Abstract
Background:
Anger is a common problem in society, and anger's relationship with mortality, in particular with cardiovascular mortality, has been studied mainly in male western population. There are no prospective studies in Japan, about the association between anger and mortality.
Materials and Methods:
This study examined the association of anger with all-cause and cause-specific mortality in a Japanese community. Data came from the Takayama Study, which recruited residents aged ≥35 years in 1992 from Takayama City in Gifu, Japan. The current study used information on anger that was obtained from the second survey in 2002. A total of 11,902 healthy participants aged ≥45 years completed a self-administered questionnaire. Anger was assessed using the Spielberger Trait Anger Scale.
Results:
The main causes of deaths during the follow-up period from 2002 to 2013 were 460 for neoplasm, 254 for cardiovascular, and 435 for other causes. After adjusting for potential confounders, we found a significant positive association between the trait anger score and the risk of cardiovascular mortality for women, with a hazard ratio for high versus low score of trait anger of 1.81 (95% confidence interval 0.91–3.63, p for trend = 0.04), but not for men.
Conclusions:
Data suggest that for Japanese women, high trait anger score may be associated with an increased risk of cardiovascular mortality. Potential gender differences in the association between trait anger and mortality should be further studied from the cultural context.
Introduction
Anger has been defined by psychologists as a feeling related to a cognitive appraisal of perceived wrongdoing. This feeling is also associated with the action tendency to undo or correct that wrongdoing. 1 There is evidence in western literature that the experience of negative emotions, including anger, is associated with an increased short-term risk of incident cardiovascular events, independent of traditional risk factors. 2,3 Early research showed that type A behavior pattern—which is primarily characterized by hostility, intense ambition, competitive drive, constant preoccupation with deadlines, and a sense of time urgency—was related to the development of coronary heart disease (CHD). 4 However, a meta-analysis of prospective studies conducted in 20015 failed to show an association between type A behavior pattern and CHD. Some researchers therefore changed their focus to anger, hostility, or related constructs—one of the key dimensions of type A behavior pattern. 6 Much attention has been paid to the association between anger and CHD. 7 A meta-analysis by Mostofsky et al. 8 observed that outbursts of anger were associated with the short-term risk of heart attacks, strokes, and disturbances in cardiac rhythm that occur in everyday life.
Anger and hostility can also be seen as a long-term risk factor. Chida Steptoe 6 made a systematic review of prospective studies and reported that anger and hostility were moderately associated with CHD in healthy populations. The overall combined hazard ratio (HR) between anger and the risk of incidence/mortality from CHD was 1.19 (95% confidence interval [CI]: 1.05–1.35) for the healthy population studies. The researchers also pointed out that the harmful effect of anger and hostility was greater in men than in women. Greater trait anger, 9,10 tendency to experience anger, 11 and hostile thoughts and acts 12 have been suggested to predict CHD in men but not in women. However, there are not many studies reporting gender-separate analysis, and the majority of studies were conducted in western population.
There are only two investigations about anger or hostility and the risk of cardiovascular disease (CVD) in Japanese population. In one, anger control was associated with a decreased risk of acute myocardial infarction in men in one study. 13 The other study observed that anger score, measured by the Muller Anger Coping Questionnaire, was associated with an increased risk of CHD. 14 However, these studies were case–control studies, and both control groups were selected from participants of a health checkup program. There have been no prospective investigations about anger and CVD, as well as general mortality among Japanese. The results from previous studies in western populations cannot be extrapolated to other cultures. There is evidence that the emotional response to the same situation can vary depending on the culture. 15 In addition, the physiological response from the body to negative emotions is different from culture to culture; for example, Miyamoto et al. 16 reported that negative emotions predicted higher interleukin 6 levels among American adults, but there was no association among Japanese adults. Therefore, in the present study, we investigated the association between anger and the risk of all-cause and cause specific, including CVD mortality in a cohort of Japanese men and women. We focused on aspects of anger measured by the Spielberger Trait Anger Scale, 17 which measures the tendency to experience anger and to be aware of experiences of anger. This scale indicates individual differences in anger proneness as a personality trait.
Materials and Methods
Study population
The Takayama study was a population-based cohort study initiated in 1992. The design and methodology of the study have been described elsewhere. 18,19 Briefly, all nonhospitalized residents aged 35 or older in Takayama City in Gifu, Japan, were invited to participate in the study. A total of 31,552 persons, a participation rate of 85.3%, completed a questionnaire on demographic characteristics, smoking and drinking habits, diet, exercise, education, and medical and reproductive histories. In July 2002, a second survey was conducted to update lifestyle and Health-Status information, including questions regarding anger, sleep disorders, smoking status, medication use, and medical histories of the participants. In this 2002 survey, the target population was restricted to respondents whose age at baseline were <70 years (n = 26,546). After exclusion of those who had died (n = 1,564), were physically unable to complete the questionnaire because of being hospitalized or seriously ill (n = 51), or had moved (n = 2,537), the study population consisted of 22,394 individuals, of whom 14,971 (66.9%) responded to the second questionnaire, yielding a follow-up rate of 56.4%. Flow chart of study population is given elsewhere. 20 This study was approved by the Ethics Board of the Gifu University Graduate School of Medicine.
Predictor variable measurements
To assess anger, we used the Japanese version 21 of the Spielberger Trait Anger Scale. 17 This scale consists of two distinct facets, namely anger temperament and anger reaction. The items measuring anger temperament are designed to refer to more frequent and intense feelings of anger that occur in a broader range of situations, whereas anger reaction more narrowly refers to experiencing anger when negatively evaluated or mistreated by others. It includes 10 questions; for each question, the participants were asked to choose one that applied to 4 responses: Almost never, sometimes, often, and almost always. These responses were scored as 1 to 4, and the sum of these scores was calculated as the trait anger score. The total score range was 10 to 40, with higher scores indicating the predisposition toward quick unprovoked anger, a fiery temper, and aroused anger in response to unfair treatment and negative critique. The trait anger score in the Japanese version correlated with all subscales of the Jenkins Activity survey, which measures Type A behavior pattern. 22
Physical activity was assessed in the study in a validated questionnaire that asked participants to report the average number of hours they had spent weekly to carry out various kinds of activities during the last year, translated into a metabolic equivalent (MET), and summed to obtain a physical activity score (METs-hour/week). The details, including its validity, are described elsewhere. 23 Sleeping disorders were measured with the Japanese version of the Pittsburgh Sleep Quality Index. 24,25 The Pittsburgh Sleep Quality Index assesses the sleep duration, sleep latency, and frequency and severity of specific sleep-related problems of the previous month; four items from the Pittsburgh Sleep Quality Index were included in the questionnaire: difficulty initiating sleep, difficulty maintaining sleep, poor perceived quality of sleep, and hypnotic medication use, as proposed by Doi et al. 25
Mortality
The deaths and their causes during the follow-up period were confirmed using death certificates provided by the Legal Affair Bureau, Japan. The causes of death were coded according to the International Classification of Diseases (ICD), 10th Revision. The end points of this study were all-cause mortality and cause-specific mortality, including mortality from neoplasm (ICD-10: C00–D48) and CVD (ICD-10: I00–I99). Information concerning individuals who moved away from the survey area was obtained from the residential registers of family registers. During the study period, 277 individuals moved out of the study area, and the date of emigration was unknown for 50 out of them. They were censored at the latest date when they were known to reside in the city.
For the present analysis, we excluded subjects who reported a medical history of CHD or stroke (n = 1,150) and cancer (n = 628) and those who did not respond to questions about Anger (n = 1,291). Hence, the population for the analysis consisted of 11,902 subjects (5,346 men and 6,556 women).
Statistical analyses
A follow-up period was calculated for each participant from the second survey (July 1st 2002) to the date of death, the date of emigration out of Takayama, or the end of the study (March 31st 2013), whichever came first. The trait anger score was divided into low, medium, or high (low: from 10 to14, moderate from 15 to 21, and high from 22 to 40 points), adopting the cutoff points used by Williams et al. 26 The low score group was a reference category. The characteristics of the study subjects according to the trait anger score were assessed by the linear regression analysis for continuous variables and the chi-square test for categorical variables. The HRs and 95% CI of all-cause and cause-specific mortalities for each category of anger score were calculated using the Cox Proportional Hazard model, and the continuous value to anger score was used to assess the linear trend. We also estimated HRs of mortalities for each 1-U increases in anger temperament and anger reaction scores. A multivariate analysis was conducted using covariates, including age (continuous), marital status (married or not married), education level (<12, 12–14, ≥15 years), body mass index (in quartile), physical activity (continuous), smoking status (never, former <30 years, former >30 years, current <30 years, and current >30 years for men; never, former, and current for women), alcohol consumption (in quartile for men and nondrinkers, categorized as those who drank no alcohol, equal to or less than the median, or more than the median for women), medical history of diabetes and hypertension (yes or no), amount of sleeping hours (continuous), and sleeping disorders assessed by the Pittsburgh Sleep Quality Index (yes or no). As women were more likely to be nondrinkers and less likely to be long-term smokers, we used different categories for men and women. All four items for the sleeping disorders were highly correlated with each other, and their potential cofounding effects were examined, including one item at the time in the models. As for alcohol consumption and education level, information from the first survey was used because we did not obtain those data in the second survey. In sensitivity analysis, we excluded deaths during the first 2 years to estimate the HRs. Statistical analyses were performed using the SAS program. Significance was defined as two-sided p < 0.05.
Results
Table 1 shows the basic characteristics of the participants at the time of the 2002 survey. Male participants with the highest score of trait anger were more likely to be young, have longer years of education, being current smokers, have higher alcohol consumption, and present a history of hypertension. Female participants with higher trait anger scores were more likely to be young, have more years of education, and have higher alcohol consumption; they were also more likely to have sleep disorders. During the 10.75-year follow-up period, there were a total of 756 male deaths and 393 female deaths.
Basic Characteristics of the Participants in the Takayama Study Divided by Punctuation in the Trait Anger Scale
BMI, body mass index; DIS, difficulty initiating sleep; DMS, difficulty maintaining sleep; HUM, hypnotic medication use; PQS, poor perceived quality of sleep; SD, standard deviation.
Tables 2 and 3 show the association between anger and mortality among male and female populations. In men, there were significant associations between trait anger score and the risk of cancer and CVD mortality in age-adjusted analyses, but these associations vanished when multiple adjusted (Table 2). In women, higher trait anger score was significantly associated with a decreasing risk of cancer mortality (p for trend = 0.04), and the HRs for high versus low score were 0.73 (95% CI: 0.37–1.47) after controlling for covariates (Table 3). However, none of specific cancer sites revealed significant association between trait anger score and mortality risk. The higher trait anger score was significantly associated with an increasing risk of CVD mortality (p for trend = 0.04); compared with low score group, the HRs for middle and high score groups were 1.26 (95% CI: 0.83–1.92) and 1.81 (95% CI: 0.91–3.63), respectively. HRs of CVD mortality for each 1-U increases in anger temperament and anger reaction scores were 1.08 (95% CI: 1.00–1.17, p for trend = 0.07) and 1.08 (95% CI: 0.98–1.20, p for trend = 0.13), respectively.
Hazard Ratios in Male Population of All-Cause and Cause-Specific Mortality According to the Trait Anger Scale Score in the Takayama Study
Person-years of follow-up are 18,984, 27,144, and 7,508 for low to high score groups, respectively.
Adjusted for age, BMI, physical activity, smoking status, alcohol consumption, education, marital status, medical history of hypertension or diabetes, amount of sleeping hours, and hypnotic medication use.
p for trend when trait anger used as a continuous variable.
HR, hazard ratio; CI, confidence interval; CVD, cardiovascular disease.
Hazard Ratios in Female Population of All-Cause and Cause-Specific Mortality According to the Trait Anger Scale Score in the Takayama Study
Person-years of follow-up are 28,208, 34,326, and 5,272 for low to high score groups, respectively.
Adjusted for age, BMI, physical activity, smoking status, alcohol consumption, education, marital status, medical history of hypertension or diabetes, amount of sleeping hours, and hypnotic medication use.
p for trend when trait anger used as a continuous variable.
Inclusion of difficulty initiating sleep, difficulty maintaining sleep, or poor perceived quality of sleep instead of hypnotic medication use as a covariate did not significantly affect the results; the HR for high versus low scores was 1.90 (95% CI: 0.95–3.80, p for trend = 0.03), 1.80 (95% CI: 0.90–3.59, p for trend = 0.04), and 1.81 (95% CI: 0.75–4.34, p for trend = 0.06), respectively. Trait anger score might have been affected by an underlying disease at the baseline. Therefore, as a sensitivity analysis, we reanalyzed the data after excluding deaths during the first 2 years of follow-up. The effect sizes did not alter substantially; although the association for CVD in women was nonsignificant, the HR of CVD mortality for high versus low score was 1.83 (95% CI: 0.88–3.80, p for trend = 0.08).
Discussion
In the present study, we found a significant association between high score in the Trait Anger Scale and an increased risk of mortality from CVD in women; however, no significant relationship was observed between high score of trait anger and mortality in men, in the city of Takayama Gifu, Japan.
Our results in women were similar to the finding among Danish women; hostility measured by the Cook-Medley Hostility Scale was associated with the increased risk of myocardial infarction. 27 The Cook-Medley Hostility Scale also measures anger experience. 28 The Spielberger Trait Anger Scale, which we used, was also utilized in the Atherosclerosis Risk in Communities Study; trait anger was associated with an increased risk of CHD 26 and stroke 29 after controlling for sex and other covariates (no interaction due to sex). Some studies observed that the Spielberger trait anger or anger reaction was associated with CVD risk factors or surrogate markers, such as diastolic blood pressure, 30 central visceral obesity, 31 blood cholesterol, 32 soluble intercellular adhesion molecule-1, 33 and carotid atherosclerosis 34 in women. Proposed mechanisms for the link between anger/hostility and CVD may be applicable to our results on trait anger and CVD mortality in women. Candidate mechanisms include increased platelet reactivity, inflammation, and autonomic dysregulation. 35 In a meta-analysis by Schum et al., 36 the anger experience was significantly associated with high systolic blood pressure that is related to dysfunction of the sympathetic nervous system.
A positive association between anger and the risk of CVD has been reported from prospective studies among western men. 8,37 The lack of association between anger score and CVD mortality in men in our study may be explained from a cultural background. There are different authors who support the theory that anger and its expression are culturally contingent. Kitayama et al. 38 observed that anger expression assessed with anger-out subscale of the Spielberger Anger Expression Inventory was positively associated with the indices of blood pressure, total/HDL cholesterol ratio, and pro-inflammatory markers in Americans, but the association was inverse in Japanese. They suggested that the expression of anger may be more likely to indicate the level of empowerment or entitlement in Japan. Park et al. 39 also utilized the anger-out scale and reported that Japanese with higher social status expressed more anger, with the relationship mediated by decision-making authority. Although Japanese people tend to inhibit or regulate negative emotions with little effort, 40 some may still express anger if they feel that their expression of anger is culturally permitted and socially justified. In this case, high social status may function as an authorization to express anger. 39 This tendency may be stronger among men than women in a male-dominated society as in Japan. However, we should keep in mind that we measured anger experience but not anger expression. Two dimensions of anger identified by factor analyses, that is, anger expression and anger experience, have been widely accepted. 36 Spielberger's trait anger and anger-out are separately assigned to anger experience and anger expression dimensions, respectively. Nonetheless, in a factor analysis conducted by Izawa et al. 41 among Japanese, both trait anger and anger-out scales had similar high loading score in the first factor. Therefore, our observed association of CVD mortality with trait anger may reflect that with anger expression. Moderately high correlations of angry temperament with anger-out scores were noted, and, as suggested by Spielberger et al., 42 individuals who have an angry temperament may be more likely to express their anger outwardly.
The strengths of our study include its prospective design, a long follow-up period, a large size of the study population, and information obtained for several confounders. However, the present study has several limitations. The assessment of trait anger relied on self-report, and some persons may have given a more desirable response, which could lead to misclassifications. However, such a bias is likely to underestimate the true association in this prospective study. In addition, this questionnaire only included trait anger score, but not other aspects such as anger-in, anger-out, or anger control. The results were not altered substantially after excluding early deaths, but underlying illness may have caused anger or more stress, and these participants may have inclined to answer more negatively on some questionnaire items. Although we took into consideration many confounders, information on personality and other emotional status was not obtained. Our results were derived from only one community, which could limit the generalizability of the study.
Conclusions
For Japanese women, we found an association between high scores of trait anger and high risk of death due to CVD, but we did not find any association in men. This can be due to the fact that anger expression is culturally modified and gender contingent. Considering the potential role of cultural factors, further studies on anger and mortality in different gender and ethnic groups are recommended.
Footnotes
Acknowledgments
The authors thank the Takayama study participants and staff for their important contributions.
Authors' Contributions
All authors contributed to the editing of the article and approved the final submitted version of the article without compensation.
Author Disclosure Statement
No competing financial interests exist.
Funding Information
The study was supported by a grant from the Ministry of Education, Culture, Sports, Science, and Technology, Japan.