Race-Gender Differences in the Association of Trait Anger with Subclinical Carotid Artery Atherosclerosis: The Atherosclerosis Risk in Communities Study

Abstract

This paper examines the association between trait anger and subclinical carotid artery atherosclerosis among 14,098 Black or White men and women, aged 48–67 years, in the Atherosclerosis Risk in Communities Study cohort, 1990–1992. Trait anger was assessed using the 10-item Spielberger Trait Anger Scale. Carotid atherosclerosis was determined by an averaged measure of the wall intimal-medial thickness (IMT) of the carotid bifurcation and of the internal and common carotids, measured by high-resolution B-mode ultrasound. In the full study cohort, trait anger and carotid IMT were significantly and positively associated (p = 0.04). In race-gender stratified analysis, the association was strongest and independent only in Black men, among whom a significant trait anger-carotid IMT relation was observed for both the overall trait anger measure (p = 0.004) and the anger reaction dimension (p = 0.001). In Black men, carotid IMT levels increased across categories of overall trait anger and anger reaction, resulting in clinically significant differences (67 μm (95% confidence interval: 23, 110) and 82 μm (95% confidence interval: 40, 125), respectively) from low to high anger. Sociodemographic, lifestyle, anthropometric, and biologic cardiovascular disease risk factors appear to mediate the relation in Black women, White men, and White women. In conclusion, these findings document disparate race-gender patterns in the association of trait anger with subclinical carotid artery atherosclerosis.

atherosclerosis, carotid arteries, continental population groups, sex, stress, psychological

There is growing evidence to indicate that anger/hostility increases a person's susceptibility to cardiovascular disease (1). Anger/hostility, a common manifestation of psychological distress, has received a great deal of research attention for its impact on cardiovascular disease and has been positively associated with premature cardiovascular disease onset (2) and mortality (3, 4), atherosclerosis (5–8), atrial fibrillation (4), and behavioral risk factors (9, 10). Anger, the emotional component of hostility, is the focus of the current investigation. Whether a person is prone to experience anger and the manner in which he or she characteristically expresses this emotion are two types of anger that are frequently investigated. Anger proneness or trait anger, as it is termed in the literature, is the predisposition to experience anger often, intensely, and for long periods of time (11). Anger-prone individuals react to numerous events in their environment with irritation, annoyance, and even full-blown rage.

In a previous Atherosclerosis Risk in Communities (ARIC) Study analysis, trait anger was positively associated with incident coronary artery disease among normotensive individuals (12); however, the precise mechanism linking anger to coronary artery disease remains unknown. There are two leading hypotheses: One is indirect, implicating deleterious health behaviors (9, 10), and the other is direct, implicating activities of the sympathetic adrenomedullary system and the hypothalamic-pituitary-adrenocortical axis and their consequential hemodynamic and neurohormonal sequelae (1). In response to psychosocial stress, excess circulating stress hormones, such as catecholamines and corticosteroids, may adversely impact blood pressure and heart rate, causing platelet aggregation and increased wall sheer stress leading to endothelial injury and arterial thrombus formation (13). Hyperresponsivity of the sympathetic adrenomedullary system and the hypothalamic-pituitary-adrenocortical axis can also catalyze a cascade of proinflammatory events, including activation of macrophages, the production of cytokines, activation of acute-phase proteins, and mast cell activation, each of which is intricately involved in atherogenesis. Thus, psychological distress may ignite a stress response and an inflammatory response, which, together, aggressively promote atherosclerosis—a disease that is now considered the result of a chronic inflammatory process (14). Increased intimal-medial thickness (IMT) is a subclinical manifestation of atherosclerosis assessable by ultrasound. Previous ARIC Study analyses have indicated a positive association between carotid IMT and prevalent cardiovascular disease (15), incident coronary artery disease (16), and incident stroke (17).

The purpose of the present analysis was to examine the cross-sectional association between trait anger and carotid IMT levels in order to elucidate a possible atherosclerotic mechanism by which trait anger and coronary heart disease are linked. Despite previous findings of a positive association between anger/hostility and subclinical carotid artery atherosclerosis (5–8), studies of this relation in large population-based samples are needed. Further, Black participants traditionally have been underrepresented in studies of the relation of psychosocial stress to atherosclerosis. An examination of this relation in the ARIC Study will help to fill the gap since the ARIC Study, a population-based, epidemiologic study, includes a large biracial cohort. We hypothesized that trait anger would be positively associated with carotid IMT, independent of established sociodemographic, behavioral, and biologic coronary heart disease risk factors. The rationale is that anger may have a direct, pathophysiologic effect on the vessels via hemodynamic and neurohormonal actions of the sympathetic adrenomedullary system and the hypothalamic-pituitary-adrenocortical axis (13). Further, we hypothesized that there would be a monotonic increase in carotid IMT levels with increased trait anger, assuming that a greater direct physiologic effect would result in a greater vascular impact.

MATERIALS AND METHODS

Study population

Participants in this analysis were members of the ARIC Study cohort, who were selected through probability sampling from the four US communities of Washington County, Maryland; suburban Minneapolis, Minnesota; Forsyth County, North Carolina; and Jackson, Mississippi. The ARIC Study is a large, population-based, prospective study of the etiology and natural history of atherosclerosis among men and women aged 45–64 years, inclusive, at baseline. Black participants were exclusively sampled in Jackson, Mississippi. A detailed description of the sampling strategy and clinic examination procedures used in the ARIC Study has been reported elsewhere (18). After baseline examinations, held from 1987 to 1989, ARIC Study participants returned for three triennial clinical examinations, and morbidity and mortality community surveillance was conducted continuously. Participants for the present analysis comprised the 14,348 ARIC Study cohort members (92.2 percent) who returned for the first clinical reexamination between 1990 and 1992 (visit 2) when the anger questionnaire was administered. Excluded from analyses were 42 participants with a racial/ethnic identity other than Black or White, 163 with incomplete responses on the anger questionnaire, and 45 Black participants from the suburban Minneapolis, Minnesota, and Washington County, Maryland, field centers because of small numbers, for a final sample of 14,098.

Assessment of trait anger

Participants were asked to complete the 10-item Spielberger Trait Anger Scale (11). Respondents rated on a Likert-type scale their typical experience with anger: almost never = 1, sometimes = 2, often = 3, and almost always = 4. Responses to each of the items were summed to yield an overall score. Internal consistency reliability coefficients for the Spielberger Scale have been reported to range from 0.81 to 0.92 (11). The Spielberger Scale correlates positively with other widely used measures of anger/hostility, such as the Buss-Durkee Hostility Inventory (r = 0.66–0.73) and the Cook-Medley Hostility Scale (r = 0.43–0.59) (11). Two subscales comprise the Spielberger Trait Anger Scale—anger temperament and anger reaction. Anger temperament is defined as the predisposition toward quick, unprovoked (or minimally provoked) anger, whereas anger reaction refers to anger aroused in response to frustration, criticism, or unfair treatment. Each subscale comprises four items, which were summed to yield an overall score. Two items from the overall trait anger scale are not included in the subscales. Internal consistency reliability coefficients have been reported to range from 0.84 to 0.89 and from 0.70 to 0.75 for the anger temperament and anger reaction subscales, respectively (11).

Assessment of carotid intimal-medial thickness

Carotid IMT was measured by high-resolution B-mode ultrasound, based on the technique validated by Pignoli et al. (19), using a Biosound 2000II-SA ultrasound system (Biosound Incorporated, Indianapolis, Indiana). Sonographers who were trained to use standardized procedures in all study centers read the ultrasound measurements. Carotid IMT estimates were defined as the average far wall thickness of the right and left sides at three different sites in 1-cm sections of the carotid bifurcation and of the internal and common carotids. Eighty-seven percent of participants were missing an IMT image for at least one carotid segment (17). Missing data were imputed by use of estimated values from regression models with the expectation-maximization algorithm (17, 20). The current analysis used the mean reader- and trend-adjusted IMT estimates, measured and imputed. A detailed report of carotid measurements has been published previously in the ARIC Study operations manual (21).

Assessment of cardiovascular disease risk factors

At the clinical examination, trained technicians used standardized protocols to make physiologic measurements and laboratory assessments. Sitting blood pressures were taken using a random-zero sphygmomanometer after 5 minutes of rest. The average of the second and third of three consecutive measurements was used to calculate systolic and diastolic blood pressure levels reported in these analyses. Hypertension was defined as diastolic blood pressure of ≥90 mmHg, systolic blood pressure of ≥140 mmHg, or use of one or more antihypertensive medications within the past 2 weeks. High-density lipoprotein (HDL) cholesterol was measured enzymatically after precipitation of low-density lipoprotein (LDL)-containing proteins using dextran-magnesium. LDL cholesterol was calculated by use of the Friedewald formula. Diabetes was defined as a fasting serum glucose level of ≥126 mg/dl, a nonfasting serum glucose level of ≥200 mg/dl, and/or a history of diabetes, insulin therapy, or oral hypoglycemic medication use. Fibrinogen was measured by the Clauss method. The waist/hip ratio was the ratio of waist girth, abdominal circumference measured at the umbilicus, to hip girth, measured at the maximal gluteal protrusion. Cigarette smoking, race/ethnicity, and level of educational attainment were self-reported.

Statistical analyses

Means and percentages were used to describe the population characteristics by coronary heart disease risk factors. The association between trait anger and carotid IMT was determined by use of multiple linear regression analysis in which the exposure variable, the overall trait anger score, was entered into the model as a linear variable. The dependent variable was carotid IMT. We began with the following potential confounders: age, race-center, educational level, diabetes, hypertension, waist/hip ratio, fibrinogen, HDL- and LDL-cholesterol, history of smoking, and pack-years of smoking. The trait anger–carotid IMT relation was examined for its two-way interactions with race-center and gender individually and with its three-way interactions with race-center and gender together. We conducted race-gender–specific regression models in order to identify potential sociodemographic, behavioral, anthropometric, and biologic mediators of the trait anger–carotid IMT relation. A factor was considered a mediator if there was a substantial reduction in the amount of variance in carotid IMT that was accounted for by trait anger (partial r²), after adjusting for the potential mediator. The final model for each race-gender subgroup excluded those variables considered as mediators. In order to examine differences in the mean carotid IMT levels at different levels of anger, trait anger scores 10–14 were categorized as low, 15–21 as moderate, and 22–40 as high, using cutpoints from previous ARIC Study analyses (12). Anger reaction scores 4–10 were categorized as low and 11–16 as high, using values at the 85th percentile as cutpoints. The p value was set at 0.05 for the main effect and 0.10 for interactive effects. All p values are two sided.

RESULTS

The overall mean trait anger score in the population was 16.1 (standard deviation (SD): 3.8). In the race-gender subgroups, the mean anger score was 15.6 (SD: 4.1) in Black men and 16.0 (SD: 4.1) in Black women. Among White participants, these scores were 16.4 (SD: 3.8) and 15.9 (SD: 3.5) in men and women, respectively. Further, mean carotid IMT levels were 801 μm (SD: 190) and 742 μm (SD: 168) for Black men and women, respectively; mean carotid IMT levels were 801 μm (SD: 210) and 706 μm (SD: 170) for White men and women, respectively. Tables 1 and 2 present race-gender–specific coronary heart disease risk factor profiles by trait anger categories. Black men and women who reported high trait anger, compared with their peers who reported low trait anger, were more likely to have less than a high school education and were more likely to be current smokers who smoked more cigarettes in a lifetime, on average. White men and women who reported high trait anger, compared with those reporting low trait anger, were younger, had higher waist/hip ratios, and were more likely to have less than a high school education. In addition, they were more likely to be former smokers who smoked more cigarettes in a lifetime, on average. In none of the race-gender subgroups was trait anger related to lipid, blood pressure, or fasting glucose levels (results not shown). With the exception of White men (p = 0.03), trait anger was also not related to fibrinogen in the subgroups. White men who reported moderate or high trait anger had higher fibrinogen levels than did their counterparts who reported low trait anger.

TABLE 1.

Distribution of population characteristics across categories of trait anger, Black participants, Atherosclerosis Risk in Communities Study, United States, 1990–1992

	Spielberger trait anger groups
	Low (scores 10–14)	Moderate (scores 15–21)	High (scores 22–40)
Participants, no. (%)
Men	572 (45.3)	590 (46.7)	101 (8.0)
Women	853 (39.2)	1,128 (51.9)	194 (8.9)
Age in years, mean (standard deviation)
Men	56.7 (6.1)	56.1 (5.7)	56.8 (6.0)
Women	56.6 (5.8)	55.8 (5.6)	56.4 (5.8)**
Education (%)**
<High school
Men	42.5	36.7	60.4
Women	39.2	35.6	51.6
High school
Men	25.6	26.7	19.8
Women	29.4	30.1	27.3
>High school
Men	31.9	36.6	19.8
Women	31.4	34.3	21.1
Cigarette smoking in pack-years, mean (standard deviation)
Men	17.6 (22.7)	18.8 (22.7)	29.0 (31.6)***
Women	6.0 (12.7)	7.7 (14.3)	12.0 (21.6)***
Smoking status (%); men*, women***
Current
Men	30.2	37.2	43.6
Women	14.0	23.3	32.0
Former
Men	41.8	36.7	36.6
Women	21.8	24.4	25.8
Never
Men	28.0	26.0	19.8
Women	64.2	52.3	42.3
Waist/hip ratio, mean (standard deviation)
Men	0.944 (0.051)	0.945 (0.053)	0.953 (0.064)
Women	0.908 (0.078)	0.913 (0.076)	0.919 (0.077)
Hypertensive (%)
Men	51.0	53.2	58.2
Women	57.5	56.3	61.1
Diabetic (%)
Men	23.0	23.2	30.7
Women	25.5	25.7	32.5

*

p < 0.05;

**

p < 0.01;

***

p < 0.0001.

TABLE 2.

Distribution of population characteristics across categories of trait anger, White participants, Atherosclerosis Risk in Communities Study, United States, 1990–1992

	Spielberger trait anger groups
	Low (scores 10–14)	Moderate (scores 15–21)	High (scores 22–40)
Participants, no. (%)
Men	1,673 (33.3)	2,888 (57.5)	464 (9.2)
Women	2,083 (37.0)	3,175 (56.3)	377 (6.7)
Age in years, mean (standard deviation)
Men	58.0 (5.6)	57.5 (5.7)	57.4 (5.8)*
Women	57.5 (5.7)	56.6 (5.5)	55.9 (5.7)***
Education (%)**
<High school
Men	16.5	15.8	24.4
Women	15.8	14.1	20.4
High school
Men	38.8	39.7	38.6
Women	49.9	52.5	54.1
>High school
Men	44.6	44.5	37.1
Women	34.3	33.4	25.5
Cigarette smoking in pack-years, mean (standard deviation)
Men	19.4 (23.3)	25.2 (25.4)	32.0 (27.3)***
Women	9.2 (16.3)	12.5 (18.3)	16.4 (20.7)***
Smoking status (%)***
Current
Men	17.5	23.0	26.1
Women	16.8	22.9	30.5
Former
Men	48.9	53.3	58.8
Women	27.9	31.3	34.8
Never
Men	33.7	23.6	15.1
Women	55.3	45.8	34.8
Waist/hip ratio, mean (standard deviation)
Men	0.966 (0.054)	0.974 (0.052)	0.982 (0.052)***
Women	0.882 (0.089)	0.887 (0.084)	0.904 (0.082)***
Hypertensive (%)
Men	31.1	31.5	33.8
Women	28.3	27.6	27.3
Diabetic (%)
Men	13.2	13.7	16.6
Women	10.1	9.7	11.9

*

p < 0.05;

**

p < 0.01;

***

p < 0.0001.

In the full ARIC Study cohort, multivariate-adjusted (age, educational level, diabetes, hypertension, waist/hip ratio, fibrinogen, HDL- and LDL-cholesterol, history of smoking, and pack-years of smoking) regression analyses showed a significant positive relation between the linear trait anger term and carotid IMT (p = 0.04) (R² = 0.21). For each one-unit increment in trait anger (approximately 1 SD), carotid IMT was higher by 1 μm (95 percent confidence interval (CI): 0, 2). Heterogeneity of association was observed for the three-way interactions with race-center and gender (p = 0.08). Thus, subsequent analyses were simultaneously stratified according to race and gender.

After full covariate adjustment, the trait anger–carotid IMT relation was strong and independent only in Black men (for overall trait anger: p = 0.004, R² = 0.15; for anger reaction: p = 0.001, R² = 0.15). In Black men, educational level and cigarette smoking met the criteria for potential mediators. However, the proportion of variance in carotid IMT was not substantially changed after controlling for these variables, suggesting that they are not mediators. Educational level and cigarette smoking also met the criteria for potential mediators in Black women. After controlling for these factors, partial r² decreased by 25 percent and 50 percent, respectively, suggesting that they mediate the trait anger–carotid IMT relation in Black women. In White men, educational level, cigarette smoking, waist/hip ratio, and fibrinogen were potential mediators. There was no reduction in partial r² after controlling for educational level, but partial r² decreased substantially when cigarette smoking, waist/hip ratio, and fibrinogen were controlled (100 percent, 50 percent, and 50 percent, respectively). In White women, educational level, cigarette smoking, and waist/hip ratio were potential mediators. After the effects of educational level were partialled out, the amount of variance in carotid IMT accounted for by trait anger did not change, whereas partial r² decreased by 100 percent when cigarette smoking and waist/hip ratio were controlled. Final regression models for each race-gender group are presented in table 3.

TABLE 3.

Parameter estimates and 95% confidence intervals for final regression models of mean carotid intimal-medial thickness on trait anger, by race-gender, Atherosclerosis Risk in Communities Study, United States, 1990–1992

	Parameter estimate (μm)	95% confidence interval	p value for model
Black men (n = 1,263)
Final model (overall trait anger): age, education, history of smoking, pack-years of smoking, waist/hip ratio, diabetes, hypertension, HDL*- and LDL*-cholesterol, fibrinogen	4	1, 7	0.004
Final model (anger reaction): age, education, history of smoking, pack-years of smoking, waist/hip ratio, diabetes, hypertension, HDL- and LDL-cholesterol, fibrinogen	8	3, 12	0.001
Final model (anger temperament): age, education, history of smoking, pack-years of smoking, waist/hip ratio, diabetes, hypertension, HDL- and LDL-cholesterol, fibrinogen	5	−1, 11	0.10
Black women (n = 2,175)
Final model (overall trait anger): age, waist/hip ratio, diabetes, hypertension, HDL- and LDL-cholesterol, fibriongen	2	0, 4	0.03
White men (n = 5,025)
Final model (overall trait anger): age, diabetes, hypertension, HDL- and LDL-cholesterol	2	1, 3	0.007
White women (n = 5,635)
Final model (overall trait anger): age, diabetes, hypertension, HDL- and LDL-cholesterol, fibrinogen	1	0, 2	0.04

*

HDL, high density lipoprotein; LDL, low density lipoprotein.

An examination of race-gender–specific mean carotid IMT levels by overall trait anger and anger reaction categories indicated 67-μm (95 percent CI: 23, 110) (figure 1) and 82-μm (95 percent CI: 40, 125) differences in the magnitude of carotid IMT from low to high anger, respectively, among Black men after full covariate adjustment. The difference for overall trait anger was markedly greater than the differences observed in White men (5 μm (95 percent CI: −13, 22)), White women (11 μm (95 percent CI: −10, 31)), and Black women (0 μm (95 percent CI: −28, 28)) after full covariate adjustment (figures 1 and 2).

FIGURE 1.

Multivariate-adjusted carotid intimal-medial thickness means (95% confidence interval) across levels of trait anger, Black men and White men, Atherosclerosis Risk in Communities Study, United States, 1990–1992.

FIGURE 2.

Multivariate-adjusted carotid intimal-medial thickness means (95% confidence interval) across levels of trait anger, Black women and White women, Atherosclerosis Risk in Communities Study, United States, 1990–1992.

DISCUSSION

This analysis indicated that trait anger was positively associated with subclinical carotid artery atherosclerosis among men and women in the ARIC Study cohort. However, heterogeneity of association was observed. The relation was strongest and independent only in Black men, among whom a significant association between trait anger and carotid IMT was observed for overall trait anger and anger reaction. In Black women and White men and women, sociodemographic, lifestyle, anthropometric, and biologic risk factors appeared to mediate the trait anger–carotid IMT link. Further, in Black men, carotid IMT levels increased monotonically with greater overall trait anger and anger reaction, resulting in clinically significant differences from low to high trait anger.

An explanation for the racial/ethnic differences in our findings is not readily apparent. Heterogeneity of the psychosocial stress–carotid IMT association by race/ethnicity has been reported in at least one other study (22). In that study, psychological distress and unfair treatment were associated with carotid IMT in Black women but not in White women. The authors asserted that the cardiovascular consequences of psychosocial stress may be more deleterious in Black individuals and may occur earlier in them than in Whites (22). The former assertion seems relevant to our findings, particularly given similar trait anger levels between Black and White participants. Moreover, in response to psychological distress, Black individuals, compared with Whites, may be more “vascular” responders, as evidenced by their tendency to exhibit more alpha-adrenergic activation (i.e., increased peripheral vasoconstriction) than cardiac activity (i.e., increased heart rate and cardiac output) (23, 24). Additionally, Black individuals have been found to have attenuated endothelium-dependent vasodilation, as evidenced by reduced nitric oxide availability (25, 26). Less nitric oxide can lead to smooth muscle proliferation, which promotes wall thickness and thus reduced vasodilation.

Another consideration is the sociocultural context in which the trait anger–carotid IMT association takes place and the impact of that contextual reality on subclinical atherosclerosis. A large proportion of Black Americans live with double jeopardy, belonging to a racial/ethnic minority group and having low socioeconomic status. As such, they may experience stressors unique to these two social positions (27, 28). In this analysis, 41 percent of Black men had a formal educational level that was below high school compared with 17 percent of White men. Conversely, 59 percent of Black men had an educational level that was at or above high school compared with 83 percent of White men. The cumulative toll exacted by the stresses and strains of everyday life, the stress that is associated with being a member of a racial/ethnic minority group, and that which is associated with low socioeconomic status may combine synergistically to produce particularly damaging cardiovascular consequences in Black men.

A third consideration is the interrelations among anger/hostility, blood pressure, hypertension, and carotid IMT, since anger/hostility is positively associated with elevated blood pressure levels (29), hypertension (30), and carotid IMT (5–8). Hypertension is a risk factor for greater carotid IMT (31), and both of these conditions were observed in greater proportions in Blacks in the current analyses. Moreover, Blacks typically have a longer, more severe hypertension trajectory, and it is possible that these unmeasured dimensions of high blood pressure were contributing factors but were not captured in the definition of hypertension used in our analysis.

We did not find an independent association between trait anger and carotid IMT among White men. Our findings differ from those of Julkunen et al. (8) and Bleil et al. (5) who reported independent anger–carotid IMT associations in Finnish men and in a sample of predominantly White (80 percent) American men. Differences in anger measurement might have accounted for the divergent findings relative to the study by Julkunen et al., since they did not use the Spielberger anger scales. Differences in sample characteristics might have contributed to the divergent findings relative to the study by Bleil et al. The sample in that report comprised mostly untreated hypertensives (86 percent), whereas the men in our study were unselected with respect to hypertensive or medication status. Given the positive associations between hypertension and carotid IMT (31) and between anger/hostility and hypertension (30), hypertensive men who had high anger might have been particularly predisposed to increased IMT levels. In addition, the average age of participants in the studies by Julkunen et al. and Bleil et al. (56 years and 54 years, respectively) was somewhat lower than that of White men in our study (57.7 years).

In our analysis, educational level and cigarette smoking appeared to mediate the trait anger–carotid IMT relation in Black women, whereas cigarette smoking and the waist/hip ratio were possible mediators in White women. Troxel et al. (22) observed that cardiovascular disease risk factors, namely, systolic blood pressure, HDL-cholesterol, and the waist/hip ratio, partially mediated the psychosocial stress–carotid IMT relation in Black women in their study. Raikkonen et al. (7) and Matthews et al. (6) reported positive associations that were also not independent between trait anger and carotid IMT in women. Raikkonen et al. found evidence for possible mediation by metabolic risk factors. In the study by Matthews et al., mode-of-anger expression (e.g., anger in) and hostile attitudes, rather than trait anger, predicted carotid IMT independently and prospectively. In both studies, participants were predominantly White (90 percent) and the Spielberger anger scales were used. Similar to the men, the average age of participants in the studies by Raikkonen et al. and Matthews et al. was somewhat lower (50.6 years and 47.3 years, respectively) than that among White (56.9 years) and Black (56.1 years) women in our study.

The results of mediational analyses suggest a causal pathway by which trait anger may be associated with carotid IMT in Black women and in White men and women, lending support to a behavioral mechanism. In these subgroups, psychosocial stress may lead to maladaptive coping strategies, such as unhealthy eating habits, physical inactivity (leading to excess body weight), and cigarette smoking, each of which has deleterious cardiovascular consequences (9, 10). In addition, angry individuals who have low educational levels may not be motivated to seek health information or to adhere to the prevailing guidelines for health promotion and disease prevention.

It is important to note the limitations and strengths of this analysis. Our findings are limited by their cross-sectional nature and thus the inherent temporal ambiguity in them. Therefore, they do not permit us to answer the question of whether trait anger preceded higher carotid IMT levels or vice versa, as the observed levels of trait anger existed at the time of the carotid IMT measurements. Further, our study assessed the extent to which participants experienced frequent and intense anger only, and therefore the degree to which mode-of-anger expression (e.g., holding in angry feelings or directing them outward) is associated with subclinical carotid artery atherosclerosis in this cohort cannot be determined from these data. For example, we did not find an independent association between trait anger and carotid IMT in White women, and it is possible, as observed in a previous study (6), that mode-of-anger expression is more predictive of atherosclerosis in this subgroup. Further, the assessment of trait anger relied on self-report, which is subject to a social desirability response bias. Such a bias may underestimate the true levels of trait anger in this population. Our findings fill a gap in the existing literature, primarily as the result of the inclusion of a large, population-based sample of both Black and White men and women. They document disparate race-gender patterns in the association of trait anger with subclinical carotid artery atherosclerosis and suggest different causal pathways.

In conclusion, the results of these analyses indicate that trait anger is independently associated with carotid IMT in Black men, whereas behavioral and social factors appear to mediate the association in Black women and in White men and women. These findings complement the existing literature on the association between psychosocial stress and subclinical carotid artery atherosclerosis.

Abbreviations

Abbreviations
• ARIC

  Atherosclerosis Risk in Communities

• CI

  confidence interval

• HDL

  high-density lipoprotein

• IMT

  intimal-medial thickness

• LDL

  low-density lipoprotein

• SD

  standard deviation

The ARIC Study is carried out as a collaborative study supported by National Heart, Lung, and Blood Institute contracts N01-HC-55015, N01-HC-55016, N01-HC-55018, N01-HC-55019, N01-HC-55020, N01-HC-55021, and N01-HC-55022.

The authors thank the staff in the ARIC Study for their important contributions.

Conflict of interest: none declared.