Triggering of Acute Myocardial Infarction Onset by Episodes of Anger
Background Many anecdotes and several uncontrolled case series have suggested that emotionally stressful events, and more specifically, anger, immediately precede and appear to trigger the onset of acute myocardial infarction. However, controlled studies to determine the relative risk of myocardial infarction after episodes of anger have not been reported.
Methods and Results We interviewed 1623 patients (501 women) an average of 4 days after myocardial infarction. The interview identified the time, place, and quality of myocardial infarction pain and other symptoms, the estimated usual frequency of anger during the previous year, and the intensity and timing of anger and other potentially triggering factors during the 26 hours before the onset of myocardial infarction. Anger was assessed by the onset anger scale, a single-item, seven-level, self-report scale, and the state anger subscale of the State-Trait Personality Inventory. Occurrence of anger in the 2 hours preceding the onset of myocardial infarction was compared with its expected frequency using two types of self-matched control data based on the case-crossover study design. The onset anger scale identified 39 patients with episodes of anger in the 2 hours before the onset of myocardial infarction. The relative risk of myocardial infarction in the 2 hours after an episode of anger was 2.3 (95% confidence interval, 1.7 to 3.2). The state anger subscale corroborated these findings with a relative risk of 1.9 (95% confidence interval, 1.3 to 2.7). Regular users of aspirin had a significantly lower relative risk (1.4; 95% confidence interval, 0.8 to 2.6) than nonusers (2.9; 95% confidence interval, 2.0 to 4.1) (P<.05).
Conclusions Episodes of anger are capable of triggering the onset of acute myocardial infarction, but aspirin may reduce this risk. A better understanding of the manner in which external events trigger the onset of acute cardiovascular events may lead to innovative preventive strategies aimed at severing the link between these external stressors and their pathological consequences.
It is well recognized that in certain cases, emotionally stressful events, and more specifically, episodes of anger, immediately precede and appear to trigger the onset of acute myocardial infarction.1 2 3 4 5 6 In approximately 4% to 18% of cases of myocardial infarction, emotional stress has been reported to occur immediately before the onset of symptoms.2 3 4 However, because these studies lack necessary control data, it has not been possible to quantify the association between emotionally stressful events, such as anger, and onset of acute myocardial infarction, nor to examine factors that might modify the risk of triggering.
Previous studies of the etiological role of psychological stress and the occurrence of coronary artery disease have almost exclusively focused on chronic risk and have yielded controversial results.7 Recent studies suggest that hostility, cynicism, and anger comprise a critical “toxic” component8 of type A behavior and are more strongly associated with the incidence of coronary artery disease than other aspects of global type A behavior.7 8 9 10 We therefore chose to focus on discrete episodes of anger as a potential trigger of acute myocardial infarction. Clarification of the role of anger as a trigger is important because a better understanding of triggering could facilitate approaches to sever the link between potential triggering activities and the transient physiological risk states that produce myocardial infarction.
The Determinants of Myocardial Infarction Onset Study (Onset Study) is a multicenter, interview-based study in patients with acute myocardial infarction.11 In this study, we used a case-crossover design12 13 to quantify the relative risk of myocardial infarction onset after discrete episodes of anger in 1623 patients with confirmed acute myocardial infarction.
The Onset Study was conducted in 22 community hospitals and 23 tertiary care centers (see “Appendix”). Between August 1989 and March 1993, 1623 patients (1122 men and 501 women; age range, 20 to 92 years) were interviewed a median of 4 days (range, 0 to 30) after myocardial infarction.
Interviewers identified eligible cases by reviewing coronary care unit admission logs and patients’ charts. For inclusion, patients were required to meet all of the following criteria: (1) elevated creatine kinase level, with positive MB isoenzymes, (2) an identifiable onset of pain or other symptoms typical of the onset of infarction, and (3) ability to complete a structured interview. The protocol was approved by the institutional review board at each participating center, and informed consent was obtained from each patient.
Interviewers were trained by personal instruction, a training manual, an instructional videocassette, and through ongoing feedback from the study coordinator. Approximately one third of the interviews were audiotaped for randomly selected quality control checks of the coding accuracy. To minimize bias in information ascertainment, the interviewers were not informed of the duration of the hypothesized hazard period.
The interview identified the time, place, and quality of myocardial infarction pain and other symptoms, the estimated usual frequency of exposure during the previous year, and the intensity and timing of exposure during the 26 hours before the onset of pain for episodes of anger and other potential triggers of the onset of myocardial infarction. Details pertaining to exposure to heavy physical exertion have been reported elsewhere.11
To study anger as a trigger of the onset of myocardial infarction, we developed the onset anger scale, a new seven-level self-report anger scale. Patients were shown a chart consisting of seven levels of anger (Table 1⇓) and were asked to estimate their usual frequency of exposure to each level during the previous year (“usual annual frequency”). They were also asked to state the timing and level of anger during each of the 26 hours before the onset of myocardial infarction. Patients were considered exposed if they reported a peak level of anger greater than or equal to level 5 (very angry, furious, or enraged) during the interval of interest.
In addition to the onset anger scale, we administered the state anger, anxiety, and curiosity subscales of the State-Trait Personality Inventory (STPI)14 to a subset of 881 patients. Patients were asked to respond to the STPI instrument, first focusing on the 2-hour period preceding symptom onset and then for the same 2-hour period on the preceding day. The STPI data were analyzed by determining the distribution of scores for each subscale in the control period. Individuals with responses above the 75th percentile were considered exposed.
A new epidemiological technique called the case-crossover design (Fig 1⇓) was developed for the Onset Study.11 12 This approach was developed to assess the change in risk of an acute event during a brief “hazard period” after exposure to a transient risk factor. The important innovative feature of this approach is that control information for each patient is based on his or her past exposure experience.11 12 13
For the onset anger scale, a 2-hour hazard period immediately preceding the onset of myocardial infarction was compared with two types of control data obtained from the patients: (1) their usual frequency of exposure to episodes of anger over the past year and (2) their actual exposure in the comparable 2-hour “control period” at the same time on the day before myocardial infarction. For the STPI subscales, data were available only for the second approach. To help maintain comparability of reporting of exposures for the hazard and control periods, the interview treated the 26-hour period before the onset of myocardial infarction as one long hazard period.
The analysis of case-crossover data is a new application of standard methods for stratified data analysis.12 13 15 16 In this analysis, the stratifying variable is the individual patient, as in a crossover experiment.
The ratio of the observed exposure frequency in the hazard period to the expected frequency (from the control information) was used to calculate estimates of the odds ratio as a measure of relative risk.11 12 13 Expected frequencies were estimated two ways: (1) usual annual frequency of episodes of anger and (2) frequency of episodes of anger in the control period on the day before onset.
Usual Annual Frequency of Episodes of Anger
This was the primary analytical method used for analysis of the onset anger scale. The amount of person-time exposed to episodes of anger was estimated by multiplying the reported usual annual frequency of exposure by its hypothesized usual duration. Unexposed person-time was then calculated by subtracting the exposed person-time in hours from the number of hours in a year. Hazard periods of various durations were analyzed using methods for cohort studies with sparse data in each stratum.12 13 17
Frequency of Episodes of Anger in the Control Period on the Day Before Onset
Relative risks were computed using standard methods for matched-pair, case-control studies. Instead of concordant and discordant pairs of subjects, however, each patient contributed a pair of intervals—a “hazard” period and a “control” period, which were either concordant or discordant for exposure.11 12 13 15 Both STPI and onset anger scale data were analyzed using this approach. Ninety-five percent confidence intervals and two-sided probability values were computed using exact methods based on the binomial distribution.15 Conditional logistic regression was used to control for within-person confounding.11 12
Modification of the relative risk was assessed by comparing relative risks in subgroups defined by different levels of the potential effect modifier. Subgroups were compared using the χ2 test for homogeneity.15 To estimate induction time (the length of time from anger to the onset of myocardial infarction), relative risks were calculated for each 1-hour period before the onset of myocardial infarction, with statistical control for subsequent exposure.18
The correlation between the onset anger scale and STPI data was assessed using Spearman’s rank correlation.19 Paired comparisons of the STPI scores in the hazard and control period were performed with the Wilcoxon signed rank test.19
Characteristics of the study sample are shown in Table 2⇓. Of the 1623 patients, 8.0% (130 patients) and 2.4% (39 patients) reported episodes of anger greater than or equal to level 5 on the onset anger scale in the 24-hour period and 2-hour period before myocardial infarction, respectively. The most frequent causes were arguments with family members (25%), conflicts at work (22%), and legal problems (8%).
Onset Anger Scale
Usual Annual Frequency of Episodes of Anger
Based on the analysis using usual annual frequency of exposure for control, the relative risk for the onset of myocardial infarction was elevated only during the first 2 hours after an outburst of anger. Relative risks associated with outbursts of anger occurring more than 2 hours before the onset of myocardial infarction were not significantly different from 1.0 (Fig 2⇓), indicating that the induction time was less than 2 hours. Therefore, all subsequent analyses were based on a 2-hour hazard period. The relative risk in the 2-hour period immediately after episodes of anger self-reported to be at least level 5 on the onset anger scale was 2.3 (95% confidence interval, 1.7 to 3.2).
Frequency of Episodes of Anger in the Control Period on the Day Before Onset
In the standard matched-pair analysis, 36 cases were exposed only in the 2-hour hazard period compared with 9 exposed only in the control period (the same 2-hour period the day before). Three subjects were exposed at both times. This analysis yielded a relative risk for myocardial infarction of 4.0 (95% confidence interval, 1.9 to 9.4). There were only 2 patients who reported both heavy exertion of at least 6 metabolic equivalents of the task (METs) and an episode of anger above level 4 on the onset anger scale during the hazard period. There were no patients exposed to both in the control period. The relative risk associated with an episode of anger was not materially altered in a conditional logistic regression analysis that statistically controlled for exposure to heavy physical exertion of at least 6 METs.
State-Trait Personality Inventory Subscales
There was a modest but statistically significant correlation between the level of anger reported on the onset anger scale in the hazard period and the anger score from the STPI state anger subscale measured for the same time period (r=.25, P=.03).
Table 3⇓ shows that in the hazard period, patients reported significantly higher scores on the anger (P=.001) and anxiety (P=.02) subscales but not on the curiosity subscale (P=.20).
The relative risk of myocardial infarction onset after exposures above the 75th percentile for the anger, anxiety, and curiosity subscales is shown in Table 4⇓. The data indicate a significantly elevated relative risk associated with higher levels of anger (1.9; 95% confidence interval, 1.3 to 2.7) and anxiety (1.6; 95% confidence interval, 1.1 to 2.2). There was no association found for the curiosity subscale (0.9; 95% confidence interval, 0.6 to 1.4).
Modification of the Relative Risk
Potential modifiers of the relative risk were examined (Table 5⇓) using the onset anger scale. The relative risk was significantly lower among regular users of aspirin (1.4; 95% confidence interval, 0.8 to 2.6) than among nonusers (2.9; 95% confidence interval, 2.0 to 4.1) (P<.05). The relative risk tended to be lower in men than women and among regular users of β-adrenergic antagonists than nonusers; however, these differences were not statistically significant. There were also trends toward lower relative risks among patients with a history of previous myocardial infarction and hypertension; however, these two groups comprised the major indications for use of aspirin and β-adrenergic antagonists in this population. There were not enough data to further stratify these subgroups to resolve whether the clinical indication, the treatment itself, or other patient characteristics such as use of other medications were responsible for the heterogeneity in the observed relative risks.
Episodes of anger as self-reported on both the Onset Study anger scale and the State anger subscale of the State-Trait Personality Inventory14 were associated with a transient doubling of the baseline risk of the onset of myocardial infarction in the subsequent 2 hours. These results were found to be consistent when two different types of control data were used for the onset anger scale.
Slightly more than 2% and 8% of the patients studied reported episodes of anger in the 2- and 24-hour periods immediately preceding the onset of symptoms, respectively—percentages similar to those reported in previous uncontrolled studies of myocardial infarction onset.3 4 One prior study reported a higher percentage of subjects exposed to emotional stress immediately before myocardial infarction onset,2 but this study included other types of emotional stress in addition to anger.
Although there are no population-based studies of the effect of outbursts of anger on the incidence of cardiac events, psychological stress in general has been reported to be associated with increases in cardiac events. Trichopoulos et al20 found an excess of cardiac deaths in the days after the 1981 Athens earthquake and proposed that this excess was due to mental stress triggering the onset of infarction.
Clarification of the mechanisms through which anger may trigger myocardial infarction is aided by the finding by our group and others that heavy physical exertion, a more easily quantifiable exposure than anger, can trigger myocardial infarction.11 21 The analogy is strengthened by the similarity of the physiological responses associated with heavy exertion and with anger.22 23 24 25
Several well-documented physiological effects of anger make the hypothesis of triggering of myocardial infarction by an outburst of anger biologically plausible. Mental stress has been shown to increase heart rate, blood pressure, and myocardial oxygen demand.22 These effects are mediated at least in part by catecholamine secretion.22 Verrier et al26 have reported that anger, induced in a dog model, elicited increases in coronary vascular resistance and ischemic ST segment changes. Mental stress has been documented to increase coronary vascular resistance in the presence of minimal angiographically detectable underlying coronary artery disease.22 27 A recent study documented that higher levels of self-reported anger induced by an anger recall task can produce coronary vasoconstriction, but only in narrowed segments.28 It has also been shown that a recalled anger task is more potent than other mental stressors or exercise in eliciting myocardial ischemia, as measured by a reduction in left ventricular ejection fraction in patients with documented coronary artery disease.29 In addition to hemodynamic effects, mental stress also has been shown to increase platelet aggregability24 as well as coagulation and opposing fibrinolytic factors.25
A proposed mechanism for triggering of myocardial infarction by anger is that onset occurs when a vulnerable but not necessarily stenotic atherosclerotic plaque disrupts in response to hemodynamic stresses; thereafter, hemostatic and vasoconstrictive forces determine whether the resultant thrombus becomes occlusive.22 30 The rarity with which a potential trigger becomes an actual trigger is probably a result of the infrequent coexistence of an atherosclerotic plaque vulnerable to disruption and other conditions required for acute occlusive thrombosis.
The relative risk tended to vary in subgroups of interest: Regular use of aspirin appeared to be protective, conferring a relative risk approximately half that observed in nonusers. Users of β-adrenergic antagonists also tended to have a lower relative risk of myocardial infarction onset after outbursts of anger; women tended to have a higher spike in risk than men.
A possible explanation for the ability of aspirin, and perhaps β-adrenergic blockers, to reduce the relative risk of myocardial infarction onset after anger is that these agents may decrease the likelihood of plaque disruption and acute occlusive thrombus formation in response to an outburst of anger. β-Adrenergic blockers may reduce the hemodynamic response that may lead to plaque disruption, whereas aspirin may prevent platelet aggregation leading to thrombus formation.
A potential limitation of the present study is recall bias, in that patients may not report episodes of anger with equal accuracy in the period immediately preceding myocardial infarction onset as in control periods. The case-crossover design helps to minimize this bias during data collection by treating the entire 26-hour interval before myocardial infarction onset as one long hazard period. The observed modification of the relative risk by regular use of aspirin also argues against significant recall bias because it is unlikely that use of this medication would alter the fidelity of reporting of anger. Furthermore, episodes of anger of level 5 or above on the onset anger scale are relatively rare events and are easy to remember and report. Additionally, the specificity observed for anger and anxiety as opposed to curiosity on the STPI subscales also argue against recall bias.
Because the case-crossover design uses self-matching, there is no variability in traditional chronic risk factors for myocardial infarction within each stratum. Thus, there can be no confounding by these chronic risk factors.11 12 16 Controlling for circadian rhythm and heavy exertion of 6 METs or higher, the only other well-documented trigger of myocardial infarction onset, did not materially alter the results. This indicates that the observed effect of anger was independent of all characteristics that are stable in an individual over time as well as circadian rhythm and exposure to heavy exertion greater than or equal to 6 METs.
The Onset Study anger scale is a new instrument that previously has not been characterized against other psychological instruments for assessing anger or hostility, such as the State-Trait Anger Scale14 or the Cook-Medley Hostility Scale.31 However, in this report, we have demonstrated a modest correlation with the STPI instrument, and the effect of anger on myocardial infarction onset was very similar when assessed with either instrument. The finding of specificity for the anger and anxiety subscales is also reassuring.
In addition, the case-crossover study design ensures that whatever each patient’s interpretation of each scale, the same interpretation will be used for case and control data, since each patient serves as his or her own control. Therefore, each scale provides a consistent measure that may be used to make internally valid comparisons between case and control exposure information.
The consistency of the relative risks obtained with the onset anger scale using two types of control data and with the state anger subscale of the STPI14 further supports the validity of the findings. The slightly higher relative risk estimated using control exposure data from the period on the day before the myocardial infarction than that obtained with control data based on usual frequency may be explained by a tendency of the patients to overreport their usual annual frequency. This would lead to a slight underestimation of the relative risk with this method. Alternatively, the patients may have underreported exposure on the day before their infarction. It is also possible that this difference occurred due to random sampling variability because estimators of relative risk are much more efficient when usual annual frequency control data are used.13 This is reflected in the narrower width of the 95% confidence intervals obtained using the usual annual frequency control data.
There is a possibility of bias caused by differential survival of cases who had myocardial infarction triggered by different mechanisms. For example, if patients whose infarctions were triggered by episodes of anger were more likely to survive than those whose infarctions were unrelated to anger, then the apparent relative risk may be overestimated. This scenario, although possible, seems unlikely.
Previous reports have shown that heavy physical exertion can trigger the onset of acute myocardial infarction.11 21 32 In this report, we provide data indicating that episodes of anger, which produce similar physiological responses,22 23 24 25 are also triggers. The finding in the present study that aspirin appears to eliminate the risk caused by discrete episodes of anger and that in our prior study that regular exertion greatly decreases the risk posed by heavy physical exertion11 are examples of the manner in which triggering research may contribute to prevention of cardiovascular disease. The goal would not be to prevent potential triggering activities but to sever the link between a potential trigger and its pathological consequences.
Hospitals and Sponsors
South Shore Hospital, South Weymouth, Mass; Charles Gaughan, MD.
New England Deaconess Hospital, Boston; Stuart Zarich, MD.
Carney Hospital, Boston; Robert Rimmer, MD.
St Vincent Hospital, Worcester, Mass; Richard Bishop, MD.
Carle Heart Center, Urbana, Ill; Bruce Handler, MD.
Burbank Hospital, Fitchburg, Mass; Paul Block, MD.
Hahneman Hospital, the Medical Center of Central Massachusetts, Worcester; John A. Ferrullo, MD; Daniel Miller, MD.
Beth Israel Hospital, Boston; Richard Pasternak, MD; Anthony Ware, MD.
Brigham and Women’s Hospital, Boston; Elliott Antman, MD.
Newton Wellesley Hospital, Newton, Mass; James Sidd, MD.
St Luke’s/Roosevelt Hospital Center, New York, NY; Judith Hochman, MD.
Memorial Hospital, the Medical Center of Central Massachusetts, Worcester; Joshua Greenberg, MD.
Norwood Hospital, Norwood, Mass; George Bero, MD.
Faulkner Hospital, Jamaica Plain, Mass; Alberto Ramirez, MD.
Washington Hospital Center, Washington, DC; Lucy Van Voorhees, MD.
New England Medical Center, Boston; Shapur Naimi, MD.
Massachusetts General Hospital, Boston; Patrick O’Gara, MD.
University of Massachusetts Medical Center, Worcester; Joel Gore, MD.
Leominster Hospital, Leominster, Mass; Nicholas Mercadante, MD.
Overlook Hospital, Summit, NJ; John Gregory, MD.
Tampa General Hospital, James A. Haley Tampa Veterans Hospital, Tampa, FL; Robert Zobel, MD, PhD.
Boston University Medical Center, Boston; Michael Klein, MD.
Rush Presbyterian/St Luke’s Medical Center, Chicago; Philip R. Liebson, MD.
Stony Brook Health Sciences Center, Stony Brook, NY; Peter Cohn, MD; Richard Friedman, PhD.
Memorial Hospital of Rhode Island, Pawtucket; Abdul Khan, MD.
Flushing Hospital Medical Center, Flushing, NY; Samuel Zoneraich, MD.
West Virginia VA Medical Center, Huntington; Robert Touchon, MD.
Milford Whitensville Regional Hospital, Milford, Mass; Arthur Sgalia, MD.
Quincy Hospital, Quincy, Mass; Alan Berrick, MD.
Montefiore Medical Center, Bronx, NY; Mark Goldberger, MD.
VA Medical Center, Long Beach, Calif; Amer Alzarka, MD.
Denver General Hospital, Denver, Colo; Koonlawee Nademanee, MD.
Concord Hospital, Concord, NH; Carl Levick, MD.
St Elizabeth’s Hospital, Brighton, Mass; Bernard Kosowsky, MD.
Danbury Hospital, Danbury, Conn; David L. Copen, MD.
Harlem Hospital Center, New York, NY; Jay Brown, MD (deceased).
Boston City Hospital; Sheilah Bernard, MD.
Metro West Medical Center—Framingham Union Hospital, Framingham, Mass; Harton S. Smith, MD.
Goddard Memorial Hospital, Stoughton, Mass; Myron Mazur, MD.
Illinois Heart Institute, Peoria; Paul Schmidt, MD.
New Britain General Hospital, New Britain, Conn; Milton Sands, MD.
New York Hospital/Cornell Medical Center, New York, NY; Robert Allan, PhD; Stephen Scheidt, MD.
We are grateful to the Onset Study interviewers for their dedication. This study was supported by a grant (HL-41016) from the National Heart, Lung, and Blood Institute. Dr Mittleman was supported by a Bourse de formation en recherche from the Fonds de la Recherche en Santé du Québec.
- Received March 1, 1995.
- Revision received April 11, 1995.
- Accepted April 16, 1995.
- Copyright © 1995 by American Heart Association
Obraztsov VP, Strazhesko ND. The symptomatology and diagnosis of coronary thrombosis. In: Vorobeva VA, Konchalovski MP, eds. Works of the First Congress of Russian Therapists. Comradeship Typography of A.E. Mamontov; 1910:26-43.
Williams RB. Psychological factors in coronary artery disease: epidemiological evidence. Circulation. 1987;76(suppl I):I-117-I-123.
Shekelle RB, Gale M, Ostfeld AM, Paul O. Hostility, risk of coronary heart disease and mortality. Psychosom Med. 1983;45:109-114.
Barefoot JC, Dahlstrom WG, Williams RB. Hostility, CHD incidence, and total mortality: a 25-year follow-up study of 225 physicians. Psychosom Med. 1983;45:59-63.
Mittleman MA, Maclure M, Tofler GH, Sherwood JB, Goldberg RJ, Muller JE, for Determinants of Myocardial Infarction Onset Study Investigators. Triggering of acute myocardial infarction by heavy exertion: protection against triggering by regular exertion. N Engl J Med. 1993;329:1677-1683.
Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol. 1991;133:144-153.
Mittleman MA, Robins JM, Maclure M. Control sampling strategies for case-crossover studies: an assessment of relative efficiency. Am J Epidemiol. 1995;142:91-98.
Spielberger CD. Preliminary Manual for the State-Trait Personality Inventory (STPI). Tampa, Fla: University of South Florida; 1971.
Rothman KJ. Modern Epidemiology. Boston, Mass: Little, Brown & Co; 1986.
Wacholder S, Silverman DT, McLaughlin JK, Mandel JS. Selection of controls in case-control studies. Am J Epidemiol. 1992;135:1019-1028.
Rothman KJ. Induction and latent periods. Am J Epidemiol. 1981;114:253-259.
Lehmann EL. Nonparametrics: Statistical Methods Based on Ranks. San Francisco, Calif: Holden-Day; 1975.
Muller JE, Tofler GH, Stone PH. Circadian variation and triggers of onset of acute cardiovascular disease. Circulation. 1989;79:733-743.
Rosing DR, Brakman P, Redwood DR, Goldstein RE, Beiser GD, Astrup T, Epstein SE. Blood fibrinolytic activity in man: diurnal variation and the response to varying intensities of exercise. Circ Res. 1970;27:171-184.
Levine SP, Towell BL, Saurez AM, Khieriem LK, Harris MM, George JN. Platelet activation and secretion associated with emotional stress. Circulation. 1985;71:1129-1134.
Verrier RL, Hagestad EL, Lown B. Delayed myocardial ischemia induced by anger. Circulation. 1987;75:249-254.