Perceived Discrimination and Coronary Artery Obstruction
We used data from the Cardiac Decision Making Study, an observational cohort study of white and black veterans who had a cardiac nuclear imaging study performed between August 1999 and January 2001 at 1 of 5 Department of Veterans Affairs (VA) Medical Centers with on-site catheterization laboratories. Overall, 2,335 of the 5,278 patients who were screened had a positive nuclear imaging study. Of these, 981 patients were excluded because they were unable to be contacted (n = 456), had a cardiac procedure in the preceding 6 months (n = 209), were not black or white (n = 102), had impaired mental status (n = 78), were in another study that would influence treatment of their coronary artery disease (n = 32), or other reasons (eg, impaired hearing, patient was a nonveteran, nuclear imaging study was performed for compensation and pension evaluation; n = 104). A total of 1,025 of the 1,354 eligible veterans agreed to participate, of which 793 (629 whites and 164 blacks) had complete data. We also collected coronary anatomy data in a subset of 311 patients (259 whites and 52 blacks) who had coronary angiography within 90 days of their imaging study.
Sociodemographic Information Patients self-reported race, age (in years), and education (0, <12 years of education; 1, ≥12 years). Only patients who responded that they were white/Caucasian (coded as 0) or black/African American (coded as 1) were included in the study.
Clinical Variables Trained nurses abstracted clinical information, including diagnosis of hypertension and diabetes, current smoking status, prior revascularization, and prior myocardial infarction. These variables were coded as 1 if the conditions were present and 0 if they were not. We also collected data on a history of a diagnosis of hypercholesterolemia and if the participants were on antiplatelet or lipid-lowering therapy.
Perceived Discrimination Perceived discrimination was assessed with an adaptation of a previously published 7-item measure asking if the participant has ever been treated unfairly, been kept from doing something, or been made to feel inferior because of his or her race, ethnic group, or skin color in several domains, including (1) school, (2) getting a job, (3) at work, (4) getting medical care, (5) getting housing, (6) from the police or in the courts, or (7) on the street or in a public setting. Because our patient population consisted of military veterans, we adapted the original measure by breaking the medical care question into "getting medical care at the VA" and "getting medical care outside the VA," and added "in the military" as an additional domain. The "yes" responses were counted to create a perceived discrimination score. Overall, the 9-item measure had good internal reliability among white (α = .75) and black (α = .88) participants. We also conducted the analyses using the original 7-item scale and found the same pattern of results, so only results using the 9-item scale are reported.
Psychosocial VariablesNegative affect was assessed using a 9-item version of the emotionality scale of the Eysenck Personality Inventory. Participants responded "yes or "no" to items such as "Are you often uneasy, feeling that there is something you want without knowing it." The number of "yes" responses was counted to create a negative affect score. Optimism was assessed using a 10-item measure that asked participants to rate their agreement on items such as "In uncertain times, I usually expect the best" on a scale of 1 (strongly disagree) to 5 (strongly agree). Responses were summed to create an optimism score. Social support was assessed using a validated 4-item scale, which asked about satisfaction with family relationships, frequency of social contact with friends and relatives, and frequency of contact with someone the patient trusts and can confide in. Scores reflect the number of items to which the respondent answered "yes." Religiosity was assessed using the sum of 3 items asking about how often patients attended religious services; watched or listened to religious programs on TV or radio; or prayed, meditated, or studied the Bible or other religious text (0 = never/almost never, 7 = daily/more often).
Risk of Severe Coronary Obstruction We categorized risk of severe coronary obstruction using a modification of the methods of Bateman et al, described in detail elsewhere. Briefly, a board-certified general internist and a cardiology fellow classified the severity of each nuclear imaging study based on review of the official report, which did not include information on patient race. We considered patients with reversible lesions in the distribution of left anterior descending coronary artery or in both the right coronary artery and left circumflex artery to be at high risk, along with patients with increased lung uptake or transient ischemic dilation with exercise or pharmacologic stress. We considered patients with reversible lesions in just 1 of the right coronary artery or left circumflex artery to be at moderate risk and patients whose defects were very small or minimally reversible to be at low risk. For the current analysis, we collapsed patients into 2 categories; patients at moderate or low risk of severe coronary obstruction and those at high risk.
We also examined severity of coronary artery obstruction among participants who underwent coronary angiography. We classified coronary obstruction as severe if there was a stenosis of ≥70% in the left main coronary artery or all 3 major coronary systems. We classified nonsevere obstruction as moderate if there was at least 70% obstruction of the proximal left anterior descending artery. We defined mild obstruction as at least 1 coronary artery obstruction >70%, but not moderate or severe. We classified obstruction as none if there was no obstruction of >70%. For the current analysis, we collapsed obstruction into 2 categories: moderate/severe and mild/none.
We first examined race differences in all study variables using t tests or χ tests. We then used hierarchical logistic regression models to examine the association between perceived discrimination and risk of severe coronary obstruction separately for blacks and whites. We chose to stratify the analyses by race because the experiences and consequences of racial discrimination are likely to be qualitatively different for blacks and whites, and the analytic strategies of including race as a covariate or attempting to adjust for variables confounded with race often can produce results that obscure racial differences. Stratifying analyses by race is a recommended analytic technique that treats race as a "marker for differential experiences and exposures" (p. 303) and is less prone to mask important race differences.
Each hierarchical model included 3 steps. Step 1 included sociodemographic and clinical characteristics that are known to differ by race or were associated with race in our sample (ie, age, education, hypertension, diabetes, prior revascularization, prior myocardial infarction, and current smoking status). We tested the impact of including additional clinical variables (receipt of antiplatelet or lipid lowering therapy, hypercholesterolemia), but they were not significant, so we omitted them from the final model. In the second step, we added several psychosocial variables that are related to cardiovascular health (ie, negative affect, optimism, social support, and religiosity). We added perceived discrimination in the final step. We repeated these steps for the analysis of coronary obstruction in the subsample of veterans who underwent coronary angiography.
The predictive power of the models was examined by use of the area under the receiver operating characteristic curve (ie, the c statistic). Values near 0.50 reflect a model with no apparent accuracy, values between 0.70 and 0.80 are considered good, values between 0.80 and 0.90 are considered excellent, and values above 0.90 are considered outstanding.
This research was supported by the Department of Veterans Affairs, Veterans Health Administration, Health Services Research and Development Service (ECV 970–022 and RCS 02–066, NR Kressin, PI); the American Heart Association and the Pharmaceutical Roundtable (9970113N, NR Kressin, PI); Veterans Affairs Health Services Research and Development postdoctoral fellowship (BJ Ayotte); Veterans Affairs Health Services Research and Development Career Development Program (RCD 06–287, LRM Hausmann). The authors are solely responsible for the design and conduct of this study, all study analyses, and drafting and editing of the manuscript. The views expressed in this article are those of the authors and do not necessarily represent the views of the Department of Veterans Affairs.
Methods
Study Population
We used data from the Cardiac Decision Making Study, an observational cohort study of white and black veterans who had a cardiac nuclear imaging study performed between August 1999 and January 2001 at 1 of 5 Department of Veterans Affairs (VA) Medical Centers with on-site catheterization laboratories. Overall, 2,335 of the 5,278 patients who were screened had a positive nuclear imaging study. Of these, 981 patients were excluded because they were unable to be contacted (n = 456), had a cardiac procedure in the preceding 6 months (n = 209), were not black or white (n = 102), had impaired mental status (n = 78), were in another study that would influence treatment of their coronary artery disease (n = 32), or other reasons (eg, impaired hearing, patient was a nonveteran, nuclear imaging study was performed for compensation and pension evaluation; n = 104). A total of 1,025 of the 1,354 eligible veterans agreed to participate, of which 793 (629 whites and 164 blacks) had complete data. We also collected coronary anatomy data in a subset of 311 patients (259 whites and 52 blacks) who had coronary angiography within 90 days of their imaging study.
Data
Sociodemographic Information Patients self-reported race, age (in years), and education (0, <12 years of education; 1, ≥12 years). Only patients who responded that they were white/Caucasian (coded as 0) or black/African American (coded as 1) were included in the study.
Clinical Variables Trained nurses abstracted clinical information, including diagnosis of hypertension and diabetes, current smoking status, prior revascularization, and prior myocardial infarction. These variables were coded as 1 if the conditions were present and 0 if they were not. We also collected data on a history of a diagnosis of hypercholesterolemia and if the participants were on antiplatelet or lipid-lowering therapy.
Perceived Discrimination Perceived discrimination was assessed with an adaptation of a previously published 7-item measure asking if the participant has ever been treated unfairly, been kept from doing something, or been made to feel inferior because of his or her race, ethnic group, or skin color in several domains, including (1) school, (2) getting a job, (3) at work, (4) getting medical care, (5) getting housing, (6) from the police or in the courts, or (7) on the street or in a public setting. Because our patient population consisted of military veterans, we adapted the original measure by breaking the medical care question into "getting medical care at the VA" and "getting medical care outside the VA," and added "in the military" as an additional domain. The "yes" responses were counted to create a perceived discrimination score. Overall, the 9-item measure had good internal reliability among white (α = .75) and black (α = .88) participants. We also conducted the analyses using the original 7-item scale and found the same pattern of results, so only results using the 9-item scale are reported.
Psychosocial VariablesNegative affect was assessed using a 9-item version of the emotionality scale of the Eysenck Personality Inventory. Participants responded "yes or "no" to items such as "Are you often uneasy, feeling that there is something you want without knowing it." The number of "yes" responses was counted to create a negative affect score. Optimism was assessed using a 10-item measure that asked participants to rate their agreement on items such as "In uncertain times, I usually expect the best" on a scale of 1 (strongly disagree) to 5 (strongly agree). Responses were summed to create an optimism score. Social support was assessed using a validated 4-item scale, which asked about satisfaction with family relationships, frequency of social contact with friends and relatives, and frequency of contact with someone the patient trusts and can confide in. Scores reflect the number of items to which the respondent answered "yes." Religiosity was assessed using the sum of 3 items asking about how often patients attended religious services; watched or listened to religious programs on TV or radio; or prayed, meditated, or studied the Bible or other religious text (0 = never/almost never, 7 = daily/more often).
Risk of Severe Coronary Obstruction We categorized risk of severe coronary obstruction using a modification of the methods of Bateman et al, described in detail elsewhere. Briefly, a board-certified general internist and a cardiology fellow classified the severity of each nuclear imaging study based on review of the official report, which did not include information on patient race. We considered patients with reversible lesions in the distribution of left anterior descending coronary artery or in both the right coronary artery and left circumflex artery to be at high risk, along with patients with increased lung uptake or transient ischemic dilation with exercise or pharmacologic stress. We considered patients with reversible lesions in just 1 of the right coronary artery or left circumflex artery to be at moderate risk and patients whose defects were very small or minimally reversible to be at low risk. For the current analysis, we collapsed patients into 2 categories; patients at moderate or low risk of severe coronary obstruction and those at high risk.
We also examined severity of coronary artery obstruction among participants who underwent coronary angiography. We classified coronary obstruction as severe if there was a stenosis of ≥70% in the left main coronary artery or all 3 major coronary systems. We classified nonsevere obstruction as moderate if there was at least 70% obstruction of the proximal left anterior descending artery. We defined mild obstruction as at least 1 coronary artery obstruction >70%, but not moderate or severe. We classified obstruction as none if there was no obstruction of >70%. For the current analysis, we collapsed obstruction into 2 categories: moderate/severe and mild/none.
Statistical Analyses
We first examined race differences in all study variables using t tests or χ tests. We then used hierarchical logistic regression models to examine the association between perceived discrimination and risk of severe coronary obstruction separately for blacks and whites. We chose to stratify the analyses by race because the experiences and consequences of racial discrimination are likely to be qualitatively different for blacks and whites, and the analytic strategies of including race as a covariate or attempting to adjust for variables confounded with race often can produce results that obscure racial differences. Stratifying analyses by race is a recommended analytic technique that treats race as a "marker for differential experiences and exposures" (p. 303) and is less prone to mask important race differences.
Each hierarchical model included 3 steps. Step 1 included sociodemographic and clinical characteristics that are known to differ by race or were associated with race in our sample (ie, age, education, hypertension, diabetes, prior revascularization, prior myocardial infarction, and current smoking status). We tested the impact of including additional clinical variables (receipt of antiplatelet or lipid lowering therapy, hypercholesterolemia), but they were not significant, so we omitted them from the final model. In the second step, we added several psychosocial variables that are related to cardiovascular health (ie, negative affect, optimism, social support, and religiosity). We added perceived discrimination in the final step. We repeated these steps for the analysis of coronary obstruction in the subsample of veterans who underwent coronary angiography.
The predictive power of the models was examined by use of the area under the receiver operating characteristic curve (ie, the c statistic). Values near 0.50 reflect a model with no apparent accuracy, values between 0.70 and 0.80 are considered good, values between 0.80 and 0.90 are considered excellent, and values above 0.90 are considered outstanding.
This research was supported by the Department of Veterans Affairs, Veterans Health Administration, Health Services Research and Development Service (ECV 970–022 and RCS 02–066, NR Kressin, PI); the American Heart Association and the Pharmaceutical Roundtable (9970113N, NR Kressin, PI); Veterans Affairs Health Services Research and Development postdoctoral fellowship (BJ Ayotte); Veterans Affairs Health Services Research and Development Career Development Program (RCD 06–287, LRM Hausmann). The authors are solely responsible for the design and conduct of this study, all study analyses, and drafting and editing of the manuscript. The views expressed in this article are those of the authors and do not necessarily represent the views of the Department of Veterans Affairs.