Siegman, Aron Wolfe PhD; Townsend, Susan T. PhD; Civelek, A. Cahid MD; Blumenthal, Roger S. MD
CHD = coronary heart disease, CVR = cardiovascularreactivity, HBI = hostile behavior index, Ho = Cook-MedleyHostility Scale, MI = myocardial infarction, RR = relativerisk, SES = socioeconomic status, SI = structured interview, SPECT = single-photon emission computed tomography, STAXI =Spielberger State-Trait Anger Expression Inventory.
This study has three foci: 1) the relationship between antagonistic behavior and CHD in men and women; 2) the relationship between dominance and CHD in men and women; and 3) the relationship between attitudinal hostility, as measured by the Ho scale (1), and CHD. These relationships were studied in a group of men and women referred for thallium stress testing.
Reviewers of the literature (2, 3) agree that of the two most popular methods for the assessment of hostility, 1 that is, the paper-and-pencil Ho scale (1) and SI-based (4) hostility ratings (5), the latter shows a more consistent pattern of significant positive relationships with CHD. A reanalysis of the Multiple Risk Factor Intervention Trials by Dembroski et al. (5) demonstrates that the interview-based clinical ratings of hostility predict incidence of CHD and that they do so prospectively. In their study, Dembroski et al. (5) used a matched case-control design for comparing the incidence of coronary death and MIs in participants as a function of clinical hostility ratings. Case subjects consisted of 192 patients with coronary death and/or documented nonfatal MI. Angina only was not counted as a CHD event but served as an exclusion criterion for the 384 matched control subjects.
Participants were assigned a hostile content score, an intensity of hostility score (which takes into account the use of emotion-laden words, profanity, and emphatic expressions), and a hostile style score (displays of antagonistic behavior toward the interviewer, characterized by argumentativeness, rudeness, boredom, uncooperativeness, condescension, and the like). Of the three hostility scores, only style (antagonistic behavior) was a significant risk factor for CHD. To explain these findings, Dembroski et al. (5) refer to a study by Siegman et al. (6), who suggest that it is primarily the outward expression of anger and hostility that constitutes a risk factor for CHD, in contrast to the mere experience of anger, or neurotic hostility, which is not a significant risk factor for CHD. This explains why the hostile style scores, which load on a factor that measures the outward expression of anger but not on an anger-in factor (7), were significant predictors of CHD. Dembroski et al. (5) point out that of the three SI-derived hostility scores, that is, content, intensity, and style, only the latter does not also reflect neurotic hostility.
A recent study by Haney et al. (8), using patients referred for coronary angiography, confirmed the hypothesized positive relationship between antagonistic behavior during the SI and CHD. Patients’ responses to the SI were scored by means of the interpersonal hostility assessment technique, which yields an HBI score. The HBI is essentially an elaboration of the hostile style score developed by Dembroski et al. (5) with an emphasis on subtle manifestations of antagonistic behavior toward the interviewer. In the Haney et al. (8) study, the correlation (r) between participants’ HBI scores and the severity of coronary occlusion was an impressive 0.58.
All participants in the Dembroski et al. (5) and Haney et al. (8) studies were men, so we know very little about the relationship between antagonistic behavior and CHD in women. There has been some speculation that the lower incidence of CHD in women, at least in premenopausal women, than in men might be related to women’s relatively lower levels of aggression. However, even if women are less prone to engage in physical aggression than are men, they are not necessarily also less prone than men to engage in the more subtle forms of antagonistic behavior. This latter category and its role in CHD was the focus of this study, in which we ascertained the relationship between antagonistic behavior, as assessed by the HBI, and CHD, in both male and female patients referred for thallium stress testing.
Dominance and CHD
Recent research has focused primarily on the role of anger and hostility in CHD, but other affective states and personality traits, such as dominance, may also be significant risk factors for CHD. One of the first studies to provide evidence in support of the dominance-CHD relationship involved a group of patients referred for coronary angiography (9). All patients were administered Rosenman’s (4) SI. The frequency with which patients attempted to “take the floor” from their interviewer during the SI correlated positively and significantly with patients’ angiographically determined severity of coronary artery occlusion scores. The relevance of this finding is that attempts to take the floor by interrupting one’s interviewer or speaking partner is characteristic of dominant and competitive individuals (10). However, the meaning of this finding is by no means unambiguous, because anger, too, is associated with frequent interruptions of one’s partner and with simultaneous speech (11, 12).
Using a cluster analytic procedure, Houston et al. (13, 14) identified six speech patterns, including dominating-competitive speech (characterized by short response latencies, rapid speech, frequent interruptions, and simultaneous speech) and angry-hostile speech (characterized by loud, explosive speech, a quickening pace, muscular tension, and marked hostility). Both types of speech were independent predictors of CHD in 8.5-year and 22-year follow-up examinations of the Western Collaborative Group Study. In the 22-year follow-up, both the dominating-competitive and angry-hostile speech patterns also correlated positively with all-cause mortality. Finally, in a recent study (15), the authors found that both anger-out, as measured by Spielberger’s (16) STAXI scale, and scores on a dominance questionnaire of an aggression inventory were independent predictors of CVR (blood pressure) during a math task in which male participants were provoked and harassed. The relevance of this finding is that heightened CVR seems to be a significant risk factor for CHD (17–19).
In the present study, we tried to cross-validate the finding of Houston et al. (13, 14) that dominance is a risk factor for CHD independent of antagonistic behavior and hostility in a group of patients made up about equally of men and women.
Ho and CHD
Because the Ho scale (1) was included in this study, albeit primarily to ascertain the relationship between the HBI (8) and attitudinal hostility, we also determined the relationship between participants’ Ho scores and their CHD status. Many studies, both prospective and cross-sectional, have obtained significant relationships between Ho scores and CHD, although null findings have also been reported (2, 3). The evidence (20) suggests that the Ho scale measures several relatively independent dimensions. Barefoot et al. (21), who conducted a rational analysis of the Ho scale items, identified the following four components: cynicism-mistrust, hostile attributions, hostile affect, and aggressive responding. In a recent large-scale prospective study, Barefoot et al. (22) reported that two of four Ho component scores (hostile attribution 2 and hostile affect) were each significantly associated with incidence of MI (at the p.03 level or better) but that the other two (cynicism and aggressive responding) failed to reach the p < .05 significance level. In another recent large-scale prospective study, Everson et al. (23) found that a Ho scale–derived measure of cynical mistrust was associated with an increased risk of MI but that this relationship was mediated through other risk factors. Yet another large-scale prospective study (24) also failed to show a significant relationship between cynicism, as measured by the Ho scale, and CHD. It would seem, then, that cynical mistrust, which has been identified as the core construct of the Ho scale (20), may not be an independent risk factor for CHD. There are, then, two issues with regard to the Ho-CHD relationship that need further clarification: 1) Precisely which components of the Ho scale confer risk for CHD? 2) Are these components independent risk factors for CHD, or are they mediated by other disease and lifestyle risk factors? These two questions were addressed in this study.
One hundred ninety-six patients (101 men and 95 women) were recruited among patients referred for thallium stress testing at the University of Maryland Medical Center and the Johns Hopkins Medical Institutions. Their mean age was 55.2 years (SD = 12.9 years), with no significant difference between the mean age of men and women patients. According to Hollingshead’s Index for Social Position, 27.6% of the patients were upper class; 34.7%, middle class; and 37.8%, lower class. Mean years of schooling was 12.6 (SD = 4.03), with males attending school longer than females (13.3 vs. 11.8 years;t = 2.51, p < .05). 3
Patients reported to their respective nuclear medicine departments for thallium scanning. Patients were recruited for this study after completing the first series of scans and while waiting for the second series. Only four patients refused to participate. While waiting for their second scan, participants were administered Rosenman’s (4) SI. Interviewees’ SI responses were scored for antagonistic hostility (HBI) according to the criteria developed by Haney et al. (8). Participants were also administered a background information questionnaire, the Ho scale (1), Spielberger’s (16) STAXI, and Bendig’s (25) Manifest Anxiety Scale. A subset of 97 participants were rated by their spouses for anger level by means of Spielberger’s (16) STAXI. 4
Thallium-201 Imaging Protocol
All patients underwent a symptom-limited regular or modified Bruce protocol treadmill exercise test. Patients fasted (NPO) for at least 4 hours before the study (with the exception of taking medication when indicated) to reduce splanchnic uptake of thallium. Patients were continuously monitored by 12-lead electrocardiography during and 5 minutes after the stress test. Blood pressure measurements were made at each stage of exercise. At peak exercise, 3.0 mCi (111.0 MBq) of thallium-201 chloride was administered intravenously, and exercise was continued for 40 to 60 seconds after injection.
Stress SPECT imaging was started within 10 minutes after thallium injection. Images were acquired with a triple-head gamma camera, in noncircular orbit, equipped with low-energy, ultrahigh-resolution, parallel-hold collimators. Acquisition parameters were 30 seconds per step (frame) and a 3° step-and-shoot mode in a 64 × 64 image matrix. Three-hour delayed SPECT reinjection studies were acquired in a similar manner 10 minutes after reinjection of 1.0 mCi of thallium-201.
SPECT images were then reconstructed following a standard protocol, which includes uniformity correction with a 150 million count flood, center of rotation correction and low-pass filtering, no attenuation correction, and reconstruction into 1-pixel-thick (5.3 mm) transaxial sections. Oblique angle reorientation and summation produced 1-pixel-thick horizontal and vertical long-axis and short-axis sections of the left ventricle.
For each study, consecutive horizontal and vertical long-axis and short-axis slices of the left ventricle were displayed and reviewed concurrently by two nuclear medicine physicians without previous knowledge of the patient’s clinical history, physical findings, laboratory data, wall motion studies, or cardiac catheterization results. Anterior, lateral, inferior, and septal walls on short-axis slices and the apical region on long-axis slices were evaluated for the distribution of activity using the following scoring scale: 0, normal; 1, mild; 2, moderate; and 3, severe reduction of thallium activity. For the purposes of the present study, categories 1, 2, and 3 were collapsed into a single impaired hypoperfusion category.
Patients were placed in one of three categories: 1) normal (ie, no previous CHD history, no ischemic electrocardiographic responses to exercise, and normal thallium scans), 2) documented CHD (ie, positive thallium scans or previous MI), or 3) equivocal (ie, no previous CHD history, normal scans, but failed to reach 85% of age-predicted maximum heart rate and/or showed ischemic electrocardiographic responses to exercise). Altogether, 44 patients were classified as normal; 119, as having documented CHD; and 33, as equivocal. The corresponding figures broken down by gender were 23, 55, and 17 for females and 21, 64, and 16 for males, respectively.
Randomly selected thallium scans of 40 patients were rated and categorized by a second experienced cardiologist. Interjudge reliability, as calculated by the κ statistic, was 0.89.
Measuring Antagonistic Behavior and Dominance From the SI
All participants were administered Rosenman’s (4) SI, which lasted about 15 minutes. The interviewer (S.T.T.) adapted a matter-of-fact professional demeanor that was neither overly warm nor excessively challenging. Only some of the SI questions focused on antagonistic or aggressive behavior, but in this scoring system, the emphasis is on subtle manifestations of interpersonal antagonism, not on content. Furthermore, previous research (8) indicates that HBI scores derived from questions focusing on anger-antagonism are not more strongly related to CHD than scores based on the other SI questions. HBI scores are the sum of the occurrences of four categories of antagonistic behavior: hostile withhold or evasion, direct challenge, indirect challenge, and irritability. Hostile withhold or evasion is scored if the interviewee’s response is uncooperative or evasive. Both content and vocal stylistics must be considered in scoring hostile withhold/evasion. Direct challenges involve an open confrontation with the interviewer (eg, “That’s a stupid question.”). Indirect challenge is scored if the respondent challenges the interview or interviewer by implication. Vocal stylistics is critical in this decision. Irritability is scored when a respondent’s voice stylistics reflect negative arousal indicative of anger or antagonism. Of the four HBI component scores, the direct challenge component represents the most overt form of antagonistic behavior, and the indirect challenge represents the least overt form of antagonistic behavior. The irritability component is considered to be a form of anger expression.
The dominance score is the summed total of respondent’s interruptions of the interviewer during the SI plus the incidence of simultaneous speech (overtalk). Interscorer reliability for this category was 0.93. The SI scorer was, of course, blind to patients’ CHD status.
All participants were also administered the 50-item Ho scale (1). This scale was scored twice, once using all 50 items and again using only the items comprising the four subscales identified by Barefoot et al. (21) as predicting survival: cynical mistrust, hostile attributions, hostile affect, and aggressive responding. Results of prior studies suggest (2) that this abbreviated Ho scale is a better predictor of CHD outcome than the total Ho scale. Because Barefoot’s (21) abbreviated Ho scale was also consistently better in predicting CHD outcome than the original 50-item Ho scale in this study, the findings to be presented for the Ho-CHD relationship are all based on Barefoot’s (21) abbreviated scoring procedure.
Logistic regression analyses were performed to determine 1) the relationship between participants’ SI-derived HBI scores and their CHD status, 2) the relationship between participants’ HBI component scores (indirect challenge and irritability) 5 and CHD status, 3) the relationship between participants’ SI-derived dominance scores and CHD status, and 4) the relationship between patients’ Ho scores (and Ho component scores) and CHD status. CHD status was always a dichotomous variable (patients with documented CHD vs. patients without CHD) with equivocals excluded from the analyses. 6
The data were submitted to multivariate logistic regressions in which the following traditional risk factors were entered into the model in the first step of the analysis: age, cholesterol, diabetes mellitus, hypertension, and SES. 7 Gender was entered in the second step. HBI, dominance, indirect challenge, irritability, or Ho scores were entered in the third step, followed by gender by HBI, gender by dominance, gender by indirect challenge, gender by irritability, or gender by Ho score in the next step. Separate regressions were performed for each of the psychological variables. The above regressions were also performed separately for males and females (of course, without gender in the analyses).
Traditional Risk Factors and CHD
Relationships between traditional risk factors and disease outcome were assessed by means of logistic regression analyses. Significant relationships were obtained in relation to diabetes mellitus, hypertension, and cholesterol level. Age and SES were also significant risk factors. All these relationships were in the expected direction, so that patients with CHD tended to be older, of a lower SES level, and had a history of diabetes mellitus, hypertension, and hyperlipidemia (Table 1).
Gender Effects in Patients’ HBI and Dominance Scores
There were no significant gender differences in relation to the patients’ HBI, indirect challenge, irritability, dominance, and Ho scores (Table 2).
HBI and CHD
In the multivariate logistic regression analysis (ie, with the traditional risk factors in the model), 8 HBI correlated positively and significantly with CHD status (RR = 1.22, p < .01) (Table 3). For the group as a whole, the odds of having CHD increased by 22% for each unit increase in HBI scores. The relationship between HBI and CHD remained significant even after controlling for dominance (RR = 1.22, p < .02).
Of the two HBI component scores, irritability and indirect challenge, only irritability correlated significantly with CHD outcome (RR = 1.22, p < .02) (Table 3). The odds of having CHD increased by 22% for each unit increase in irritability scores.
Separate logistic regressions were conducted for men and women as planned (Tables 4 and 5). HBI and indirect challenge correlated significantly with CHD outcome in women (RR = 1.29 and 1.56, respectively;p < .05), but the relationship between irritability and CHD was clearly not significant (Table 4). For men, there was a positive relationship between HBI and CHD and between irritability and CHD, but both relationships fell short of clearcut significance (RR = 1.22 and 1.48, respectively;p < .10 and .06, respectively) (Table 5).
Dominance and CHD
Patients’ dominance scores correlated positively and significantly with CHD status (RR = 1.47, p < .03) (Table 3). The odds of having CHD increased by 47% for each unit increase in dominance scores. This relationship was significant even after controlling for HBI (RR = 1.45, p < .03).
There was no significant interaction with gender. Separate regressions for men and women yielded a nearly significant relationship for men (RR = 1.56, p = .06) and a nonsignificant relationship for women (RR = 1.40, p > .15).
Hostility and CHD
In relation to the Ho score, unlike the other psychological variables, there was a clearcut difference between the results of the univariate analysis and the results of the multivariate analysis. In the univariate analysis, the relationship between patients’ Ho scores and CHD status was clearly significant (p = .034) and remained so even after adjusting for age and gender (β = 0.081, RR = 1.08). However, this relationship was no longer significant after adjusting for the remaining traditional risk factors (RR = 1.03, p > .4). In fact, merely adjusting for SES or other disease risk factors (eg, hypertension or diabetes mellitus) was sufficient to eliminate the significant Ho-CHD relationship (RR = 1.05 and 1.06, respectively;p > .15).
Similar findings were obtained when we analyzed the relationships between participants’ Ho component scores and CHD status. The relationships between aggressive responding and CHD and between hostile affect and CHD were marginally significant (β = 2.61 and 0.224, respectively; RR = 1.29 and 1.25, respectively;p < .10), but they, too, were wiped out when we adjusted for traditional risk factors (RR = 0.89 and 0.93, respectively;p > .4 and .5, respectively).
Correlations (r Values) Between HBI, Dominance, and Ho Scores
Neither patients’ HBI scores nor their HBI component scores correlated significantly with their dominance scores (Table 6). There were no significant correlations between patients’ HBI, irritability, indirect challenge, dominance, and Ho scores (Table 6).
HBI and CHD
For the group as a whole, the odds of having documented CHD increased by 22% with each unit increase in HBI scores. Furthermore, this HBI-CHD relationship was independent of patients’ dominance scores.
Although there was no significant interaction between gender by HBI on CHD, separate analyses revealed that the HBI-CHD relationship was significant for women (RR = 1.29, p < .05) and only marginally significant for men (RR = 1.22, p < .10). To the best of our knowledge, this is the first published study to report a significant positive relationship between HBI and CHD in women.
In this study, patients’ HBI scores essentially consisted of two components: irritability and indirect challenge. These two components shared very modest common variance (r (195) = 0.22, p < .002) (Table 6). After controlling for traditional risk factors, only irritability remained a significant risk factor (RR = 1.22, p < .02). It should be noted that irritability loads significantly on an anger-out factor (27).
Although there were no significant interactions between gender and the two component scores, there is some indication that indirect challenge and irritability may relate differently to CHD in men than in women. In women, the indirect challenge scores correlated significantly with CHD (RR = 1.56, p < .05), but the correlation between irritability and CHD was clearly not significant (p > .2). In men, the relationship between irritability and CHD was marginally significant (RR = 1.48, p = .06), but the relationship between indirect challenge and CHD was not significant (p > .40).
According to Haney et al. (8), of the original four HBI component scores, indirect challenge is the most subtle and indirect expression of antagonistic behavior. This suggests that only subtle expressions of antagonism are a significant risk factor for CHD in women. The reverse seems to be the case with men. For them, irritability, that is, the outward expression of anger, was a marginally significant risk factor for CHD, but indirect challenge, a subtle manifestation of interpersonal antagonism, was clearly not a significant risk factor.
There is one obvious discrepancy between the results of this study and those obtained by Haney et al. (8). In the Haney et al. study, the relationship between the men’s HBI scores and CHD was clearly significant, whereas in this study it was not (.05 >p < .10). Perhaps the discrepancy is a consequence of the fact that there were many more incidents of direct challenges and withhold/evade responses (ie, direct expressions of antagonistic behavior) in the Haney et al. study (8) than in ours, in which there were almost none.
Dominance and CHD
The hypothesized positive relationship between dominance, as indexed by the interruption measure, and CHD was confirmed. For a one-unit increase in dominance scores, the odds of having CHD increased by 47%. Moreover, the effect of dominance was independent of HBI. This study, then, is the third study to suggest that dominance is a risk factor for CHD. The first study (9) had angiographically documented coronary occlusion as an end point. The second study (13, 14), a large-scale prospective study, had documented MI and sudden death as end points. Finally, this case-control study had documented CHD as an end point. The above studies also differ in terms of how dominance was assessed. Dominance was assessed behaviorally (number of interruptions) in the first and present studies and by interviewer ratings in the second study. Until recently, most research on the role of psychosocial factors in CHD has focused on hostility and anger. The above studies suggest that dominance, too, may be a risk factor for CHD. It is interesting to note that competitiveness or dominance is a major constituent of the Type A construct (4).
Although there was no significant interaction of gender by interviewee interruptions, separate analyses conducted for females and males revealed a marginally significant relationship between dominance and CHD in males but not in females.
What Is Measured by the HBI and Dominance Indices?
Psychologists distinguish between anger, hostility, and aggression, with anger referring to an emotional state, hostility to attitudes, and aggression to behavior (28, 29). Furthermore, factor analytic studies of anger questionnaires have identified two factorially orthogonal modes of anger expression: anger-out and anger-in (26, 28, 30). Anger-out is defined as the tendency to express anger outwardly, either physically or verbally (ie, aggressive behavior). Anger-in refers to the inward experience of anger and/or the tendency to withhold or inhibit the outward expression of anger (ie, anger suppression rather than repression in the psychoanalytic sense). The evidence suggests that it is primarily the full-blown outward expression of anger, in all of its paraverbal intensity (eg, angry facial expression, loud voice, and clenched fists), that is associated with exceedingly high levels of CVR (31, 32) and that puts one at risk for CHD (6, 26, 31, 33). Finally, social psychologists distinguish between impulsive, anger-driven aggression and deliberate, instrumental aggression (29). Other terms occasionally used to denote these two types of aggressive behavior are “hot” and “cold” aggression.
As pointed out above, the spouses of 97 participants in the present study rated their wives or husbands on various anger dimensions using Spielberger’s (16) STAXI. 4 Participants’ HBI scores correlated positively and significantly with their STAXI anger-out, angry temperament, and angry expression scores that were assigned to them by their spouses (r (96) = 0.251, 0.202, and 0.196, respectively;p < .05). All these STAXI subscales load on an anger-out factor (26). In contrast, there was no significant correlation between participants’ HBI scores and the STAXI spouse-assigned anger-in scores. These findings suggest that the HBI scores are related to anger-out rather than to anger-in. Furthermore, because there was no significant relationship between antagonistic hostility, as measured by HBI, and hostility (as measured by the Ho scale), the reference to HBI scores as an index of hostility is probably a misnomer and can be a potential source of confusion.
The pattern of correlations between participants’ dominance scores and the STAXI scores assigned to them by their spouses differed from the pattern of correlations between participants’ HBI and spouse-assigned STAXI scores. For the group as a whole, there were no significant correlations between participants’ dominance scores and any of the STAXI scores, although separate correlations for males and females revealed a negative correlation between dominance and anger-out scores in males (r = −0.25, p < .10) and a significant positive correlation between these two indices in females (r = 0.30, p < .05).
At least for males, then, dominance is not anger driven, nor does it correlate significantly with hostility, as measured by the Ho scale. Perhaps it is more an expression of cold, instrumental aggression than of anger-driven aggression. Although by some definitions, dominance may not qualify as a manifestation of aggression because it may lack intent to hurt, it should qualify as a manifestation of aggressiveness. (For a discussion of the distinction between these two terms, see Ref. 34). In fact, in a recent factor analytic study, a dominance questionnaire loaded significantly on an instrumental aggression factor (27).
Pathophysiology of the Dominance-CHD Relationship
Given the evidence that dominance may be a significant risk factor for CHD, at least in males, we need to know more about the mechanism and the pathophysiology that may be involved in this relationship. A study recently completed in our laboratory found that both anger-out and dominance-driven aggression were independently associated with heightened levels of CVR during a serial subtraction task (15). These findings suggest that heightened levels of CVR may mediate the dominance-CHD relationship as they apparently do the anger-CHD relationship (18–19, 31, 32). Similarly, Smith et al. (35, 36) report that effortful attempts to exert dominance are associated with heightened CVR. Additionally, hyperlipidemia may be involved in the dominance-CHD relationship: One study (37) found a significant positive relationship between questionnaire-assessed dominance and serum triglyceride concentration but not with serum cholesterol concentration. Another study (38), however, found that questionnaire-assessed dominance did correlate positively and significantly with total serum cholesterol levels in individuals low in physical fitness. Research on the CVR and neuroendocrine correlates of hostility and anger in provoked and angered individuals has contributed to our understanding of the mechanisms that are involved in the anger/hostility-CHD relationship (39). We now need to proceed with a similar research program on the CVR and neuroendocrine correlates of dominance behaviors.
Role of Gender
Of the two risk factors examined in this study, antagonistic behavior and dominance, the former seems to have been the strongest risk factor in our female population, and the latter seems to have been the strongest in our male population. This should not be interpreted to mean that only dominance is a risk factor for CHD in men. There is strong evidence, some based on prospective data, of a significant positive relationship between the outward expression of anger in men and CHD (6, 9, 24, 26, 31, 33). In the present study, too, there was a nearly significant relationship between the male participants’ irritability scores and documented CHD. The relevance of this finding is that irritability items have been found to load on an anger-out factor (27). More significantly, in this study, as reported elsewhere (26), spouse ratings of their husbands’ anger-out levels (anger-out factor scores) correlated significantly with their husband’s CHD status. On the other hand, spouse ratings of their wives ’ anger-out levels (anger-out factor scores) did not correlate significantly with their wives’ CHD status. Taken together, these findings suggest that different dimensions of the anger-antagonistic behavior domain may be CHD risk factors for men than for women. Taken together, the findings of this study suggest that the full-blown outward expression of anger and dominance are risk factors for CHD in men and that the more subtle expressions of antagonism (indirect challenge) are risk factors for CHD in women. However, this conclusion must be tempered by the fact that notwithstanding these indications of gender differences, these differences were not strong enough to yield significant gender by HBI, gender by indirect challenge, or gender by irritability interactions. Clearly, the role of gender as a mediator of the anger/antagonism-CHD relationship, which has been reviewed by Stoney and Engebretson (40), needs further systematic study.
Worthy of note is the finding that gender was not a significant source of variance in any of our anger/antagonism measures. The widely shared belief that men express anger primarily outwardly and women primarily inwardly is not supported by our data. What may have been true in the past may not reflect current reality.
Attitudinal Hostility and CHD
Age- and gender-adjusted Ho scores were associated with a significant increased risk for CHD. However, this relationship was no longer significant after making further adjustments for SES or disease risk factors (hypertension and diabetes mellitus), suggesting that the hostility-CHD relationship was primarily mediated through these other risk factors. Furthermore, in this study, there was a marginally significant relationship between two Ho component scores (hostile affect and aggressive responding) and CHD (p < .10), but these relationships, too, were completely wiped out when we adjusted for SES or hypertension and diabetes mellitus. The finding that adjusting only for SES attenuates the Ho-CHD relationship so that it is no longer significant casts a cloud on the Ho-CHD studies that have failed to make such an adjustment. In this study, there was an inverse relationship between SES and Ho scores (r = −0.29, p < .001), which accounts for the finding that once we adjusted for SES, the Ho-CHD relationship was no longer significant. (For a more complete discussion of the role of SES in hostility, see Ref. 41.)
Even if attitudinal hostility turns out not to be an independent risk factor for CHD, it does not follow that attitudinal hostility is a benign factor as far as health is concerned. Attitudinal hostility may be a significant risk factor for diseases other than CHD. Thus, Everson et al. (23) found that hostility was an independent risk factor for carotid artery atherosclerosis, and in this study, the abbreviated Ho scores were associated with an increased risk of essential hypertension, even after adjusting for SES and the other traditional risk factors (β = 0.102, SE = 0.040, p = .01, RR = 1.11). Furthermore, recent evidence (42) suggests that even when it is not an independent risk factor, hostility can interact with other risk factors to produce heightened risk of CHD.
SES and CHD
Of the various CHD risk factors examined in this study (including the traditional risk factors), SES was, from a statistical perspective, one of the most significant. That low SES is a significant risk factor for CHD, as it is for morbidity and mortality in general, has been known for some time, but precisely what mediates this relationship is a matter of debate (43, 44). One major contender, as far as the SES-CHD relationship is concerned, is poor health. For example, in this study, SES correlated inversely and significantly with both hypertension (r = −0.25, p < .001) and diabetes mellitus (r = −0.16, p < .03). However, when we covaried these two risk factors (ie, hypertension and diabetes mellitus) from the SES-CHD correlation, the relationship remained significant (RR = 0.542, p < .02). Nor did covarying the other traditional risk factors from the SES-CHD relationship diminish its significance (RR = 0.479, p = .02). Perhaps the stress that is associated with a low SES lifestyle is what puts one at risk for CHD (45). This is a hypothesis that needs to be carefully tested.
Partialling out the effect of SES from the HBI-CHD relationship, or from the irritability-CHD, indirect challenge-CHD, or dominance-CHD relationship, did not alter their significance, but doing the same for the Ho-CHD relationship reduced it from p < .03 to p = .40. It would seem, then, that in this study, SES accounted for the significant positive univariate Ho-CHD relationship. (The correlation between SES and Ho was −0.29, p < .001). This finding implies that in studying the effects of hostility on CHD, or other diseases, it is important to control for SES.
Hostility, as measured by the Ho scale (1) and SI-based hostility ratings (5), refers not only to hostile attitudes but also contains affective (anger) and behavioral (antagonistic behavior and aggression) components. Cited Here...
It should be noted that the hostile attribution items have a distinct paranoid flavor. Cited Here...
Thirty-three patients were excluded from the statistical analyses reported in Results because of their uncertain CHD status. The mean age and years of schooling of the remaining 133 patients were very close to those of the original 196 recruits (56.2 and 12.7 years, respectively), as was their SES distribution (26% upper class, 36% middle class, and 37% lower class). Cited Here...
The correlations between patients’ STAXI self-ratings, spouse anger ratings, Bendig’s Manifest Anxiety Scale ratings, and CHD status have been reported elsewhere (26). (The STAXI items were rephrased to make them appropriate for the spouse anger ratings.) Cited Here...
There were insufficient scorable direct challenge and withhold/evasion responses to permit separate regression analyses of these two categories. Cited Here...
We also analyzed our data with the equivocals combined with the CHD group. In general, the results were very much like those obtained when equivocals were excluded from the analyses, except they were consistently somewhat less significant. When we analyzed the data with the equivocals combined with the normals, only HBI remained a significant risk factor. These findings support our assumption that the equivocals are a mixed group and contain at least some CHD cases, which is why they were excluded from our analyses. Cited Here...
All these risk factors correlated significantly with CHD in this study. Cited Here...
Covarying traditional risk factors consistently strengthened the relationships slightly as compared with the univariate analyses. Given the similarity of the two sets of findings, we present only the multivariate results, except for the Ho-CHD relationship, in which there was a clear difference between the univariate and the multivariate analyses. Cited Here...
1. Cook W, Medley D. Proposed hostility for pharisaic-virtue skills of the MMPI. J Appl Psychol 1954; 38:414–8.
2. Helmers KF, Posluszny DM, Krantz DS. Associations of hostility and coronary artery disease: a review of studies. In: Siegman AW, Smith TW, editors. Anger, hostility, and the heart. Hillsdale (NJ): Erlbaum Associates; 1994. p. 67–96.
3. Miller TQ, Smith TW, Turner CW, Guijarro ML, Hallet AJ. A meta-analytic review of research on hostility and physical health. Psychol Bull 1996; 119:322–48.
4. Rosenman RH. The interview method of assessment of the coronary-prone behavior pattern. In: Dembroski TM, Weiss SM, Shields JL, Haynes SG, Feinleib M, editors. Coronary-prone behavior. New York: Springer-Verlag; 1978. p. 55–70.
5. Dembroski TM, MacDougall TM, Costa PT, Grandits GA. Components of hostility as predictors of sudden death and myocardial infarction in the Multiple Risk Factor Intervention Trial. Psychosom Med 1989; 51:514–22.
6. Siegman AW, Dembroski TM, Ringel N. Components of hostility and the severity of coronary artery disease. Psychosom Med 1987; 49:127–35.
7. Musante L, MacDougall JM, Dembroski TM, Costa PT Jr. Potential for hostility and dimensions of anger. Health Psychol 1989; 8:343–54.
8. Haney TO, Maynard KE, Hoseworth SJ, Scherwitz LW, Williams RB, Barefoot JC. Interpersonal hostility assessment technique: description and validation against the criterion of coronary artery disease. J Pers Assess 1996; 66:386–401.
9. Siegman AW, Feldstein S, Tommaso C, Ringel N, Lating J. Expressive vocal behavior and the severity of coronary artery disease. Psychosom Med 1987; 49:545–61.
10. Ferguson N. Simultaneous speech, interruptions and dominance. Br J Soc Clin Psychol 1977; 16:295–302.
11. Siegman AW. Expressive correlates of affective states and traits. In: Siegman AW, Feldstein S, editors. Multichannel integrations of nonverbal behavior. Hillsdale (NJ): Erlbaum Associates; 1985. p. 37–68.
12. Siegman AW, Feldstein S, editors. Nonverbal behavior and communication. Hillsdale (NJ): Erlbaum Associates; 1978.
13. Houston BK, Chesney MA, Black GW, Cates DS, Hecker MHL. Behavioral clusters and coronary heart disease risk. Psychosom Med 1992; 54:447–61.
14. Houston BK, Babyak MA, Chesney MA, Black G, Ragland DR. Social dominance and 22-year all-cause mortality in men. Psychosom Med 1997; 59:5–12.
15. Boyle S, Siegman AW. Angry aggression, instrumental aggression, and cardiovascular reactivity. Proceedings of the Fifth International Congress of Behavioral Medicine; 1998 Aug 19–22; Copenhagen, Denmark.
16. Spielberger CD. Manual for the State-Trait Anger Expression Inventory (STAXI). Odessa (FL): Psychological Assessment Resources, Inc.; 1988.
17. Kaplan JR, Botchin MB, Manuck SB. Animal models of aggression and cardiovascular disease. In: Siegman AW, Smith TW, editors. Anger, hostility, and the heart. Hillsdale (NJ): Erlbaum Associates; 1994. p. 127–48.
18. Kral BG, Becker LC, Blumenthal RS, Aversano T, Fleisher LA, Yook RM, Becker DM. Exaggerated reactivity to mental stress is associated with exercise-induced myocardial ischemia in an asymptomatic high-risk population. Circulation 1997; 96:4246–53.
19. Keys A, Taylor AL, Blackburn H, Bloxek J, Anderson JT, Simonson E. Mortality and coronary heart disease among men studied for 23 years. Arch Intern Med 1971; 128:201–14.
20. Barefoot JC, Lipkus IM. The assessment of anger and hostility. In: Siegman AW, Smith TW, editors. Anger, hostility, and the heart. Hillsdale (NJ): Erlbaum Associates; 1994. p. 43–6.
21. Barefoot JC, Dodge KA, Peterson BL, Dahlstrom WG, Williams RB Jr. The Cook-Medley Hostility Scale: item content and ability to predict survival. Psychosom Med 1989; 51:46–57.
22. Barefoot JC, Larsen S, von der Lieth L, Schroll M. Hostility, incidence of acute infarction, and mortality in a sample of older Danish men and women. Am J Epidemiol 1995; 142:477–84.
23. Everson SA, Kauhanen J, Kaplan GA, Goldberg DE, Julkunen J, Tuomilekto J, Salonen JT. Hostility and increased risk of mortality and acute myocardial infarction: the mediating role of behavioral risk factors. Am J Epidemiol 1997; 146:142–52.
24. Kawachi I, Sparrow D, Spiro A III, Vokonas P, Weiss ST. A prospective study of anger and coronary heart disease: the Normative Aging Study. Circulation 1996; 94:2090–5.
25. Bendig AW. The development of a short form of the Manifest Anxiety Scale. J Consult Psychol 1962; 20:384.
26. Siegman AW, Townsend ST, Blumenthal RS, Sorkin JD, Civelek AC. Dimensions of anger and CHD in men and women: self-ratings versus spouse ratings. J Behav Med 1998; 21:315–36.
27. Siegman AW, Boyle S, Kawachi I, Sparrow D, Kubzansky LD. The AHA syndrome and coronary heart disease: the Normative Aging Study. Proceedings of the 1999 Annual Stress and Anxiety Research Society (STAR) Meeting; 1999 July 12–14; Krakow, Poland.
28. Spielberger CD, Reheiser EC, Sydeman SJ. Measuring the experience, expression, and control of anger. In: Kassinove H, editor. Anger disorders: definitions, diagnosis, and treatment. Washington DC: Taylor and Francis; 1995. p. 49–67.
29. Berkowitz L. Aggression: its causes, consequences, and control. New York: McGraw Hill; 1993.
30. Buss AH, Durkee A. An inventory for assessing different kinds of hostility. J Consult Psychol 1957; 42:155–62.
31. Siegman AW. Cardiovascular consequences of expressing, experiencing, and repressing anger. J Behav Med 1993; 16:539–69.
32. Siegman AW, Snow S. The outward expression of anger, the inward experience of anger and CVR: the role of vocal expression. J Behav Med 1997; 20:29–45.
33. Mittleman MA, Maclure M, Sherwood JB, Mulry RP, Tofler GH, Jacobs SC, Friedman R, Benson H, Muller JE. Triggering of acute myocardial infarction onset by episodes of anger. Circulation 1995; 92:1720–6.
34. Evans RI. Konrad Lorenz: the man and his ideas. New York: Harcourt Brace Jovanovich; 1975. p. 35–48.
35. Smith TW, Allred KD, Morrison CA, Carlson SD. Cardiovascular reactivity and interpersonal influence: active coping in a social context. J Pers Soc Psychol 1989; 56:209–18.
36. Smith TW, Baldwin M, Christensen A. Interpersonal influence as active coping: effects of task difficulty on cardiovascular reactivity. Psychophysiology 1990; 27:429–37.
37. Fowkes FGR, Lang GC, Donnan PT, Deary IJ, Riemersma RA, Housley E. Serum cholesterol, triglycerides, and aggression in the general population. Lancet 1992; 340:995–8.
38. Greene RA, Houston BK, Holleran SA. Aggressiveness, dominance, developmental factors, and serum cholesterol level in college males. J Behav Med 1995; 18:569–79.
39. Williams RB Jr. Biological mechanisms mediating the relationship between behavior and coronary heart disease. In: Siegman AW, Dembroski TM, editors. In search of coronary-prone behavior: beyond type A. Hillsdale (NJ): Erlbaum Associates; 1989.
40. Stoney CM, Engebretson TO. Anger and hostility: potential mediators of the gender difference in coronary heart disease. In: Siegman AW, Smith TW, editors. Anger, hostility, and the heart. Hillsdale (NJ): Erlbaum Associates; 1994. p. 215–38.
41. Siegler IC. Hostility and risk: demographic and lifestyle variables. In: Siegman AW, Smith TW, editors. Anger, hostility, and the heart. Hillsdale (NJ): Erlbaum Associates; 1994. p. 199–214.
42. Knox SS, Siegmund KD, Weidner G, Ellison RC, Adelman A, Paton C. Hostility, social support and coronary heart disease in the National Heart, Lung and Blood Institute Family Heart Study. Am J Cardiol 1998; 82:1192–6.
43. Susser M, Watson W, Hopper K. Sociology in medicine. 3rd ed. Oxford, United Kingdom: Oxford University Press; 1985.
44. Adler NE, Boyce T, Chesney MA, Cohen S, Folkman S, Kahn RL, Syme SL. Socioeconomic status and health: the challenge of the gradient. Am Psychol 1994; 49:15–24.
45. McLeod JD, Kessler RC. Socioeconomic status differences in vulnerability to undesirable life events. J Health Soc Behav 1990; 31:162–2.