Study Design
This population-based cross-sectional study was carried out at the Institute for Clinical Evaluative Sciences (ICES) using multiple, linked health and administrative databases (Fig. 1) through a research agreement with the Ontario Ministry of Health and Long-Term Care. Research ethics board approval was obtained from The Hospital for Sick Children and Sunnybrook Health Sciences Centre in Toronto, Ontario.
Dataset Sources
The Ontario Health Insurance Plan (OHIP) is the single-payer, universal funding plan for medically necessary physician and hospital services in Ontario. Permanent residents admitted to Canada as immigrants or refugees are typically eligible for OHIP within 3 months of living in Ontario. OHIP eligible persons are entered into Ontario’s health care registry, the Registered Persons Database (RPDB), which contains socio-demographic information about residents including their age, sex, and postal code. Their OHIP number is encoded and linked to a number of other health administrative and demographic databases. The Canadian Institute for Health Information (CIHI) Discharge Abstract Database (DAD) and National Ambulatory Care Reporting System (NACRS) provide standard data collection and reporting tools to capture hospitalization and emergency department visit information, respectively, and include the main diagnosis. Linkage rates to the RPDB are greater than 97% [21]. These were used to obtain injury-related information where a hospital or ED visit occurred. The Ontario Registrar General – Death (Vital Stats) was used to identify in and out of hospital injury-related deaths in Ontario. Neighbourhood income quintile was obtained using Statistics Canada’s Postal Code Conversion File to link a patient’s postal code at the dissemination area level (400 to 700 persons) based on the 2006 Canadian Census [22].
The Permanent Resident Data System, a federal database maintained by Immigration, Refugees and Citizenship Canada (IRCC), holds socio-demographic and immigration information on all permanent residents landing in Ontario from January 1st, 1985 to the present. Permanent residents are immigrants who have been granted to the right to stay and work in Canada without limitations on their stay. It does not include data on temporary or undocumented immigrants. Information is collected from official immigration documents upon landing. The MOMBABY database is an ICES derived database that pairs all mothers with their newborns, delivered in hospital in Ontario through linkage of the CIHI-DAD inpatient admission records of delivering mothers and their newborns [21]. This was used to identify newborns born in Canada to immigrant mothers.
Study Population
The study population included children and youth from birth to 24 years of age who were immigrants, or were born in Canada to immigrant mothers. To be included, individuals must have had an IRCC permanent resident record or be the child of a mother with an IRCC record and have a linked mother-baby dyad in the ICES MOMBABY database. Maternal immigrant status was included to reflect the influence of family immigrant status, rather than that of only the child. Parental health and risk-taking behaviours are particularly important for a child’s risk of injury (e.g. safe sleeping and cooking environments, car seat safety, etc.) Moreover, maternal immigrant status is associated with risk of injury in children [19, 23]. Individuals must have been residing in Ontario with a valid Ontario Health Insurance Plan (OHIP) number between January 1st 2011 and December 31st, 2012.
Outcome Measures
The main outcome measure was an unintentional injury-related visit to an ED or a hospitalization, or an unintentional injury related death (in and out of hospital) during the study window (2011 to 2012). The International Classification of Disease 10-CM External Cause of Injury Codes were used to identify and group injuries by cause [24]. Multiple events by the same patient were included although only one event per patient per day was included. Where individuals had an ED visit with a subsequent hospitalization, only the hospitalization was counted. Duplicate, overlapping, or transferred ED visits were not double counted so that only the first record for any given injury event was considered.
Exposure Variables
The main exposure was duration of residence in Canada. Immigrants were grouped as recent (0 to 5 years in Canada), intermediate-term (6 to 10 years in Canada), and long-term (greater than 10 years in Canada). Duration of residence was determined from the difference of the number of years from the date of landing to December 31st, 2011 (study mid-point). If the patient was born in Canada to an immigrant mother, the duration of residence was calculated based on the maternal immigration information. The secondary exposure was region of origin, based on the country of birth, using modified IRCC region groupings [20].
Covariates
A number of covariates that have been shown to be associated with risk of unintentional injury were considered. Older age (adolescence) is a strong predictor of injury risk, as is male sex [17, 25]. Both age and sex were therefore included as covariates. In a number of populations, low income has been associated with an increased risk of both unintentional and intentional injury [23, 26]. We therefore included postal code based neighbourhood income quintiles within dissemination areas (areas of 400 to 700 individuals) from health records as a covariate.
Statistical Analysis
Descriptive statistics were performed for the main independent variables, outcome variables, and covariates. The crude injury rates were calculated, as were the total number of events. To compare outcomes of immigrants by duration of residence, multiple variable Poisson regression models adjusting for overdispersion were used to compute rate ratios with 95% confidence intervals. A sensitivity analysis excluding children and youth of immigrant mothers was also performed. For each model, variables were selected a priori and included in the regression analysis. The cause-specific rate ratios of injury by duration of residence was also calculated using multiple variable Poisson regression models adjusting for overdispersion. All statistical modelling was carried out using SAS Enterprise Guide, version 6.1 (SAS Institute Inc., Cary, NC).