and hospitalizations found in this study (6% and 2%, respectively) are lower than this; however, this is to be expected, as cases of AGI are largely under-reported to healthcare and public health agencies [Reference Flint20]. Moreover, research has demonstrated that only 12–20% of AGI cases seek healthcare [Reference Schmidt21]. Out of everyone who seeks healthcare, the proportion seeking hospital-level care is likely even lower than the reported 12–20% as most cases can be managed in the outpatient setting. With this in mind, the true incidence of AGI cases across Canada is likely significantly higher than what we are reporting in this study.
This study identified novel risk factors and inversely associated factors, as well as reproduced similar findings for established risk factors. Novel risk factors which have not yet been thoroughly examined in the literature included living alone, low education level, cigarette smoking, and poor self-perceived health. Novel inverse associations were living with children, being a racial minority and/or immigrant to Canada, not having a regular healthcare provider, and being compliant with dietary guidelines for the daily consumption of fruits and vegetables. Identified risk factors which are consistent with previous research include female sex, old age, low income, and the presence of chronic conditions (asthma, arthritis, bowel disease, cancer, diabetes, or heart disease). Four main factors appear to influence the associations identified in this study: structural inequities, physical health, access to healthcare, and individual behaviours.
Structural inequities
The effects of household income and race/ethnicity are not attributable to innate biological factors [Reference Weller22]. Rather, they are likely due to complex structural inequities which could not be captured in this study [Reference Weller22]. Specifically, the effect of income may be influenced by socioeconomic marginalization, whereas the effect of race/ethnicity is likely influenced by racial biases at the institutional level. In this study, household income was associated with both outcomes. Specifically, those in the lowest income quintiles had an increased risk of incurring both ED visits and hospitalizations due to AGI. This finding may be attributable to the types of food that those with lower incomes have access to [Reference Weller22]. People with low income often live in food deserts and rely primarily on smaller grocers, convenience and fast-food retailers (compared to super markets) for their food supply [Reference Gordon23]. The food sold by the retailers in food deserts tend to be of lower quality and have been found to be more likely to harbour foodborne pathogens compared to retailers outside of food deserts, which may have the potential to increase the risk of illness in these populations [Reference Gordon23]. However, in the multivariable model series, we noted that these income quintiles were inversely associated with requiring hospital-based care for AGI. This may suggest that the effects of income are confounded by or related to other variables in the model, and thus, the potential mechanism of the increased risk in the unadjusted and age- and sex-adjusted models may not be entirely relevant.
A reduced risk of requiring an ED visit or hospitalization was observed among visible minorities, i.e., racialized people, except for when adjusting for all variables, after which case they had an increased risk of requiring a hospitalization. The relationship between race and ethnicity and AGI is complex and not well described in the literature, particularly in the Canadian context. There is some evidence to suggest that the relationship is pathogen dependent, with visible minorities being disproportionately impacted by certain pathogens and not others [13]. Differences in health care access may contribute to the relationship between race/ethnicity and AGI [Reference Weller22]. Previous research from the United States reported that racialized patients were less likely to be admitted to hospital for their presenting illnesses or injuries compared to White patients [Reference Wilson24]. Thus, racial minorities may appear to have reduced risk of AGI-related hospitalizations due to racial biases, which influence clinicians’ willingness to admit them. As for AGI-related ED visits, disparities in how racial minorities are treated in this setting may affect their decision to seek this type of care [Reference Schrader and Lewis25]. This may result in a reduced incidence and risk of requiring AGI-related ED care. However, given complexities and nuances of structural inequities, we believe our work has set the stage for further research aimed at exploring this association more thoroughly.
Physical health
Contrary to income and race, there are a number of associations identified in this study which may be explained by the physical health status of the respondents. These factors are immigrant status, BMI, self-perceived health, fruit and vegetable intake, and old age. Being an immigrant to Canada was inversely associated with ED visits or hospitalization due to AGI. There does not appear to be any literature that explores the effect of immigration status on the risk of developing AGI, thus this is one of the first studies to examine the relationship. While immigrant status is not a physical health characteristic, its inverse association with the incidence of AGI-related hospital care could be related to the healthy immigrant effect, which refers to the often-better health status of immigrants compared to those born in Canada [18]. Therefore, it is possible that better health among immigrants may result in less severe AGI, thus preventing the need for hospital-based care. It may also be possible that the reduced risk among immigrants is related to their care seeking behaviours. Specifically, research by Wu Penning, & Schimmele [Reference Wu, Penning and Schimmele26] found that immigrants were less likely to report unmet healthcare needs compared to non-immigrants [Reference Wu, Penning and Schimmele26]. Thus, it is possible that immigrants to Canada are equally impacted by AGI but are not seeking hospital-based care for it.
Respondents who were normal weight had a decreased risk of requiring an ED visit or hospitalization due to AGI compared to those who were overweight or obese. The risk associated with being underweight was variable and not statistically significant. This reduced risk in normal-weight respondents compared to overweight and obese respondents was not surprising, as obesity is a known risk factor for AGI. A possible explanation for this may be the chronic low-grade inflammation observed in some people with obesity, which increases their susceptibility to more severe illness requiring hospital-level care [Reference De Heredia, Gómez-Martínez and Marcos7, Reference Bennett8].
Having poor self-perceived health was associated with an increased risk of requiring both ED visits and hospitalizations for AGI. This is in line with one previous study that evaluated the effect of perceived health on hospitalization for AGI and found the same increased risk [Reference Chen27]. As both the indicators of chronic conditions and self-rated health were associated with risk, our finding is consistent with prior work showing those with underlying health challenges are at higher risk for severe outcomes with AGI [Reference Dobner and Kaser6].
Consuming the recommended daily intake of fruits and vegetables was associated with a decreased risk of requiring an ED visit or hospitalization due to AGI. Though the consumption of raw fruits and vegetables has been linked to a number of outbreaks [Reference Simms and Gilbart9], it is also associated with better physical health and reduced incidence of chronic disease [Reference Slavin and Lloyd28]. Therefore, the health status of people who consume a sufficient quantity of fruits and vegetables per day likely reduces their risk of experiencing severe illness.
Older adults (65+ years) were found to have an increased risk of requiring both ED visits and hospitalizations due to AGI. Older adults are a well-established high-risk group for AGI due to age-related decline in their immune systems [Reference Barkley5], and as a result, they are susceptible to more severe illness and increased morbidity.
Healthcare access
Interestingly, not having access to a regular healthcare provider was associated with a decreased risk of requiring hospital care due to AGI. This is one of the first studies reporting a relationship between regular healthcare access and the outcome of hospital-based care due to AGI. Despite this, it is important to note that the state of primary care in Canada was vastly different during this study’s follow-up period compared to the present day. Specifically, most respondents in this study had a regular healthcare provider regardless of case status. This is in contrast to recent federal data from 2023, which reports that 17% of Canadians 18 years of age and older do not have a regular healthcare provider [29]. In light of this information, the reduced risk associated with a lack of a regular healthcare provider is likely due to differences in how many Canadians had regular access to care between the study follow-up period and the present.
Behaviours
Respondents’ behaviours, particularly as they relate to health, healthcare seeking, and food handling may explain the associations observed between educational level, living arrangements, sex, young age, sexuality, cigarette smoking, and alcohol consumption and the incidence of AGI-related hospital care. Those with less than a secondary school education had an increased risk of requiring a hospital visit compared to those with post-secondary education. Educational level may have an influence on one’s health behaviours, beliefs, and awareness of issues related to their health, which are considered predictors of healthcare system usage [Reference Tam, Rodrigues and O’Brien30]. People with post-secondary education may have enhanced health knowledge, which allows them to better assess the severity of their illness or be more aware of how to manage it [Reference Tam, Rodrigues and O’Brien30]. Therefore, it is possible that people with less than a secondary school education may not have the same health knowledge necessary to manage their AGI. As a result, people with lower levels of education may be more reliant on hospital-based care. The level of education that one has achieved may also have an effect on their food safety knowledge and handling behaviours [Reference Ruby31], which would affect their ability to safely prepare and consume food. However, additional research is needed to test this hypothesis, as direct information on food safety knowledge and behaviours was not available in the linked databases used in this study.
Living alone was associated with an increased risk of requiring an ED visit or hospitalization for AGI, whereas living with children was associated with a decreased risk. One study from France also reported living alone as a risk factor of developing AGI [Reference Ecollan32]. This finding may relate to health-seeking behaviours – people who live alone may be less likely to seek medical care for new onset AGI, resulting in a delay in care and potential increase in illness severity. Conversely, the effect of living with children might at least partially be explained by increased vigilance around food safety. Studies from the United States have indicated that parents of children are often highly concerned about protecting them from AGI and take additional steps to ensure their food is prepared safely [Reference Charlesworth, Mullan and Moran33], reducing risk for all household members.
This study found a reduced risk of requiring an ED visit or inpatient hospitalization due to AGI among males compared to females. This is consistent with a number of other studies which have reported that the incidence and risk of AGI is increased in females [Reference Hall34, Reference Majowicz35]. This reduction of risk is likely due to gender related differences such as eating patterns and health behaviors [Reference Bertakis36], as opposed to biological sex differences that would decrease males’ susceptibility.
With respect to age, we found that adults 18–24 years had an increased risk of requiring an ED visit due to AGI compared to adults ages 25–64. Previous studies have found that young adults tend to have poor food safety knowledge and perceive their risk of acquiring AGI to be low [Reference Lazou37, Reference Sanlier and Konaklioglu38]. This lack of knowledge may partially explain the increased risks of severe AGI noted in this age group in this study. Conversely, we found that young adults had a decreased risk of requiring a hospitalization due to AGI compared to those 25–64 years. This is not surprising, as this age group tends to be healthier overall compared to adults and seniors.
An increased risk of requiring hospitalizations for AGI was observed among MSM. However, a decreased risk of requiring an ED visit due to AGI was observed among this group. The increased risk of requiring hospitalization due to AGI is well supported by a number of studies which have reported a higher incidence and risk of contracting AGI in MSM [Reference Van Cauteren11, Reference Marchand-Senécal39]. This finding may be attributable to specific sexual practices (i.e., anilingus) [Reference Simms and Gilbart9]; however, MSM are not the only group of people to engage in these behaviours. With respect to ED visits, the decreased risk may be explained by the complex relationship that sexual minorities like MSM have with the healthcare system [Reference Hatzenbuehler10]. Specifically, MSM and other sexual minorities often face barriers to accessing the healthcare system [Reference Hatzenbuehler10]. Thus, it is possible that those who are affected by AGI are less likely or able to access ED care for their symptoms. In any case, applying the finding of increased risk in community settings requires nuance and consideration of ongoing stigma for MSM and other sexual minorities [Reference Hatzenbuehler10]. Given the mechanism is not fully elucidated, further work on risk factors within this community (e.g., food safety practices) could be warranted.
Cigarette smoking was associated with an increased risk of requiring ED visits and hospitalizations for AGI. This finding is consistent with two other studies that reported cigarette smoking or exposure to tobacco in the environment as predictors for the development of AGI [Reference Kim40, Reference Kum-Nji41]. Cigarette smoking damages the gastrointestinal tract by increasing mucosal permeability, negatively altering the microbiome population, and impairing the mucosal immune responses [Reference Gui, Yang and Li42]. As a result, cigarette smoking may increase the risk of contracting AGIs, as the immune system might not be able to respond to enteric pathogens as effectively.
Alcohol consumption of any kind was associated with a decreased risk of requiring ED visits or hospitalizations due to AGI. Though there is not much research on the effect of alcohol consumption on the occurrence of AGI, a study by Kim et al. [Reference Kim40] found that the prevalence of AGI remained high in heavy alcohol consumers, despite the overall prevalence decreasing in other groups over their study period [Reference Kim40]. In this study, it is difficult to ascertain the true relationship between alcohol consumption and the outcome of AGI-related hospital care. This is largely due to the fact that one cannot determine the CCHS respondents’ dependency on alcohol, or how consuming it affects their consumption of food. Therefore, there are likely other factors that are contributing to the reduced risk of requiring hospital care for AGI in people who consume alcohol.
Limitations
Key limitations of this study include challenges to case ascertainment, target population database coverage, and analytic assumptions. Firstly, the cohort only captured cases of AGI for which CCHS respondents sought care at a hospital, which represents the most severe cases. Therefore, the findings of this study may not be representative of milder, self-limiting cases of AGI that do not require an ED visit or hospitalization. Another limitation of this study is that only unweighted statistics were reported, therefore these findings may not be entirely representative of the entire Canadian population. However, as this was a cohort study, we were more interested in the characteristics of the baseline cohorts as opposed to estimating their effects at the national level. As well, given that the CCHS is cross-sectional and voluntary, biases may be introduced. The CCHS overall has a high participation, with a response rate as high as 85.1% [43]. By limiting our analytic cohort to those respondents eligible for linkage to national health databases (which particularly excludes Québec residents), the generalizability of our findings is limited. The proportional hazards assumption was supported for some, but not all variables assessed in this study (Supplementary Material 5). Therefore, the HRs produced should be interpreted as the average risk over time [Reference Stensrud and Hernán44]. Our approach to model construction may mean there are inter-relationships between a range of variables. Therefore, any associations we found should be examined in context with minimally and unadjusted models, and we hope to see future work on which sociodemographic indicators have the most relevant covarying relationships with AGI outcomes. CIHI databases used in this study do not cover Quebec, so respondents from this province were not captured. Lastly, the CCHS was not linked to the Canadian Mortality Database in this study, so there is the potential that some of the non-cases may have died before the end date of 31 December 2017. As a result, the person-time they contributed to the study may have been overestimated and there is a possibility that the hazard rate has been biased downward, resulting in an underestimation of the true estimates produced in this study.
Conclusion
This retrospective cohort study described the associations between sociodemographic and health status characteristics and the incidence of emergency department visits or inpatient hospitalizations due to AGI in CCHS respondents from 2003 to 2017. This study identified novel sociodemographic factors associated with the occurrence of hospital-based care for AGI. These findings highlight the need for further research with the high-risk groups identified in this study to ascertain their food safety knowledge, behaviours, and specific exposures related to their illness risks. The present study and future work can be used to help inform future educational and outreach needs with the identified vulnerable groups.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0950268825100666.
Data availability statement
The datasets used in this study are under the custodianship of Statistics Canada and can be accessed via the Canadian Research Data Centre Network (CRDCN) for eligible researchers, including security clearance. See this link for eligibility and process to request access: https://www.statcan.gc.ca/eng/rdc/index).
Author contribution
Conceptualization: I.Y., M.A.H.; Methodology: I.Y., M.A.H., L.G., A.J.G.; Data curation and analyses: A.J.G.; Manuscript drafting: A.J.G.; Review & editing: I.Y., M.A.H., L.G., A.J.G.
Funding statement
This work was supported by the Ontario Graduate Scholarship (A.J.G., 2023–2024, 2024–2025).
Competing interests
The authors declare none.
Open access
