Coronary Heart Disease Incidence and Survival in African-American Women and Men: The NHANES I Epidemiologic Follow-up Study

Methods

The first National Health and Nutrition Examination Survey (NHANES I) collected data from 1971 to 1975 on a nationwide probability sample of the civilian noninstitutionalized population of the United States (excluding Alaska, Hawaii, and reservation lands of American Indians) aged 1 to 74 years [10, 11]. Details of the plan, sample design, response rates, and the manner in which the study was conducted are published elsewhere, as are the procedures used to obtain informed consent and maintain confidentiality [10, 11]. Elderly persons, women of childbearing age, and persons residing in poverty areas were oversampled.

The NHANES I Epidemiologic Follow-up Study is a longitudinal study of participants in NHANES I who were 25 to 74 years of age when the survey was conducted [12-17]. The personal interviews and physical and laboratory examinations of NHANES I provided the baseline data for the Follow-up Study. This analysis is based on four waves of follow-up data collection done in 1982 to 1984, 1986, 1987, and 1992. Information was obtained from an interview, medical records from health care facilities for the period between baseline and last follow-up interview, and death certificates for all decedents. A study that used both Medicare data and data from the follow-up study found that the follow-up study detected 80% of the hospital stays for coronary heart disease that were detected by either source [17].

This analysis included African-American and white persons who were 25 to 74 years of age at baseline for NHANES I. Of the 14 183 persons who were in this age group at baseline, 775 (5.5%) were lost to follow-up because they or their proxy was not interviewed during any follow-up wave or because no death certificate was available. Excluded from all analyses were 983 persons with unknown baseline history of heart disease, systolic blood pressure, serum cholesterol level, smoking status, family income, or level of education. Also excluded from analyses of coronary heart disease were 1019 persons who had a history of heart disease at baseline; these patients were defined as those who had ever been told by a physician that they had had a heart attack or heart failure or who had used any medicine, drugs, or pills for a weak heart during the 6 months before the baseline interview. After all exclusions, 11 406 persons examined at NHANES I survey locations 1 through 99 remained for analyses of coronary heart disease. The length of follow-up in the NHANES I Epidemiologic Follow-up Study for survivors who were free of coronary heart disease ranged from 8.0 to 22.1 years (median, 19.2 years).

Incident cases of coronary heart disease met at least one of the following criteria: 1) the death certificate on which the underlying or primary cause of death was coded with ICD-9 [International Classification of Diseases, Ninth Revision] codes 410 through 414 or 2) there were one or more hospital stays during the follow-up period for which any discharge diagnosis was coded with ICD-9-CM (ICD-9, Clinical Modification) codes 410 to 414. An ICD-9-CM code of 410 served to identify incident cases of acute myocardial infarction that required hospitalization among the discharge diagnoses from the first recorded hospitalization in which coronary heart disease was the discharge diagnosis. Persons who were hospitalized for coronary heart disease that was not related to acute myocardial infarction were not considered to be at risk for subsequent incident cases of acute myocardial infarction requiring hospitalization. The date of occurrence was estimated as the date of first hospital admission with a diagnosis of coronary heart disease or the date of death for persons who died of coronary heart disease and whose hospital records did not list a diagnosis of coronary heart disease. For mortality analyses, cause of death was defined by the underlying cause of death: coronary heart disease with ICD-9 codes 410 to 414.

Baseline Variables

Blood samples were obtained, and frozen serum was sent to the Centers for Disease Control and Prevention for measurement of serum total cholesterol levels [10, 18]. The baseline questionnaire on medical history included questions about selected conditions and about medications used for these conditions during the preceding 6 months [10]. Ethnic group (“race,” designated “black,” “white,” and “other”) was determined by interviewer observation. A revised race variable was created during the 1982-1984 follow-up period to resolve discrepancies between the race observed by the NHANES I interviewer and the race reported by the respondent during the 1982-1984 follow-up period. Revised race codes were based on a case-by-case adjudication. At the beginning of the baseline physical examination, the physician measured the blood pressure with the examinee seated [10]. Information on smoking status was obtained at baseline for the subsample (approximately 50% of the whole) that underwent a more detailed baseline examination, as described elsewhere [10, 11]. For the remaining persons, information on smoking status at baseline was derived from responses to follow-up questions on lifetime smoking history or was imputed [19, 20]. The validity of this approach has been documented [19, 20]. Other baseline variables were measured as described elsewhere [10].

Statistical Analysis

For the 25- to 54-, 55- to 64-, and 65- to 74-year-old age groups, approximate plots of cumulative probabilities of survival without coronary heart disease were based on Kaplan-Meier statistics calculated using the LIFETEST procedure in SAS software [21, 22]. Age-adjusted and risk-adjusted estimates of the risk for coronary heart disease in African-American persons relative to white persons were derived from Cox proportional-hazards regression models that were computed by using the PHREG procedure in SAS software [23, 24]. Separate analyses were done for groups categorized by age at baseline and sex. Results are shown for models that included age at baseline in single years and age plus other confounders, entered as continuous variables (systolic blood pressure, serum cholesterol level) or as indicator variables (history of current smoking, level of education < 12 years, family income < $5000). Modification of the effect of ethnic group according to age or education was assessed in regression models and is reported only if the modification was significant. In the incidence analyses, duration of follow-up was calculated as the time from the date of examination to the date of coronary event or, for participants who did not have a coronary event, the date of last follow-up interview or death. For analyses of coronary disease-related mortality, participants whose underlying cause of death was a condition other than coronary heart disease were censored on their date of death; survivors were censored on the date of the last follow-up interview with the participant or proxy [24].

For analysis of survival after first diagnosis of coronary heart disease, 1858 persons who were hospitalized at least once and had a discharge diagnosis of coronary heart disease (ICD-9-CM codes 410 to 414) were followed from the date of the first recorded hospitalization for which the discharge diagnosis was coronary heart disease until death or the last follow-up contact. Duration of follow-up ranged from 0 days to 20.4 years (median, 4.2 years). Kaplan-Meier curves were examined and Cox proportional-hazards regression models were computed to obtain estimates of risk for death from all causes for African-American persons relative to white persons; these models controlled for age at first hospitalization for coronary heart disease and a discharge diagnosis of acute myocardial infarction at the first hospitalization for coronary heart disease. Separate models were run for men and women, for admission ages 25 to 64 years and 65 years and older, and for all admission ages combined. Survivors were censored on the date of the last follow-up interview with the participant or proxy.

For analysis of coronary procedures, incidence rates per 1000 person-years were computed among persons who were ever hospitalized with a diagnosis of coronary heart disease. Data are presented for 1) coronary revascularization (coronary bypass surgery [ICD-9-CM code, 36.1]) or removal of coronary artery obstruction (percutaneous transluminal coronary angioplasty [ICD-9-CM code, 36.0]) and 2) cardiac catheterization (ICD-9-CM code, 37.21 to 37.23) or coronary angiography (ICD-9-CM code, 88.55 to 88.57). The age-, sex-, and risk-adjusted relative risks among African-American persons compared with white persons were computed by using Cox proportional-hazards regression analysis for coronary procedures that occurred after the date of the first recorded hospital admission for which the discharge diagnosis was coronary heart disease.

To assess the effect of the sample weights on the results, weighted Cox regression analyses were done using the PROC SURVIVAL procedure in SUDAAN [25, 26]. Results of weighted analyses were generally consistent with the conclusions derived from the unweighted estimates from SAS, which have smaller variances than SUDAAN estimates because of weighting and use of a different variance estimator. In women, for example, the relative percentage difference in weighted and unweighted risk-adjusted relative risk for coronary heart disease was 8.6% at ages 25 to 54 years and 15.7% at ages 65 to 74 years. In men, the values were −10.4% and −8.4%,respectively. Therefore, only the data from the unweighted Cox regression models are presented [26].

Results

Survival Free of Coronary Heart Disease

Table 1 shows the number of incident cases of coronary heart disease, acute myocardial infarctions, and deaths from coronary heart disease according to ethnic group, sex, and age at baseline. Table 2 shows the mean levels or prevalence of selected risk factors according to coronary heart disease status at follow-up. Figure 1 and Figure 2 show the cumulative proportion of women and men who survived free of coronary heart disease during follow-up (based on Kaplan-Meier analysis).

View larger version:

• In this window
• In a new window

Figure 1. Solid lines represent African-American persons, and dashed lines represent white persons. Probability of surviving free of coronary heart disease in women aged 25 to 74 years according to ethnic group and age at baseline.

View larger version:

• In this window
• In a new window

Figure 2. Solid lines represent African-American persons, and dashed lines represent white persons. Probability of surviving free of coronary heart disease in men aged 25 to 74 years according to ethnic group and age at baseline.

Adjusted Risks for Coronary Heart Disease, Acute Myocardial Infarction, and Coronary Death

Among women aged 25 to 74 years, a significant interaction between age and ethnic group was seen for incidence of coronary heart disease (P = 0.001). Consistent with the results of Kaplan-Meier analyses, proportional hazards analyses in women revealed a significantly higher age-adjusted risk for coronary heart disease in African-American persons than in white persons at ages 25 to 54 years (age-adjusted relative risk, 1.76 [95% CI, 1.36 to 2.29]) (Table 3). Most of the excess risk among younger African-American women was explained statistically by their higher level of standard risk factors; the risk-adjusted relative risk did not significantly differ from 1.0. At ages 65 to 74 years, however, African-American women had a significantly lower risk for coronary heart disease than white women, after controlling for multiple risk factors (Table 3). In men, African-American ethnic group was associated with an age-adjusted risk for coronary heart disease that was lower than that seen in white persons across the three age groups, significantly so at ages 55 to 64 and at ages 25 to 74 years combined (Table 3). After we controlled for multiple risk factors, the relative risks were further reduced to levels significantly below 1.00 for each age group.

In women, the age-adjusted risk for acute myocardial infarction did not significantly differ between ethnic groups, either within age groups or for all age groups combined. However, if risk factor levels in African-American women aged 25 to 74 years were equal to those in white women, African-American women would have only about two thirds of the risk for acute myocardial infarction seen in white persons (indicated by the risk-adjusted relative risk shown in Table 3). Age-adjusted risk for acute myocardial infarction was lower in African-American men than in white men at each age, significantly so for all ages combined. Adjustment for all risk factors had little effect on relative risks (Table 3).

Similar to findings for total incidence of coronary heart disease, an almost significant interaction between age and ethnic group was found for death from coronary heart disease in women (P = 0.07). Only among women aged 25 to 54 years did the age-adjusted risk for death from coronary heart disease differ significantly between African-American persons and white persons (relative risk, 2.25 [CI, 1.19 to 4.24]); most of this difference was eliminated after we controlled for other risk variables (Table 3). The risk-adjusted relative risks indicate that if risk factor levels in African-American men were equal to those in white men, African-American men would have 40% to 70% of the risk seen in white men (Table 3).

Survival after First Diagnosis of Coronary Heart Disease

Survival after the first recorded hospitalization for coronary heart disease was determined for 225 African-American persons and 1633 white persons. One hundred twenty African-American persons and 914 white persons died after the first hospitalization. A proportional hazards regression model that was done for all participants who had been hospitalized with coronary heart disease and that included ethnic group, age at hospitalization for the incident case of coronary heart disease, sex, and diagnosis of acute myocardial infarction at hospitalization for the incident case of coronary heart disease (categorized as yes or no) revealed no significant association between African-American ethnic group and death (relative risk, 1.10 [CI, 0.91 to 1.33]). To determine whether early follow-up experience was similar to the overall experience, the analysis was repeated for the first 10 years of follow-up. The results were essentially the same, a finding consistent with the robustness of this analysis.

Coronary Artery Bypass Grafting and Other Procedures

Among persons with no history of coronary heart disease at baseline, the incidence of coronary artery bypass grafting or angioplasty was 27.8 (n = 245) per 1000 person-years in white persons but only 5.1 (n = 6) per 1000 person-years in African-American persons (ratio, 5.5). The incidence of cardiac catheterization or coronary angiography was 40.3 (n = 355) per 1000 person-years in white persons and 25.6 (n = 30) per 1000 person-years in African-American persons (ratio, 1.6). For any of these procedures, the rates were 47.0 (n = 414) per 1000 person-years in white persons and 26.5 (n = 31) per 1000 person-years in African-American persons (ratio, 1.8). Among African-American persons, the age- and sex-adjusted relative risk for undergoing any coronary procedure at any time after the date of a first hospital admission for coronary heart disease (n = 1858) was 0.40 (CI, 0.16 to 0.99). The relative risk was 0.15 (CI, 0.02 to 1.05) for any type of coronary revascularization and 0.53 (CI, 0.21 to 1.34) for any cardiac catheterization or coronary angiography. Controlling for multiple risk factors and presence of acute myocardial infarction did not alter these results.

Discussion

In a national cohort that had no medical history of coronary heart disease at baseline, African-American women aged 25 to 54 years had higher age-adjusted risk for coronary heart disease and death from coronary heart disease than white women of the same ages; African-American men aged 25 to 74 years had lower age-adjusted rates of coronary heart disease and acute myocardial infarction than white men of the same ages. The increased risk in African-American women could be explained statistically by their higher risk factor levels. Among African-American men, however, the results of controlling for multiple risk factors suggested that if risk factor levels in African-American persons were reduced to levels seen in white persons, the incidence rates of coronary heart disease would be substantially lower in African-American men than in white men and would be similar to those seen more generally in the United States before the 1970s [1-4]. Controlling for body mass index in addition to other risk factors did not alter the findings (data not shown). No significant differences between ethnic groups in survival after the first hospitalization for coronary heart disease could be shown. However, the incidence of coronary procedures was markedly lower in African-American persons than in white persons.

Higher levels of high-density lipoprotein cholesterol and other protective factors in African-American persons compared with white persons may have previously reduced the relative incidence of coronary heart disease and acute myocardial infarction, resulting in much lower national rates of death from and incidence of coronary heart disease in African-American persons before the 1970s [1, 2, 4, 27, 28]. However, the rapid adoption of unhealthful lifestyles since the 1960s and slower adoption of heart-healthy behaviors in the 1970s and 1980s by African-American persons may have caused an almost complete loss of this protection in African-American women and a partial loss in African-American men by the 1990s [1, 2, 4, 9, 27, 28].

High rates of early death from competing noncoronary causes in African-American men and older African-American women could also play a role in the lower rates of coronary death in these groups. Kaplan-Meier analyses that used death as the end point (data not shown) and that were done for all-cardiovascular mortality, all-noncardiovascular mortality, and all-cause mortality revealed that African-American persons had worse survival than white persons in almost every age and sex group. The differences were larger and more consistent in women than in men (data not shown).

Results of the few cohort studies reported before 1980 have been reviewed elsewhere [1, 2]. Most of these studies indicated a lower incidence of acute myocardial infarction and coronary heart disease in African-American men than in white men and a lower or similar incidence in African-American women compared with white women. More recent reports have often focused on extended follow-up of small cohorts from the southern part of the United States [29, 30] or follow-up for mortality only in men [31-35]. Statistically significant differences between ethnic groups were inconsistently found, but the studies tended to show lower rates in African-American men than in white men [7, 29-39]. Most previous cohort studies have found that elevated systolic blood pressure and, less consistently, serum cholesterol levels, history of diabetes, and cigarette smoking were independent risk factors for coronary heart disease in African-American men, African-American women, or both [1, 2, 7, 29-39]. Lower rates of coronary procedures in African-American persons than in white persons have been reported repeatedly since this phenomenon was discussed in an early review in 1982 [1, 8, 9, 40]. Recent studies suggest underutilization of these procedures by African-American persons but have not ruled out overutilization by white persons [40]. Relatively few published data address the relation between ethnic group and rate of survival after diagnosis of coronary heart disease or acute myocardial infarction; the existing analyses have suggested that African-American persons have worse survival after acute myocardial infarction than do white persons, especially among women, and worse survival after coronary angiography [5, 37, 40-45]. More data are needed on this question.

Limitations of our study include possible bias arising from loss to follow-up and, for incidence, missing hospital data at follow-up (for example, differential underascertainment of events in African-American men could produce lower rates). Another source of possible bias is misclassification due to inaccuracy of diagnoses coded on death certificates and hospital records. Analyses for which the definition of coronary heart disease included deaths or hospitalizations with an ICD-9 code of 429.2 (“arteriosclerotic cardiovascular disease”) or an interview history of heart attack or coronary bypass surgery without supporting hospital records (260 additional cases) produced results essentially similar to those reported here. When an ICD-9 code of 429.2 was included, age-adjusted and risk-adjusted relative risks among African-American persons were slightly closer to 1.0, were generally not significant in men and women 55 years of age and older, and were slightly further from 1.0 and significant in women aged 25 to 54 years. The number of cases in some subgroups was relatively small, as reflected by the wide CIs; results of risk-adjusted analyses in these subgroups must be interpreted with caution.

In conclusion, data from the NHANES I Epidemiologic Follow-up Study revealed that African-American women aged 25 to 54 years had a higher age-adjusted risk for first diagnosis of coronary heart disease than did white women of the same ages. However, no significant difference was seen between the ethnic groups at older ages. The excess risk seen at ages 25 to 54 years was probably explained by higher levels of identified risk factors in African-American women. In men, the age-adjusted risk was lower in African-American persons at each age, significantly so for ages 55 to 64 years and for all ages combined. Results obtained after controlling for multiple risk factors suggested that if risk factor levels in African-American persons were reduced to those seen in white persons, the incidence of coronary heart disease in African-American women would be similar to that in white women and the incidence of coronary heart disease in African-American men would be no more than two-thirds of that in white men, similar to the pattern of death rates observed in the United States before the 1970s. No significant differences were seen between ethnic groups in survival after diagnosis of coronary heart disease. However, the rate of coronary procedures was markedly lower in African-American persons. Further cohort studies of more African-American persons are needed to replicate the findings for women and for survival after diagnosis of coronary heart disease and to examine the incidence of sudden cardiac death [1-9].

On the basis of these findings, clinicians should vigorously work to reduce the high levels of cardiovascular risk factors in African-American women younger than age 65 years. Physicians must also take extra care to promote the appropriate use of cardiac catheterization and coronary revascularization among African-American patients, given the repeatedly observed disparities in utilization of these procedures.

Dr. Madans: Office of the Center Director, Centers for Disease Control and Prevention, National Center for Health Statistics, 6525 Belcrest Road, Room 1140, Hyattsville, MD 20782.