Self -Perceived Psychological Stress and Incidence of Coronary Artery Disease in Middle-Aged Men Annika Rosengren, MD, GBsta Tibblin, MD, and Lars Wilhelmsen, MD
Self-perceived psychological stress as a risk factor for coronary artery disease (CAD) was evaluated in a general population study comprising 6,935 men aged 47 to 55 years at baseline without previous myocardial infarction. In 1970 to 1973, the men answered a question about psychological stress defined as a feeling of tension, irritability or anxiety, or as having sleeping difficulties as a result of conditions at work or at home. Psychological stress was graded as follows: (1) never experienced stress; (2) 21 period of stress; (3) I1 period of stress during the last 5 years; (4) several periods of stress during the last 5 years; and (5 to 6) permanent stress during the last year or the last 5 years. After a mean follow-up of 11.8 years, 6% of the men with the lowest 4 stress ratings (n = 5,665) had either devw a nonfatal myocardial infarction or died from CAD, with no increase in risk from grade 1 to 4. The corresponding flgure among the men with the highest 2 stress ratings (n = 1,070) was 10%; the odds ratio was 1.5 (95% conffdence interval 1.2-1.9) after controlling for age and other risk factors. Similar, independent associations were seen with stroke, and with death from cardiovascular disease and from all causes, but not with death from cancer. With respect to CAD, no decrease in the effect of stress at baseline could be seen over time. No relation between life events and self-perceived psychologieal stress was found in another sample of 732 fifty-year-okl men. (AmJCardiol1991;9&1171-1175)
From the Section of Preventive Cardiology, Department of Internal Medicine, Ostra Hospital, University of Giiteborg, Goteborg, Sweden, and the Department of Family Medicine, University of Uppsala, Sweden. This study was supported by the Swedish National Bank Tercentenary Fund, Stockholm, the Swedish National Association Against Heart and Cheat Diseases,Stockholm, the Giiteborg Medical Society, Gateborg, and the Knut and Alice Wallenberg Foundation, Stockholm. Manuscript received February 14, 1991; revised manuscript received June 12,1991, and accepted June 14. Address for reprints: Annika Rosengren, MD, Department of Medicine, Gstra Hospital, S-416 85 GBteborg, Sweden.
n recent years, epidemiologic research, concerning the etiology of coronary artery disease (CAD) has been increasingly directed toward environmental and psychosocial factors, such as socioeconomic status,1-2 social ~upport~-~ and occupational stress.5-6Inevitably, the concept of stress enters either implicitly or explicitly into most of this research. In the public eye, stress is commonly held to be one of the more important causes of myocardial infarction. A vast body of literature connects various psychosocial factors with CAD7; however, the evidence connecting CAD with stress, per se, is not very extensive. Stress research has traditionally focused primarily on the consequences of stressful stimuli for individual persons, not populations.* Recognizing the possible importance of stress as a risk factor, a question on stress was asked in 1970 to 1973 of all participants in a large prospective study on CAD in middle-aged men. In the present study, we analyzed cardiovascular and other end points during 11.8 year follow-up in relation to self-perceived stress at baseline.
I
METHODS The study population of the first part of this study comprises the intervention group of the Multifactor Primary Prevention Trial in Goteborg, the main results of which have been previously published.9 This study began in 1970 and included all men in Giiteborg who were born between 1915 and 1925, with the exception of those born in 1923. The intervention group of 10,000 men represents a random third of the men in the trial with 2 equally large control groups. In all, 7,495 men aged 47 to 55 years (75% of those invited) attended a screening examination that took place between January 1970 and March 1973. A second screening began in 1974 and was completed in 1977. All surviving men from the intervention group who still lived in Goteborg (n = 9,411) were invited to this second screening; the response rate was 76%. The same methods were used for both screenings. AI1 participants completed a postal questionnaire before the examination. The questionnaire contained an item in which stress was defined as feeling tense, irritable or filled with anxiety, or as having sleeping difficulties as a result of conditions at work or at home. The men were asked whether they had experienced this and to check 1 of the following choices: (1) never experienced stress; STRESS AND CAD
1171
TABLE I Incidence of Coronary Self-Perceived Stress Stress Rating Coronaj artery disease Stroke Death from: Cardiovascular disease Cancer Other causes All causes
Artery
Disease,
Stroke,
and Death from Cancer,
1 n = 1,162 % (n)
2 n = 1,494 % (n)
3 n = 1,300 %(n)
4 n = 1,909 %(n)
5 n = 281 % (n)
6.0 (70) 1.4 (16)
6.7 (100) 1.3 (20)
6.8 (89) 1.3 (17)
6.0 (114) 1.7 (33)
11.4 (32) 1.1 (3)
3.9 3.2 2.6 9.6
3.3 (49) 3.1(47) 1.8 (27) 8.2 (123)
3.8 3.0 2.0 8.8
3.1 2.5 2.6 8.2
5.0 4.3 4.3 13.5
(45) (37) (30) (112)
(50) (39) (26) (115)
(59) (48) (49) (156)
*Self-perceived stress rating 5 to 6 vs 1 to 4, unadjusted. tSelf-perceived stress rating 5 to 6 vs 1 to 4, adjusted forage, systolic blood pressure, serum cholesterol, occupational class, marital state, leisure time physical activity and registration for alcohol abuse. Cl = confidence interval.
(2) I: 1 period of stress; (3) some period of stress during the last 5 years; (4) several periods of stress during the last 5 years; (5) permanent stress during the last year; or (6) permanent stress during the last 5 years. The questionnaire also contained questions on smoking habits, physical activity during leisure, diabetes, previous myocardial infarction, symptoms of cardiovascular disease and family history of myocardial infarction. Physical activity during leisure was graded 1 to 4, with 1 denoting sedentary leisure time activity. Occupational class was defined as previously described.1o Data on alcohol abuse” were collected from official registers with special permission. At the screening examination, blood pressure was measured to the nearest 2 mm Hg after 5 minutes’ rest with the subject seated. Body mass index was calculated as wt/h@. Serum cholesterol was determined according to standard laboratory procedures. Follow-up extended through March 1983 (mean follow-up from the first screening 11.8 years, and from the second screening 7.1). All men with previous myocardial infarction according to the questionnaire, as well as those with incomplete data, were excluded from the analyses concerning the first, longer follow-up period, leaving 6,935 men. Only men with complete data from the second screening and no myocardial infarction before the second screening were included in the later, shorter follow-up period (n = 6,003). All nonfatal myocardial infarction and stroke cases in Giiteborg are recorded according to specific criteria. 12,13Death certificates for the men in the study were continuously collected, and the Swedish national causespecific death register was matched against the computer file for all the men. Cause-specific mortality was coded by 2 physicians according to the eighth revision of the International Classification of Diseases. CAD was defined as all subjects with nonfatal myocardial infarction or death attributed to CAD, or both, during follow-up. Angina pectoris is not included.
1172
THE AMERICAN
Other Causes and All Causes During
JOURNAL
OF CARDIOLOGY
VOLUME
68
(14) (12) (12) (38)
6 n = 789 % (r-d
9.3 (73) 2.9 (23)
6.1 1.8 4.3 12.2
(48) (14) (34) (96)
Follow-Up
by
Odds Ratio* (95% Cl)
Odds Ratio? (95% Cl)
1.6 (1.3-2.0) 1.7 (1.1-2.6)
1.5 (1.2-1.9) 1.8(1.1-2.8)
1.7 0.8 2.0 1.5
1.7 (1.2-2.4) 0.8 (0.5-1.3) 1.4 (0.97-2.1) 1.4(1.1-1.7)
smoking, body mass index, diabetes,
(1.3-2.3) (0.5-1.3) (1.4-2.7) (1.2-1.9)
family history of myocardial
1
infarction,
The Primary Prevention Study was designed as an intervention study, with special measures for smokers, as well as for men with hypertension or hypercholesterolemia. There were no significant differences in pattern of risk factors or outcome with respect to cardiovascular disease, cancer or mortality between the intervention group and 2 equally large control groups9 which means that the intervention group may be regarded as representative of the male middle-aged population in Goteborg during this period of time. To evaluate the possible association between stress and stressful life events, a second study was performed in a smaller sample of men, all born in 1933 and resident in Giiteborg. The men in the sample (n = 1,016) all aged 50 years, were invited to a screening examination in 1983 to 1984. An item regarding stress, identical to the one in the Primary Prevention Study, was used in a questionnaire that was filled out by 732 men with no history of myocardial infarction. At the examination, another questionnaire on life events was distributed. After a 6 year follow-up ending on December 31, 1989, a total of 26 of the respondents had either developed a nonfatal myocardial infarction or died from CAD. Possible associations between continuous and graded variables were tested using a nonparametric permutation test. l4 Odds ratios were calculated for the 2 highest stress ratings (i.e., permanent stress during either the last year or the last 5 years), with men with the 4 lowest stress ratings as a comparison group. Adjusted odds ratios were calculated by logistic regression. l 5 RESULTS
Table I lists the various end points of the study in relation to psychological stress. No increase in risk for any end point was detectable for men with self-perceived stress ratings of 1 to 4. For men with the highest stress ratings (i.e., permanent stress during either the last year or the last 5 years), there was a significant increase in risk of CAD (fatal and nonfatal). The same NOVEMBER
1, 1991
TABLE Ill incidence of Coronary Artery Disease and Mortality from All Causes During 7.1 Years Follow-Up from the Second Screening in 1974 to 1977 by Self-Perceived Stress Ratings at Both Screenings
TABLE II Risk Factors in Men with High and Low Self-Perceived Stress Ratings Self-Perceived Risk Factor Serum cholesterol, mmol/l (SD) Systolic blood pressure, m m Hg (SD) Body mass index, kg/ m2 (SD) Smoking, % (n) Diabetes, % (n) Married, % (n) Family history of myocar-
1 to 4 (n = 5,865) 6.45C1.16) 148.8
‘i
Stress
(22.0)
5to 6 (n = 1,070)
P Value
6.42
0.51
146.6
(1.12) (22.0)
0.0024
(3.5)
0.084
25.5C3.2)
25.7
47.7 1.8 85.9 19.9
59.0 (631) 2.1 (23) 80.0(856) 23.0 (246)
0.0001 0.40 0.0001 0.023
10.8(116)
0.0001 0.0005
(2799) (104) (5036) (1165)
Incidence Coronary Disease Self-Perceived Stress Rating (no. of men) Low* ratings at both screenings (II = 4,728) High* stress rating at first screening, low at second (n = 533) Low stress rating at first screening, high at second (n = 387) High stress ratings at both screemngs (n = 355) Total (n = 6,003)
dial infarction, % (n) Registered for alcohol abuse, % (n) Low occupational class (1 to 2), % (n) Low physical activity cl), % (n)
6.OC3.53) 46.6
(2732)
49.9
(534)
22.9
(1345)
37.4(400)
0.0001
pattern was apparent for deaths from cardiovascular disease and from all causes, whereas stroke had an increase among men with the highest stress rating only. Because the relation between CAD and psychological stress was nonlinear, with only men with the highest 2 stress ratings displaying an increase in risk, these men will be considered together in the following. Table II shows that men with stress rating 5 or 6 were more frequently smokers and registered for alcohol abuse, and were also less active during leisure time than the rest of the men. They also had slightly lower blood pressure levels. Adjustment for other risk factors when comparing men with stress ratings 5 to 6 with those
of Artery
Mortality from All Causes
Odds Ratio (95% Cl)
% (n)
Odds Ratio (95% Cl)
% (n) 5.7 (271)
4.5 (215)
7.1 (38)
1.6 (1.1-2.3)
6.7 (26)
1.5 (0.99-2.3)
7.6 (27)
1.7 (1.1-2.6)
9.9 (53)
10.6
(41)
9.0(32)
5.1(306)
1.8(1.3-2.5)
1.9 (1.4-2.8)
1.6 (1.1-2.4)
6.6(397)
“Low stress rating: 1 to 4; high stress rating: 5 to 6. Cl = confidence Interval.
with ratings 1 to 4 with respect to outcome resulted in small decreases in odds ratios (Table I). Psychological stress was assessed at both the first screening in 1970 to 1973 and the second screening 4 years later. Data from both examinations were used to
Percent
10 -
FIGURE 1. Cum&tive annual hidewe of coronary arlery disease during 1970 to 1992 (first 3 months of 1983 not in-).
6-
1970
1972
1974
1976
1978
1980
1982
STRESS AND CAD
Year
1173
TABLE IV “Life Events” and Coronary Artery Disease in 1983 to 1984. Study of Men Born in 1933”
DISCUSSION
by Stress
We found self-perceived psychological stress to be an independent risk factor for CAD in this population Low Stress High Stress Rating Rating Odds of middle-aged men. The limitations of this study are (1 to4) (5 to 6) Ratio obvious and will be discussed later; the strength of our n = 635 n = 97 (95% CJ) findings is in the prospective design and in the sample Serious illness in family size, which was sufficiently large to detect this relatively member modest effect. Past year 15.9 (101) 16.5 (16) 1 .O (0.6-l .9) At any time 50.7 (322) 59.8 (58) 1.4 (0.9-2.2) The most serious limitation is the definition of psySerious concern about chological stress. The item concerning stress in the family member Past year 30.9 (196) 40.2 (39) 1.5 (0.97-2.3) questionnaire was conceived on a purely intuitive basis, At any time 51.5 (327) 63.9 (62) 1.7 (1.1-2.6) focusing on various psychological problems such as Death of family member anxiety, tension and sleeping difficulties. Psychological Fast year 15.4 (98) 18.6 (18) 1.2 (0.7-2.2) At any time 74.6 (474) 74.2 (72) 1.0 (0.6-1.6) factors such as obsessional traits and propensity to worDivorce or separation ry have been found to predict CAD.16 On the other Past year 2.2 (14) 2.1(2) 0.9 (0.2-4.2) hand, stress is often described as a psychological proAt any time 21.9 (139) 16.5 (16) 0.7 (0.4-1.2) Forced to move house cess, defined as the internal state of a person who perPast year 3.6 (23) 6.2 (6) 1.8 (0.7-4.4) ceives threats to his or her physical and psychological At any time 16.5 (105) 16.5 (16) 1.0 (0.6-1.8) well-being.’ Our definition fits part of this description, Forced to change job Past year 4.7 (30) 8.2 (8) 1.8 (0.8-4.1) but is also broad and nonspecific. Some of the subjects At any time 19.1 (121) 22.7 (22) 1.2 (0.7-2.1) who answered the question may have affirmed the presBeen made redundant ence of symptoms, rather than of stress. In this case, Past year 4.3 (27) 6.2 (6) 1.5 (0.6-3.7) At any time 12.6 (80) 13.4 (13) 1.1 (0.6-2.0) the stress question may have been a measurement more Feelings of insecurity at of intraindividual psychological problems than of stress work Past year 13.7 (87) 24.7 (24) 2.1 (1.2-3.5) caused by external factors. At any time 28.2 (179) 39.2 (38) 1.6 (1.1-2.6) Some retrospective case-control studies have found Serious financial trouble stress, defined in various ways, to be associated with Past year 5.2 (33) 18.6 (18) 4.2 (2.2-7.7) At any time 13.4 (85) 27.8 (27) 2.5 (1.5-4.1) myocardial infarction, 17,i8 but in general there seems to Been legally prosecuted be very little prospective data linking CAD to stress. Past year 2.2 (14) 6.2 (6) 2.9 (1.1-7.8) When examined in epidemiologic and clinical investigaAt any time 17.2 (109) 14.4 (14) 0.8 (0.4-1.5) Coronary artery disease 3.3 (21) 5.2 (5) 1.6 (0.6-4.3) tions, life stress has often been defined as the numerical during follow-up accumulation of major life events.19*20This method of*Figures are percentages (number of men). fers a clear advantage in that measurement of events that are easily identifiable minimizes the chance of subevaluate whether the effect of stress was transient or jective variation in the responses, but in retrospective not, with follow-up extending from the second exami- studies the occurrence of myocardial infarction may innation only (Table III). Men with permanent stress at fluence how patients respond to questions of life events. both examinations had the same rate as men with per- However, it is clear from analysis of the second, younmanent stress at only 1 of the screenings. The same ger group in our study that psychological stress, as we applies to mortality from all causes. The cumulative in- defined it, has very little to do with life events such cidence of CAD for men with stress ratings 5 and 6 as bereavement, separation or unemployment, and is compared with those with stress ratings 1 to 4 is shown much more closely related to worries. in Figure 1. The effect of stress did not decrease over With the exception of life events, the stress questionnaire that we used has not so far been validated against time. other measures of stress. However, outcome with reData on life events in relation to psychological stress from the study of men born in 1933 are listed in Table spect to CAD was remarkably similar in both screenIV. Important negative life events were only weakly or ings, as well as in the investigation of men born in not at all related to heavy stress. Items more concerned 1933, In this sample also, men with a stress rating of 5 with worry were more firmly linked to stress. There or 6 had a 50% increase in CAD incidence, but owing were few CAD events; however, the association be- to the small number of infarcts, this increase was not tween CAD and stress, even though not significant, was significant. None of the items on the life event scale of the same order of magnitude as in the Primary Pre- predicted CAD, but again, the number of cases was small. vention Study.
1174
THE AMERICAN
JOURNAL
OF CARDIOLOGY
VOLUME
68
NOVEMBER
1, 1991
Much of the research on the association between stress and CAD has been directed toward the influence of occupational stress.s,6p21The question on stress that we used asked about occupational stress as well as about stress induced by conditions at home; these 2 entities were not separated. In a previous analysis22 of this study population, only 2 single occupational groups had a significant increase in risk of CAD; bus and tram drivers, and taxi drivers. Neither group had an increase in stress. Consequently, if some form of occupational stress was responsible for the increase in CAD risk in the 2 groups it could obviously not be measured in this relatively simple manner. The small group of men with disability pensions (n = 105) had a large proportion of subjects with a stress rating of 5 or 6 (40% compared with 15 among the rest; p
Self-perceived psychological stress as a risk factor for coronary artery disease (CAD) was evaluated in a general population study comprising 6,935 men aged 47 to 55 years at baseline without previous myocardial infarction. In 1970 to 1973, the men answered a question about psychological stress defined as a feeling of tension, irritability or anxiety, or as having sleeping difficulties as a result of conditions at work or at home. Psychological stress was graded as follows: (1) never experienced stress; (2) 21 period of stress; (3) I1 period of stress during the last 5 years; (4) several periods of stress during the last 5 years; and (5 to 6) permanent stress during the last year or the last 5 years. After a mean follow-up of 11.8 years, 6% of the men with the lowest 4 stress ratings (n = 5,665) had either devw a nonfatal myocardial infarction or died from CAD, with no increase in risk from grade 1 to 4. The corresponding flgure among the men with the highest 2 stress ratings (n = 1,070) was 10%; the odds ratio was 1.5 (95% conffdence interval 1.2-1.9) after controlling for age and other risk factors. Similar, independent associations were seen with stroke, and with death from cardiovascular disease and from all causes, but not with death from cancer. With respect to CAD, no decrease in the effect of stress at baseline could be seen over time. No relation between life events and self-perceived psychologieal stress was found in another sample of 732 fifty-year-okl men. (AmJCardiol1991;9&1171-1175)
From the Section of Preventive Cardiology, Department of Internal Medicine, Ostra Hospital, University of Giiteborg, Goteborg, Sweden, and the Department of Family Medicine, University of Uppsala, Sweden. This study was supported by the Swedish National Bank Tercentenary Fund, Stockholm, the Swedish National Association Against Heart and Cheat Diseases,Stockholm, the Giiteborg Medical Society, Gateborg, and the Knut and Alice Wallenberg Foundation, Stockholm. Manuscript received February 14, 1991; revised manuscript received June 12,1991, and accepted June 14. Address for reprints: Annika Rosengren, MD, Department of Medicine, Gstra Hospital, S-416 85 GBteborg, Sweden.
n recent years, epidemiologic research, concerning the etiology of coronary artery disease (CAD) has been increasingly directed toward environmental and psychosocial factors, such as socioeconomic status,1-2 social ~upport~-~ and occupational stress.5-6Inevitably, the concept of stress enters either implicitly or explicitly into most of this research. In the public eye, stress is commonly held to be one of the more important causes of myocardial infarction. A vast body of literature connects various psychosocial factors with CAD7; however, the evidence connecting CAD with stress, per se, is not very extensive. Stress research has traditionally focused primarily on the consequences of stressful stimuli for individual persons, not populations.* Recognizing the possible importance of stress as a risk factor, a question on stress was asked in 1970 to 1973 of all participants in a large prospective study on CAD in middle-aged men. In the present study, we analyzed cardiovascular and other end points during 11.8 year follow-up in relation to self-perceived stress at baseline.
I
METHODS The study population of the first part of this study comprises the intervention group of the Multifactor Primary Prevention Trial in Goteborg, the main results of which have been previously published.9 This study began in 1970 and included all men in Giiteborg who were born between 1915 and 1925, with the exception of those born in 1923. The intervention group of 10,000 men represents a random third of the men in the trial with 2 equally large control groups. In all, 7,495 men aged 47 to 55 years (75% of those invited) attended a screening examination that took place between January 1970 and March 1973. A second screening began in 1974 and was completed in 1977. All surviving men from the intervention group who still lived in Goteborg (n = 9,411) were invited to this second screening; the response rate was 76%. The same methods were used for both screenings. AI1 participants completed a postal questionnaire before the examination. The questionnaire contained an item in which stress was defined as feeling tense, irritable or filled with anxiety, or as having sleeping difficulties as a result of conditions at work or at home. The men were asked whether they had experienced this and to check 1 of the following choices: (1) never experienced stress; STRESS AND CAD
1171
TABLE I Incidence of Coronary Self-Perceived Stress Stress Rating Coronaj artery disease Stroke Death from: Cardiovascular disease Cancer Other causes All causes
Artery
Disease,
Stroke,
and Death from Cancer,
1 n = 1,162 % (n)
2 n = 1,494 % (n)
3 n = 1,300 %(n)
4 n = 1,909 %(n)
5 n = 281 % (n)
6.0 (70) 1.4 (16)
6.7 (100) 1.3 (20)
6.8 (89) 1.3 (17)
6.0 (114) 1.7 (33)
11.4 (32) 1.1 (3)
3.9 3.2 2.6 9.6
3.3 (49) 3.1(47) 1.8 (27) 8.2 (123)
3.8 3.0 2.0 8.8
3.1 2.5 2.6 8.2
5.0 4.3 4.3 13.5
(45) (37) (30) (112)
(50) (39) (26) (115)
(59) (48) (49) (156)
*Self-perceived stress rating 5 to 6 vs 1 to 4, unadjusted. tSelf-perceived stress rating 5 to 6 vs 1 to 4, adjusted forage, systolic blood pressure, serum cholesterol, occupational class, marital state, leisure time physical activity and registration for alcohol abuse. Cl = confidence interval.
(2) I: 1 period of stress; (3) some period of stress during the last 5 years; (4) several periods of stress during the last 5 years; (5) permanent stress during the last year; or (6) permanent stress during the last 5 years. The questionnaire also contained questions on smoking habits, physical activity during leisure, diabetes, previous myocardial infarction, symptoms of cardiovascular disease and family history of myocardial infarction. Physical activity during leisure was graded 1 to 4, with 1 denoting sedentary leisure time activity. Occupational class was defined as previously described.1o Data on alcohol abuse” were collected from official registers with special permission. At the screening examination, blood pressure was measured to the nearest 2 mm Hg after 5 minutes’ rest with the subject seated. Body mass index was calculated as wt/h@. Serum cholesterol was determined according to standard laboratory procedures. Follow-up extended through March 1983 (mean follow-up from the first screening 11.8 years, and from the second screening 7.1). All men with previous myocardial infarction according to the questionnaire, as well as those with incomplete data, were excluded from the analyses concerning the first, longer follow-up period, leaving 6,935 men. Only men with complete data from the second screening and no myocardial infarction before the second screening were included in the later, shorter follow-up period (n = 6,003). All nonfatal myocardial infarction and stroke cases in Giiteborg are recorded according to specific criteria. 12,13Death certificates for the men in the study were continuously collected, and the Swedish national causespecific death register was matched against the computer file for all the men. Cause-specific mortality was coded by 2 physicians according to the eighth revision of the International Classification of Diseases. CAD was defined as all subjects with nonfatal myocardial infarction or death attributed to CAD, or both, during follow-up. Angina pectoris is not included.
1172
THE AMERICAN
Other Causes and All Causes During
JOURNAL
OF CARDIOLOGY
VOLUME
68
(14) (12) (12) (38)
6 n = 789 % (r-d
9.3 (73) 2.9 (23)
6.1 1.8 4.3 12.2
(48) (14) (34) (96)
Follow-Up
by
Odds Ratio* (95% Cl)
Odds Ratio? (95% Cl)
1.6 (1.3-2.0) 1.7 (1.1-2.6)
1.5 (1.2-1.9) 1.8(1.1-2.8)
1.7 0.8 2.0 1.5
1.7 (1.2-2.4) 0.8 (0.5-1.3) 1.4 (0.97-2.1) 1.4(1.1-1.7)
smoking, body mass index, diabetes,
(1.3-2.3) (0.5-1.3) (1.4-2.7) (1.2-1.9)
family history of myocardial
1
infarction,
The Primary Prevention Study was designed as an intervention study, with special measures for smokers, as well as for men with hypertension or hypercholesterolemia. There were no significant differences in pattern of risk factors or outcome with respect to cardiovascular disease, cancer or mortality between the intervention group and 2 equally large control groups9 which means that the intervention group may be regarded as representative of the male middle-aged population in Goteborg during this period of time. To evaluate the possible association between stress and stressful life events, a second study was performed in a smaller sample of men, all born in 1933 and resident in Giiteborg. The men in the sample (n = 1,016) all aged 50 years, were invited to a screening examination in 1983 to 1984. An item regarding stress, identical to the one in the Primary Prevention Study, was used in a questionnaire that was filled out by 732 men with no history of myocardial infarction. At the examination, another questionnaire on life events was distributed. After a 6 year follow-up ending on December 31, 1989, a total of 26 of the respondents had either developed a nonfatal myocardial infarction or died from CAD. Possible associations between continuous and graded variables were tested using a nonparametric permutation test. l4 Odds ratios were calculated for the 2 highest stress ratings (i.e., permanent stress during either the last year or the last 5 years), with men with the 4 lowest stress ratings as a comparison group. Adjusted odds ratios were calculated by logistic regression. l 5 RESULTS
Table I lists the various end points of the study in relation to psychological stress. No increase in risk for any end point was detectable for men with self-perceived stress ratings of 1 to 4. For men with the highest stress ratings (i.e., permanent stress during either the last year or the last 5 years), there was a significant increase in risk of CAD (fatal and nonfatal). The same NOVEMBER
1, 1991
TABLE Ill incidence of Coronary Artery Disease and Mortality from All Causes During 7.1 Years Follow-Up from the Second Screening in 1974 to 1977 by Self-Perceived Stress Ratings at Both Screenings
TABLE II Risk Factors in Men with High and Low Self-Perceived Stress Ratings Self-Perceived Risk Factor Serum cholesterol, mmol/l (SD) Systolic blood pressure, m m Hg (SD) Body mass index, kg/ m2 (SD) Smoking, % (n) Diabetes, % (n) Married, % (n) Family history of myocar-
1 to 4 (n = 5,865) 6.45C1.16) 148.8
‘i
Stress
(22.0)
5to 6 (n = 1,070)
P Value
6.42
0.51
146.6
(1.12) (22.0)
0.0024
(3.5)
0.084
25.5C3.2)
25.7
47.7 1.8 85.9 19.9
59.0 (631) 2.1 (23) 80.0(856) 23.0 (246)
0.0001 0.40 0.0001 0.023
10.8(116)
0.0001 0.0005
(2799) (104) (5036) (1165)
Incidence Coronary Disease Self-Perceived Stress Rating (no. of men) Low* ratings at both screenings (II = 4,728) High* stress rating at first screening, low at second (n = 533) Low stress rating at first screening, high at second (n = 387) High stress ratings at both screemngs (n = 355) Total (n = 6,003)
dial infarction, % (n) Registered for alcohol abuse, % (n) Low occupational class (1 to 2), % (n) Low physical activity cl), % (n)
6.OC3.53) 46.6
(2732)
49.9
(534)
22.9
(1345)
37.4(400)
0.0001
pattern was apparent for deaths from cardiovascular disease and from all causes, whereas stroke had an increase among men with the highest stress rating only. Because the relation between CAD and psychological stress was nonlinear, with only men with the highest 2 stress ratings displaying an increase in risk, these men will be considered together in the following. Table II shows that men with stress rating 5 or 6 were more frequently smokers and registered for alcohol abuse, and were also less active during leisure time than the rest of the men. They also had slightly lower blood pressure levels. Adjustment for other risk factors when comparing men with stress ratings 5 to 6 with those
of Artery
Mortality from All Causes
Odds Ratio (95% Cl)
% (n)
Odds Ratio (95% Cl)
% (n) 5.7 (271)
4.5 (215)
7.1 (38)
1.6 (1.1-2.3)
6.7 (26)
1.5 (0.99-2.3)
7.6 (27)
1.7 (1.1-2.6)
9.9 (53)
10.6
(41)
9.0(32)
5.1(306)
1.8(1.3-2.5)
1.9 (1.4-2.8)
1.6 (1.1-2.4)
6.6(397)
“Low stress rating: 1 to 4; high stress rating: 5 to 6. Cl = confidence Interval.
with ratings 1 to 4 with respect to outcome resulted in small decreases in odds ratios (Table I). Psychological stress was assessed at both the first screening in 1970 to 1973 and the second screening 4 years later. Data from both examinations were used to
Percent
10 -
FIGURE 1. Cum&tive annual hidewe of coronary arlery disease during 1970 to 1992 (first 3 months of 1983 not in-).
6-
1970
1972
1974
1976
1978
1980
1982
STRESS AND CAD
Year
1173
TABLE IV “Life Events” and Coronary Artery Disease in 1983 to 1984. Study of Men Born in 1933”
DISCUSSION
by Stress
We found self-perceived psychological stress to be an independent risk factor for CAD in this population Low Stress High Stress Rating Rating Odds of middle-aged men. The limitations of this study are (1 to4) (5 to 6) Ratio obvious and will be discussed later; the strength of our n = 635 n = 97 (95% CJ) findings is in the prospective design and in the sample Serious illness in family size, which was sufficiently large to detect this relatively member modest effect. Past year 15.9 (101) 16.5 (16) 1 .O (0.6-l .9) At any time 50.7 (322) 59.8 (58) 1.4 (0.9-2.2) The most serious limitation is the definition of psySerious concern about chological stress. The item concerning stress in the family member Past year 30.9 (196) 40.2 (39) 1.5 (0.97-2.3) questionnaire was conceived on a purely intuitive basis, At any time 51.5 (327) 63.9 (62) 1.7 (1.1-2.6) focusing on various psychological problems such as Death of family member anxiety, tension and sleeping difficulties. Psychological Fast year 15.4 (98) 18.6 (18) 1.2 (0.7-2.2) At any time 74.6 (474) 74.2 (72) 1.0 (0.6-1.6) factors such as obsessional traits and propensity to worDivorce or separation ry have been found to predict CAD.16 On the other Past year 2.2 (14) 2.1(2) 0.9 (0.2-4.2) hand, stress is often described as a psychological proAt any time 21.9 (139) 16.5 (16) 0.7 (0.4-1.2) Forced to move house cess, defined as the internal state of a person who perPast year 3.6 (23) 6.2 (6) 1.8 (0.7-4.4) ceives threats to his or her physical and psychological At any time 16.5 (105) 16.5 (16) 1.0 (0.6-1.8) well-being.’ Our definition fits part of this description, Forced to change job Past year 4.7 (30) 8.2 (8) 1.8 (0.8-4.1) but is also broad and nonspecific. Some of the subjects At any time 19.1 (121) 22.7 (22) 1.2 (0.7-2.1) who answered the question may have affirmed the presBeen made redundant ence of symptoms, rather than of stress. In this case, Past year 4.3 (27) 6.2 (6) 1.5 (0.6-3.7) At any time 12.6 (80) 13.4 (13) 1.1 (0.6-2.0) the stress question may have been a measurement more Feelings of insecurity at of intraindividual psychological problems than of stress work Past year 13.7 (87) 24.7 (24) 2.1 (1.2-3.5) caused by external factors. At any time 28.2 (179) 39.2 (38) 1.6 (1.1-2.6) Some retrospective case-control studies have found Serious financial trouble stress, defined in various ways, to be associated with Past year 5.2 (33) 18.6 (18) 4.2 (2.2-7.7) At any time 13.4 (85) 27.8 (27) 2.5 (1.5-4.1) myocardial infarction, 17,i8 but in general there seems to Been legally prosecuted be very little prospective data linking CAD to stress. Past year 2.2 (14) 6.2 (6) 2.9 (1.1-7.8) When examined in epidemiologic and clinical investigaAt any time 17.2 (109) 14.4 (14) 0.8 (0.4-1.5) Coronary artery disease 3.3 (21) 5.2 (5) 1.6 (0.6-4.3) tions, life stress has often been defined as the numerical during follow-up accumulation of major life events.19*20This method of*Figures are percentages (number of men). fers a clear advantage in that measurement of events that are easily identifiable minimizes the chance of subevaluate whether the effect of stress was transient or jective variation in the responses, but in retrospective not, with follow-up extending from the second exami- studies the occurrence of myocardial infarction may innation only (Table III). Men with permanent stress at fluence how patients respond to questions of life events. both examinations had the same rate as men with per- However, it is clear from analysis of the second, younmanent stress at only 1 of the screenings. The same ger group in our study that psychological stress, as we applies to mortality from all causes. The cumulative in- defined it, has very little to do with life events such cidence of CAD for men with stress ratings 5 and 6 as bereavement, separation or unemployment, and is compared with those with stress ratings 1 to 4 is shown much more closely related to worries. in Figure 1. The effect of stress did not decrease over With the exception of life events, the stress questionnaire that we used has not so far been validated against time. other measures of stress. However, outcome with reData on life events in relation to psychological stress from the study of men born in 1933 are listed in Table spect to CAD was remarkably similar in both screenIV. Important negative life events were only weakly or ings, as well as in the investigation of men born in not at all related to heavy stress. Items more concerned 1933, In this sample also, men with a stress rating of 5 with worry were more firmly linked to stress. There or 6 had a 50% increase in CAD incidence, but owing were few CAD events; however, the association be- to the small number of infarcts, this increase was not tween CAD and stress, even though not significant, was significant. None of the items on the life event scale of the same order of magnitude as in the Primary Pre- predicted CAD, but again, the number of cases was small. vention Study.
1174
THE AMERICAN
JOURNAL
OF CARDIOLOGY
VOLUME
68
NOVEMBER
1, 1991
Much of the research on the association between stress and CAD has been directed toward the influence of occupational stress.s,6p21The question on stress that we used asked about occupational stress as well as about stress induced by conditions at home; these 2 entities were not separated. In a previous analysis22 of this study population, only 2 single occupational groups had a significant increase in risk of CAD; bus and tram drivers, and taxi drivers. Neither group had an increase in stress. Consequently, if some form of occupational stress was responsible for the increase in CAD risk in the 2 groups it could obviously not be measured in this relatively simple manner. The small group of men with disability pensions (n = 105) had a large proportion of subjects with a stress rating of 5 or 6 (40% compared with 15 among the rest; p
