| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Department of Psychiatry (S.H.B., R.B.W., B.H.B., I.C.S., M.J.H., J.C.B.), Duke University Medical Center, Durham, North Carolina; and the Division of Cardiology (D.B.M.), Department of Medicine, Durham, North Carolina.
Address correspondence and reprint requests to Stephen H. Boyle, PhD, Behavioral Medicine Research Center, Duke University Medical Center, Department of Psychiatry, Box 2969, Durham, NC. E-mail: shboyle{at}duke.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: Nine hundred thirty-six patients (83% were male; mean age = 51.48) with CAD who were followed for an average of 14.9 years. The ACM consisted of the combination of the cynicism, hostile attribution, hostile affect, and aggressive responding subscales that were identified in an earlier study (Barefoot et al. [1989]) by a rational analysis of the item content. The relation between hostility and survival was examined with Cox proportional hazard models (hazard ratios [HRs] based on a two standard deviation difference).
RESULTS: Controlling for disease severity, the ACM was a significant predictor for both CHD mortality (HR = 1.33, p < .009) and total mortality (HR = 1.28, p < .02). The total CMHS was only a marginally significant predictor of either outcome (p values < 0.06).
CONCLUSION: The results of this study suggest that hostility is associated with poorer survival in CAD patients, and it may be possible to refine measures of hostility in order to improve prediction of health outcomes.
Key Words: hostility, • mortality, • coronary artery disease.
Abbreviations: CMHS = Cook-Medley Hostility Scale;; CHD = coronary heart disease;; CAD = coronary artery disease;; MI = myocardial infarction;; ACM = abbreviated Cook-Medley Hostility Scale;; SD = standard deviation;; HR = hazard ratio;; CI = confidence interval.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
A number of factors (e.g., small sample sizes and insufficient follow-up) may explain the lack of significant findings reported in these studies, but another issue is the use of the total CMHS as a measure of hostility.
An examination of the content of the CMHS reveals a wide variety of items, with the majority reflecting the construct of hostility. However, a subset of items appears to reflect other constructs. The reason for this may be that this scale was originally constructed to distinguish between teachers with good and bad rapport with their students, rather than as a measure of hostility. Because other variables are likely to be associated with student rapport, it is not surprising that constructs other than hostility are represented in this scale. Based on a rational analysis of the item content, Barefoot et al. (7) identified four item subsets reflecting the cognitive (cynicism and hostile attributions), affective (hostile affect), and behavioral (aggressive responding) dimensions of hostility. Two additional subsets of items (social avoidance and other) were identified as measuring constructs other than hostility. They found that the combination of the cynicism, aggressive responding, and hostile affect subscales was a stronger predictor of total mortality than the total scale in a prospective study of 118 lawyers. Although it was not a significant predictor, the hostile attribution subscale was also viewed as a component of hostility.
More recently, Barefoot et al. (4) found that an abbreviated form of the CMHS (ACM) consisting of items from the cynicism, hostile attribution, aggressive responding, and hostile affect subscales was a significant predictor of MI and total mortality in a sample of Danish men and women. It is important to note that the two subscales measuring constructs other than hostility were not significant predictors of mortality in either study. Finally, Helmers et al. (13) found that a short version (i.e., cynicism, hostile affect, and aggressive responding) of the CMHS was a stronger predictor of some indicators of myocardial ischemia than the total scale. These findings suggest that an ACM consisting of items that more clearly reflect hostility may be a better predictor of health outcomes than the total scale.
This article reports the results of a reanalysis of a study that examined the relation between the CMHS and survival (8) in patients with documented CAD. As noted previously, this earlier study failed to find a significant relationship between the CMHS and survival. Since publication of that study, there have been two developments that justify a reanalysis of the data. First, there have been significant improvements in the measurement of hostility that suggest that an ACM should be as good if not better at predicting health outcomes than the total score. Second, there has been additional follow-up of this sample, which now includes a much larger number of events and provides greater statistical power.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Hostility Measurement
Hostility was measured by the CMHS (1). In addition to the total scale, a 39-item ACM was also tested as a predictor of survival. This version contains the items from the four subscales of the CMHS identified by Barefoot et al. (7) as reflecting the hostility construct. Cynicism items are statements of negative beliefs about the trustworthiness or other qualities of people (e.g., "I think most people would lie to get ahead"). Items in the hostile attribution subscale reflect suspicion that others intend harm to the respondent (e.g., "I tend to be on my guard with people who are somewhat more friendly than I had expected").
The hostile affect subscale contains items that reflect the experience of negative emotions associated with interpersonal relationships (e.g., "Some of my family have habits that bother and annoy me very much"). Finally, aggressive responding items reflect a tendency to be aggressive or to endorse aggression as an appropriate way to solve problems (e.g., "I have at times had to be rough with people who were rude or annoying").
Participants completed the CMHS before receiving the results of the angiogram (Table 1). The response format for the CMHS was true/false. Participants who failed to complete >20% of the items on a scale were deleted from analyses with that scale (N = 61 for the ACM and N = 66 for the CMHS). Scores for subjects with 
20% of the items missing were prorated. The mean score on the ACM was 16.58 (SD = 6.41) and the mean score on the CMHS was 20.38 (7.59).
|
2. All models included disease severity as a covariate. Disease severity was represented by a "hazard score" (14). That was derived by a Cox proportional regression analysis of the survival of the entire population of patients with CAD seen at Duke University Medical Center from 1969 to 1984. Thus, it represents the summary of the baseline variables that have been associated with survival in this population. The variables used to derive the hazard scores include left ventricular ejection fraction, electrocardiographic abnormalities, number of vessels with >75% narrowing, and various indicators of myocardial damage. Over the course of follow-up, a number of patients had coronary artery bypass surgery (N = 549). Because surgical intervention is an important determinant of prognosis, this variable was included as a time-dependent covariate. An assumption of the proportional hazards model is that the effect of the predictor, in this case hostility, is constant over time. Including a time-dependent covariate consisting of the interaction of the predictor (ACM or CMHS) and the log of survival time in the model tested this assumption. The results of these tests indicated that there were not significant violations of the proportional hazards assumption.
Follow-up
Procedures for the documentation of cause of death and validation of follow-up methods have been detailed elsewhere (14). Deaths attributed to CHD were determined on the basis of information provided by the patient’s physician. These decisions were made by a mortality committee that had no knowledge of the patient’s score on the ACM. Total mortality and CHD mortality were the endpoints used in this study. Follow-up of the patients was conducted at 6 and 12 months after catheterization and annually thereafter. For these analyses, March 10, 1994 was considered the end of follow-up. This resulted in an additional 10 years of follow-up. The increased follow-up yielded 435 (350 from CHD) events compared with 133 (115 from CHD) from the 1989 report. Most of the non-CHD cases were attributed to a variety of medical conditions such as cancer, renal disease, and chronic obstructive pulmonary disease. The average length of follow-up was 14.9 years (SD = 2.75) for participants who did not experience an event.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
2 = 6.88, p < .009) and total mortality (2 = 6.61, p < .02). The hazard ratio (HR) associated with a 2-SD difference in ACM scores was 1.33 (95% confidence interval [CI] = 1.07–1.64) for CHD mortality and 1.28 (95% CI = 1.06–1.55) for total mortality. Unadjusted models yielded essentially the same results. To further illustrate the relationship between hostility and CHD mortality, see Figure 1 for Kaplan-Meier curves comparing patients scoring in the upper, middle, and lower tertiles of the ACM.
|
2 [1] = 3.65, p < .06) and total mortality (2 [1] = 3.64, p < .06). The HR associated with a 2-SD difference was 1.23 (95% CI = 1.00–1.51) for CHD mortality and 1.20 (95% CI = 1.00–1.45) for total mortality. The results of unadjusted models were essentially the same. Thus, the ACM was a somewhat better predictor of survival than the total scale. This revealed an association not apparent in the 1989 analyses. The increased follow-up was also an important factor in the significant findings observed in this study. When we examined these associations in the earlier data set, the pattern of results were similar, but neither the ACM nor the CMHS was a significant predictor of mortality. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
The increased follow-up was also an important factor in the significant findings observed in this study, because the ACM was not a significant predictor of survival when using the 1989 data set. It should be noted that the current study used a larger sample and lengthier follow-up than most previous investigations. Previous studies may not have had adequate power to demonstrate a significant effect of the magnitude reported in this study. Thus, the risk attributed to hostility is of moderate but nevertheless meaningful size from a public health perspective, given the large number of people with CHD.
A number of mechanisms may account for these findings (15). The relationship between hostility and survival may be mediated by excessive, repeated, and/or prolonged activation of the sympathetic adrenal medullary system (16–19). These physiological processes may play a role in the progression of atherosclerosis and also in triggering coronary events in vulnerable individuals. For example, studies suggest that intense outbursts of anger are potent triggers of heart attacks (20,21) and are capable of inducing myocardial ischemia (22,23) and ventricular arrhythmias (24). Hostile individuals also display reduced heart rate variability in response to stress (25).
Finally, treatment adherence may also help explain the association between hostility and survival. Brummett et al. (26) found hostility was associated with a higher risk of continuing to smoke in a sample of CAD patients. The association between hostility and treatment adherence should be investigated in future studies.
Thus, these data suggest that associations demonstrated in healthy populations extend to those with established disease. Measurement and sample size limitations may explain inconsistencies in previous literature. Finally, interventions designed to lower hostility might improve prognosis in CHD patients (27).
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Received for publication June 4, 2003.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  H. Nabi, M. Kivimaki, M. Zins, M. Elovainio, S. M Consoli, S. Cordier, P. Ducimetiere, M. Goldberg, and A. Singh-Manoux Does personality predict mortality? Results from the GAZEL French prospective cohort study Int. J. Epidemiol., April 1, 2008; 37(2): 386 - 396. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. H. Boyle, J. E. Michalek, and E. C. Suarez Covariation of Psychological Attributes and Incident Coronary Heart Disease in U.S. Air Force Veterans of the Vietnam War Psychosom Med, November 1, 2006; 68(6): 844 - 850. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. D. Kubzansky, K. W. Davidson, and A. Rozanski The Clinical Impact of Negative Psychological States: Expanding the Spectrum of Risk for Coronary Artery Disease Psychosom Med, May 1, 2005; 67(Supplement_1): S10 - S14. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. K. Schulman, P. R. Muskin, and P. A. Shapiro Psychiatry and Cardiovascular Disease Focus, April 1, 2005; 3(2): 208 - 224. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
