Unfortunately, these studies lacked sufficiently detailed clinical information to determine whether these differences were a function of differing clinical characteristics or even whether the rates were higher than appropriate in whites, lower than appropriate in African-Americans, or both. For example, African-Americans have a significantly lower rate of hospital discharge with acute myocardial infarction as the primary diagnosis [8]. Although ischemic heart disease is the leading cause of death in African-Americans as well as in whites, the effect of race on clinical course and outcomes (that is, on the natural history of ischemic heart disease) in patients with chest pain has not been well studied [9]. Thus, further information is needed to understand racial differences in the rates of cardiac procedures.

To address this need, we merged detailed clinical and resource utilization data on 3031 patients from two hospitals to measure the effect of race on the clinical presentation, the natural history of disease, and the resource utilization of patients coming to the emergency department because of acute chest pain. Our detailed patient-specific information permitted us to adjust for clinical characteristics and complications to understand the basis for any differences in overall procedure rates.

Methods

Top Methods Results Discussion Author & Article Info References

Study Design

The Multicenter Chest Pain Study, a cooperative investigation of the clinical characteristics and outcomes of emergency-department patients with acute chest pain, was conducted between December 1983 and October 1986 [9-18]. All patients 30 years or older who came to one of the participating emergency departments with a chief complaint of anterior, precordial, or left lateral chest pain that could not be explained by obvious local trauma or abnormalities on a chest radiograph were eligible. The emergency-department physician, who was usually an intern or a resident, recorded clinical data at the time of presentation, including the patient's age; sex; findings from the history, physical examination, and electrocardiogram; and results of cardiac enzyme assessments. The emergency-department physician or a research nurse recorded these data on a standardized data form that was part of the permanent medical record at a time when he or she had no knowledge of, and could not be biased by, the patient's subsequent course.

Data on race were collected prospectively in 94% of patients at the University of Cincinnati Hospital and retrospectively, by chart review, in 96% of patients at the Brigham and Women's Hospital. Patients from these two hospitals composed our study group. The term African-American is used to describe blacks residing in the continental United States who are of African descent.

Each patient visit was considered an individual event. Only the first three visits for each patient during the study period were included in the analysis. Patients who experienced cardiac arrest in the emergency department were excluded from the study.

During the study period, 4173 potentially eligible patient visits occurred at the two hospitals (Figure 1). A trained study nurse reviewed the charts of all admitted patients, recording data on cardiac enzyme levels, official discharge diagnoses, and dates and times of complications and procedures.

View larger version (19K): [in this window] [in a new window]   Figure 1. Derivation of the study population of 3031 patients from the 4173 patients who came to the emergency department with acute chest pain.

Consent was obtained from 1991 (79%) of the 2522 discharged patients. The other 531 nonconsenting, discharged patients did not differ significantly from the 1991 consenting patients in terms of gender or age.

Consenting, nonadmitted patients were retained in the study only if the investigators, who were blinded to all of the patient's emergency-department data, thought that the follow-up information was sufficient to place the patient into a diagnostic category, as defined below. Sixty-two percent of the consenting, nonadmitted patients included in the final study sample had a follow-up assessment of cardiac enzyme levels, and 66% had either assessment of cardiac enzyme levels or a follow-up visit. Those nonadmitted patients who did not return for follow-up evaluation were contacted by telephone to assess their clinical status. In such cases, patients considered to be at a higher risk for myocardial infarction because of a suspicious pain description, electrocardiographic abnormalities, or continuing symptoms were encouraged by telephone to return for an evaluation; if they were still unwilling to return, further follow-up was obtained by telephone and contact with the patient's personal physician. Patients who returned for a follow-up visit or enzyme assessment were similar to patients who received telephone follow-up with regard to sex and age, but African-Americans were more likely than whites to have enzyme assessments, electrocardiograms, or both (71% compared with 61%, P < 0.0001).

Race was identified in 1593 (97%) of the 1640 admitted patients and 1676 (94%) of the 1787 consenting, discharged patients with adequate follow-up. The 226 Hispanic and 12 Asian patients were excluded because of their small number. Therefore, the final study sample included 3031 patients: 1508 admitted patients (35% African-American, 49% male) and 1523 consenting, nonadmitted patients (55% African-American, 39% male).

Data on zip-code and insurance status were available from one of the two hospitals. Household income was estimated by matching zip codes with 1980 census data on median household income in that geographic location.

Diagnoses

A final diagnosis of acute myocardial infarction was made in both admitted and nonadmitted patients on the basis of one of the following criteria: 1) characteristic evolution of serum enzyme levels, as defined by the detection of creatine kinase MB isoenzyme levels in more than trace amounts by the qualitative electrophoretic assay or as at least 5% of an elevated total creatine kinase level and showing a typical rise and fall by the quantitative assay; by a lactic dehydrogenase isoenzyme 1 level that was greater than the isoenzyme 2 level in the absence of hemolysis or renal infarction; or, if isoenzymes were not assayed, by serial total creatine kinase levels demonstrating a typical rise and fall, with a peak value exceeding twice the usual upper limit of normal; 2) an electrocardiogram showing development of pathologic Q waves [at least 0.04 seconds in duration] and at least a 25% decrease in the amplitude of the following R wave compared with that of the emergency-department electrocardiogram; and 3) sudden unexplained death within 72 hours of presentation.

In the Chest Pain Study, the purpose of which was to define optimal management strategies for patients with acute chest pain, discharged patients with sudden death within 72 hours of admission were classified as having had an acute myocardial infarction unless contradicting data were found, because admission to an intensive or coronary care unit would have been the preferred triage strategy had this outcome been anticipated. Admitted patients who had sudden cardiac arrests and died before cardiac enzyme assessments could be done were similarly classified unless other explanations for the arrest emerged. In all of these patients, the presenting characteristics, electrocardiograms, and clinical courses were most consistent with the diagnosis of acute myocardial infarction. All patients who underwent coronary reperfusion with thrombolytic therapy showed electrocardiographic or enzymatic evidence of acute myocardial infarction before undergoing the procedure. Patients who showed creatine kinase and creatine kinase MB isoenzyme elevations after but not before an invasive procedure (for example, coronary arteriography, coronary angioplasty, or coronary artery bypass grafting) were considered not to have had infarctions, because such procedures were done at the discretion of the physicians who were caring for the patients, and acute myocardial necrosis might not have occurred if the patients had been managed conservatively.

The final diagnosis of acute ischemic heart disease was made if the patient was found to have either acute myocardial infarction or unstable angina. Unstable angina was diagnosed if 1) the patient's original emergency-department chest pain syndrome was either new or worse [in frequency, severity, or duration] than his or her chronic anginal syndrome; and 2) the diagnosis of angina was made by the senior clinician associated with the case.

Statistical Analysis

Racial distribution differed by hospital, with African-Americans accounting for 37% of the patients at Brigham and Women's Hospital and for 62% of the patients at the University of Cincinnati Hospital. All analyses were therefore adjusted for hospital location.

Frequencies of clinical characteristics, triage decisions, diagnostic outcomes, and patterns of resource utilization were compared between whites and African-Americans using the chi-square test of association for discrete variables. Continuous variables were compared using the Student t-test or the Wilcoxon rank-sum test. Within each racial group, odds ratios (with 95% CIs) for acute myocardial infarction were calculated for individual clinical features by dividing the odds of acute myocardial infarction among patients with the feature by the odds of acute myocardial infarction among patients without the feature. The odds ratios of the two racial groups were compared using a chi-square test of homogeneity. Logistic regression and linear regression analyses, for discrete and continuous outcomes, respectively, were used to measure the effect of race while controlling for possible confounding clinical characteristics. All variables included in the multivariate models are listed in the corresponding tables. Variables were divided into demographic characteristics, description of chest pain, findings on physical examination, pertinent past medical history, findings from the electrocardiographic evaluation done in the emergency department, and events during the hospital course. Missing values for confounding variables were imputed using predicted values obtained from linear and logistic regression models based on available data. In the multivariate models used to predict missing data, the confounder with missing data was the dependent variable and the remaining confounders were the independent variables.

Results

Top Methods Results Discussion Author & Article Info References

Clinical Characteristics

Of the 3031 patients in the analysis, 1374 (45%) were African-American and 1657 (55%) were white. African-Americans and whites had a mean age of 53 years and 58 years, respectively (P < 0.0001). The African-Americans were significantly more likely than whites to be female (68% compared with 47%, P < 0.0001) and less likely to have a past history of angina or myocardial infarction or to have had a previous cardiac catheterization or coronary artery bypass surgery (Table 1). Given the lower prevalence of these risk factors among African-Americans, it was not surprising that they were less likely to have an acute myocardial infarction, unstable angina, or a major complication or to die.

African-Americans who had acute myocardial infarctions generally had presenting signs and symptoms that were similar to those of whites with the same diagnosis. Thus, the odds ratios for acute myocardial infarction in relation to characteristics such as chest pain lasting 30 minutes or more; chest pain described as a pressure sensation; chest pain that radiated to the left arm, left shoulder, neck, or jaw; diaphoresis; rales on physical examination; or an electrocardiogram showing ST-segment elevation, ST-segment depression, or T-wave inversion suggestive of ischemia were all significantly greater than 1.0 in both races but not significantly different between whites and African-Americans (Table 2).

Race no longer had a statistically significant association with the diagnosis of acute myocardial infarction after adjustments were made for presenting clinical signs and symptoms using logistic regression analysis; the odds ratio for acute myocardial infarction in African-American patients was 0.77 (CI, 0.54 to 1.1) (Table 3). Race was an independent predictor of acute ischemic heart disease, with African-Americans being less likely to have a final diagnosis of acute ischemic heart disease after adjustment using multivariate analysis (odds ratio, 0.75; CI, 0.59 to 0.95). In an analogous analysis, the adjusted odds ratio of death or of experiencing a major complication for African-Americans was 0.79 (CI, 0.45 to 1.4). The only significant factors positively associated with death or major complications after adjustment for confounding variables using logistic regression analysis were the presence of rales on the admission physical examination (odds ratio, 1.8; CI, 1.1 to 3.0), acute myocardial infarction (odds ratio, 8.2; CI, 4.6 to 15), acute ischemic heart disease (odds ratio, 5.8; CI, 2.5 to 13), and age (odds ratio for 10-year increase, 1.2; CI, 1.0 to 1.5). When the emergency-department electrocardiogram was excluded from the logistic regression models, the point estimate for the odds ratio of race as a multivariate correlate of acute myocardial infarction, acute ischemic heart disease, or death or major complications remained virtually unchanged.

Medical Management

African-Americans with acute myocardial infarction tended to come to the emergency department at a later time after their chest pain began to subside compared with whites (median times to presentation after onset, 6 hours and 3 hours, respectively [P = 0.08]). In univariate analyses, African-Americans were more likely than whites to come to the emergency department more than 6 hours after the onset of chest pain (61% compared with 50%, P < 0.0001) and were less likely to be admitted to the hospital. Once admitted, both African-Americans and whites were equally likely to be admitted to the coronary care unit and to undergo cardiac catheterization (Table 4).

After adjustments were made for presenting symptoms and signs Table 5, African-Americans were still more likely than whites to come to the emergency department more than 6 hours after the onset of chest pain (odds ratio, 1.3; CI, 1.1 to 1.5) and were less likely to be admitted to the hospital (odds ratio, 0.69; CI, 0.56 to 0.84). Once admitted, race was a borderline independent predictor of triage to the coronary care unit, with an adjusted odds ratio of 0.81 (CI, 0.65 to 1.0) for African-Americans (see Table 5).

Full data were available for 324 of the 335 patients who underwent cardiac catheterization (97%): These 324 patients included 112 (35%) African-Americans and 212 (65%) whites. Race was not an independent predictor of undergoing cardiac catheterization when indicators of severity of illness, such as major complications and recurrent ischemia, were included in the analysis (odds ratio, 0.86; CI, 0.64 to 1.2).

Among the patients who underwent coronary angiography, African-Americans had less severe coronary artery disease than whites, both before and after stratification by sex (P < 0.0001) (Table 6). No significant coronary artery disease was found in 54 (48%) of the African-Americans compared with only 35 (17%) of the whites.

Although African-Americans and whites were equally likely to undergo cardiac catheterization, African-Americans were less likely to undergo coronary artery bypass surgery (see Table 6; Figure 2). For each level of severity of coronary artery disease, African-Americans had an odds ratio of less than 1.0 for undergoing coronary artery bypass surgery. The Mantel-Haenszel summary statistic over all four strata showed that African-Americans had an odds ratio of 0.36 [CI, 0.16 to 0.80] for undergoing coronary artery bypass surgery (see Figure 2). Additional adjustment for clinical factors, such as age, gender, presenting symptoms and signs, past history of cardiac catheterization or coronary artery bypass surgery, and complications, did not alter this result [odds ratio, 0.24; CI, 0.08 to 0.71] (see Table 5).

View larger version (15K): [in this window] [in a new window]   Figure 2. Odds ratios for undergoing coronary artery bypass surgery in African-Americans when compared with whites. Patients were stratified by severity of coronary artery disease. Although the 95% CIs include 1.0 for one- and three-vessel and left main coronary artery disease, the odds ratio is less than 1.0 when calculated using the Mantel-Haenszel summary statistic.

Among admitted patients, the mean length of stay (±SD) for African-Americans and whites was 7.0 ± 24 days and 7.8 ± 14 days, respectively, whereas the median length of stay was 5 days for both groups. Length of stay also did not differ according to race after adjusting for age, presenting symptoms and signs, diagnosis, procedures done, and medical complications during the hospitalization.

Adjustment for Income and Insurance

When household income, which was estimated at one hospital with available zip-code data for patients undergoing a major cardiology procedure, was divided into quartiles, African-Americans who underwent cardiac catheterization or coronary artery bypass surgery were more likely than whites to have a median income in the lower two quartiles (odds ratio, 24; CI, 8.0 to 77). At this hospital, more African-Americans were uninsured or insured by Medicaid when compared with whites (18% compared with 3%, P < 0.0001; and 21% compared with 3%, P < 0.0001, respectively). Neither income nor insurance status was an independent predictor of undergoing coronary artery bypass surgery at this hospital. After the two factors were added to the clinical data in the multivariate analysis, the correlation of race with coronary artery bypass graft surgery still met the usual criteria for statistical significance (P = 0.03). Zip-code and insurance status were not available at the second hospital site.

Discussion

Top Methods Results Discussion Author & Article Info References

Among our patients, the unadjusted rates of myocardial infarction and unstable angina differed for African-Americans and whites. However, after adjusting for multiple clinical predictors, we found no significant differences in the presentation, natural history, or final diagnosis of acute myocardial infarction between African-Americans and whites who presented to the emergency department with acute chest pain. Although both groups had similar adjusted rates of acute myocardial infarction, presentation, and hospital course, race was a significant predictor of undergoing coronary artery bypass surgery.

The lower unadjusted rates of myocardial infarction and unstable angina in African-Americans may be explained by differences in the populations presenting to the emergency department. The lower adjusted rate of acute ischemic heart disease and the less severe coronary artery disease found in the African-Americans may also reflect the differing risk of coronary artery disease among the population samples coming to the emergency department with acute chest pain. African-Americans may be more likely to use the emergency department as a primary care facility or may not be as knowledgeable about the signs and symptoms of ischemic heart disease, raising the possibility of decreased access to care for African-Americans. This disparity in disease rates may also have an effect on emergency-department physicians' triage decisions, given that African-Americans were less likely to be admitted and triaged to the coronary care unit after controlling for clinical factors in multivariate analyses.

Race was not predictive of in-hospital death or complications in our sample. Data from the Multicenter Investigation of the Limitation of Infarct Size study showed a significantly higher mortality rate for African-American women [19]. Using data from the National Hospital Discharge Survey, Roig and colleagues [20] reported a higher in-hospital fatality rate for African-Americans younger than 70 years. Other studies have shown higher postdischarge mortality rates for African-Americans with acute myocardial infarction [21]. It has also been suggested that African-Americans have a higher rate of sudden death before coming to the hospital [22]. These differences between our findings and those of other investigators may reflect differences in the populations studied. Also, out-of-hospital sudden death before presentation to the emergency department was not evaluated in our study.

An analysis of retrospectively collected data showed a delay in seeking medical attention among African-American patients with acute myocardial infarction [23]. Our prospectively collected data showed a similar trend. Delays in seeking medical attention are even more important in this era of thrombolytic therapy. It is important to note that African-Americans were less likely than whites to come to the emergency department within 6 hours of the onset of their chest pain, the generally accepted time limit in which thrombolytic therapy is known to be of benefit. Other investigators found delays in seeking medical care to be associated with increased length of stay and worsened health outcomes, perhaps because of lack of factual information regarding the symptoms of ischemic heart disease or lack of access to health care [24, 25].

The finding that the African-Americans in our study sample underwent fewer coronary artery bypass procedures is consistent with other reported data [5-8, 26-28]. Most recently, Goldberg and colleagues [7], in an analysis of national Medicare data from 1986, found lower rates of coronary artery bypass grafting procedures in African-Americans than in whites. Maynard and coworkers [6] reported higher rates of angioplasty and coronary artery bypass surgery in whites than in African-Americans in the Myocardial Infarction Triage and Intervention Registry. Similarly, Wenneker and Epstein [5] found that among patients discharged from Massachusetts hospitals in 1985 with a diagnosis of chest pain or circulatory disease, whites were more likely than blacks to undergo cardiac catheterization and coronary artery bypass procedures; odds ratios were 1.29 and 1.89 (P <.05), respectively, after adjusting for clinical status, insurance type, and income [5].

However, many analyses have been missing critical clinical data, such as those relating to severity of disease and symptoms. The Multicenter Chest Pain Study database included detailed, prospectively collected clinical data to address this problem. In our analysis, African-Americans had an odds ratio for undergoing coronary artery bypass surgery of.24 (95% CI, 0.08 to 0.71), after adjusting for clinical characteristics and for the severity of coronary artery disease found on cardiac catheterization; this finding suggests that African-Americans may have decreased access to cardiology procedures. This lack of access was not explained by socioeconomic status, because after including income and insurance status in the multivariate analysis, race continued to be a statistically significant independent predictor of undergoing coronary artery bypass surgery at one institution. The Multicenter Chest Pain Study database has many strengths, including the prospective nature of the data collection and the detailed clinical and resource utilization data available. However, the database has limitations. For example, from 1984 through 1986, the use of percutaneous transluminal coronary angioplasty was lower than it is now. In addition, the number of patients with three-vessel or left main disease was small for both African-American and white groups, and some predictors of undergoing coronary artery bypass surgery may not have been included in this analysis. Although it is difficult to judge appropriateness, the lower rate of coronary artery bypass surgery in African-Americans with left main disease and with three-vessel disease suggests that African-Americans' access to beneficial care was reduced.

Our study was limited to two hospitals, which may not reflect practices nationwide. Further, in the absence of prospective information on why patients did or did not undergo coronary artery bypass grafting, we cannot comment on the relative effect of patients' preferences; physicians' attitudes; or differences in distal coronary anatomy, left ventricular function, comorbidity, intensity of precatheterization medical therapy, or other unmeasured factors on surgical decisions. We also cannot conclude definitively whether operative rates were higher than optimal in whites, lower than optimal in African-Americans, or both [29].

Ischemic heart disease continues to be the leading cause of death in both African-Americans and whites. Therefore, it is of major importance to optimize care and resource utilization for this disease process. Our data suggest that there is a significant difference between African-Americans and whites in the rates of coronary artery bypass surgery procedures done, after adjusting for clinical factors. It is necessary to study larger numbers of patients and to understand the preferences of both patients and physicians to assess the appropriateness and reasons for these differences.