Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Jun 1.
Published in final edited form as: Psychosom Med. 2021 Jun 1;83(5):402–409. doi: 10.1097/PSY.0000000000000948

Specificity in associations of anger frequency and expression with different causes of mortality over 20 years

Claudia Trudel-Fitzgerald 1, Laetitia R Reduron 2, Ichiro Kawachi 1, Laura D Kubzansky 1
PMCID: PMC8178222  NIHMSID: NIHMS1694794  PMID: 33901055

Abstract

Objective:

Although evidence has linked anger and hostility with all-cause mortality risk, less research has examined whether anger frequency and expression (outwardly expressing angry feelings) are linked to all-cause and cause-specific mortality.

Methods:

In 1996, men (N=17,352) free of medical conditions from the Health Professionals Follow-Up Study reported anger frequency and aggressive expression levels. Deaths were ascertained from participants’ families, postal authorities, and death registries. Cox proportional hazards regression models estimated hazard ratios (HR) and 95% confidence intervals (CI) of mortality risk until 2016 with a 2-year lag, adjusting for a range of relevant covariates.

Results:

There were 4,881 deaths throughout follow-up. After adjustment for sociodemographics and health status, moderate and higher (versus lower) levels of anger frequency and aggressive expression were generally unrelated to risk of death from all-cause, neurological, or respiratory diseases. However, cardiovascular mortality risk was greater with higher anger frequency (HR=1.17, 95%CI=1.01–1.34), whereas cancer mortality risk was greater with higher anger expression (HR=1.14, 95%CI=0.98–1.33). Results were similar after including all covariates and stronger when considering anger expression’s interaction with frequency.

Conclusions:

In this cohort of men, experiencing angry feelings and expressing them aggressively were related to increased risk of dying from cardiovascular disease and cancer but not from other specific causes, over two decades. These results suggest that not only the experience of negative emotions but also how they are managed may be critical for some but not all health outcomes, highlighting the importance of considering causes of death separately when investigating psychosocial determinants of mortality.

Keywords: anger, mortality, cancer, coping, cardiovascular disease, emotion regulation, longevity, respiratory disease, neurological disease, survival

INTRODUCTION

Whether anger influences health outcomes, including mortality, has long been a topic of interest (1). To study this issue, the well-documented affective, behavioral, and cognitive dimensions of anger (2, 3) have been considered. The affective dimension of anger arises out of feelings of injustice and is considered a stable disposition (“trait anger”), often captured by the frequency/intensity of angry affect. Regarding the behavioral dimension, anger can be regulated in various ways. Strategies that are potentially maladaptive include expressing anger outwardly (e.g., aggression toward others) or not expressing it (e.g., suppression), while more adaptive forms comprise constructive expression (e.g., discussion). Hostility, the cognitive dimension, refers to the tendency to make hostile attributions or have a cynical attitude. Among studies evaluating anger’s role in death, most have focused on the cognitive dimension (4–7). While further investigating whether the frequency/intensity of angry feelings (affective dimension) relate to mortality will be informative, considering how one regulates these feelings (behavioral dimension) may offer more comprehensive insight into the relationship.

In relation to health, angry feelings have been mostly conceptualized as being aversive and detrimental (3); yet, whether anger yields positive or negative consequences may depend on how anger is regulated. Engaging in maladaptive strategies, like losing one’s temper (aggressive expression) or keeping emotions to oneself (suppression), tend to maintain anger’s intensity, potentially increasing deteriorative effects on health (8, 9). Conversely, more adaptive strategies (e.g., discussing feelings) can help down-regulate the emotion and, consequently, may be beneficial for health by avoiding the cumulative physiologic toll of persistent anger (8, 9). Furthermore, it is unclear if maladaptive strategies on their own might lead to higher mortality risk or if effects depend mostly on anger frequency/severity. For example, aggressive expression may be harmful in individuals feeling sustained, but not sporadic, anger.

Existing anger-mortality studies rarely accounted for anger’s multidimensional nature, and in those that did, the possible interaction between its dimensions was not evaluated. For instance, one study examined affective and behavioral dimensions, separately, with mortality risk in 3,682 adults (10). Each one standard deviation (SD) increase in trait anger was related to a 10-year elevated risk of all-cause mortality in men (ndeaths=175; relative risk=1.2, 95%CI=1.1–1.4), but not in women (ndeaths=92), while maladaptive (suppression, aggressive expression) and adaptive (discussion) strategies were unrelated to mortality in both groups. Although these findings may imply that the affective rather than the behavioral dimension matters for longevity, it is also possible that suppression and aggressive expression are detrimental for some but not all levels of anger frequency. Conversely, in another study neither trait anger nor aggressive expression, studied separately, were related to 5-year all-cause mortality in 2,679 adults (ndeaths=242) (7). While these inconsistent results could be due to different follow-up lengths, targeting all-cause mortality may have obscured associations as effects of anger may not be uniform across all causes of death.

Behavioral and biological sequelae of anger may matter for some more than other health outcomes and have been postulated as key mechanisms linking anger to such outcomes (9, 11, 12). Previous investigations of the anger-lifestyle linkage, albeit limited, suggested higher trait anger was associated with subsequent heavy drinking (13) while greater aggressive expression was related to future smoking relapse (14). Prior biological studies showed relationships of severe/frequent angry feelings with arrhythmia (15) and less favorable blood pressure patterns (16) hours later, and higher aggressive expression levels with 4-year hypertension risk (17) and elevated glucose 9 years later (18). Because biobehavioral factors predict certain outcomes more strongly than others (e.g., cardiovascular disease [CVD] versus respiratory diseases) (19, 20), studying anger’s role in specific causes of death is critical.

To date, authors who investigated cause-specific mortality have mainly considered CVD-related deaths. A meta-analysis reported that greater trait anger was related to an elevated although not statistically significant CVD-mortality risk (HR=1.13; 95%CI=0.94–1.36) among population studies of disease-free individuals (12). In parallel, other authors have shown associations between suppressing anger and increased CVD-death risk (21, 22). In previous work in the male-only Health Professionals Follow-up Study (HPFS), greater use of anger aggressive expression was associated with a marginally or statistically significant lower risk of CVD and stroke, respectively, over 2 years (23). Using a longer follow-up of 15 years, another study of men and women showed that incident CVD risk was elevated with greater trait anger, reduced with greater anger control, and unrelated to anger expression and suppression, after adjusting for sociodemographics, health status, and health behaviors (24). However, these two studies considered associations with only incident disease (including fatal and non-fatal cases), and thereby did not capture potential effects of anger on disease progression or secondary events including mortality. Limited and inconsistent evidence is available for anger’s association with cancer-related mortality. For instance, both positive (21) and null results (22) were found with anger suppression, although caution is warranted since <35 deaths, all cancers combined, were documented in each study. When considering cancer incidence specifically, prior findings from a large cohort study suggested that anger control increased risk in some (e.g., prostate) but not all (e.g., colorectal) cancers (25), also hinting to the importance of considering cancer sites separately. To our knowledge, anger’s role in risk of other distinct causes of deaths has not been examined.

Altogether, evidence that the affective and behavioral dimensions of anger may lead to increased mortality risk remains limited and mixed, partly due to the high heterogeneity across studies with regard to anger dimensions considered, causes of death included in the mortality endpoint, and also because of small samples with a limited number of specific events. Moreover, examining associations among men is important, as they tend to engage more in aggressive behaviors, because of general beliefs that expressing anger this way is more acceptable for men than women (26). Such aggressive expression may result in different health consequences of anger for men compared to women (1, 10, 12).

Therefore, the present study evaluated the association of anger frequency and aggressive expression, separately and in combination, with all-cause and cause-specific mortality risk among men from the HPFS over 20 years. We examined the most frequent causes of death in this cohort (CVD, cancer, respiratory, neurological) with these dimensions of anger, adjusting for potential covariates following prior research (10, 21, 22). In sensitivity analyses, we additionally adjusted for markers of anxiety and depression, based on a persisting theoretical question of whether anger influences health outcomes independently of other negative affect dimensions (11, 27). We hypothesized higher versus lower levels of anger frequency and aggressive expression would be related to greater mortality risk, with the magnitude of associations varying across causes of deaths, with stronger estimates for diseases that prior work suggests are particularly susceptible to effects of anger, including CVD. We further postulated that frequent angry feelings combined with regular use of aggressive expression to manage those feelings would relate to increased mortality risk (also with various magnitude of associations). We did not hypothesize direction of effects for other combinations of anger frequency and aggressive expression given the lack of prior work on this topic.

METHODS

Participants

HPFS is a prospective ongoing cohort study launched in 1986 among 51,529 male health professionals aged 40–75 years. Participants have completed biennial questionnaires on lifestyle, medical history and newly diagnosed medical conditions, with sustained high response rates (70–90% across cycles) (28, 29). The anger measures were administered in 1996, which constitutes the analytic baseline for this study. We excluded participants who died before the start of study follow-up (n=5,872), or who did not complete the anger measures (n=11,584), reported a major chronic condition (cardiometabolic diseases [myocardial infarction, angina, diabetes, stroke], cancer [all sites except non-melanoma skin cancer], respiratory disease [chronic obstructive pulmonary disease], neurological diseases [Parkinson, multiple sclerosis, epilepsy, amyotrophic lateral sclerosis]; n=9,753), or had missing data on sociodemographics (n=331), health status (n=3,296), and behavioral covariates described below (n=3,341), leading to an analytic sample of 17,352. Participants with major chronic conditions at baseline were excluded to mitigate potential confounding by health status and because individuals with severe health conditions are more likely to be angry (30). Participants who did not versus did complete the anger measures were more likely to be overweight/obese, current smokers, and less physically active, but did not differ meaningfully on other factors. The study protocol was approved by the institutional review boards of the Brigham and Women’s Hospital and Harvard T.H. Chan School of Public Health, and those of participating registries as required.

Measures

Anger frequency and aggressive expression.

In 1996, participants were asked how often they felt angry using a single item; response choices ranged from 1 (almost never) to 6 (≥2 times/day). As done previously in this cohort (23), responses were categorized in tertiles: “almost never” (Lower=29.1%), “1–2 times/month” (Moderate=31.8%), and “≥1 times/week” (Higher=39.1%). Aggressive expression of anger was assessed using the validated 8-item Spielberger Anger-Out Expression Scale (31), which queries how often individuals behaved in a specific way when feeling angry (e.g., screaming, insulting, slamming doors) on a 4-point scale from 1 (almost never) to 4 (almost always). A total score ranging from 8 to 32 was obtained by summing up the ratings for each item, where higher scores indicate a greater frequency of outwardly and aggressively expressing anger. Based on previous research (23), individuals who completed ≤50% of the items were set to missing. Scores were categorized into three levels according to cutoff points used following prior work (23, 31), where Lower=8–9 (18.6%), Moderate=10–12 (41.4%), and Higher=≥13 (40.1%). Other research has found this subscale is highly correlated with objective assessments of anger manifestations, including verbal and non-verbal indices (32). Internal consistency was acceptably high in the current sample (α=0.74).

Consistent with correlations observed in previous work (2, 3), the correlation between our two anger measures was moderate (r=.46), indicating the two constructs are fairly independent. The coefficient is also similar to that reported elsewhere between the Spielberger Trait Anger, a standardized measure of anger frequency, and Anger Out Scales (rmales=0.52) (31), supporting the convergent validity of our one-item assessment of anger frequency.

Covariates.

Following prior research (10, 21, 22), potential confounders and pathways included sociodemographics, health status, and health behaviors, which were self-reported in 1996 unless otherwise noted. Validity of behavioral-related measures are detailed in Supplemental Text 1, Supplemental Digital Content. Sociodemographics included age (continuous) and profession (queried in 1986; dentist, osteopath, pharmacist, veterinarian, others [optometrist, podiatrists]), while health status factors were high blood pressure (yes, no), high cholesterol (yes, no), and body mass index (BMI; ≤25kg/m2, >25kg/m2). Health behaviors included smoking (never, former, current), physical activity (<150min/week, ≥150min of moderate-to-vigorous activity), alcohol consumption (queried in 1998; ~1 drink/day, otherwise) and diet quality, based on the validated Alternative Healthy Eating Index (33) (queried in 1998; continuous). As no symptom measures were available at the analytic baseline, anxiety and depression were captured via self-reported regular use of anxiolytics and antidepressants, separately, over the past two years (yes, no).

Mortality.

Mortality data was collected from state vital records and the National Death Index, and supplemented by reports from family members and postal authorities, leading to 98% mortality follow-up (34). Study physicians blind to study hypotheses ascertained cause of death from death certificates, supplemented by medical records, according to the International Classification of Diseases, Eighth Revision. The most frequent causes of death in this cohort (cardiovascular diseases [heart disease, stroke], cancer [all cancer sites], respiratory disease [chronic obstructive pulmonary disease], neurological diseases [Parkinson, multiple sclerosis, epilepsy, amyotrophic lateral sclerosis]) were selected for this study. Deaths were identified through June 1, 2016.

Statistical Analyses

Analyses were conducted using SAS®9.4 with a two-sided 0.05 p-value. We examined the covariate distribution (age-standardized) across baseline anger levels.

Primary analyses.

Cox proportional hazard models evaluated hazard ratios (HRs) and 95% confidence intervals (CIs) of all-cause and cause-specific mortality risk, until the end of follow-up (2016), across anger frequency and aggressive expression levels, separately, in three models to examine the role of covariates. Model 1 was stratified by age and time period and adjusted for sociodemographics. Model 2, the core model, further accounted for health status. As behavioral factors may lie on the pathway, an exploratory model, Model 3, additionally included behavioral factors. Continuous standardized scores (1-SD) of anger frequency and aggressive expression were also considered separately with the same models. All analyses introduced a 2-year lag to address possible concerns for reverse causation, whereby underlying health problems might influence the experience/report of anger levels (30).

Secondary analyses.

Three additional sets of analyses were conducted, which all used the core model (Model 2). Firstly, to evaluate the potential synergistic effect of affective and behavioral dimensions of anger on mortality risk, we created the following categories: “low-moderate frequency/low expression,” “low-moderate frequency/high expression,” “high frequency/low-moderate expression,” and “high frequency/high expression,” and tested their association with mortality risk as done in primary models. Secondly, to acknowledge the role of other psychological factors, regular use of anxiolytics and antidepressants, separately (as proxy for anxiety and depression), were further added to the primary models in sensitivity analyses. Lastly, associations of anger frequency and aggressive expression, separately, were evaluated with the most common specific CVD and cancer mortality subtypes in this cohort (heart disease, stroke; prostate, lung, pancreatic, colorectal cancer) a posteriori to verify whether certain diseases were driving the primary associations.

RESULTS

Baseline Characteristics

Table 1 presents the distribution of covariates by baseline anger levels. Participants were on average 61.40 years old (SD=8.59; range=49–87) and the majority were married (89.97%) and dentists (56.49%). Additionally, few had high blood pressure (20.43%) or cholesterol (24.08%), while the vast majority were not current smokers (94.78%) and had favorable alcohol consumption habits (~1 drink/day; 65.97%). However, approximately half of the sample had an unhealthy BMI (56.35%) and fewer engaged in moderate-to-vigorous physical activity ≥150min/week (46.92%). Overall, the distribution of characteristics was not meaningfully different across levels of anger exposures. However, men with higher versus lower levels of anger frequency and aggressive expression were less likely to be never smokers and more likely to be overweight. Of note, a slightly smaller proportion of current smokers was noted among men with higher versus lower anger frequency levels specifically. Over the 20-year follow-up period, 4,881 deaths were documented.

Table 1.

Age-standardized characteristics of participants by levels of anger frequency and expression (N=17,352)

Anger frequency levels Anger expression levels
Lower (n=5045) Moderate (n=5523) Higher (n=6784) Lower (n=3219) Moderate (n=7178) Higher (n=6955)
Age, mean (SD)* 64.2 (8.9) 61.4 (8.4) 59.3 (7.9) 64.5 (9.1) 61.8 (8.6) 59.6 (7.9)
Married, % 88.8 90.5 90.9 89.2 89.9 91.0
Profession
- Dentist, % 55.7 55.7 57.7 54.0 56.9 57.2
- Osteopath, % 3.1 3.6 3.9 2.9 3.2 4.2
- Pharmacist, % 9.8 9.7 7.6 10.3 8.9 8.4
- Veterinarian, % 21.8 21.2 22.0 22.7 21.4 21.5
- Other (optometrist or podiatrist), % 9.5 9.8 8.8 10.1 9.5 8.6
Prevalent high blood pressure, % 19.1 20.6 21.4 19.2 20.3 21.4
Prevalent high cholesterol, % 22.5 23.6 25.6 21.9 24.4 24.9
Body mass index >25kg/m2, % 55.4 55.8 57.2 51.8 54.6 60.0
Smoking status
- Never smoker, % 53.9 50.9 46.8 58.9 51.7 44.5
- Former smoker, % 40.5 43.6 48.7 36.0 43.0 50.3
- Current smoker, % 5.7 5.5 4.5 5.1 5.3 5.2
Moderate to vigorous physical activity <150min/week, % 53.4 52.5 52.9 55.5 52.9 52.2
Unfavorable alcohol intake (less or more than 1 drink/day), % 32.0 35.2 34.5 33.1 33.6 35.2
Diet quality (from 0 [poorer diet] to 100 [healthier diet]) 50.3 (10.8) 50.2 (10.6) 49.7 (10.4) 49.7 (10.9) 50.1 (10.6) 50.0 (10.5)
Open in a new tab

Notes. Although anger frequency and expression are presented in the same table, they represent distinct analyses.

Percentages or ns or both for categorical variables, and are standardized to the age distribution of the study population.

Values of polytomous variables may not sum to 100% due to rounding.

*

Value is not age adjusted.

Relationships of anger frequency with mortality

Compared to lower levels, moderate and higher levels of anger frequency were not clearly related to all-cause, cancer, and neurological mortality risk (see Figure 1; Supplemental Table S1). For example, higher levels were marginally associated with increased risk of all-cause mortality when adjusting for sociodemographics (HRModel 1=1.07, 95%CI=0.99–1.15), and further adjustment for health status (Model 2) barely weakened this association. However, higher anger frequency levels were more strongly related to increased CVD-mortality risk (HRmodel 2=1.17, 95%CI=1.01–1.34), and marginally reduced respiratory-mortality risk (HRmodel 2=0.79, 95%CI=0.61–1.02). Further including potential behaviors pathways did not change the magnitude of these relationships. Considering continuous levels of anger frequency led to similar results with CVD and respiratory-related deaths (see Table 1; per 1-SD increase, in Models 2: HRCVD=1.07, 95%CI=1.01–1.14; HRrespiratory=0.90, 95%CI=0.80–1.01), and with all-cause mortality (Model 2: HRall-cause=1.03, 95%CI=0.99–1.06).

Figure 1.

Figure 1.

Open in a new tab

Association of anger frequency and aggressive expression levels, respectively, with risk of all-cause and cause-specific mortality.

Notes. CVD=cardiovascular disease, Respir.=Respiratory, Neuro.=Neurological.

Anger levels: Lower=Reference Group, Moderate=Light grey circle, Higher=Dark grey circle; vertical bars represent 95% confidence intervals.

All models are stratified by age and time period, and adjusted for marital status, profession, high blood pressure, high cholesterol, and body mass index (Model 2).

Relationships of anger aggressive expression with mortality

Overall, moderate and higher versus lower anger aggressive expression levels were unassociated with increased all-cause, respiratory, and neurological mortality risk (see Figure 1; Supplemental Table S2). Greater anger aggressive expression was weakly associated with CVD-related mortality in the sociodemographics model (HRmodel 1=1.14, 95%CI=0.98–1.33), but adjustment for health status further weakened the point estimate (HRmodel 2=1.10, 95%CI=0.94–1.28). An increased risk of cancer mortality was observed with higher anger aggressive expression levels, after controlling for baseline sociodemographics and health status (HRmodel 2=1.14, 95%CI=0.98–1.33). Further adjustment for behavioral factors in Model 3 meaningfully attenuated this relationship (HRhigher=1.09, 95%CI=0.93–1.27), with smoking showing the strongest association with cancer death in this model, among all behaviors included as covariates (HRformer=1.56, 95%CI=1.39–1.76; HRcurrent=3.25, 95%CI=2.66–3.96). When considering continuous levels of anger aggressive expression, risk was elevated for CVD and cancer mortality (see Table 1; per 1-SD, in Models 2: HRCVD=1.06, 95%CI=1.00–1.13; HRcancer=1.06, 95%CI=1.00–1.12).

Synergistic effects of anger frequency and aggressive expression

Combining anger frequency and aggressive expression levels yielded results that were of similar directions but larger magnitude than those from the primary models, with the strongest associations evident among in men with “high frequency/high expression” versus those with “low-moderate frequency/low-moderate expression” (reference group; see Supplemental Table S3). Overall, the two other groups, “low-moderate frequency/high expression” and “high frequency/low-moderate expression,” did not show clear associations with mortality outcomes. More specifically, in Model 2, the “high frequency/high expression” group had a 10% increased all-cause mortality risk (95%CI=1.01–1.19) and a 30% increased CVD-related mortality (95%CI=1.12–1.52); estimates did not reach statistical significance for the cancer-related mortality, although the magnitude was comparable to the one of all-cause mortality (HRModel 2=1.10, 95%CI=0.95–1.27). Consistent with the primary results, an inverse association was noted with respiratory-related mortality for the “high frequency/high expression” group (HRModel 2=0.75; 95%CI=0.55–1.03), but unlike the primary results, an inverse association was also apparent for neurological-related deaths (for the “low-moderate frequency/high expression” group only; HRModel 2=0.61, 95%CI=0.38–0.96).

Role of psychotropics and differences across CVD/cancer subtypes

Results from sensitivity analyses were similar to those obtained in the primary analyses. For instance, after adding anxiolytics and antidepressant use to Model 2, higher versus lower levels of anger frequency were unrelated to risk of all-cause, cancer, and neurological mortality but remained associated with CVD and respiratory deaths in similar directions and magnitude (HRCVD=1.15, 95%CI=1.00–1.33; HRrespiratory=0.77, 95%CI=0.59–0.99). Findings were also robust when adding these indicators of psychological distress to the anger aggressive expression models (e.g., higher versus lower levels: HRcancer=1.14, 95%CI=0.98–1.32).

In analyses stratified by subtypes of CVD- or cancer-related death (Supplemental Table S4), findings indicated that the elevated CVD mortality risk for higher anger frequency levels was mainly due to heart disease (HRhigher vs. lower=1.21, 95%CI=1.03–1.41). Of note, estimates were also elevated for prostate and pancreatic cancer among men with moderate levels of anger frequency (HRprostate=1.46, 95%CI=0.98–2.17; HRpancreatic=1.51, 95%CI=1.00–2.28). Regarding anger aggressive expression, results from the primary analyses appeared mainly driven by heart disease (HRhigher vs. lower=1.15, 95%CI=0.97–1.37) and pancreatic cancer (HRhigher vs. lower=1.49, 95%CI=0.92–2.42). Although a somewhat reduced stroke risk was obtained with greater aggressive expression, it did not reach statistical significance (HRmoderate vs. lower=0.80, 95%CI=0.59–1.09; HRmoderate vs. lower=0.94, 95%CI=0.68–1.31).

DISCUSSION

Prior research has investigated the association of anger-related constructs, particularly hostility, with all-cause mortality risk (4–7). However, limited work has specifically focused on other dimensions of anger, individually or in combination, and their potential role in risk of cause-specific mortality, beyond CVD-related death. Such nuances are critical given that anger’s dimensions and biobehavioral sequelae may contribute differently to these distinct causes of mortality. Consistent with our initial hypotheses, the current results suggested that experiencing higher (versus lower) levels of angry feelings was related to a 17% increased risk of CVD death, whereas greater aggressive expression of those feelings was associated with a 14% elevated risk of cancer death over two decades, after controlling for sociodemographics and health status. These primary findings appeared mainly driven by heart disease and pancreatic cancer, respectively. However, except for respiratory death, levels of anger frequency and aggressive expression when considered separately were unrelated to other specific causes of mortality among these midlife/older men. When using continuous measures of anger in the models, results were similar; a marginal 3% elevated risk of all-cause mortality was also noted per 1-SD increase in anger frequency and aggressive expression. Associations were robust to further adjustment for psychotropic medication. Analyses evaluating the synergy between anger frequency and aggressive expression generally showed that detrimental effects of anger on mortality risk are greatest when anger is both frequently experienced and aggressively expressed. Furthermore, estimates were often larger than those obtained in the primary models, hinting to the predictive potency of combining affective and behavioral dimensions of anger when studying their role in health outcomes.

The primary results are consistent with existing research also demonstrating a modest relationship of all-cause mortality, and in some cases, cancer- and CVD-mortality, with trait anger (10) and suppression (21, 22). Conversely, elevated risk for all-cause, cancer- and CVD-related mortality among individuals reporting greater use of aggressive expression is somewhat at odds with prior, although scarce, findings of no association between this specific behavioral strategy and mortality risk (7, 10). However, the current analytic sample included more deaths (nall-cause=4,881; nCVD=1,324; ncancer=1,378) than did prior studies, which may have facilitated detecting these small associations. Similar to findings from the prior study with HPFS, we found anger expression was associated with reduced stroke risk (23); however, in the present study, this association did not reach statistical significance. Important to note is that the current study is not a replication of the earlier one with more follow-up time (20 versus 2 years). In fact, findings between the two studies may differ because the outcomes differed (incident events including fatal and non-fatal cases versus mortality including secondary events) and the current analyses introduced a 2-year lag to lower concerns about reverse causation. More generally, the association of greater aggressive expression with higher mortality risk is consistent with evidence pointing to the detrimental role of aggressively expressing anger in various chronic disease risk and unfavorable biobehavioral processes involved in mortality risk (13, 14, 18). Other maladaptive strategies for regulating anger, like suppression, have also been related to greater mortality risk (21, 22). Of note, scoring low on aggressive expression may not necessarily indicate suppressing; accordingly, in prior work aggressive expression and suppression were only modestly correlated (e.g., r=.20 in a sample of midlife U.S. adults (35)). Individuals reporting limited use of aggressive expression may in fact express anger constructively or suppress it; thus, additional information is needed to determine more definitively how they regulate this emotion.

The apparent protective role of anger frequency, when studied separately and in combination with aggressive expression, in respiratory death remains more puzzling. To our knowledge, no previous research has examined the prospective association of anger frequency or aggressive expression with death from respiratory causes, but two prior cross-sectional studies noted that greater levels of trait anger, suppression, and, to a lesser extent, aggressive expression, were associated with more concurrent respiratory problems (36, 37). Our unexpected result may be due to the small proportion of current smokers at the analytic baseline (5%) and their lower prevalence among men with higher anger frequency levels. Because smoking is an established risk factor for respiratory-related conditions (e.g., COPD, lung cancer), having few current smokers might have altered our results. Since these unexpected findings might be spurious or due to the composition of this cohort (i.e., health professionals may be less likely to engage in unhealthy behaviors like smoking), replication is warranted. The absence of clear associations between individual anger dimensions and neurological mortality in the primary models may either reflect true null effects or be due to the smaller number of these deaths in our sample (n=220) making a small association more difficult to detect. Similarly, caution is warranted when interpreting the lower risk of neurological deaths obtained among men with “low-moderate frequency/high expression” given the limited number of cases in this group (n=25).

Given primary findings that both anger frequency and aggressive expression are associated with increased risk for certain causes of mortality, some consideration of pathways is warranted. Interestingly, in the current study distribution of health behaviors at baseline was not substantially different across levels of anger exposures. Moreover, in Model 3, further statistical control for these lifestyle factors attenuated the association of anger with cancer- but not CVD-mortality. This phenomenon suggests that the anger-cancer mortality association may be partly explained by the adoption of detrimental habits, like unhealthy drinking and smoking; however, the anger-CVD mortality relationships may be attributed to other potential mechanisms. In the current study, health status indicators (BMI, high blood pressure, cholesterol) were generally evenly distributed across levels of anger exposures at baseline and controlling for them did not attenuate existing relationships of anger dimensions with most mortality endpoints, hence lowering their likelihood of being major pathways. Yet, higher BMI was observed among higher anger aggressive expression levels and confidence intervals of the association between anger aggressive expression and CVD death became wider after adjustment for health status covariates, including BMI. Other evidence suggests that being overweight is a risk factor for pancreatic cancer (38), for which mortality risk was elevated among men with higher (versus lower) anger aggressive expression in the current study. Altogether, these findings reiterate the importance of considering causes of mortality separately.

This study has several limitations. Firstly, the HPFS cohort is restricted to midlife/older male health professionals, who represent a relatively high socioeconomic status; participants are also predominantly White and particularly healthy at baseline. Although such homogeneity enhances the study internal validity, it restricts the generalizability of findings. In fact, accumulating evidence suggests that anger has different health consequences for men versus women (1, 10, 12). Additionally, prior research showed that the experience and aggressive expression of anger would be more pronounced in younger vs. midlife/older individuals, especially among those from lower socioeconomic status (39, 40), suggesting that the current results might underestimate anger’s role in mortality. Moreover, existing preliminary results hinted to the importance of race and culture. For instance, a study found greater aggressive expression of anger was related to a poorer biological profile in Americans but with a healthier one in Japanese (35), perhaps because among Japanese, expressing anger symbolizes authority and empowerment, whereas among Americans it is more often used for venting irritation and annoyances (41). Moreover, in prior work investigating anger frequency and aggressive expression in relation to interleukin-6 and fibrinogen levels, distinct patterns were noted between educated African Americans versus Whites (42). Secondly, both anger measures were queried only at baseline, which may overlook potential changes in anger frequency and aggressive expression over the lifecourse, and additional anger constructs that may relate to various causes of death were not evaluated in this cohort (e.g., suppression). Thirdly, anger frequency was assessed with a single item; yet, single items capturing either anger suppression (21) or other negative emotion frequency (43) have been found to be predictive of mortality in other samples.

These limitations are balanced by numerous strengths, including a large sample of participants from a well-characterized prospective cohort, who were free of chronic diseases at the study onset. The availability of multiple covariates allowed the consideration of factors that could either confound or potentially mediate the association of anger with mortality risk. Moreover, the 20-year follow-up permitted the occurrence of long-term processes that may eventually lead to death, like atherogenesis and carcinogenesis, as well as the inclusion of a 2-year lag to reduce concerns about reverse causation. Lastly, considering aggressive expression and its interaction with frequency of angry feelings offered a more comprehensive understanding of whether and how this emotion is associated with mortality risk.

In conclusion, anger frequency and aggressive expression were unrelated to increased risk of several causes of mortality among midlife/older men. However, regularly experiencing angry feelings and expressing them aggressively were related to increased CVD and cancer mortality risk, beyond sociodemographics and health status. Behavioral factors appeared to partly explain some of these associations and effect estimates became larger when considering potential synergistic effects of anger frequency and aggressive expression. Although the magnitude of effects was generally modest in these midlife/older men, in the context of high levels of injustice and conditions that give rise to anger (e.g., structural racism, unequal access to protection during the pandemic), the impact at the population level is potentially meaningful.

The current study suggests that not only the experience of negative emotions but also the way they are managed may be critical for longevity. Considering causes of death separately when investigating psychosocial risk factors of mortality will also be important to truly understand the relationship. Given prior work demonstrating health protective effects of more adaptive ways of regulating anger (e.g., constructive or assertive expression) (44, 45), future research should consider if such strategies may mitigate potentially toxic effects of anger frequency in relation to cause-specific mortality across adulthood. Considering the context in which emotion regulation strategies are used (e.g., in relation to a spouse’s coping style, when facing an authority figure versus a loved one) will also be critical for understanding how anger may influence mortality risk (14, 22, 46).

Supplementary Material

FINAL PRODUCTION FILE: SDC

Table 2.

Association of anger frequency and aggressive expression, respectively (per 1-SD), with risk of all-cause and cause-specific mortality.

Person-years (death cases) Model 1: sociodemographics HR (95%CI) Model 2: M1 + health status HR (95%CI) Model 3: M2 + health behaviors HR (95%CI)
Anger frequency
All-cause 284,683 (4881) 1.03 (1.00–1.06)† 1.03 (0.99–1.06)† 1.02 (0.99–1.06)
CVD 284,683 (1324) 1.08 (1.02–1.15)** 1.07 (1.01–1.14)* 1.07 (1.01–1.14)*
Cancer 284,683 (1378) 1.04 (0.98–1.10) 1.04 (0.98–1.10) 1.03 (0.97–1.09)
Respiratory 284,683 (422) 0.90 (0.81–1.01)† 0.90 (0.80–1.01)† 0.90 (0.80–1.00)†
Neurological 284,683 (250) 1.07 (0.93–1.23) 1.07 (0.93–1.23) 1.06 (0.92–1.22)
Anger aggressive expression
All-cause 284,683 (4881) 1.03 (1.00–1.06)† 1.03 (0.99–1.06) 1.01 (0.98–1.04)
CVD 284,683 (1324) 1.07 (1.01–1.14)* 1.06 (1.00–1.13)† 1.05 (0.99–1.12)
Cancer 284,683 (1378) 1.06 (1.00–1.12)* 1.06 (1.00–1.12)* 1.03 (0.98–1.09)
Respiratory 284,683 (422) 0.97 (0.86–1.08) 0.96 (0.86–1.08) 0.93 (0.83–1.04)
Neurological 284,683 (250) 0.94 (0.81–1.09) 0.94 (0.81–1.09) 0.94 (0.81–1.08)
Open in a new tab

Notes.

†

p<.10

*

p<.05

**

p<.01; CI=confidence interval, CVD=cardiovascular disease, HR=hazard ratio, SD=standard deviation.

Model 1: stratified by age and time period, and adjusted for marital status and profession;

Model 2: M1 + high blood pressure, high cholesterol, body mass index;

Model 3: M2 + smoking, alcohol consumption, physical activity, diet quality.

ACKNOWLEDGMENTS

We would like to thank the participants and the staff of the Health Professionals Follow-Up Study from the Channing Division of Network Medicine, Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School, Boston (MA), USA for their valuable contributions, as well as the following state cancer registries for their help: AL, AZ, AR, CA, CO, CT, DE, FL, GA, ID, IL, IN, IA, KY, LA, ME, MD, MA, MI, NE, NH, NJ, NY, NC, ND, OH, OK, OR, PA, RI, SC, TN, TX, VA, WA, WY. The authors assume full responsibility for analyses and interpretation of these data. The Health Professionals Follow-Up Study is supported by grants U01 CA167552 and R01 HL35464 by the National Institutes of Health. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Source of Funding: The Health Professionals Follow-Up Study is supported by grants U01 CA167552 and R01 HL35464 by the National Institutes of Health. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

LIST OF ABBREVIATIONS

CI

confidence interval

CVD

cardiovascular disease

HPFS

Health Professionals Follow-up Study

HR

hazard ratio

SD

standard deviation

Footnotes

Conflict of Interest

The authors report no conflict of interest.

REFERENCES

  • 1.Smith TW. Hostility and health: Current status of a psychosomatic hypothesis. Health Psychol. 1992;11(3):139–50. [DOI] [PubMed] [Google Scholar]
  • 2.Martin R, Watson D, Wan CK. A three-factor model of trait anger: Dimensions of affect, behavior, and cognition. Journal of Personality. 2000;68(5):869–97. [DOI] [PubMed] [Google Scholar]
  • 3.Suls J Anger and the heart: Perspectives on cardiac risk, mechanisms and interventions. Prog Cardiovasc Dis. 2013;55(6):538–47. [DOI] [PubMed] [Google Scholar]
  • 4.Appleton KM, Woodside JV, Arveiler D, Haas B, Amouyel P, Montaye M, Ferrieres J, Ruidavets JB, Yarnell JW, Kee F, Evans A, Bingham A, Ducimetiere P, Patterson CC, group Ps. A role for behavior in the relationships between depression and hostility and cardiovascular disease incidence, mortality, and all-cause mortality: the Prime Study. Ann Behav Med. 2016;50(4):582–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Tindle HA, Chang YF, Kuller LH, Manson JE, Robinson JG, Rosal MC, Siegle GJ, Matthews KA. Optimism, cynical hostility, and incident coronary heart disease and mortality in the Women’s Health Initiative. Circulation. 2009;120(8):656–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Boscarino JA, Figley CR. The impact of repression, hostility, and post-traumatic stress disorder on all-cause mortality: A prospective 16-year follow-up study. J Nerv Ment Dis. 2009;197(6):461–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Klabbers G, Bosma H, van den Akker M, Kempen GI, van Eijk JT. Cognitive hostility predicts all-cause mortality irrespective of behavioural risk at late middle and older age. Eur J Public Health. 2013;23(4):701–5. [DOI] [PubMed] [Google Scholar]
  • 8.DeSteno D, Gross JJ, Kubzansky L. Affective science and health: The importance of emotion and emotion regulation. Health Psychol. 2013;32(5):474–86. [DOI] [PubMed] [Google Scholar]
  • 9.Trudel-Fitzgerald C, Gilsanz P, Mittleman MA, Kubzansky LD. Dysregulated blood pressure: Can regulating emotions help? Curr Hypertens Rep. 2015;17(12):92. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Eaker ED, Sullivan LM, Kelly-Hayes M, D’Agostino RB, Sr., Benjamin EJ. Anger and hostility predict the development of atrial fibrillation in men in the Framingham Offspring Study. Circulation. 2004;109(10):1267–71. [DOI] [PubMed] [Google Scholar]
  • 11.Suls J, Bunde J. Anger, anxiety, and depression as risk factors for cardiovascular disease: the problems and implications of overlapping affective dispositions. Psychological bulletin. 2005;131(2):260–300. [DOI] [PubMed] [Google Scholar]
  • 12.Chida Y, Steptoe A. The association of anger and hostility with future coronary heart disease: A meta-analytic review of prospective evidence. Journal of the American College of Cardiology. 2009;53(11):936–46. [DOI] [PubMed] [Google Scholar]
  • 13.Bensley KM, Seelig AD, Armenta RF, Rivera AC, Peterson AV, Jacobson IG, Littman AJ, Maynard C, Bricker JB, Boyko EJ, Rull RP, Williams EC. Posttraumatic stress disorder symptom association with subsequent risky and problem drinking initiation. J Addict Med. 2018;12(5):353–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Bongard S, Olson L, Nakajima M, al’Absi M. Anger expression style predicts the domain of the first smoking relapse after a quit attempt. Subst Use Misuse. 2016;51(13):1810–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Mostofsky E, Penner EA, Mittleman MA. Outbursts of anger as a trigger of acute cardiovascular events: A systematic review and meta-analysis. European heart journal. 2014;35(21):1404–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Thomas KS, Nelesen RA, Dimsdale JE. Relationships between hostility, anger expression, and blood pressure dipping in an ethnically diverse sample. Psychosomatic medicine. 2004;66(3):298–304. [DOI] [PubMed] [Google Scholar]
  • 17.Everson SA, Goldberg DE, Kaplan GA, Julkunen J, Salonen JT. Anger expression and incident hypertension. Psychosomatic medicine. 1998;60(6):730–5. [DOI] [PubMed] [Google Scholar]
  • 18.Shen BJ, Countryman AJ, Spiro A, 3rd, Niaura R. The prospective contribution of hostility characteristics to high fasting glucose levels: The moderating role of marital status. Diabetes Care. 2008;31(7):1293–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Lacombe J, Armstrong MEG, Wright FL, Foster C. The impact of physical activity and an additional behavioural risk factor on cardiovascular disease, cancer and all-cause mortality: A systematic review. BMC Public Health. 2019;19(1):900. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Lu Y, Monaco G, Camous X, Andiappan AK, Rotzschke O, Ng TP, Larbi A. Biomarker signatures predicting 10-year all-cause and disease-specific mortality. J Gerontol A Biol Sci Med Sci. 2019;74(4):469–79. [DOI] [PubMed] [Google Scholar]
  • 21.Chapman BP, Fiscella K, Kawachi I, Duberstein P, Muennig P. Emotion suppression and mortality risk over a 12-year follow-up. Journal of psychosomatic research. 2013;75(4):381–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Harburg E, Julius M, Kaciroti N, Gleiberman L, Schork MA. Expressive/suppressive anger-coping responses, gender, and types of mortality: A 17-year follow-up (Tecumseh, Michigan, 1971–1988). Psychosomatic medicine. 2003;65(4):588–97. [DOI] [PubMed] [Google Scholar]
  • 23.Eng PM, Fitzmaurice G, Kubzansky LD, Rimm EB, Kawachi I. Anger expression and risk of stroke and coronary heart disease among male health professionals. Psychosomatic medicine. 2003;65(1):100–10. [DOI] [PubMed] [Google Scholar]
  • 24.Haukkala A, Konttinen H, Laatikainen T, Kawachi I, Uutela A. Hostility, anger control, and anger expression as predictors of cardiovascular disease. Psychosomatic medicine. 2010;72(6):556–62. [DOI] [PubMed] [Google Scholar]
  • 25.White VM, English DR, Coates H, Lagerlund M, Borland R, Giles GG. Is cancer risk associated with anger control and negative affect? Findings from a prospective cohort study. Psychosomatic medicine. 2007;69(7):667–74. [DOI] [PubMed] [Google Scholar]
  • 26.Gibson DE, Callister RR. Anger in organizations: Review and integration. Journal of Management. 2010;36(1):66–93. [Google Scholar]
  • 27.Suls J. Toxic affect: Are anger, anxiety, and depression independent risk factors for cardiovascular disease? Emotion Review. 2018;10(1):6–17. [Google Scholar]
  • 28.Rimm EB, Stampfer MJ, Colditz GA, Giovannucci E, Willett WC. Effectiveness of various mailing strategies among nonrespondents in a prospective cohort study. Am J Epidemiol. 1990;131(6):1068–71. [DOI] [PubMed] [Google Scholar]
  • 29.Tsai AC, Lucas M, Sania A, Kim D, Kawachi I. Social integration and suicide mortality among men: 24-year cohort study of U.S. health professionals. Ann Intern Med. 2014;161(2):85–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Gill TK, Price K, Dal Grande E, Daly A, Taylor AW. Feeling angry about current health status: Using a population survey to determine the association with demographic, health and social factors. BMC Public Health. 2016;16:588. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Spielberger CD. State-Trait Anger Expression Inventory: Professional manual. . Odessa (FL): Psychological Assessment Resources; 1988. [Google Scholar]
  • 32.Jasinski MJ, Lumley MA, Latsch DV, Schuster E, Kinner E, Burns JW. Assessing anger expression: Construct validity of three emotion expression-related measures. J Pers Assess. 2016;98(6):640–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.McCullough ML, Feskanich D, Stampfer MJ, Giovannucci EL, Rimm EB, Hu FB, Spiegelman D, Hunter DJ, Colditz GA, Willett WC. Diet quality and major chronic disease risk in men and women: Moving toward improved dietary guidance. Am J Clin Nutr. 2002;76(6):1261–71. [DOI] [PubMed] [Google Scholar]
  • 34.Rich-Edwards JW, Corsano KA, Stampfer MJ. Test of the National Death Index and Equifax Nationwide Death Search. Am J Epidemiol. 1994;140(11):1016–9. [DOI] [PubMed] [Google Scholar]
  • 35.Kitayama S, Park J, Boylan JM, Miyamoto Y, Levine CS, Markus HR, Karasawa M, Coe CL, Kawakami N, Love GD, Ryff CD. Expression of anger and ill health in two cultures: An examination of inflammation and cardiovascular risk. Psychol Sci. 2015;26(2):211–20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.VanderVoort DJ, Ragland DR, Syme SL. Expressed and suppressed anger and health problems among transit workers. Current Psychology: Developmental • Learning ° Personality ° Social Summer. 1996;15(2):179–93. [Google Scholar]
  • 37.Russell MA, Smith TW, Smyth JM. Anger expression, momentary anger, and symptom severity in patients with chronic disease. Ann Behav Med. 2016;50(2):259–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.American Cancer Society. Pancreatic cancer risk factors: American Cancer Society; 2020. [Available from: https://www.cancer.org/cancer/pancreatic-cancer/causes-risks-prevention/risk-factors.html. [Google Scholar]
  • 39.Phillips LH, Henry JD, Hosie JA, Milne AB. Age, anger regulation and well-being. Aging & mental health. 2006;10(3):250–6. [DOI] [PubMed] [Google Scholar]
  • 40.Schieman S. Socioeconomic status and the frequency of anger across the life course. Sociological Perspectives. 2003;46(2):207–22. [Google Scholar]
  • 41.Park J, Kitayama S, Markus HR, Coe CL, Miyamoto Y, Karasawa M, Curhan KB, Love GD, Kawakami N, Boylan JM, Ryff CD. Social status and anger expression: The cultural moderation hypothesis. Emotion. 2013;13(6):1122–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Boylan JM, Lewis TT, Coe CL, Ryff CD. Educational status, anger, and inflammation in the MIDUS national sample: Does race matter? Ann Behav Med. 2015;49(4):570–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Lefevre T, Singh-Manoux A, Stringhini S, Dugravot A, Lemogne C, Consoli SM, Goldberg M, Zins M, Nabi H. Usefulness of a single-item measure of depression to predict mortality: The GAZEL prospective cohort study. Eur J Public Health. 2012;22(5):643–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Davidson KW, Mostofsky E. Anger expression and risk of coronary heart disease: Evidence from the Nova Scotia Health Survey. Am Heart J. 2010;159(2):199–206. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Muller MM, Rau H, Brody S, Elbert T, Heinle H. The relationship between habitual anger coping style and serum lipid and lipoprotein concentrations. Biol Psychol. 1995;41(1):69–81. [DOI] [PubMed] [Google Scholar]
  • 46.Bourassa KJ, Sbarra DA, Ruiz JM, Karciroti N, Harburg E. Mismatch in spouses’ anger-coping response styles and risk of early mortality: A 32-year follow-up study. Psychosomatic medicine. 2019;81(1):26–33. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

FINAL PRODUCTION FILE: SDC
NIHMS1694794-supplement-FINAL_PRODUCTION_FILE__SDC.docx (48.5KB, docx)

RESOURCES