1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

ACADEMIC EMERGENCY MEDICINE 2011; 18:836–843 © 2011 by the Society for Academic Emergency Medicine


Objectives:  The relative effects of socioeconomic status (SES) and health status on emergency department (ED) utilization are controversial. The authors examined this in a setting with universal health coverage.

Methods:  For Ontario participants age 20–74 years, Canadian Community Health Survey 2000 to 2001 responses were linked to Ontario Health Insurance Plan (OHIP) physician utilization data for 1999 to 2001 and the National Ambulatory Care Reporting System (NACRS) for ED utilization in 2002. SES was defined primarily according to high school completion and secondarily according to income. The primary outcome was less urgent ED visit, defined as Canadian Triage and Acuity Scale (CTAS) 4 or 5 and not admitted to hospital.

Results:  The weighted sample was 9,323,217. Overall, 31.4% of the sample used an Ontario ED in 2002. The majority of visits (59.1%) were classified as less urgent. Fair or poor self-perceived health was the largest predictor of ED use, regardless of visit urgency. Respondents with low education were more likely to have both less urgent visits (odds ratio [OR] = 1.65, 95% confidence interval [CI] = 1.35 to 1.94) and more urgent visits (OR = 1.39, 95% CI = 1.09 to 1.68) after controlling for age, sex, income, self-perceived health, urban or rural location, regular doctor, and non-ED physician visits. Education was not associated with having less urgent versus more urgent visits (OR = 0.92, 95% CI = 0.68 to 1.14).

Conclusions:  In a setting with universal health insurance, worse health status is the largest predictor of ED utilization, but low SES is independently associated with increased use of the ED, regardless of visit urgency. This study lends support to findings in other health systems that those using EDs are more ill and more disadvantaged.

Emergency department (ED) use continues to rise in North America, and worsening crowding compromises patient care.1,2 Understanding contributing factors and reasons for use, therefore, has health policy relevance. In the United States, previous studies conducted at single sites indicated that those who use the ED were more likely to be the uninsured, patients without primary care physicians, visible minorities, and other “vulnerable populations.”3–6 However, recent studies that examined population-based samples in the United States found that those using EDs were simply more likely to be in poor health and to have experienced disruptions in care.7,8 Canada’s health system differs from that of the United States in that health insurance is universal; thus, access to health care should not vary by age, income, or other factor, after accounting for health needs.9 Assuming equal access to health care, use of EDs should theoretically not vary by socioeconomic status (SES), especially for low urgency conditions that potentially can be managed in non-ED settings. There is no population-based information from Canada to document the characteristics of those who use the ED compared with those who do not and whether the principles of Canadian universal health insurance hold true.

Examining waiting times for health services is currently a priority for the Canadian health system, incited by the 2003 First Ministers’ Accord.10 A 2005 report by the Canadian Institute for Health Information investigating ED wait times found that 57% of ED visits are for less urgent or nonurgent reasons.1 An Ontario-based report found a lack of variation in use of primary care and specialist physician services for disadvantaged groups compared with nondisadvantaged groups, despite worse health status.11 The definition of disadvantaged used in this report included SES, unmet health needs, rural residence, immigration, and ethnicity. The report also demonstrated higher frequencies of ED use specifically for those of lower SES. These trends were hypothesized to represent a barrier that disadvantaged populations, particularly those with low SES, experience in accessing a universal health care system, which results in increased use of EDs.

Lower SES is understood to result in worse health status, which predictably leads to increased health care utilization.12 In addition to this, individuals of lower SES may be more likely to encounter barriers in navigating the health care system.11,13 This could result in proportionately more ED visits than those with higher SES, with the ED representing the “safety net” when other providers are inaccessible. Specifically, with increasing shortages of family physicians, those of lower SES may be less likely to obtain the care of a family physician. They may, therefore, have to turn to the ED for care, especially for less urgent or lower-acuity reasons, although these can often be managed in non-ED settings.14

The purpose of this study was to examine a population-based sample from Ontario, Canada, for variation of ED utilization across socioeconomic groups. The primary outcome measure was less urgent ED use, and the secondary outcomes were overall ED utilization and more urgent utilization. Our hypothesis was that individuals with lower SES would be more likely to have increased ED utilization for less urgent conditions.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

Study Design

This was a secondary analysis of data from the Canadian Community Health Survey (CCHS)–Cycle 1.1. The study was approved by the institutional research ethics board at Sunnybrook Health Sciences Center.

Study Setting and Population

The study setting of Ontario is Canada’s largest province with almost 12 million people. The target population of the CCHS included household residents in all provinces and territories, with the principal exclusion of populations in First Nations reserves, Canadian Forces Bases, and some remote areas. The CCHS employed a multistage stratified cluster design and the Ontario portion consisted of 37,681 respondents.

Our study population was restricted to individuals between the ages of 20 and 74 years to avoid proxy responses that could be assigned to children and older seniors. The sample was stratified by sex and age for descriptive purposes.

Study Protocol

The CCHS was a national survey conducted by Statistics Canada in 2000–2001. It was designed to provide timely cross-sectional estimates of health determinants, health status, and health system utilization at a subprovincial level (health region or combination of health regions). CCHS respondents were asked to provide their Ontario health card numbers and to consent to linkage with their health care utilization data. Those consenting were linked to the Ontario Registered Persons Database (RPDB), the province’s health care registry. Once linked with the RPDB, health card numbers were used to link respondents with physician fee-for-service claims to the Ontario Health Insurance Plan (OHIP) for 2000–2001, which corresponded to one fiscal year (approximately 94% of all physician encounters in the province are included in this database) and to records of ED visits for 2002 from the National Ambulatory Care Reporting System (NACRS), for which close to 100% of ED claims in the province are included. The year 2002 was chosen for NACRS data since 2002 was the first largely complete year for the database. The data were accessed as part of a comprehensive research agreement with the Ontario Ministry of Health and Long-Term Care.

Definitions.  Socioeconomic status as a measure of disadvantage can be conceptualized in different ways. For the purposes of this study, the main predictor variables were as follows. Low education was the primary independent variable and was defined as not having graduated from high school. Low income was viewed as an important predictor of SES and was defined by the CCHS categories of lowest, lower middle, and middle income. Income, however, was viewed as a secondary independent variable due to the potential for missing values. The lowest, lower middle and middle income categories were defined as a household income of ≤$29,999, ≤$39,999, or ≤$59,999, respectively.

Covariates from the CCHS for inclusion in the analysis were determined a priori. The CCHS variable of self-perceived health was defined as the primary indicator of health need. This variable has been validated and reliably found to represent an individual’s health needs.15,16 Living in a rural area as defined by the CCHS was also considered a health need variable, due to the reduced availability of some health care services and providers in rural areas of Ontario.11

Use of non–emergency physician health services was included as an additional measure of health care need as well as overall health care-seeking behavior. This was measured for family physician/general practitioner (FP/GP) visits, as well as specialist visits based on OHIP health insurance claims. The CCHS variable of “regular medical doctor” was used to control for having a regular source of care.

Outcome Measures

The primary outcome measure was less urgent visits to an ED. Less urgent was defined according to the Canadian Triage and Acuity Scale (CTAS) values of 4 (less urgent) and 5 (nonurgent) and excluded those who were admitted to hospital.17 Secondary outcome measures were 1) more urgent ED visits, defined as CTAS 1, 2, and 3 (resuscitation, emergent, and urgent, respectively) and including all patients admitted to hospital from any triage category; and 2) ED visits of any triage category. For those with repeat visits to EDs in the study year, we included only the first ED visit in our analysis; thus, demographic characteristics were analyzed once per individual. CTAS values were obtained from the NACRS database of ED records and are based on the score assigned by a triage nurse at the time of ED triage.

Data Analysis

Regression analysis was used to test the association between the independent variables and the outcomes, while controlling for the potential confounders described above. All covariates of interest were entered into the model simultaneously. Multivariate logistic regression was used to fit the model for the dichotomous outcomes. The dichotomous outcomes used were more urgent, less urgent, and all ED visits, each separately compared with the group that had no ED visits. Last, more urgent and less urgent visit groups were compared with each other. Poisson regression was used to model the number of ED visits. Statistical significance was defined at the 0.05 level. The analysis was conducted using SAS Version 9.1 (SAS Institute Inc., Cary, NC).

All analyses were performed with the weight variable provided by Statistics Canada to reflect sampling design and variation.18 Unweighted counts of less than 30 are not reported. Estimates with a coefficient of variation of >33.3% are not reported, while those between 16.5 and 33.3% are reported as marginal estimates to be interpreted accordingly, as recommended by Statistics Canada.18 All confidence intervals (CIs) and the coefficients of variation were calculated by means of bootstrap methods with 500 replications using bootstrap weights provided by Statistics Canada. Model fit for regression models was assessed using the ratio of deviance to degrees of freedom, with good fit being close to 1.

Sensitivity Analyses.  Visits to EDs were examined by diagnostic category for the comparison groups according to the major categories used by the International Classification of Diseases revision 10 (ICD-10).19 This was done by age strata of 20 to 34, 35 to 49, 50 to 64, and 65 to 74 years to address possible confounding by age. In addition, day of the week (weekend or holiday versus weekday) and time of day (daytime defined as 08:00 to 20:00 and after hours as 20:00 to 08:00) were examined to account for any differences in times for seeking care that could occur between comparison groups. CTAS 3 is an intermediate category, and for that reason we repeated primary analyses using solely CTAS 1 and 2 as more urgent and CTAS 3, 4, and 5 as less urgent.

It was anticipated that some aspects of health need may not be captured by a combination of self-perceived health and non–emergency physician visits. For this reason, additional covariates from the CCHS of self-reported disability, depression, having two or more chronic diseases, and unmet health needs in the past 12 months, were considered a priori as secondary indicators of health need.

The Johns Hopkins Adjusted Clinical Group (ACG) Case-Mix System was also used to account for patient comorbidity.20 This method of adjusting for case-mix has previously been used in Canada, in addition to having been validated in the United States.21–23 The system uses individual-level data to assign measures of resource use and comorbidity from diagnoses during a specified time period, obtained from patient records. In this study, OHIP health insurance claims and hospital admissions from 2001 were used to determine case-mix using the ACG software.20 The variables from the ACG software that were input into the model were adjusted clinical groups and resource utility bands.

Finally, it was hypothesized that education and income could interact with self-perceived health and that this may vary by sex or age. Consequently, education and income were tested for an interaction with self-perceived health and examined for both sexes and by age category.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

Characteristics of Study Subjects

In Ontario, 87.2% of CCHS respondents provided their health card numbers, consented to the linkage, and were successfully linked to the RPDB. The weighted sample of study subjects of ages 20 to 74 years that were linked by health card represented 9,323,217 Ontarians. The largest proportion of missing values (10.3%) occurred for the income variable. The mean age of the overall sample was 46.0 years (SD ±14.7 years) and the proportion of males was 46.3% overall.

Characteristics of the study subjects are shown in Table 1. According to Statistics Canada policy, only nationally weighted population estimates are provided. Overall, 31.4% of the sample used an Ontario ED in 2002. An estimated 59.1% of visits were classified according to CTAS as less urgent or nonurgent. Of all ED visits, approximately 40% occurred during daytime hours Monday to Friday (excluding holidays).

Table 1.    Characteristics of CCHS Respondents and Their ED and Other Health Care Utilization in 2002
CharacteristicN = 9,323,317*Males (%)Females (%)
20–49 yrs (n = 3,088,763)50–74 yrs (n = 1,500,347)20–49 yrs (n = 3,132,362)50–74 yrs (n = 1,601,845)
  1. All values are percentages.

  2. CCHS = Canadian Community Health Survey; CTAS = Canadian Triage and Acuity Scale; FP/GP = family physician/general practitioner; NACRS = National Ambulatory Care Reporting System; OHIP = Ontario Health Insurance Plan; SES = socioeconomic status.

  3. *Representing nationally weighted sample.

  4. †Obtained from CCHS self-reported data.

  5. ‡Low education defined as not having completed secondary school.

  6. §Low income defined according to CCHS categories of lowest, lower middle, and middle income.

  7. ||Obtained from NACRS for ED visits.

  8. ¶CTAS definition: 1 = emergent, 2 = urgent, 3 = semiurgent, 4 = less urgent; and 5 = nonurgent.

  9. **Marginal estimate with coefficient of variation between 16.5 and 33.3%.

  10. ††Includes after hours (20:00—08:00 hr) that occur on weekends.

  11. ‡‡Obtained from OHIP claims.

SES indicator†
 Low education‡18.714.029.811.531.1
 Low income§34.631.034.034.342.7
Health indicator†
 Self-rated health fair or poor14.
 Two or more chronic diseases43.128.952.640.267.4
 Has regular medical doctor90.984.295.592.297.1
 Rural residence15.214.717.514.615.0
ED utilization||
 ED visits in 200231.430.832.031.332.2
 CTAS¶ 1–28.75.814.86.1**13.3**
 CTAS 332.225.337.933.437.6
 CTAS 441.545.431.945.536.0
 CTAS 517.623.615.515.113.1
 Weekend visits30.331.729.129.430.6
 After-hours visits††30.833.025.932.129.1
Other health care utilization‡‡
  No visits3.
  One to four visits 28.544.521.424.414.3
  More than four visits 67.849.874.773.482.6
  No visits10.318.
  One to four visits 36.946.929.238.223.9
  More than four visits 52.734.864.952.673.9

Regarding utilization of other sources of health care services, 90.9% reported having a regular medical doctor and 67.8% had more than four visits to a FP/GP. More than 50% of individuals also had more than four visits to a specialist physician.

Individuals with low education and low income were more likely to visit the ED at least once (rate ratios [RRs] = 1.37 and 1.30, respectively; Table 2) and even more likely to have multiple visits (RR = 1.69 and 1.62, respectively). There was no statistically significant difference in ED triage score by education or income. Those living in a rural area were more likely to visit EDs, but had visits of lower urgency. Individuals with greater self-reported health needs, including fair or poor health, disabilities, and multiple chronic diseases, were more likely to visit the ED (RR = 1.74, 1.54, and 1.44, respectively) and their visits were of higher urgency. The results were similar whether more-urgent visits were defined as CTAS 1 and 2 or as CTAS 1, 2, and 3 (results for CTAS 1 and 2 not shown). Goodness of fit was adequate with the ratio of deviance to degrees of freedom equaling 0.98.

Table 2.    Frequency of ED Visits by SES, Location and Health Needs Characteristics
 Any ED Visit vs. No VisitsMore Than One ED Visit vs. No VisitsMore Urgent* Triage for One or More ED Visits
%RR† (95% CI)%RR† (95% CI)%RR† (95% CI)
  1. Low education = did not complete high school; the lowest, lower middle, and middle income categories were defined as a household income of ≤$29,999, ≤$39,999, or ≤$59,999, respectively.

  2. CTAS = Canadian Triage and Acuity Scale; RR = rate ratio; SES = socioeconomic status.

  3. *More urgent in this study defined as CTAS 1–3 based on first visit to an ED in 2002.

  4. †Comparing low SES to high SES and greater health needs to lower health needs.

 Low 26.81.3711.11.6941.91.09
Rural residence
Self-perceived health
 Fair or poor33.31.7416.12.5951.51.41
 Good to excellent19.2(1.60–1.88)6.2(2.25–2.93)36.6(1.26–1.56)
≥2 chronic diseases

Regression Analysis: Less Urgent Visits Compared With No ED Visits

Low education was independently associated with a higher risk of having a less urgent ED visit compared with having no ED visits (adjusted odds ratio [AOR] = 1.65, 95% CI = 1.35 to 1.94; Table 3). Individuals with fair or poor self-perceived health had the highest independent association with risk of using the ED for less urgent presentations (AOR = 1.80, 95% CI = 1.41 to 2.19). Having a regular medical doctor was associated with decreased risk of using the ED (AOR = 0.75, 95% CI = 0.56 to 0.93). These trends persisted in a model that examined those with multiple ED visits (data not shown). The ratio of deviance to degrees of freedom was 1.5.

Table 3.    Logistic Regression Modeling of ED Visit Outcomes
VariableAORs for ED Visits (95% CI)
Less Urgent* vs. No ED VisitMore Urgent† vs. No ED VisitED Visit of Any Triage Category vs. No VisitMore Urgent vs. Less Urgent* ED Visit
  1. AORs = adjusted odds ratios; CTAS = Canadian Triage and Acuity Scale; FP/GP = family physician/general physician; low education = did not complete high school; the lowest, lower middle, and middle income categories were defined as a household income of ≤$29,999, ≤$39,999, or ≤$59,999, respectively.

  2. *Less urgent in this study defined as CTAS 4 or 5 ED visit not resulting in admission to hospital.

  3. †More urgent in this study defined as CTAS 1–3 based on first visit to an ED or admitted patient in 2002.

 Age (years)0.97(0.97–0.98)1.00(0.99–1.00)0.98(0.98–0.99)1.02(1.01–1.03)
 Low education1.65(1.35–1.94)1.39(1.09–1.68)1.53(1.30–1.76)0.92(0.68–1.14)
 Low income1.42(1.23–1.62)1.36(1.10–1.62)1.38(1.21–1.56)0.93(0.73–1.13)
 Self-perceived health fair or poor1.80(1.41–2.19)2.78(2.22–3.33)2.25(1.91–2.58)1.51(1.07–1.95)
 Rural vs. urban1.72(1.47–2.04)1.02(0.83–1.19)1.42(1.25–1.64)0.62(0.50–0.81)
 Regular medical doctor0.75(0.56–0.93)0.97(0.65–1.28)0.81(0.65–0.98)1.39(0.84–1.94)
FP/GP visits
 0 vs. >40.39(0.26–0.52)0.31(0.18–0.43)0.36(0.26–0.46)0.84(0.39–1.28)
 1–4 vs. >40.63(0.52–0.73)0.49(0.40–0.59)0.58(0.50–0.66)0.81(0.60–1.03)
Visits to specialists
 0 vs. >40.50(0.38–0.61)0.31(0.21–0.40)0.42(0.34–0.50)0.62(0.37–0.87)
 1–4 vs. >40.67(0.58–0.77)0.51(0.42–0.60)0.60(0.53–0.67)0.79(0.60–0.97)

Regression Analysis: More Urgent Visits Compared With No ED Visits

Low education also had a significantly higher risk of making a more urgent ED visit (AOR = 1.39, 95% CI = 1.09 to 1.68). Age, geographic location, and having a regular doctor were not significantly associated with increased risk for more urgent visits. Fair or poor self-perceived health remained the single largest predictor of risk for making an ED visit. The ratio of deviance to degrees of freedom was 1.5.

Regression Analysis: More Urgent Compared With Less Urgent ED Visits

Low education was not associated with an increased risk of having a more urgent visit compared with a less urgent visit (AOR = 0.92, 95% CI = 0.68 to 1.14). Fair or poor self-perceived health was, however, associated with increased risk of a more-urgent ED visit (AOR = 1.51, 95% CI = 1.07 to 1.95). Rural residence was associated with less risk of having a more urgent visit (AOR = 0.62, 95% CI = 0.50 to 0.81). The risk of an ED visit was lower among those making fewer FP/GP and specialist visits, but this effect was not significant for FP/GP visits for the comparison of more urgent to less urgent visits. The ratio of deviance to degrees of freedom was 2.2.

Graphs of self-perceived health by education are shown in Figures 1 and 2 for males and females, respectively. The figures demonstrate higher ED use among those with low education, a finding that is more pronounced among those with better self-reported health and among women. Those with the poorest health status have the highest ED use, regardless of education level.


Figure 1.  Percentage of males with ED visits versus self-perceived health.

Download figure to PowerPoint


Figure 2.  Percentage of females with ED visits versus self-perceived health.

Download figure to PowerPoint

Sensitivity Analyses

With the persistent association of low education and income with an increased risk of both low-urgency and high-urgency ED, the additional predetermined indicators of health need were added to the models: self-reported disability, depression, two or more chronic diseases, and unmet health needs in the past 12 months. After these variables were added into the model, low education continued to be significantly associated with a higher risk of visiting the ED. Similarly, adjusting for individual-level comorbidity using the ACG Case-Mix system did not appreciably alter the strength of the association between low education and ED visits. There were no statistically significant differences between education or income groups for weekend or after-hours visits.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

We examined a population-based sample from Ontario, Canada, and found that low education and income, conceptual measures of SES, were independently associated with a modest increase in risk of both high-urgency and low-urgency ED visits. As expected, worse health status was the largest determinant of ED use across all acuity levels, while having a regular medical doctor was associated with reduced use of the ED for less urgent visits. Having no contact with non-ED health care providers was not associated with increased ED utilization.

To our knowledge, this is the first study that links population-based individual-level health survey data with actual ED utilization and the first Canadian population-based study on the subject. Several recent U.S. population-level studies of ED utilization found that those with worse health status, low income, and unmet health needs were more likely to use the ED.7,8 The same studies found that ED users were no more likely to be uninsured, nor lack a usual source of care, than non-ED users, which contradicted prior hospital-based studies that found the opposite.3–8 The Canadian health care system differs from the United States in that health insurance for hospital and physician care is publicly funded and universal. In this environment of fewer direct financial barriers to medical care, we found that individuals who are sicker are more likely to use the ED, and those who use the ED are also more likely to be users of other providers of health care such as family physicians and specialists. Unlike previous studies, however, we found that low education and income were associated with increased ED utilization, even after controlling for health status and access to primary and specialist care.

Although our data allowed us to determine the triage categories and main diagnoses for ED visits, we did not attempt to define “appropriateness” of particular visits, nor whether ED visits could have been managed in other settings. Our aim was to use the triage score and disposition to define a group of less urgent patients whose visit to the ED may indicate access barriers to ambulatory care and compare their characteristics with the general, non–ED-using population.

We hypothesized that in Ontario, where a shortage of primary care physicians is recognized across the province, less urgent presentations to the ED could be a reflection of immediate access barriers to alternative sites of care such as family doctors or walk-in clinics.14,24,25 This hypothesis is supported by single-site surveys from Canada that have demonstrated that while less urgent ED use is not associated with lack of a family doctor, it is associated with an inability to access timely primary care.26,27 A Canadian population-based study demonstrated that recent primary care reform efforts in Ontario have led to increased rostering of individuals of higher SES than expected for the population, also suggesting reduced primary care access for those with lower SES.28 We further hypothesized that people with low SES would be more likely to experience barriers due to difficulty navigating the health care system. The finding that individuals with low education and income, which we used to represent SES, are more likely to make ED visits, irrespective of acuity level and having a regular doctor, does not support this hypothesis. To our knowledge, this is the first time that this relationship has been reported in a Canadian population; however, this concurs with findings from other countries with public health insurance that report an association between SES and overall ED utilization.29 The finding that after-hours and weekend ED visits do not vary by education or income also does not support an access barrier hypothesis, unless these barriers apply more or less equally to all SES groups.

If access barriers and health status are not a primary cause of increased ED use among people with low SES in our setting, then what is the explanation for this finding? It is understood that a greater burden of disease rests on those who are poorly educated and have lower income.30 We attempted to control for health needs in both primary and sensitivity analyses, but none of these analyses resulted in reduced effects of education or income. While residual confounding by need may be an explanation, it is also possible that perception of urgency and preference for source of care vary by SES. These are both concepts that have been demonstrated to be associated with less urgent ED use in the literature.26,27,31,32 Patients with lower education or income may have less knowledge of or access to alternatives to EDs, such as tele-health phone lines (e.g., lack of knowledge or telephone) or walk-in clinics (e.g., distance from home) or self-care for self-limited conditions (e.g., over-the-counter medications and supplies). Our finding that low-education groups have higher ED utilization even among those who report good or excellent health supports this possible explanation. Additionally, measures of health need may not predict injuries and higher injury rates among those with low SES, which may also be contributing to our findings.

Since the inception of EDs as organized and integral sites of care within the health care system, they have been recognized for their desirable features.33 Health care workers and laboratory and radiology services are on-site to provide comprehensive care 24 hours a day, 7 days a week. The accessibility of ED services in Canada and elsewhere may contribute to increased use by low SES groups due to their “one-stop” nature. Visits to EDs may obviate the need for scheduled office visits or multiple scheduled appointments for further investigations and/or consultations. This factor may be important for working people with low education or income due to lack of flexibility to plan ahead, constraints from inflexible work hours, child care availability, and ability to organize transportation. Such ED utilization is often considered less desirable from a health system perspective; however, low-urgency or low-complexity patients do not contribute significantly to overall ED wait times or overcrowding.34,35 In addition, such utilization of EDs may be the most appropriate course of action for the individual and may lead to fewer missed work days, lower stress, and lower incurred costs in the long term. In addition to services provided, there may be more subtle reasons for preferring EDs as sites of care. If EDs are equally welcoming than office settings to individuals from all social, cultural, and economic backgrounds, then low SES groups may have a preference for this type of care. Further research into the contributing factors to ED use is needed.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

The CCHS excluded certain groups and had a degree of non-response, all of which could influence the generalizability of these results. Statistics Canada applies both sampling and poststratification weights to minimize these potential biases. The independent variables used in this study are based on self-report, which is susceptible to social desirability, nonresponse, and related biases. For example, just over 10% of respondents did not complete the question regarding income.

The databases that were used to provide health care utilization data are administrative in nature and were not designed specifically for research purposes. However, since Ontario’s health system uses a single payer, the use of insurance claims is a nearly complete reflection of actual utilization.

Triage categories in CTAS were designed to direct timely ED care to certain groups. Our use of these categories to distinguish more urgent from less urgent visits has not been externally validated.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

We found that worse health status is the largest predictor of ED utilization in Ontario, but low education and income were independently associated with increased ED visits. The increased use seen in low education and income groups is evident across acuity levels, suggesting that increased health care needs are underpinning this utilization. This study provides support to findings in other countries demonstrating that ED users are sicker and socially disadvantaged and are not relying on EDs simply as a primary care provider. Our findings support an international policy direction that should, in the short term, better understand the factors leading to a preference for ED-based care among low socioeconomic status patients. In the long-term, improved health promotion, preventive health care and, chronic disease management, coupled with efforts to reduce socioeconomic disparities, may help to alleviate the greater burden of disease that is carried by disadvantaged populations.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References

The authors acknowledge Mohammed Agha for his assistance with the data sets.


  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. Limitations
  7. Conclusions
  8. Acknowledgments
  9. References
  • 1
    Canadian Institute for Health Information. Understanding Emergency Department Wait Times. Ottawa, Ontario: Canadian Institute for Health Information, 2005.
  • 2
    Institute of Medicine. The Future of Emergency Care in the United States Health System. Ann Emerg Med. 2006; 48:11520.
  • 3
    Richardson L, Hwang U. Access to care: a review of the emergency medicine literature. Acad Emerg Med. 2001; 8:10306.
  • 4
    Grumbach K, Keane D, Bindman A. Primary care and public emergency department overcrowding. Am J Public Health. 1993; 83:3728.
  • 5
    Torrens P. Variation among emergency room populations: a comparison of four hospitals in New York City. Med Care. 1970; 8:6075.
  • 6
    Pane G, Farner M, Salness K. Health care access problems of medically indigent emergency department walk-in patients. Ann Emerg Med. 1991; 20:7303.
  • 7
    Weber E, Showstack J, Hunt K, Colby D, Callaham M. Does lack of a usual source of care or health insurance increase the likelihood of an emergency department visit? Results of a national population-based study. Ann Emerg Med. 2005; 45:412.
  • 8
    Zuckerman S, Shen YC. Characteristics of occasional and frequent emergency department users: do insurance coverage and access to care matter? Med Care. 2004; 42:17682.
  • 9
    Madore O. The Canada Health Act: Overview and Options? Available at: Accessed May 23, 2011.
  • 10
    HealthCanada. 2003 First Ministers’ Accord on Health Care Renewal. Available at: Accessed May 23, 2011.
  • 11
    Glazier RH, Tepper J, Agha M, Moineddin R. Primary care in disadvantaged populations. In: JaakkimainenL, UpshurR, SchultzS, eds. Primary Care in Ontario. Toronto: Institute for Clinical Evaluative Sciences, 2006.
  • 12
    Winkleby M, Jatulis D, Frank E, Fortmann S. Socioeconomic status and health: how education, income, and occupation contribute to risk factors for cardiovascular disease. Am J Public Health. 1992; 82:81620.
  • 13
    Schoen C, Doty M. Inequities in access to medical care in five countries: findings from the 2001 Commonwealth Fund International Health Policy Survey. Health Policy. 2004; 67:30922.
  • 14
    College of Physicians and Surgeons of Ontario. 2005 Survey of Ontario’s Physicians: Access Challenges Ahead. Toronto: College of Physicians and Surgeons of Ontario, 2006.
  • 15
    Ware J, Snow K, Kosinski M, et al. SF-36 Health Survey Manual and Interpretation Guide. Boston, MA: New England Medical Center, 1993.
  • 16
    McHomey C, Ware J, Raczek A. The MOS 36-item short-form health status survey (SF-36) II: psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care. 1993; 31:24763.
  • 17
    Grafstein E, Innes G, Westman J, Christenson J, Thorne A. Inter-rater reliability of a computerized presenting-complaint-linked triage system in an urban emergency department. Can J Emerg Med. 2003; 5:3239.
  • 18
    Statistics Canada. Canadian Community Health Survey (CCHS): Cycle 1.1: extending the wealth of health data in Canada. Available at: Accessed May 23, 2011.
  • 19
    World Health Organization. International Statistical Classification of Diseases and Related Health Problems: Tenth Revision, 2nd edn. Geneva, Switzerland: World Health Organization, 2004.
  • 20
    The Johns Hopkins University. Johns Hopkins ACG Case-Mix Adjustment System. Available at: Accessed May 23, 2011.
  • 21
    Reid R, MacWilliam L, Verhulst L, Roos N, Atkinson M. Performance of the ACG case-mix system in two Canadian provinces. Med Care. 2001; 39:8699.
  • 22
    Starfield B, Weiner J, Murla P. Ambulatory care groups: a categorization of diagnoses for research and management. Health Serv Res. 1991; 26:5374.
  • 23
    Weiner J, Starfield B, Steinwachs D, Abramson J. Development and application of a population-oriented measure of ambulatory care case-mix. Med Care. 1991; 29:45272.
  • 24
    Ontario College of Family Physicians. Where Have All the Family Doctors Gone? Toronto, Ontario: Ontario College of Family Physicians, 2001.
  • 25
    Schultz SE, Tepper J, Guttman A, Jaakkimainen L. Characteristics of primary care practice. In: JaakkimainenL, UpshurR, Klein-GeltinkJE, et al., eds. Primary Care in Ontario: ICES Atlas. Toronto: Institute for Clinical Evaluative Sciences, 2006.
  • 26
    Steele S, Anstett D, Milne WK. Rural emergency department use by CTAS IV and V patients. Can J Emerg Med. 2008; 10:20914.
  • 27
    Field S, Lantz A. Emergency department use by CTAS Levels IV and V patients. Can J Emerg Med. 2006; 8:31722.
  • 28
    Glazier RH, Redelmeier DA. Building the patient-centered medical home in Ontario. JAMA. 2010; 303:21867.
  • 29
    Shah SM, Cook DG. Socio-economic determinants of casualty and NHS Direct use. J Public Health (Oxf). 2008; 30:7581.
  • 30
    Marmot M, Smith G, Stansfeld S, et al. Health inequalities among British civil servants: the Whitehall II Study. Lancet. 1991; 337:138793.
  • 31
    Tsai JC, Liang YW, Pearson WS. Utilization of emergency department in patients with non-urgent medical problems: patient preference and emergency department convenience. J Formos Med Assoc. 2010; 109:53342.
  • 32
    Backman AS, Blomqvist P, Lagerlund M, Carlsson-Holm E, Adami J. Characteristics of non-urgent patients. Cross-sectional study of emergency department and primary care patients. Scand J Prim Health Care. 2008; 26:1817.
  • 33
    Pollock D. Barriers to health care access: what counts and who’s counting? Acad Emerg Med. 2001; 8:10168.
  • 34
    Schull M, Kiss A, Szalai J. The effect of low-complexity patients on emergency department waiting times. Ann Emerg Med. 2007; 49:25764.
  • 35
    Asplin B, Magid D. If you want to fix crowding, start by fixing your hospital. Ann Emerg Med. 2007; 49:2734.