One of the strongest and most persistent nonbiological predictors of cardiovascular-related death is social class. Cardiovascular-related death has been inversely related to measures of social class, including education [7, 12-14], occupation [15-18], and income [8, 19]. For example, in a general population study of men in Finland [19] and in a U.S. study of patients undergoing cardiac catheterization [8], low income conferred a twofold increase in the risk for death. In another study of survivors of infarction with arrhythmia, men with 8 or fewer years of education had three times the risk for sudden coronary death of men who had more education [12]. Neither risk factors for coronary disease nor clinical characteristics known to affect prognosis accounted for the observed differences.

Social class has also been associated with myocardial infarction [20] and with the presence of cardiovascular risk factors, including hypertension, elevated cholesterol level, cigarette smoking, obesity, diabetes, and hemostatic factors [21-26]. However, disparities in mortality rates among social classes are only partially explained by these other risk factors or by health-damaging behaviors, access to health care, and adherence to medical recommendations [27-30].

To date, no studies have isolated the effect of social class on risk for myocardial infarction from the effect of social class on the extent of recovery after infarction. Most patients survive, and the extent of recovery varies from complete independence to severe limitations in daily activities. Functional recovery is of critical importance to patients, for whom limitations can be as distressing and debilitating as the primary illness itself [31, 32]. In addition, functional limitations can result in longer hospital stays, greater utilization of health services, and decreased likelihood that a patient will return to work and other productive activities.

The primary objective of our study was to determine whether social class remained independently associated with functional recovery, even after controlling for clinical, demographic, and psychosocial factors known to influence outcomes after myocardial infarction. Using data from a large cohort of men who initially survived infarction, we examined the prognostic influence of social class on functional recovery.

Methods

Top Methods Results Discussion Author & Article Info References

Beta-Blocker Heart Attack Trial

Data for these analyses came from the Beta-Blocker Heart Attack Trial (BHAT), a randomized, double-blind, placebo-controlled trial designed to test the efficacy of propranolol given to patients after myocardial infarction. Between June 1978 and October 1980, 3837 patients hospitalized with acute infarction (documented by presence of symptoms and enzymatic and electrocardiographic changes) were enrolled from 31 clinical centers in the United States and Canada. Details and major results of this study have been published elsewhere [33, 34].

Participants were enrolled 5 to 21 days after infarction. Baseline examination included medical history, assessment of current medical condition, and electrocardiographic data. "Baseline" refers to the period after infarction when patients were clinically stable but before randomization into the clinical trial. Patients were followed prospectively every 3 months for at least 12 months. Clinical data for these analyses were obtained from baseline and 12-month clinical assessments. Prospective assessment enhanced the reliability of the data and minimized observer bias.

Health Insurance Plan Substudy

Additional data were obtained from interviews conducted within the Health Insurance Plan (HIP) substudy. Of the 31 BHAT centers, 25 participated in the HIP substudy. Among participating centers, 90% of eligible patients (2320 of 2572) were enrolled in the HIP substudy; women were excluded. Face-to-face structured interviews were completed 6 weeks after infarction and included the information on social class and psychosocial factors described below [7].

Baseline Patient Characteristics

Clinical Factors

To account for preexisting medical conditions, any history of myocardial infarction, stroke, diabetes, and hypertension was documented if a patient reported that a physician had ever said that the patient had any of these conditions. These variables were dichotomized (yes or no) for inclusion in the statistical analyses. Current cigarette smoking and use of ß-blockers (as per the randomly assigned treatment in the trial [ß-blocker or placebo]) were also considered.

Baseline severity of myocardial infarction was assessed by the occurrence of congestive heart failure or electrical events during hospitalization for the index infarction but before randomization. Congestive heart failure was defined as the presence of pulmonary edema or signs and symptoms that required therapy with digitalis or diuretics; basilar rales or S₃ gallop on baseline physical examination; or the BHAT physician's opinion that the patient had experienced congestive heart failure. Electrical events were defined as complete atrioventricular block, ventricular tachycardia (3 successive ventricular premature beats), ventricular fibrillation, or atrial fibrillation or flutter. Variables that measured severity were dichotomized (yes or no) for inclusion in analyses.

Psychosocial Factors

Stress was measured as the self-report of stress associated with work, finances, divorce, or victimization (for example, as a result of mugging or robbery). Isolation was measured as lack of involvement in any club or organizations, not visiting friends, and limited communication with friends or family members. Depression was measured as negative affect (feeling "blue"), hopelessness about the future, and sleep problems. For each psychosocial characteristic, patients were characterized as "high" if they reported at least 2 symptoms within each symptom set (as listed here). They were categorized as "low" if they endorsed 0 to 1 items. This categorization resulted in a set of dichotomous psychosocial variables (low or high) for use in statistical analyses.

Social Class

Social class was defined by a composite of education (years of school completed) and occupation [35]. Patients were initially classified into one of the following occupational categories: 1) professional, technical; 2) managers, officials, proprietors; 3) clerical, sales; 4) craftsmen, foremen; 5) manufacturing, transportation; 6) protective service; 7) other service; and 8) laborers. Three categories of social class were hierarchically ordered; these categories reflected the patient's social class relative to the cohort. The lowest social class group (III) included patients with grade-school education only, regard less of occupation. The middle social class group (II) included patients who had some high school education and an occupation outside of the professional or managerial categories (that is, categories 3 through 8). The highest social class group (I) included patients who had at least a high school degree and a professional or managerial occupation (that is, categories 1 and 2).

Primary Study Outcome

Functional status was measured by using New York Heart Association (NYHA) ratings; these ratings reflect activity limitations and discomfort when performing daily activities. Class I represents the highest functional status (no activity limitations and no discomfort). Class II represents slightly limited physical activity (patients are comfortable at rest, but ordinary activity results in fatigue, palpitation, dyspnea, or anginal pain). Class III represents markedly limited physical activity (patients are comfortable at rest, but less than ordinary activity results in fatigue, palpitation, dyspnea, or anginal pain). Class IV represents the lowest functional status (patients cannot perform any activity without discomfort, symptoms of cardiac insufficiency or anginal syndrome are present at rest, and discomfort is increased with activity).

Our primary outcome measure was change in NYHA functional status between baseline and 12-month follow-up. For purposes of analysis, this variable was dichotomized as "improved" (indicating improvement in 1 category between baseline and follow-up) or "not improved" (no change, decline in 1 category, or death). Patients who were rated as having the highest functional status at baseline (NYHA class I) and who maintained that status 1 year later were considered part of the "improved" group; this categorization places patients with consistently highest functional status into the group with better health outcomes and averts a "ceiling effect."

The NYHA rating is among the most commonly used ratings for functional status in patients with cardiovascular disease. The rating has been an important predictor of cardiac-related illness and death [36-39]. It has also been an outcome measure in studies of ventricular dysfunction [40, 41], in studies of treatment of atrial septal defect [42] and tricuspid regurgitation [43], and in a meta-analysis of clinical trials of pharmacologic treatment of chronic congestive cardiac failure [44]. Some investigators have expressed concerns about the reliability of this measure among patients with coronary disease; potential limitations are addressed in our Discussion section.

Statistical Analysis

The association of social class with baseline cardiovascular risk factors was examined. Bivariate logistic regression analyses were then conducted to determine whether social class and other demographic, clinical, and psychosocial factors were associated with change in functional status. Finally, multiple logistic regression analyses were conducted to examine whether social class exerted an independent effect on functional recovery after controlling for clinical, demographic, and psychosocial factors [45]. Analyses were designed to predict a change in functional status, and the results are presented separately for the comparisons of low social class with high social class and of middle social class with high social class. Associations between variables were estimated as odds ratios and were tested for significance with 95% CIs. Analyses were performed by using a mainframe-computer version of SAS software (SAS Institute, Cary, North Carolina).

Results

Top Methods Results Discussion Author & Article Info References

Study Cohort

Of the original 2320 patients enrolled in the HIP substudy, 175 patients (7.5%) were excluded because of missing data. Thus, the final sample for our analyses consisted of 2145 men. Characteristics of our sample are presented in Table 1. Most patients were white, and age ranged from 29 to 69 years (mean, 54 years). The proportions of patients in high, middle, and low social class groups were 35.0%, 44.2%, and 20.8%, respectively.

A history of hypertension was most frequently reported (37.6% of patients). Histories of myocardial infarction (16.2%), diabetes (10.3%), and stroke (2.0%) were reported by small proportions of participants. One fifth of the participants were current smokers. As would be anticipated from the design of the randomized trial, half of the patients had received ß-blockers. Congestive heart failure and electrical events were documented during baseline examination in 18.4% and 32.6% of the participants, respectively; this finding indicates that many patients had complications during the initial hospitalization.

Association of Social Class with Baseline Risk Factors and Functional Status

To begin to examine the effects of social class on health, the associations between social class and baseline risk factors were documented (Table 1). Black persons were over-represented in the low social class group. For three of the clinical risk factors considered (history of myocardial infarction, smoking status, and baseline electrical events), a linear trend indicated that patients in the low social class group were most likely and patients in the high social class group were least likely to have reported cardiovascular risk factors at baseline. Histories of hypertension, diabetes, and stroke did not differ by social class categories, nor did the experience of congestive heart failure. Social class was strongly associated with all psychosocial features: Patients in the low social class group were most likely and patients in the high social class group were least likely to have reported high levels of stress, social isolation, and depression at baseline.

A clear inverse relation was seen between social class and baseline functional status: As social class declined, the proportion of patients with functional limitations increased. Social class was also associated with change in functional status from baseline to follow-up 1 year after infarction (Table 2).

Predictors of Improved Functional Status after Myocardial Infarction

Bivariate logistic regression analyses were conducted to identify the baseline demographic, clinical, and psychosocial factors associated with change in functional status (Table 3). As patients moved down one level of social class (from high to middle or from middle to low), the odds of having no functional improvement 1 year after infarction increased by 39%, as indicated by the odds ratio of 1.39. Older age and black ethnicity were also associated with an increased likelihood of no functional improvement.

Clinical features had varied associations with functional recovery. History of myocardial infarction was the strongest predictor of recovery. Patients who had previously had an infarction were more than twice as likely not to have functional improvement after the index infarction (odds ratio, 2.31). A history of diabetes and hypertension and the presence of congestive heart failure and current smoking at baseline were each associated with an approximate 40% to 50% increase in the odds of having no functional improvement. History of stroke, electrical instability at baseline, and randomization to ß-blockers were unrelated to improved functional status.

For psychosocial factors, patients who reported high levels of stress and depression were significantly more likely to have no improvement in functional status 1 year after infarction (odds ratios of 1.40 and 1.99, respectively).

Social Class and Recovery after Infarction: Adjustment for Clinical, Demographic, and Psychosocial Factors

To examine the key question of whether social class affected functional recovery after we controlled for clinical, demographic, and psychosocial factors, we performed a logistic regression analysis to predict change in functional status. Results are presented separately for the comparisons of low with high social class and middle with high social class.

As shown in Table 4, the unadjusted odds ratio for low compared with high social class was 1.92, indicating that patients in the low social class were nearly two times more likely to have no improvement in functional status than were patients in the high social class. After adjustment for clinical factors (medical history and severity of myocardial infarction), the odds ratio decreased to 1.78. In a model that added age and race, the odds ratio decreased to 1.62; when all variables, including the psychosocial variables, were included in the model, the odds ratio was 1.51 and remained statistically significant.

For the comparison of middle with high social class, the unadjusted odds ratio was 1.45, indicating that patients in the middle social class were 45% more likely not to have improvement in functional recovery than were patients in the high social class. After adjustment for clinical factors, the odds ratio decreased to 1.40. In a model that added demographic characteristics, the odds ratio was essentially unchanged at 1.39; when all variables, including the psychosocial variables, were included in the model, the odds ratio changed slightly to 1.34 and remained statistically significant.

Social class maintained its independent effect on improved functional status, even after we controlled for selected risk factors. Patients in the low and middle social classes were significantly more likely not to have improved functional status 1 year after infarction than were patients in the high social class. Certain clinical, demographic, and psychosocial features were related to recovery, but the effect of social class could not be fully explained by these additional features. In a repeated analysis, baseline NYHA functional class was included as a covariate to minimize bias in calculating NYHA improvements; no differences were seen in the results. Additional analyses were conducted to determine whether there was any bias in the variance estimates as a function of restricting these analyses from the original 31 centers to the 25 centers that participated in the HIP substudy. In a conditional logistic regression analysis in which each hospital was a stratum, estimates for both sets of analyses (low and middle compared with high social class) were unchanged.

Discussion

Top Methods Results Discussion Author & Article Info References

Our findings confirm and extend previous data linking social class to cardiovascular outcomes, including functional recovery after myocardial infarction. Patients of low social class had higher clinical risk at baseline and had more adverse psychosocial profiles. However, disparities in functional recovery according to social class could not be explained by the selected demographic, clinical, or psychosocial factors examined in our study. Moreover, our results were not strongly influenced by access to health care or differential quality of care. Because patients were enrolled in a clinical trial, health status was evaluated regularly. Medical interventions were outlined in the protocol requirements for adverse events, although treatment decisions could be made by a patient's personal physician.

Recent reviews [27, 30, 46, 47] and empirical studies [48, 49] have documented the strong association between social class and cardiovascular health. Although occupation has been identified as a key indicator in some studies [48, 49], education has been the strongest and most consistent indicator in direct comparisons with measures of occupation and income [50]. We used a composite measure to try to capture the variance explained by both indicators. In addition to individual measures of social class, investigators might also consider household and neighborhood characteristics [51].

Given the powerful effect of social class on recovery that we observed, the principal challenge is the development of testable models to enhance understanding of these class differences [52]. Variables that were measured (such as smoking and previous infarction) and not measured (such as exercise and alcohol use) in our study could be among the important mechanisms by which social class may operate. Despite decades of research on the effects of social class on morbidity and mortality, our understanding of the mechanism by which social class "gets under the skin" to have effects is still limited.

One biological mechanism that can account for some of the association between social class and cardiovascular illness is atherosclerotic vascular disease; an inverse relation between social class and atherosclerosis is evident in early stages of disease and even among a healthy subgroup of men [53]. Stress and social isolation have been identified as psychosocial mechanisms to explain the association between education and infarction-related death among men in BHAT [7]. To clarify the determinants of social class differences in cardiovascular mortality and recovery, it will be necessary to better understand the interactions between clinical and psychosocial characteristics.

Our findings move beyond a focus on morbidity and mortality. We highlight factors that influence the extent of recovery for most patients who survive infarction. Previous research indicated that different factors predicted decline versus improvement in productive activity among older adults; moreover, when an individual factor contributed to both decline and improvement in activity, the magnitude of the associations varied [54]. Future research should focus on "recovery trajectories," identifying factors that might affect the rate and extent of recovery for individual patients. Attempts should also be made to identify protective factors that enable some persons to stay healthy despite the increased risk associated with low social class [55].

Using data from BHAT had many advantages. A large and socially diverse group of persons who survived infarction was followed. The data permitted a prospective evaluation and simultaneous examination of the role of demographic, clinical, and psychosocial factors on recovery after myocardial infarction.

However, our study also had some limitations. Participants in BHAT represent a select group of men; further research must be conducted to replicate these findings in and extend them to other groups, including women. In addition, we were restricted to the data collected as part of BHAT and to the measurements used in that study. For example, some researchers have suggested that the use of NYHA ratings of functional status for patients with coronary disease may have limited validity and interrater reliability. Unfortunately, the literature on the performance characteristics of this classification scheme is sparse. The NYHA ratings have been used as a standard of classification in cardiology and have a high level of face validity. Gorkin and colleagues [56] found that the NYHA ratings were reliable and internally consistent and correlated well with physician ratings of functional status; they also confirmed that patients in the highest class (NYHA I) scored significantly better on such health indicators as dyspnea, physical limitations, and health perceptions than did patients in NYHA class II or III (although they found no clear distinction between classes II and III). If misclassification had been a problem in our study, the statistical power to detect differences based on social class would have been diminished. The fact that the associations were consistent with change in NYHA ratings as a primary outcome indicates the robustness of these findings despite measurement concerns.

Rudolf von Virchow, a founder of cellular pathology, concluded more than 150 years ago that the cause of truly serious epidemics was primarily social [57]. Social class has clearly emerged as one of the most persistent predictors of illness and death from an array of diseases. Data on social class should be included as part of the case mix so that they can help form the best estimates of patients' prognosis [31]. Some routinely collected data have much less explanatory power for clinical research and clinical care. In our study, social class had an effect on myocardial infarction recovery that was similar in magnitude to the effect of cigarette smoking or a history of diabetes or hypertension. A better understanding of how social class influences health in general and recovery from myocardial infarction in particular can assist in the development and implementation of interventions to influence modifiable features (such as smoking and hypertension) associated with both low social class and poor health outcomes.

Health inequities are rooted in broader socioeconomic inequities. Unless we begin to address the primary causes of poverty and poor education, large disparities in the health of our citizens will persist [58]. Beyond broad social interventions, individual physicians can help reduce this disparity and improve the health and health care of patients with cardiovascular disease by recognizing that lower social class confers independent risk, by providing treatment recommendations that are sensitive to these differences, and by engaging in research to identify the mechanisms that explain the association between social class and health.

Presented in part during the plenary session of the Association of American Physicians, May 1996, Washington, D.C.