OUP user menu

Chronic psychosocial stress predicts long-term cardiovascular morbidity and mortality in middle-aged men

B Öhlin, P.M Nilsson, J-Å Nilsson, G Berglund
DOI: http://dx.doi.org/10.1016/j.ehj.2004.03.003 867-873 First published online: 2 May 2004


Aims To prospectively investigate the association between self-reported psychosocial stress and long-term cardiovascular (CV) morbidity and mortality in a population-based cohort.

Methods and results The Malmö Preventive Project is a population-based screening and intervention programme for cardiovascular disease (CVD) risk factors. Between 1974 and 1980, a total of 13,609 (2741 women) individuals, mean age 45 years, had self-reported chronic stress determined by questionnaire. CV morbidity and mortality were followed up in national registries. Median follow-up time was 21 years.

The risk ratio (RR) for a fatal or nonfatal CV incident in the men and women of the group reporting chronic stress was 1.27 (95% CI 1.15–1.39). After stepwise adjustments for known CV risk factors, the RR was reduced to 1.14 (1.02–1.28). The highest RR was found for fatal stroke in men reporting chronic stress, 2.04 (1.07–3.88). For women alone, there was no significant increase in risk after adjustments.

Conclusion Self-reported chronic stress is an independent risk factor for CVD, particularly fatal stroke, in middle-aged men; it continues to be a risk factor after adjustment for several other known risk factors. The adjustment itself might reflect mechanisms whereby psychosocial stress directly or indirectly exerts its effects on the body, indicating a possible over-adjustment.

  • Cardiovascular
  • Epidemiology
  • Mortality
  • Morbidity
  • Stress


Atherosclerosis and its consequences are the major cause of death in the Western world.1 It has long been suggested that psychosocial factors play an important role in the development of atherosclerosis,2 as well as in the acute onset of cardiovascular events, for example, angina pectoris and myocardial infarction.3 The mechanisms whereby psychosocial stress causes cardiovascular disease (CVD) are not yet fully understood, although several investigators have contributed data and hypotheses.4–9 (see Fig. 1).

Fig. 1

Chronic stress and the risk of suffering a CVD endpoint. Men and Women. Unadjusted. Embedded Image.

There is a growing body of epidemiologic evidence that psychosocial factors are linked to CVD.10–12 In a systematic review of prospective cohort studies of psychosocial factors and coronary heart disease (CHD), it was concluded that there is strong evidence of an aetiologic role of psychosocial factors, especially depression and lack of social support.13 There is, however, a report in which no correlation was found between self-reported stress and CHD in a population of Scottish men.14 Relatively few large, prospective, population-based trials with clinical endpoints have been made. Rosengren and Harmsen found that self-perceived stress predicted CHD15 and stroke16 in a population-based cohort of men in Gothenburg, Sweden. More recently, May et al.17 reported that psychologic distress increases the risk of suffering a fatal stroke. At follow-up, however, this community-based study of 2201 Welsh middle-aged men only registered 17 fatal strokes; after full adjustment, the increase in risk ceased to be significant. In a larger study from Japan that included 73,242 individuals with follow-up during 7.9 years, perceived stress was significantly associated with increased mortality from stroke among women, and with CHD for both men and women.18

In summary, based on growing epidemiologic evidence there seems to be an association between psychosocial stress and CVD. However, the results are somewhat disparate, for instance, the Japanese report on different risk scenarios for men compared to women and the Scottish report of no relationship. The risk of residual confounding due to social class has been pointed out.14 Since Macleod's report, there has been debate on the methodology of studies based on self-reported exposure and outcomes.19 After reviewing these controversies, we proposed to prospectively investigate the capability of self-reported chronic stress to predict CVD risk, subclassified into acute coronary syndromes (ACS), ischaemic heart disease (IHD), and stroke, in a Swedish urban population using strict endpoint criteria and controlling for known confounders, including a separate stratified analysis for socio-economic status.



The Malmö Preventive Project Math was a population-based screening and intervention programme 1974–1992 for CVD risk factors. The main results have been published earlier.20 Selected birth cohorts were invited for screening and the mean participation rate was 71%. All participants answered questions about perceived stress (between 1974 and 1980); a total of 13,609 individuals were included at baseline (74% of those invited, 10,868 men, mean age 46 years, median follow-up time 21.3 years and 2741 women, mean age 42 years, median follow-up time 21.2 years). A total of 329 subjects were excluded from the final analysis due to a self-reported history of myocardial infarction, angina pectoris, stroke, or cancer before the baseline examination.


The questionnaire was computerised in such a way that participants were shown one question at a time on a screen and had to select a “yes” or “no” answer before they could move on to the next question. The participants could not omit a question, that is, the questionnaire did not allow blank answers. Several questions targeted the level of stress experienced. Based on earlier research15,16 and our interest in chronic stress, we selected two questions to create a combined-stress score: (1) “Have you experienced permanent stress (defined as a feeling of tension, irritability, or anguish) during the last year?” and (2) “Have you experienced permanent stress (defined as a feeling of tension, irritability, or anguish) during the last five years?” Affirmative answers for both questions resulted in a score of 2 and two negative answers yielded a score of 0. A score of 1 was assigned to individuals who had experienced stress only during the last year. Thus, three stress-groups were created; 0 (score 0=low stress), 1 (score 1=medium stress), 2 (score 2=high stress).

When all the individual data were reviewed, 681 individuals had answered in a seemingly illogical manner: They answered “no” to question 1 but “yes” to question 2. We interpreted this as meaning that the individual acknowledged a heavy burden of stress during the last five years, with improvement in the last year. Despite this hypothetical last-year improvement, we decided that in terms of risk factors the only reasonable score for these individuals should be 2.

Family history of CVD (a first-degree relative with either stroke or acute myocardial infarction), smoking habits, alcohol abuse, marital status, occupational class, and leisure time physical activity were determined by the subjects' answers to the questions. There were five questions on leisure-time physical activity. A compound variable was constructed with a possible score of 0–4, with the high score indicating a high level of leisure-time physical activity. To evaluate the extent of problematic drinking habits, a variable based on Mm MAST-questions 21 was in the questionnaire. If the subject gave more than two affirmative answers, the subject was considered to be at risk of alcohol abuse. Smoking habits were characterised as “nonsmoker” or “smoker” based on the subjects' questionnaire answers. The term “marital status” was modified to include in the group “living together” married persons and unmarried couples living together. All others were assigned to the group “living alone”.

Prevalent disease at baseline was also determined by the subject's answers. For angina pectoris, a surrogate variable was used: “Do you use nitro-glycerine medication?” This variable was assumed to give a more accurate picture of the prevalence of angina than less precise questions like “Have you ever had chest pain?”

The socio-economic index (SEI) was estimated by occupation, as reported in Hedblad 2002.22 This variable has been transformed into three categories: white-collar workers, blue-collar workers, and others (farmers, unemployed, unclassified, and self-employed).

Physical examination

All subjects were examined for height (m) and weight (kg) in light indoor clothing. The body mass index was calculated (kg/m2). Blood pressure (BP, mmHg) and pulse rate (beats min−1) were measured twice after a 10-min rest in supine position using appropriate technical equipment (a sphygmomanometer with a proper cuff width and a mercury manometer), and the mean value was recorded.

Laboratory investigations

The following fasting variables were analysed using routine methods at the Department of Clinical Chemistry, Malmö University Hospital: serum total cholesterol, serum triglycerides, serum γ-glutamyl transferase (GGT), and blood glucose.

Registry follow-up

All subjects were followed up in local as well as national registries for total mortality, cause-specific mortality, nonfatal ischaemic heart disease, and nonfatal stroke. Registry data recorded up to 31 December 1999 were used.

Acute coronary syndrome (ACS) was defined as diagnostic code 410 of the 8th and 9th revisions of the International Classification of Diseases (ICD). From the ICD-9 revision, code 411B was added. From ICD-10, codes I 200, I 21 and I 22 were used. Within the concept of ischaemic heart disease (IHD), in-hospital diagnosis of angina pectoris (ICD-8 and ICD-9: 413; ICD-10: I 20) was included. For mortality analyses, the following diagnoses of IHD were included: ICD-8 and ICD-9: 412, 414, 427E-F (ICD-9) and ICD-10: I 23, I 25, I 461, I 469 and R 960.

Stroke mortality and morbidity were defined by ICD-8 and ICD-9 codes 431, 433, 434 and 436. In ICD-10, I 61, I 629, I 63 (not I 636) and I 64 were used. Transitory ischemic attack (TIA) was not included as an endpoint.

A cardiovascular event was defined as a stroke or an IHD event, rendering the diagnosis as specified above. The incidence of CV events was reported with (IHD+stroke) and without angina pectoris (ACS+stroke). Only the first CV event was taken into account and only the main diagnosis was used.

The Committee of Ethics in Research, University of Lund, Sweden, approved the registry follow-up.

Statistical analysis

The statistical analyses were carried out with SPSS for Windows, version 10.0.5 (1999). Baseline differences in means (SD) and proportions were tested with the Mann–Whitney Math-test for discrete variables and Spearman correlation coefficient (Math) for continuous variables. Socio-economic status was analysed with the Kruskal–Wallis test. All tests were two-sided. The risk ratios (RR) with 95% confidence intervals (CI) were calculated using the Cox proportional risk hazard model. Confidence intervals were calculated based on the chi-square distribution. In survival analyses, stress groups 1+2 were compared to stress group 0 in an attempt to separate subjects with chronic stress, irrespective of duration, from subjects without self-reported chronic stress. For continuous covariates, risk ratios for the subgroups under and above the median value for each group were calculated. The risk ratios are in the same range, with overlapping confidence intervals, supporting the assumption of linearity. Also, log-minus-log plots of the survival curves were made, plotting the log-scale on the Math-axis; these lines were approximately straight, indicating a proportional hazard between MathMath.

For each of the discrete covariates, Kaplan–Meier plots indicate that the assumption of proportional hazard is valid.

Stepwise adjustments were made, starting with unmodifiable risk factors (Model A): family history of CVD and age. Thereafter, further adjustments were made for social (occupation, marital status) and lifestyle factors (smoking, alcohol, leisure time physical activity) (Model B). Finally, adjustments for known mediating “biologic” risk factors (lipids, diastolic BP, and BMI) (Model C) were added. Triglyceride concentrations were log-transformed. In the analyses of the entire cohort (men and women together) adjustments were made for sex.

It has been speculated that individuals with high self-reported psychologic stress would report symptoms more often.14,23 This could result in more extensive medical investigations that would increase the probability of receiving a diagnosis. Presumably, these diagnoses would be “softer” and angina pectoris would be one of these diagnoses. Therefore, we report the results with and without angina diagnoses made after in-hospital care. A Math-value of less than 0.05 was considered significant.


Baseline characteristics

At baseline, significant differences were found between the stress-score groups for males and females in family history of CVD, smoking habits, alcohol risk profile, “living alone”, and GGT (Tables 1 and 2). White-collar workers of both sexes were over-represented in the chronic stress groups (stress 1 and 2), but only significantly amongst men. In the male cohort, significant differences at baseline were also found in leisure physical activity, BMI, pulse rate, and diastolic blood pressure (Table 1).

View this table:
Table 1

Baseline characteristics of men with different degrees of reported psychosocial stress

Risk factorsLow stress Embedded ImageMedium stress Embedded ImageHigh stress Embedded ImageEmbedded Image for trend
Age, years46 (4.7)45 (5.5)46 (4.3)0.12
Family historya0.370.360.450.01
Smokers, %5053560.001
At risk of alcohol abuse, %
Physical activityb2.40 (0.81)2.31 (0.85)2.25 (0.87)0.01
Living alone, %21.524.829.00.01
Occupational class, %
Blue collar45.438.141.1
White collar45.749.746.30.001
BMI, kg/m224.8 (3.3)24.6 (3.4)25.1 (3.5)0.05
Pulse rate, beats/min67.9 (10.2)68.1 (10.7)68.7 (10.6)0.05
BP syst, mmHg129.1 (15.8)129.3 (16.3)129.5 (16.0)0.36
BP diast, mmHg86.8 (10.2)86.9 (10.2)88.0 (10.3)0.01
Chol, mmol/l5.72 (1.07)5.59 (1.14)5.76 (1.42)0.43
TG, mmol/l1.57 (1.02)1.61 (0.95)1.70 (1.17)0.01
GT, μkat/l0.71 (0.92)0.81 (1.52)0.88 (1.31)0.01
Fasting glucose, mmol/l4.93 (1.00)4.98 (1.00)4.95 (1.03)0.08
Prevalent disease, Embedded Image (%)
AMI61 (0.7)7 (1.2)30 (1.8)0.001
Angina51 (0.6)10 (1.8)26 (1.6)0.001
Stroke16 (0.2)2 (0.2)9 (0.5)0.01
Cancer57 (0.7)6 (1.1)17 (1.0)0.06
  • Mean values (SD) and proportions (%) for different groups.

  • a Family history of AMI, stroke (number of first-degree relatives, mean in each group).

  • b 0=low physical activity, 4=high.

View this table:
Table 2

Baseline characteristics of women with different degrees of reported psychosocial stress

Risk factorsLow stress Embedded ImageMedium stress Embedded ImageHigh stress Embedded ImageEmbedded Image for trend
Age, years42 (8.6)39 (8.1)42 (8.8)0.52
Family historya0.350.380.470.01
Smokers, %4055500.001
% at risk of alcohol abuse0.2 Embedded ImageEmbedded Image2.7 Embedded Image0.001
Physical activityb2.36 (0.78)2.31 (0.82)2.26 (0.84)0.06
Living alone, %27.337.838.40.01
Occupational class, %
Blue collar43.934.648.2
White collar52.259.948.20.96
BMI, kg/m223.0 (3.8)22.8 (3.9)23.5 (4.0)0.19
Pulse, beats/min69.9 (9.6)70.1 (9.9)70.1 (10.6)0.83
BP syst, mmHg120.4 (15.3)117.4 (13.3)120.6 (15.4)0.29
BP diast, mmHg80.3 (9.6)78.7 (9.2)81.2 (10.0)0.97
Chol, mmol/l5.31 (1.03)5.13 (0.89)5.37 (1.09)0.93
TG, mmol/l1.18 (0.60)1.2 (0.56)1.29 (0.81)0.08
GT, μkat/l0.40 (0.56)0.39 (0.27)0.53 (0.86)0.01
Fasting glucose, mmol/l4.84 (0.72)4.83 (0.73)4.92 (0.86)0.08
Prevalent disease, Embedded Image (%)
Cancer47 (2.2)7 (3.6)16 (4.3)0.01
  • Mean values (SD) and proportions (%) for the different groups.

  • a Family history of AMI, stroke (number of first-degree relatives, mean in each group).

  • b 0=low physical activity, 4=high.

In both men (Table 1) and women (Table 2), a previous history of myocardial infarction, angina pectoris, stroke, or cancer was more common among subjects who reported a perceived stress of level 1 or 2 than in those without perceived stress. Persons with a history of these diagnoses were excluded from further analyses.

Chronic stress and risk of CVD

The risk ratio (RR) for suffering a fatal or nonfatal CV event in the entire cohort of men and women, stress groups 1 and 2 (medium-high) combined versus stress group 0 (low stress), after full adjustment, was 1.14 (SD 1.008–1.30). Use of ACS criteria did not change the outcome. The fully adjusted RR for IHD was 1.15 (1.02–1.31) and for ACS, 1.10 (0.95–1.27). For fatal or nonfatal stroke, the fully adjusted RR was 1.29 (1.04–1.60), and for fatal stroke the corresponding RR was 2.39 (1.31–4.35).

In the male cohort, 2504 first CVD events were registered. The RR for suffering an IHD event (fatal and nonfatal) in the chronic stress-group was 1.30 (1.16–1.44), adjusted for age and family history of CVD. The fully adjusted RR was 1.17 (1.02–1.33). The risk of receiving a diagnosis of ACS was slightly increased for men reporting chronic stress, although only in Model A. The unadjusted RR for fatal and nonfatal stroke was 1.42 (1.18 –1.71) in the chronic stress-group, and the fully adjusted RR was 1.25 (0.97–1.58). The greatest increase in risk in this study after full adjustments was for fatal stroke in the male chronic stress-group, with RR 2.04 (1.07–3.88). In general, men reporting chronic stress had an increased risk of suffering any of the CV endpoints, although the risk of ACS and stroke (fatal and nonfatal) ceased to be statistically significant after full adjustment (Table 3).

View this table:
Table 3

Numbers Embedded Image and proportions (%) of first cardiovascular events and deaths in groups of men Embedded Image with different degrees of self–reported chronic stress, Embedded Image (%)

Total Embedded Image person–yearsLow stress Embedded ImageMedium stress Embedded ImageHigh stress Embedded ImageModel AModel BModel C (fully adjusted)
IHD1475 (17.4)111 (20.2)335 (21.2)1.30 (1.16–1.44)1.19 (1.04–1.35)1.17 (1.02–1.33)
ACS1140 (13.4)72 (13.1)261 (16.5)1.24 (1.10–1.40)1.14 (0.97–1.32)1.12 (0.97–1.30)
Stroke438 (5.2)30 (5.5)115 (7.3)1.42 (1.18–1.71)1.30 (1.03–1.65)1.25 (0.97–1.58)
CV (IHD+stroke)1775 (21)128 (23.3)409 (25.8)1.29 (1.17–1.42)1.21 (1.07–1.36)1.18 (1.04–1.32)
CV (ACS+stroke)1481 (17.4)97 (17.7)349 (22.0)1.29 (1.16–1.43)1.19 (1.04–1.35)1.16 (1.02–1.32)
IHD478 (5.6)33 (6.0)110 (6.9)1.25 (1.04–1.51)1.09 (0.86–1.38)1.12 (0.88–1.42)
Stroke49 (0.6)7 (1.3)16 (1.0)1.97 (1.20–3.23)2.09 (1.10–3.99)2.04 (1.07–3.88)
IHD+stroke527 (6.2)40 (7.3)126 (8.0)1.32 (1.11–1.57)1.17 (0.93–1.46)1.20 (0.96–1.49)
All-cause mortality1605 (18.9)108 (19.7)361 (22.8)1.22 (1.10–1.36)1.05 (0.92–1.19)1.06 (0.93–1.20)
  • Risk ratios of medium stress + high stress versus low stress, 95% CI.

    Adjustment models: A=unmodifiable risk-factors: age, family history of CVD. B=A+social factors (occupational class, marital status) and lifestyle factors (smoking, drinking, leisure time physical activity), C=A+B+biologic parameters (BMI, diastolic BP, cholesterol and TG). ACS: acute coronary syndromes.

In the female cohort, the total number of first CVD events was only 227. For coronary disease there were no significant correlations with chronic stress. After adjustments for age and family history, a twofold greater risk of suffering stroke (fatal and nonfatal) was observed in women reporting chronic stress. After further adjustments, however, the statistical significance was lost (Table 4).

View this table:
Table 4

Numbers Embedded Image and proportions (%) of first cardiovascular events and deaths in groups of women Embedded Image with different degrees of self-reported chronic stress, Embedded Image (%)

Total Embedded Image person–yearsLow stress Embedded ImageMedium stress Embedded ImageHigh stress Embedded ImageModel AModel BModel C (fully adjusted)
Morbidity and mortality
IHD130 (6.1)7 (3.7)30 (8.5)1.10 (0.71–1.70)1.14 (0.73–1.78)1.04 (0.66–1.63)
ACS57 (2.7)4 (2.1)10 (2.8)0.88 (0.47–1.64)0.98 (0.54–1.78)0.85 (0.46–1.58)
Stroke41 (1.9)3 (1.6)16 (4.5)1.99 (1.16–3.44)1.68 (0.95–2.97)1.48 (0.83–2.65)
CV (IHD+stroke)130 (6.1)7 (3.7)30 (8.5)1.16 (0.81–1.68)1.18 (0.82–1.72)1.08 (0.74–1.58)
CV (ACS+stroke)94 (4.4)6 (3.2)23 (6.5)1.27 (0.84–1.92)1.16 (0.75–1.79)1.04 (0.67–1.62)
IHD17 (0.8)1 (0.5)5 (1.4)
Stroke2 (0.1)1 (0.5)4 (1.1)
Stroke+IHD19 (0.9)2 (1.1)9 (2.5)2.35 (1.12–4.96)
All-cause mortality150 (7.1)14 (7.4)28 (7.9)1.16 (0.82–1.63)1.02 (0.70–1.47)0.97 (0.67–1.41)
  • Risk ratios medium+high stress versus low stress, 95% CI.

    Adjustments: A=inert factors: age, family history of CVD. B=A+social factors (occupational class, marital status) and lifestyle factors (smoking, drinking, leisure time physical activity), C=A+B+biologic parameters (BMI, diastolic BP, cholesterol and TG). ACS: acute coronary syndromes. Due to the small number of cases, the risk analyses of mortality in IHD and stroke were limited.

Very high increases in the risk of fatal stroke were observed amongst women reporting chronic stress, but these risks were based on very few cases (Table 4).

Social status

To elucidate whether social status as indicated by occupation influenced the frequency of self-reported chronic stress, a stratified analysis was made for different occupational groups. In an analysis of occupational groups for CVD incidents (fatal and nonfatal, adjustments for age and family history of CVD), it turned out that male white-collar workers (RR 1.37; 1.17–1.60 [medium+high stress vs. low stress]) were just as vulnerable to chronic stress as blue-collar workers (RR 1.25; 1.07–1.46). Similar findings were obtained in the female cohort. A stricter CVD classification using only ACS as the endpoint for coronary events did not markedly change the outcome (data not shown).


In this prospective, population-based, follow-up study of 13,280 individuals and 2731 verified events with a median follow-up time of 21 years, we found that self-reported chronic stress significantly increased the risk of suffering CV, IHD, or stroke events. In particular, stroke mortality seemed to be increased in individuals reporting chronic stress, with a more than twofold greater risk, which remained significant in the group as a whole and in the male cohort after full adjustments.

The limitations of our study were largely in the field of validity. Our baseline measurement of stress is based solely on unvalidated questions, although the questions have been used by other groups that obtained similar results.15,16 Since there were no follow-up investigations of stress after baseline, we do not know anything about the subjects' “stress status” after baseline. Repeated measurements of stress over time would have given us a more reliable estimate of chronic stress. However, it is unlikely that the design contributed to false-positive correlations. Rather, it is likely that the proposed relationship between chronic stress and CVD has been diluted as a result of the above-mentioned limitations.

The possibility of residual confounding due to social class has been pointed out.14 However, stratified analysis for the different occupational groups showed no significant difference in vulnerability due to stress in relation to social class in our study groups. Although occupation might be a somewhat crude indicator of social class, the stratified analysis makes it unlikely that our analyses were confounded by social disadvantage.

The baseline prevalence of myocardial infarction, angina pectoris, stroke, and cancer (criteria for exclusion) was surprisingly high. These self-reported data are probably an overestimate of the true prevalence of concurrent disease. It seems, however, unlikely that excluding these individuals from follow-up analysis would bias the results towards stronger correlations.

Despite the limitations of our method of estimating stress, we argue that there might be an advantage in our simplistic approach since it can be easily used in a clinical setting. However, there is an obvious need to develop more sensitive, validated instruments to detect individuals at risk. Marmot has pointed out the problems associated with use of the term “stress”.24 There is no academic consensus on the definition of stress and, subsequently, no “golden standard” stress test. Of course, this makes clinical research on stress-related outcomes per se difficult. While awaiting the development of more precise instruments, we feel that the method of operationalising stress (i.e., accepting the statement that a subject feels stressed at face value) is a reasonable alternative.

The risk ratios in our study are lower compared to similar studies with regard to ACS.15 This could be due to our longer follow-up time, which is known to dilute the influence of risk factors at baseline.25 The relative risk for stroke in individuals with chronic stress was comparable to other studies.16,17

The highest RR was observed for fatal stroke, whereas the increase in total stroke was moderate. Although the RR remained significant after full adjustment, the number of fatal strokes in the male cohort was rather low (73) and, after all, could have been a chance finding. Nonetheless, this finding is supported by other investigators.17,18 Instead of self-reported stress, May used the General Health Questionnaire (GHQ) as a baseline investigation for what was called “psychologic distress”.17 It is not self-evident that our baseline questions about chronic stress are comparable to the results of the GHQ, although it seems likely that these different instruments mirror to some extent the same feeling of “psychologic discomfort”. Since May's report of a specific association between psychologic distress and fatal stroke, but not nonfatal stroke, was the first of its kind, it is interesting to note how similar findings were obtained in a different population. There were 73 fatal strokes in our male cohort, whereas there were 17 in May's cohort.

Macleod et al.14 in a cohort of 5577 Scottish men, found no significant correlation between increased stress and CHD mortality or morbidity. Using only ACS criteria, our findings are also nonsignificant after full adjustment. This can represent a true noncorrelation, with an increase in the risk of IHD due to reporting bias. Alternative explanations could be that long-term follow-up diluted the effects of some baseline predictors25 and/or that the statistical power was insufficient. Also, the adjustments included some of the pathophysiologic pathways whereby stress may exert its effects. However, our data give no substantial evidence that ACS is linked to chronic stress.

It is notable that the RR for CVD risk in women reporting chronic stress was not significant, although the trend was toward increased risk. The women in this study had a lower mean age and fewer events than the men at inclusion time, which seems to be the most obvious explanation for the findings. Although interpretations should be made carefully due to the relatively small number of events, the impression is that chronic stress in women might be more correlated to stroke than IHD.

What are the determinants of chronic stress reactions? Not only is the cumulative stress load itself important, but also individual susceptibility determined by personality traits, such as hostility,24 early childhood experiences,26 and, possibly, genetically determined vulnerability to stress hormones.27 In adult life, stressors such as strained relations, long-term unemployment, and financial problems add to the stress load and may negatively impact the so-called allostatic load for physiologic reactions.28

Possible mechanisms for CV risk of stress have been described previously.4–9 Mechanisms that have been proposed and, to some extent, proven include hypertension, dyslipidemia, and an unhealthy lifestyle, including increased use of tobacco and alcohol, and less physical activity. Difficulties in managing stressful situations have been reported to be associated with stroke in hypertensive men.29

One might speculate that adjustments for the above-mentioned CV risk factors could inadequately reduce the observed risk ratios since they may be part of the chain of events leading from chronic stress to CVD. On the other hand, since the risk remains increased even after adjustment, it is plausible to assume that chronic stress also acts through pathophysiologic pathways that were not included in the adjustments.

In conclusion, our results indicate that self-reported chronic stress is an independent risk factor for cardiovascular disease in middle-aged men. The increase in risk was especially high for fatal stroke. For women, the trends are similar but nonsignificant after full adjustment. This can be due to a relatively small number of endpoints in the female cohort, but other explanations are possible. Our study was population-based with a high participation rate and it is reasonable to think that our data can be extrapolated to other large, middle-aged, Western populations. This epidemiologic study adds to the evidence that chronic psychosocial stress is linked to CVD in general and to fatal stroke in particular.


This study was supported by the City of Malmö, Swedish Medical Research Council, Swedish Heart and Lung Foundation, the Lundström Foundation, and the Hulda and E Conrad Mossfelt Foundation.


  • This study was funded by the City of Malmö, Swedish Medical Research Council, Swedish Heart and Lung Foundation, Ernhold Lundström Foundation, and the Hulda and E Conrad Mossfelt Foundation.


  1. [1]
  2. [2]
  3. [3]
  4. [4]
  5. [5]
  6. [6]
  7. [7]
  8. [8]
  9. [9]
  10. [10]
  11. [11]
  12. [12]
  13. [13]
  14. [14]
  15. [15]
  16. [16]
  17. [17]
  18. [18]
  19. [19]
  20. [20]
  21. [21]
  22. [22]
  23. [23]
  24. [24]
  25. [25]
  26. [26]
  27. [27]
  28. [28]
  29. [29]
View Abstract