Socioeconomic Status, Service Patterns, and Perceptions of Care Among Survivors of Acute Myocardial Infarction in Canada FREE

Canada's universal health insurance system provides comprehensive coverage for most medical and hospital services without user fees at point of service. Under such provisions, patients are entitled to equitable access to health care services based on medical need, regardless of affluence, education, social position, or race. Known as "Medicare," this social program has iconic status in Canada and is unique among industrialized nations in banning any parallel private insurance of publicly insured services.¹

Proponents of the Canadian Medicare program have long argued that Canada's unusually restrictive approach to private insurance ensures that the socioeconomic elite remain in the public system.^2- 4 Because they have high expectations of service and have influence beyond their numbers, the elite are perceived to elevate the standard of service for all citizens.⁵ Critics respond by suggesting that the increased demands of those most affluent are as likely to be excessive as appropriate, thereby placing pressure on the public system by preferentially accessing an already limited supply of specialty services.^6- 8

No study has formally evaluated how service patterns and perceptions of care vary across socioeconomic strata in Canada for an acute life-threatening disease. Our objective was to explore how Canadian patients from different socioeconomic backgrounds perceive their medical care and to correlate patients' backgrounds and care received with perceptions of health care access, quality, and outcomes.

Data were obtained from the Socio-Economic Status and Acute Myocardial Infarction study (SESAMI). SESAMI is an ongoing prospective longitudinal observational study of patients hospitalized with acute myocardial infarction (AMI) throughout Ontario, Canada.⁹ The data elements collected in SESAMI pertinent to this analysis include a 13-item patient-completed questionnaire addressing risk factors for atherosclerosis as well as socioeconomic status (SES) and ethnic characteristics and 30-day follow-up telephone interviews addressing functional status, quality of life, utilization of specialized cardiac services, medication use, and perceptions of health care access, satisfaction, and quality. Mortality was identified through the Ontario Registered Persons Data Base by data linkage using encrypted health card numbers. Participation in SESAMI required written patient consent. SESAMI received research ethics approval from each of the participating institutions.⁹

SESAMI recruited English-speaking Ontario residents who were admitted for AMI through the emergency departments in 53 of 57 large-volume Ontario hospitals (≥100 AMI admissions per year) between December 1, 1999, and June 1, 2002. Trained nurses identified eligible AMI patients through chart surveillance while patients were hospitalized in the coronary or intensive care unit. The diagnosis of AMI was based on any 2 of 3 criteria: symptoms, electrocardiographic (ECG) abnormalities, or elevated cardiac enzymes. Patients were excluded if they were younger than 19 or older than 101 years, had no valid Ontario health card number, or were transferred into the recruiting site. Patients who died early (eg, within 24 hours), those with severe illness (eg, receiving ventilatory support), persons with language barriers, and those discharged or transferred early after presentation were ineligible for study participation given their requirement to complete the self-administered baseline survey at study entry.⁹

In total, 3335 of the 4474 consecutive eligible AMI patients approached agreed to participate in baseline surveys and comprised the original SESAMI cohort. Among these patients, 21.4% were ineligible for 30-day telephone interviews due to death, recurrent hospitalization, or insufficient follow-up information, leaving 2620 patients available for follow-up, of which 2256 patients consented to 30-day telephone interviews (86% participation rate). Ineligible patients (due to early death, recurrent hospitalization, or insufficient follow-up information) were less affluent, older, and significantly more likely to have preexisting heart disease than were those who participated in 30-day interviews. However, the sociodemographic and clinical risk profiles of ineligible patients were generally similar to those who were approached but refused participation in the 30-day telephone follow-up interview. Moreover, socioeconomic risk factor gradients were similar in relative magnitudes between ineligible and eligible (consented and nonconsented) patients.

SESAMI collected information on household annual income (7-level categorical variable ranging from Can <$15 000 to ≥$80 000) and education (5-level categorical variable ranging from incomplete high school to university degree). In total, 94% of patients divulged their income, while 98% of participants provided information related to education. To ensure similar sample sizes across socioeconomic subgroups, both income and education data were each reaggregated into 3-level categorical variables (income: ≤$29 999; $30 000-$59 999; ≥$60 000; education: incomplete high school; completed high school or any incomplete postsecondary school education; degree in university, college, or trade school). Similar aggregation levels for income and education have been used elsewhere and shown to be important determinants of disease risk.^10- 14 Income cutoffs also corresponded to Canadian federal and provincial year 2000 income tax rate thresholds of $30 000 and $60 000.¹⁵ All income levels reported herein are in Canadian dollars (Can $1 = US $1.34 as of February 4, 2004).

Information on ethnicity was obtained through self-report. Patients could select from 1 or more categories of a possible 13 ethnic/racial subgroups.¹⁰ For the purposes of this study, ethnic/racial data were reaggregated a priori into 5 variables: white, black, South Asian, First Nations (Canadian Native Indians), and other (eg, East Asian/Chinese). Each of these categories has been shown to be associated with variations in disease severity, treatment patterns, or both.^16- 18

Previous studies by our group and others have demonstrated that geographic and admitting hospital characteristics are important determinants of cardiac specialty service use in Canada.^19- 20 Patients were therefore categorized by urban vs rural place of residence and admitting hospital characteristics (ie, onsite coronary angiography capacity). Urban-rural information was obtained from patient postal codes using a Statistics Canada conversion algorithm.²¹

Data regarding risk factors and preexisting heart disease were obtained on all patients at study entry and were categorized by the presence or absence of the following self-reported factors: diabetes mellitus, hypertension, hyperlipidemia, current smoking, family history, and preexisting heart disease (and duration if applicable). The reliability of self-reported risk factors was evaluated through hospital chart audits for 1609 patients. Agreement levels were similar across SES strata, ranging from 73% (hyperlipidemia) to 95% (diabetes).

Patients were asked to report at 30 days on 3 measures of specialty service use: (1) coronary angiography, (2) early referral for cardiac rehabilitation, and (3) cardiology visits following discharge from the index AMI hospitalization. We also examined post-AMI discharge visits to a primary care physician (general practitioner/family physician) as a comparator. The reliability of self-reporting for invasive procedures was evaluated through linkage to administrative data for 1937 consecutive patients. Agreement levels were again similar across SES strata, ranging from 74% (angiography) to 98% (coronary artery bypass surgery).

Patients were asked a series of questions with a 5-item response scale from 1 (poor) to 5 (excellent) plus the option "don't know." Upon completion of the interview, patients were asked to respond (yes, no, uncertain) to 2 supplementary questions about perceptions regarding privatization and willingness to pay. The first applied to anyone willing to pay: "Based on the care you received over the past 4 weeks, are you in favour of a two-tiered health care system, whereby those who are willing to pay could have quicker access to care and a wider selection of treatments and services, over and above standard care?" The second was narrower and applied only to the respondent's willingness to pay himself/herself: "Based on the care that you received over the past 4 weeks, would you be willing to pay for quicker access to services and a wider choice of treatment and hospital options?" Because the questions were added following study initiation, only 1815 of 2256 consecutive patients were asked about their perceptions regarding privatization and willingness to pay. The ethics review boards of the participating institutions were informed of the supplemental questions. All but 1 hospital (consisting of 57 study patients) consented to the additional question, leaving 1758 eligible patients to respond, of whom 1684 (95.8%) did so.

We examined crude differences in baseline characteristics, specialty service use, perceptions of care, and outcomes across income and education categories using the Mantel-Haenszel χ² test for trend for categorical data and analysis of variance for continuous data. Because perceptions of care were skewed per usual toward the positive extreme, the relationship between SES and perceptions was examined for most satisfied (vs less so) and least satisfied (vs more so). Multiple logistic regression techniques examined for differences in the perception and utilization of specialty cardiac care across SES, after adjusting for age, sex, ethnic/racial differences, cardiac risk factors, preexisting coronary artery disease, onsite catheterization capacity at the admitting hospital, and place of residence (urban vs rural), using backward stepwise regression comparing –2 log likelihood ratios between models. Multiple logistic regression models were also used to examine for any interactions between SES, participation status, and clinical risk factor profiles. Cox proportional hazard models examined the relationship between SES and mortality at 1 year following the AMI, after adjusting for each of the baseline characteristics described above, using stepwise regression. For all analyses, income and education were examined as 3-level categorical variables, with each SES variable adjusting for the presence of the other in multivariate analyses. Diagnostic testing for collinearity revealed that no explanatory variable had an associated variance inflation factor of greater than 5 (maximum in our analysis was 1.88). Odds ratios were converted to relative risks (RRs) using the method described by Zhang and Yu.²² Statistical significance was defined as P<.05. SAS version 8.2 (SAS Institute, Cary, NC) was used for all analyses.

The median age of the study sample was 64 years; 30% were women. Table 1 compares the distribution of baseline characteristics between SESAMI patients who consented and those who did not consent to telephone follow-up. Consenting patients were younger, more affluent, more educated, and less likely to have preexisting heart disease.

Less affluent patients were significantly older, more likely to be women, more likely to be South Asian, reside in rural communities, and have diabetes, hypertension, and preexisting heart disease (P<.001 for all comparisons) than were more affluent patients. Socioeconomic risk factor gradients among study participants were most marked for diabetes (range, 30.9% to 16.7% from lowest to highest income tertile; P<.001), preexisting heart disease (range, 49.1% to 30.6% from lowest to highest income tertile; P<.001), and hypertension (range, 58.0% to 40.7% from lowest to highest income tertile; P<.001). The corresponding socioeconomic risk factor gradients among eligible patients who refused participation in the telephone follow-up were similar in relative magnitudes. There were no significant interactions between SES and participation status in their associations with any of the risk factor profiles examined (interaction terms, P>.20).

Crude mortality rates in the interval between 30 days and 1-year follow-up were higher among patients with lower household incomes (5.1% vs 1.9% for earnings ≤$29 999 vs earnings ≥$60 000, respectively; P = .001) and lower education (4.4% and 2.9% for those with incomplete high school vs those with completed high school and postsecondary degrees/diplomas, respectively; P = .07). After adjusting for baseline characteristics, neither income (adjusted risk ratio, 1.25; 95% confidence interval [CI], 0.68-2.31 for earnings ≤$29 999 vs ≥$60 000) nor education (adjusted risk ratio, 0.88; 95% CI, 0.51-1.50 for incomplete high school vs postsecondary degrees/diplomas) was significantly associated with mortality at 1 year following hospitalization for AMI.

While the vast majority of patients who received interventions did so during the index hospitalization regardless of their SES (88%), patients with higher incomes and education levels were significantly more likely to have been referred for coronary angiography (P<.001), cardiac rehabilitation (P<.001), and to a cardiologist (P = .04 for education, P<.001 for income) following discharge from AMI hospitalization. In contrast, general practitioner follow-up did not vary by patient income or education (Figure 1 and Table 2).

After adjusting for baseline characteristics, education was a stronger and more consistent determinant of specialty cardiac service use than was income. Compared with patients with postsecondary degrees/diplomas, patients who did not complete high school were less likely to receive any specialty cardiac services by 30 days following AMI (Table 2). After adjusting for baseline income and other variables, persons not completing high school education had significantly lower probabilities of receiving coronary angiography and cardiac rehabilitation than individuals who completed postsecondary education (Table 2).

Socioeconomic service gradients were similar in relative magnitudes when confining the analysis to urban patients regardless of whether onsite cardiac catheterization capacity was available at the admitting institution.

Table 3 illustrates the univariate associations between SES and perceptions of care. After adjusting for baseline demographic and clinical factors, income was the strongest predictor of responses on the affordability of medications, while education was the strongest determinant of perceptions of care (Table 4). Compared with higher income categories, patients earning less than $30 000 per year were less likely to be able to afford all necessary medications. Compared with patients who did not finish high school, those with postsecondary degrees/diplomas were more likely to be extremely dissatisfied with access to specialty care (adjusted RR, 2.02; 95% CI, 1.20-3.32), with provider "skill, concern, and caring" (adjusted RR, 2.40; 95% CI, 0.91-6.13), and with overall care (Table 4).

In total, 20% of AMI patients stated that they favored allowing private care for anyone who was willing to pay. When asked about their own willingness to pay, the proportion of those in favor of out-of-pocket payments for more expedient or a wider selection of treatment or hospital services increased marginally to 25%. Patients who were most satisfied with access to specialty care (21.4% vs 28.0%, P = .007), with provider concern (22.2% vs 26.7%, P = .05), and with the overall care received (22.1% vs 25.9%, P = .09) were less willing to pay for medical services compared with those with lesser degrees of satisfaction.

Perceptions regarding privatization and willingness to pay varied significantly more widely across income than education (Figure 2). After adjusting for baseline factors including education, cardiac risk factors, rural-urban status, and onsite procedural capacity, patients with higher household incomes were significantly more likely to prefer privatization of services (adjusted RR, 1.65; 95% CI, 1.24-2.14 for patients earning ≥$60 000; adjusted RR, 1.25; 95% CI, 0.93-1.65 for patients earning $30 000-$59 999), and acknowledged a greater willingness to pay for faster service or more service choices (adjusted RR, 2.11; 95% CI, 1.70-2.58 for patients with household incomes ≥$60 000; adjusted RR, 1.50; 95% CI, 1.16-1.90 for patients with household incomes of $30 000-$59 999). Preferences for private payment options were as marked for urban patients admitted to hospitals with onsite procedural capacity as they were for urban patients admitted to hospitals without, and persisted after adjusting for medication affordability.

We found significant socioeconomic differences in the provision and perceptions of care among patients with AMI in Canada. Despite receiving more intensive specialty cardiac services, patients in higher socioeconomic strata were more dissatisfied with access to and quality of care and much more willing to pay out-of-pocket for expedited services or for a wider array of treatment options. In the latter respect, we found that one quarter of all AMI survivors were willing to pay out-of-pocket for either expedited services or for a wider array of treatment options, with proportions ranging from just over 15% for persons with household incomes of less than $30 000 annually to more than 30% for those with household incomes of $60 000 and above. While overall satisfaction with care was high across all subgroups, those with highest SES were more likely to be dissatisfied than patients of lower SES.

These findings suggest that Canadian restrictions on parallel private insurance may serve to keep the upper-middle class "canaries in the coalmine," thereby providing a check on the amount of deterioration in the publicly funded system that would be tolerated. However, the question may be raised as to whether their advocacy truly enhances the overall system or is primarily self-serving. A significant minority of these individuals are also willing to pay for services outside the system. Absent a ban on private insurance, these persons may gain even more advantages in consumption of health services. But it is also arguable that these persons are competing successfully against other socioeconomic groups for scarce resources in the public system and might just as well pay privately for the unfair advantages they are deriving within the public system.

The argument is further illuminated by data on physician service utilization from other countries. Van Doorslaer et al²³ recently reviewed 1990s survey data from Organisation for Economic Co-Operation and Development nations and found that, after adjusting for age, sex, and indices of morbidity, wealthier respondents visited specialists more often in 13 of 14 nations studied. In contrast, use of general practitioner services was generally income-neutral after adjusting for need. All these nations except the United States offered some form of national health service or universal health insurance, and only Canada proscribes parallel or supplementary private insurance for physician services. There was no evidence that the differentials in service consumption were smaller in Canada, notwithstanding the variable uptake of private insurance in other nations.

Historically, Canadian data suggest that the initiation of universal health care coverage reduced socioeconomic disparities in consumption of health services but never fully eliminated them.²⁴ Our study adds to a small but growing body of work demonstrating that access to specialist services and procedures continues to vary with SES in Canada.^{14,19,25- 26} For example, our data suggest that patients who completed a postsecondary education were almost twice as likely to access cardiac rehabilitation as those who had not completed high school. The issue accordingly is not whether Medicare has enhanced equity in access to health care, but whether Canada's uniquely restrictive approach to private insurance is defensible and sustainable.

On the positive side of the ledger, our study demonstrates that only a minority of patients were willing to pay for faster or more varied services. Furthermore, only small numbers of patients were extremely dissatisfied with access to specialists, waiting times, or the skill, caring, and concern of clinicians. This may reflect the phenomenon that patients who have recently accessed health services tend to be more positive than those who are awaiting care or who have not themselves been helped by the health system in the recent past.^27- 29

Our study has several important limitations. First, patients were required to provide informed consent, patient enrollment was confined to large-volume hospitals, and attrition of the initial sample following discharge from the index AMI hospitalization may have biased our sample. However, the overall consent rate for the original SESAMI cohort was excellent (75%)⁹; 86% of eligible patients consented to participate in 30-day follow-up interviews; and socioeconomic risk factor gradients were similar between participants and nonparticipants (either because they had refused participation or because they were ineligible for telephone follow-up due to early death, recurrent hospitalization, or insufficient follow-up information). Second, SESAMI was restricted to Ontario, and the generalizability of our results to other Canadian provinces is unknown. However, Ontario comprises one third of the nation's population, and socioeconomic AMI–cardiac service gradients have been demonstrated in other Canadian provinces.³⁰ Finally, information on clinical parameters such as infarct size or severity was not known. However, the available clinical data suggest that patients with less education or lower incomes would, if anything, have more serious infarcts and a greater need for cardiac rehabilitation and secondary prevention.³¹

In conclusion, as evidenced by patients hospitalized for AMI, upper-income and better-educated Canadians were less likely to be satisfied with access to and quality of health care received, despite receiving more specialty cardiac services than those in lower socioeconomic groups. In contrast, citizens in the lower and middle tertiles of income and education received fewer specialty services but were more satisfied with services and less interested in paying for services privately. These findings speak to continuing tensions in the financial framework for Canada's Medicare system. In turn, these tensions are likely to shape continuing debate about the organization and finance of the Canadian health care system in the years ahead.