Health Care Use and System Costs Among Pediatric Refugees in Canada

We conducted a population-based matched cohort study among resettled refugee and Ontario children eligible for provincial health insurance using health administrative datasets linked at ICES (formerly Institute for Clinical Evaluative Sciences), a not-for-profit research institute whose legal status under Ontario’s health information privacy law allows it to collect and analyze health data without consent in Toronto, Canada. Ontario, Canada’s most populous province (population ∼14 million) receives ∼50% of Canada’s refugees each year, of which >98% settle in urban areas.¹³ 

We included all resettled refugee children aged 0 to 17 years who landed in Ontario between April 1, 2008 and March 31, 2017. We did not include undocumented immigrants or refugees from the In-Canada Asylum Program,¹⁴  in which refugee applications are made by those already in Canada. We excluded those without a full year of publicly funded provincial health insurance coverage from the landing date and those without a valid health card number. We further excluded children with a missing census metropolitan area. Resettled refugees were matched 1:4 with Canadian-born children on birth date (+/− 30 days), sex, income quintile, and census metropolitan area. Refugees who could not be matched were excluded from analyses. Matched Ontarians were assigned an index date corresponding to their refugee match’s landing date. For greater context, Supplemental Table 5 shows characteristics of the matched Ontario-born population and the general population of Ontario children and youth.

We used multiple health and demographic databases available and linked at ICES. Individual person-level records were linked by using a unique encoded identifier, derived from the health card number of Ontario residents eligible for provincial health insurance, including resettled refugees. The Ontario portion of the Immigration, Refugees and Citizenship Canada Permanent Resident database contains demographic information on those who landed in Ontario since 1985 with permanent residency status (those granted the right to live, stay, and work in Canada).¹⁵  We used this database to ascertain refugees’ resettlement model, country of citizenship, world region of origin, and landing date. Refugees were also grouped into “pre-Syrian” era (before November 2015) and “Syrian” era (November 2015 onward) time periods that delineated Canada’s large and specific commitments to resettle Syrian refugees.¹⁶  Linkage of Immigration, Refugees and Citizenship Canada data to the Registered Persons Databases, the provincial health insurance registry, has been validated with linkage rates of 86%.¹⁷  The Ontario Health Insurance Plan physician billings database captured outpatient visits to primary care physicians and specialists. We ascertained visits to physicians and nurse practitioners in community health centers (CHC) using the CHC dataset. CHCs are nonprofit, community-based health organizations that provide primary health care, health promotion, and disease prevention services to those facing barriers.¹⁸  The Canadian Institute for Health Information Discharge Abstract Database and Same Day Surgery Databases identified hospitalizations and surgeries, whereas the National Ambulatory Care Reporting System identified emergency department (ED) discharge data. We used the Ontario Marginalization material deprivation index, a census-based measure, to categorize neighborhood-level marginalization closest to the year of landing.¹⁹  Such neighborhood measures of socioeconomic status have been associated with health outcomes and health system utilization.²⁰  Individual-level measures of marginalization and income are not available in linked datasets. The Registered Persons Database was used for demographic characteristics (age, sex, postal code, and residence).

Our main exposure was the refugee resettlement model in 1 of 3 programs: (1) GAR, (2) PSR, or (3) BVOR (a program started in 2013).³  Under all 3 programs, resettled refugees are immediately eligible for provincial health care. GARs are assigned a case worker who helps them navigate health card applications and facilitates access to health care services. Both BVORs and PSRs are resettled by private citizens who take responsibility for this role.

Our secondary exposure was country of citizenship. Country of citizenship categories were chosen because of specific and large resettlement commitments by the federal government during the study period: Afghanistan, Iraq, Iran, Syria, Bhutan, Myanmar, Somalia, Eritrea, Congo, other African countries, and “other” countries.^21–23 

We compared health care utilization among resettled refugees and matched Ontario-born children in the first year after landing. Primary care visits included those to a family physician, nurse practitioner, or pediatrician, and specialist visits included both overall and to pediatric subspecialists. We described ED visits by acuity level (high, low) and hospitalizations by type (medical vs surgical and elective vs nonelective) and facility type (pediatric academic, nonpediatric academic, and community). We enumerated the most common visit diagnoses across all settings using the International Classification of Diseases 10^th Revision chapter and compared them across resettlement categories.

We described health care costs overall, by sector (eg, outpatient, hospitalizations), and by resettlement model as a proxy for health system resource intensity. Outpatient health care costs were based on unit costs of services provided to patients during an episode of care, paid by provincial insurance (Ontario Health Insurance Plan) to eligible health care providers. We calculated hospitalization, same-day surgery, and ED visit costs using case-mix methodology, in which the cost of a patient encounter is based on the intensity of resources used during the episode of care.²⁴  Costs do not include primary care visits within CHCs.

We measured aggregate morbidity at 1 year after arrival in Canada using the Johns Hopkins Adjusted Clinical Group (ACG) System aggregated diagnosis groups (ADGs) case-mix adjustment system. Case-mix assignment is based on several factors, including age, sex, and diagnoses received through outpatient and inpatient care settings, over a 1-year period. This measure predicts an individual’s morbidity and long-term health care needs and has been validated for use with health administrative data in pediatric populations.^25–29  We grouped ADGs into (1) any or 2+ major illnesses (using the ACG System ADGs of time-limited major; chronic medical, unstable; psychosocial, recurrent, persistent or unstable; progressive or likely to recur; or malignancy) and (2) minor (using the ACG System ADGs such as those with minor adverse effects, signs, and symptoms, and less likely to recur).

We calculated frequencies and percentages to describe baseline characteristics of refugee groups and Ontario residents and used standardized differences to compare between groups.³⁰  An imbalance in groups was defined as an absolute value >0.10. For health care utilization and costs, we used P values from a 1-way ANOVA, median (interquartile range) from the Kruskal-Wallis test, and, for categorical values, we used χ² test. Multivariable logistic regression tested the relationship between having at least 1 major ADG and resettlement model among resettled refugees, adjusted for age, sex, and neighborhood level of material deprivation. As a secondary analysis, we compared aggregate morbidity across country cohorts using all resettled refugees as a referent group.

The Research Ethics Board at The Hospital for Sick Children, Toronto, Ontario approved this study.

From April 1, 2008 to March 31, 2017, 23 450 resettled refugee children and youth landed in Ontario. After exclusions, the cohort consisted of 23 287 resettled refugees and 93 148 matched Ontario-born children (Supplemental Fig 3). Among resettled refugees, there were 13 360 (57.4%) GARs, 1544 (6.6%) BVORs refugees, and 8383 (36.0%) PSRs (Table 1). Compared with the general population of Ontario-born children, refugees and the matched Ontario-born cohort had greater proportions of individuals in the lowest income quintile, highest material deprivation quintile, and urban settings (Supplemental Table 5).

Baseline demographic characteristics of resettled refugees and matched Ontario-born children are shown in Table 1. PSRs were older than BVOR refugees, and GARs and had equal proportions of males and females. The proportion of youth living in the lowest income quintile was highest among GARs (57.3%) and lowest among BVOR refugees (43.4%). Almost all (>99%) resettled refugees resided in urban locations, with exception of BVOR refugees, in which a proportion lived in rural areas (12.3%). PSRs were mainly from Iraq (31.6%), Syria (27.6%), and Afghanistan (12.1%), whereas a higher proportion of GARs and BVORs refugees were from Syria (42.2% and 79.8%, respectively). The majority (59.8%) of PSRs landed in the pre-Syrian era⁵  and almost all (93.8%) BVOR refugees arrived during the Syrian era.

Health care use in the first year of landing differed among resettled refugees (Table 2). Primary care visits were high among resettled refugees compared with matched Ontario-born children (median visits [interquartile range], GARs 4 [2–6]; BVORs refugees 3 [2–5]; PSRs 3 [2–5]; Ontario-born 2 [1–4]) with a similar distribution for any outpatient visits to all providers. The proportion of children with an outpatient visit to a physician or nurse practitioner for mental health concerns was similar between resettled refugees (10.5%) and matched Ontario-born children (11.3%) but was lower in PSRs (6.0%) compared with GARs (13.1%) and BVOR refugees (12.1%) (P <.001).

ED visits were more common among GARs and BVOR refugees compared with PSRs (GARs 19.2%; BVORs refugees 23.4%; PSRs 13.8%, P <.001). The proportions hospitalized in the first year between resettled refugees and matched Ontario-born children were similar (2.0% and 1.8%, respectively) with marked differences by groups (GARs 2.5%; BVORs refugees 3.2%; PSRs 1.1%). GARs had the greatest proportion admitted to pediatric academic hospitals compared with PSRs (67.8% vs 43.2%). Few resettled refugees required hospitalization in an ICU (61 patients, 0.3%). Reasons for visiting the ED and hospitalizations were similar between refugee resettlement groups and matched Ontario-born children, with visits most commonly for injuries and respiratory tract illness (Supplemental Tables 6 and 7). Specialist visits for GARs were most often for infectious diseases and neurodevelopmental concerns, both common among PSRs and BVOR refugees. For Ontario-born children, specialist visits were most often for behavioral and mental health concerns (Supplemental Tables 8 and 9).

Mean health system costs (Fig 1, Table 2) were higher among resettled refugees (mean [SD]: $1009 [$6068]) than in matched Ontario-born youth (mean [SD]: $852 [$9227]), and significantly higher among GARs (mean [SD]: 1278 [7475]) versus other resettled refugees (P <.001). Across all groups, outpatient visits were the highest contributor to costs.

The odds of a major illness, as measured by 1 or more major ADG, in the first year since landing was highest among GARs (10.0%), followed by BVOR refugees (9.8%) and PSRs (6.6%) (GARs adjusted odds ratio [aOR] 1.63, 95% confidence interval [CI] 1.47–1.81; BVORs refugees aOR 1.52, 95% CI 1.26–1.84) (Tables 3 and 4). In comparing morbidity by country cohort, children and youth from Syria had greater odds of having at least 1 major illness (aOR 1.36, 95% CI 1.23–1.50), whereas those from Eritrea (aOR 0.58, 95% CI 0.43–0.79) and Myanmar (aOR 0.69, 95% CI 0.50–0.95) had lower odds of having at least 1 a major illness compared with the overall population of resettled refugees (Fig 2, Supplemental Table 9).

In this population-based matched cohort study, we observed that ambulatory care utilization, including primary care, was high across all resettled refugee groups, with the highest use observed among GARs. PSRs consistently used acute care services less than other resettled refugee and Ontario-born peers, whereas BVOR refugees’ use of acute care was highest. Among refugees, PSRs had the lowest 1-year aggregate morbidity and used fewer resources. Compared with all resettled refugees, refugees from Eritrea and Myanmar had the lowest aggregate 1-year morbidity, with the highest morbidity observed among refugees from Syria. Although BVOR refugees and GARs generally had greater morbidity and health system utilization than PSRs, at the current levels of Canadian resettlement, refugee children and youth do not place substantial demands on the health care system above and beyond utilization by their Ontario-born peers, even in the first year of resettlement.

GARs, PSRs, and BVOR refugees have access to health care services and either formal or informal settlement support services after arrival in Canada. How refugees are guided through the health and social systems differs by resettlement model (Table 1). The differences in the orientation to health services after settlement and the subsequent ability of refugees to navigate and understand health care services may contribute to our findings. Compared with standalone clinics often accessed by PSRs, integrated community-based primary care with partnerships between dedicated refugee health clinics, local reception centers, and community providers available to GARs may improve timely access to care and better match referrals with needs and services.³¹  Such care integration may have been a factor in observed high utilization, including for subspecialty care, among GARs. Although observed differences may be, at least in part, related to differences in assistance with health system navigation, our data also indicate that GARs have higher morbidity and more medical complexity than BVOR refugees or PSRs. Others have reported that GARs from Syria have a 2.6 higher odds of unmet health care needs up to 12 months after resettlement compared with PSRs.³²  This implies that the magnitude of quality resources required to meet the needs of GARs may actually be greater than our data suggest. This also highlights that there may be inequitable health service responses for different populations based on need.

Reports of refugee children indicate they have high medical needs, mostly related to communicable diseases, including tuberculosis, hepatitis B, and malaria,^8,10,12  micronutrient deficiencies,^33,34  and dental problems.^11,35,36  Because such illnesses are often manageable in primary care, early engagement is essential.⁹  Our results reveal that almost all resettled refugees have at least 1, and often more, visits to primary care within their first year in Canada, with a larger proportion of GARs seeing subspecialists for infection-related concerns. Connecting refugees to services for noncommunicable disease assessments is also important, especially for children with neurodevelopmental concerns. For all resettled refugee groups, besides infections, neurodevelopmental concerns were one of the top reasons for subspecialty visits. In contrast, for Ontario-born children, specialist visits were most often for behavioral and mental health concerns. Despite the well-described trauma and mental distress many refugees have experienced^5,37–40  and self-reported unmet mental health care needs,⁴¹  the provision of mental health care in this study was low, especially for PSRs. For some, psychological support may come from community resources, although others may have an unmet need.⁴²  Measuring physician visits alone likely underestimates the true need for mental health services for refugees.^32,43,44 

For acute care visits, we report Ontario-born children and resettled refugees have similar diagnoses, including high proportions of visits for respiratory tract illnesses and injuries. Others have also reported similar visit reasons in refugee children with comparable proportions visiting the ED in the early resettlement period.^8,45  The low proportion of low-acuity visits among refugees, and especially among PSRs, suggests access to primary care may be adequate. The higher number of hospitalizations observed in GARs and BVOR refugees compared with PSRs suggests there may be greater morbidity in these populations.

There are important limitations to this study. We included resettled refugees with immediate access to provincial health care coverage; however, refugees may still use the Interim Federal Health Program until a provincial health card number is issued.⁵  Although likely infrequent, we may have missed some early health care use. Our data do not have measures of community and social support (eg, housing, settlement workers) or allied health (eg, psychologists, social workers), so underestimates may have been made in total health care costs. Health information before migration is not available and, therefore, we have no measures of baseline health status. This study does not measure unmet health care needs, and we are unable to measure the quality of care received. There are differences in refugees’ settlement models in terms of social supports provided and the presettlement experience that may be associated with health system utilization that were not captured in available data and are an important consideration for contextualizing findings.⁵ 

Canada has a unique refugee resettlement model and resettles among the highest per capita number of refugees. Pediatric refugees who come to Canada through private sponsorship have lower aggregate morbidity with lower health system utilization in the first year after arrival compared with refugees arriving through the GAR and BVOR models. While refugees in these latter 2 models have a greater intensity of health care utilization than those resettling through private sponsorship and Ontario-born, the overall excess demand on the health system for all resettled pediatric refugees is minimal. Resettled refugees in Canada are similar to those in other countries. Taken together, these findings can assist health system planners and immigration ministries from countries across the globe to understand optimal models of refugee resettlement and facilitate planning for how best to support the early health care needs of pediatric refugees.

aOR

adjusted odds ratio

ADG

aggregated diagnosis group

CHC

community health centers

CI

confidence intervals

BVOR

Blended Visa Office-Referred program

ED

emergency department

GAR

government-assisted refugee

PSR

privately sponsored refugee