Psychological Medicine, 1992, 22, 939-949. Copyright
1992 Cambridge University Press
Printed in Great Britain
A population survey of ischaemic heart disease and minor psychiatric disorder in men S. A. STANSFELD, 1 D. S. SHARP, J. E. J. GALLACHER AND J. W. G. YARNELL From the Academic Department of Psychiatry, University College and Middlesex Hospital School of Medicine; and MRC Epidemiology Unit (South Wales), Penarth
Associations between ischaemic heart disease and psychiatric morbidity in hospital recruited samples may be confounded by differential referral of patients with co-morbidity. Associations of angina, past history of myocardial infarction, blood pressure, and electrocardiographic evidence of ischaemia with psychiatric disorder can best be examined in community samples as reported here in 2204 middle-aged men from the Caerphilly Collaborative Study. There was a strong association between past history of myocardial infarction, non-specific chest pain, Angina Grade II and psychiatric disorder measured by the 30-item General Health Questionnaire. Electrocardiographic evidence of ischaemia alone was not significantly associated with psychiatric disorder. It is suggested that non-specific chest pain is a symptom of psychiatric disorder; conversely in severe angina psychiatric disorder is secondary to the pain, restricted activity and threat to life which angina implies. SYNOPSIS
INTRODUCTION The finding of psychiatric morbidity among patients attending a cardiac clinic or complaining of chest pain attributed to the heart has long been established (Lewis, 1917; Wood, 1941; Mayou, 1989). These studies have been refined by the aid of coronary angiography providing an objective measure of the extent of coronary atherosclerosis. Such studies have demonstrated an association between established ischaemic heart disease and psychological disorder (Murrell et al. 1983; Carney et al. 1987; Young et al. 1988), but have demonstrated even stronger associations in patients with chest pain and normal coronary angiography (Bass & Wade, 1984; Katon et al. 1988; Carney et al. 1990). Investigations using detailed measures of psychiatric disorder have demonstrated associations between ischaemic heart disease and panic disorder (Bass & Wade, 1984; Beitman et al. 1987 a) as well as depressive disorder (Beitman et al. 1987/?). 1 Address for correspondence: Dr S. A. Stansfeld, Academic Department of Psychiatry, University College and Middlesex Hospital, School of Medicine, Wolfson Building, Riding House Street, London WIN 8AA.
Although there have been studies in primary care settings (Goldberg & Bridges, 1988), the size and nature of associations between cardiac disease and psychiatric disorder has not been set in the context of the general population (Chambers & Bass, 1990). Associations between angina, myocardial infarction and psychiatric disorder in studies of hospital patients may be confounded by differential referral to hospital for those with co-morbidity. It may be that highly anxious or depressed patients with chest pain are more likely to be referred for angiography by their general practitioners than patients without psychological symptoms, thus, potentially inflating the association between chest pain and psychological morbidity. This argument of referral bias has been used to cast doubt upon the association between chest pain and psychiatric disorder in those with normal coronary angiograms and alternatively to postulate a pathogenic mechanism in terms of abnormal function of the coronary microcirculation (microvascular angina) (Cannon, 1988; Wielgosz, 1988). Ischaemic heart disease is a major chronic health problem in industrialized nations, and the delineation of the size of the associations between the different mani-
939
940
S. A. Stansfeld and others
festations of ischaemic heart disease and psychological morbidity has important public health implications for planning rational psychological management of ischaemic heart disease in the community. This paper presents the cross-sectional findings from the Caerphilly Collaborative Heart Disease Study examining the association of psychological morbidity and ischaemic heart disease in men. It aims to clarify the nature of this association in a population sample and provide initial clues to the mechanisms underlying this association unbiased by referral selection factors. METHOD The Caerphilly Collaborative Heart Disease Study includes cross-sectional and prospective studies of risk factors for ischaemic heart disease and their determinants for Caerphilly and Speedwell Collaborative Group, 1984). Sample All men between the ages of 45 and 64 years, living in Caerphilly, South Wales, and its environs were invited to attend a screening clinic where physiological measurements were obtained and questionnaires completed. The electoral roll and general practitioner records were used as the sampling frame. Morning clinics from 6 a.m. and evening clinics were used to encourage working men to attend. All nonattenders were followed up by a home visit. Instruments All men completed a battery of questionnaires which included the 30-item General Health Questionnaire (GHQ) (Goldberg, 1972), the Spielberger Trait Anxiety Inventory (Spielberger et al. 1970) and brief social support measures (Marmot, 1983). The 30-item General Health Questionnaire was validated in a subsample of 97 men using the Clinical Interview Schedule (Goldberg & Blackwell, 1970). As a result of Receiver Operating Characteristic analysis respondents scoring 0-4 on the GHQ, were considered non-cases and those scoring 5 + were considered possible cases of minor psychiatric disorder (Stansfeld et al. 1991). Angina was measured by the London School of Hygiene Chest Pain Questionnaire (LSHTM) using the original author's definition of angina
(Rose, 1962; Rose et al. 1977). Angina was subdivided into Grade I (chest pain on walking uphill only) and Grade II (chest pain on walking on the level). Type ' A ' behaviour was measured by the Framingham Type A Scale (Haynes et al. 1978 a), the 21-item Jenkins Activity Survey (Jenkins et al. 1971) and 14-item Bortner Scale (Bortner, 1969). Past history of myocardial infarction was taken from the single question from the LSHTM chest pain questionnaire ' Have you ever had a severe pain across the front of your chest lasting for half an hour or more?' As at least 50% of those answering 'yes' to the single question had no evidence of myocardial infarction, additional criteria needed to be fulfilled for qualification for' past history of myocardial infarction'. These were that the respondent had (a) seen a doctor because of the pain, (b) was admitted to hospital and (c) had been told a diagnosis of myocardial infarction had been made (Yarnell et al. 1988). Demographic data collected included age, marital status, employment status, and socioeconomic status classified according to the Registrar General Social class index. Physiological measurements Blood pressure was measured by a medical observer using either a Hawksley random zero sphygmomanometer or a regular mercury sphygmomanometer after a short rest period in a sitting position. A 12-lead electrocardiogram was recorded and coded by two experienced coders using the Minnesota code. Two classes of ECG ischaemia were adopted, as in the Whitehall Study (Reid et al. 1974); possible ischaemia 1-3 (minor Q waves), 4-1, 4-2, 4-3, < M ( S T wave changes), 5-1, 5-2, 5 -3, (T wave changes) and 7-1 (left bundle branch block) and probable ischaemia 1-1, 1-2 (major Q waves). RESULTS The response rate for the study was 89%. All those who did not attend clinic were seen at home and 50% of these men agreed to an electrocardiogram. There seemed more physical illness among the 11 % of men not attending clinic but apart from this there was nothing to suggest their GHQ scores would differ from other participants. In total, 2204 men attending clinics were given General Health Question-
941
Ischaemic heart disease and psychiatric disorder
Table 1. Psychiatric disorder on the general health questionnaire and non-specific chest pain and angina No chest pain Psychiatric disorder Nol a 'case' (GHQ 0 4) 'Case' (GHQ & 5) Total (2198)
Non-specific chest pain
Angina Grade II
Angina Grade I
(%)
N
(%)
N
(%)
N
(%)
751
(86-9)
799
(73-6)
107
(74-8)
52
(49-5)
113
(131)
287
(26-4)
36
(25-2)
53
(50-5)
N
864
(100)
1086
(100)
143
(100)
105
(100)
'= 1020, df = 3, P< 00001.
Table 2. Odds ratio (95% CI) for psychiatric disorder for three types of chest pain relative to no chest pain (N = 2178)
No adjustment for covariates Adjustment for age Social class Employment status Framingham Type ' A ' status All 4 covariates
Non-specific chest pain
Angina Grade I
Angina Grade II
2-42 (1-9-3-1) 2-39 (1-9-3-0) 2-39 (1-9-3-0) 2-22 (1-7-2-8) 216 (1-7-2-8) 1-87 (1-5-2-4)
2-28 (1-5-3-5) 2-31 (1-5-3-5) 2-43 (1-6-3-7) 1-90 (1-2-3-0) 202 (1-3-3-1) 1 89 (1-1-3-2)
719 (4-9-10-6) 7-45 (5-0-11-0) 7-50 (5-0-11-2) 504 (3-1-8-1) 7-25 (4-8-11-0) 4-64 (2-7-8-0)
naires, 8'1 % (N = 194) of questionnaires were possible psychiatric cases among men with nonnot completed and 0-8 % and 0-1 % missed 1 or specific chest pain (defined as a positive response 2 items, respectively (0-7% of 'low' scorers; to: 'Have you ever had any pain or discomfort 2-2% of'high' scorers missed 1 or 2 items). All in the chest?' If ' n o ' - ' a n y pressure or heavisubjects who scored on 28 items or more on the ness?') with a comparable proportion (25%) GHQ were included. Those with 28 or 29 items among men with Angina Grade I. If non-specific had their raw score adjusted by a multiplication chest pain was incorporated into the 'no chest factor of either 30/28, or 30/29, respectively. pain' group the level of possible psychiatric Eighty-eight per cent of the total sample were caseness rose from 13-1% to 20-5% but the married, 5 % were single, 3 % widowed, and 4 % difference between 'no chest pain' and angina divorced. Fifty-three per cent were employed or still remained highly significant (x2 = 52-5, self employed, 15% unemployed, and 32% d f = 2 , P > 0-0001). retired. Thirty-five per cent were between 45 and In order to test for possible confounding by 55 years, 32 % 55 and 60 years and 33 % 60 and socio-demographic variables, odds ratios for 64 years. each of the three successive levels of chest pain relative to 'no chest pain' were examined both Chest pain and psychiatric disorder separately and in combination with relevant There was a highly significant association be- covariates (Table 2). The odds ratios were not tween chest pain and minor psychiatric disorder significantly different upon examining the Bres(Table 1). Over 50% of men with Angina Grade low-Day tests for homogeneity of the odds II were identified as possible psychiatric cases. ratios after; (a) separate stratification for social There was also a notable proportion (26%) of class (non-manual, manual), employment status
S1. A. Stansfeld and others
942
Table 3. Psychiatric disorder and angina stratified by Type 'A' behaviour: for psychiatric disorder for angina adjusted by employment
Angina Grade II
Angina Grade I Odds ratio adjusted for employment status
GHQ 'Cases' (%)
Framingham scale
odds ratios ( 9 5 % Cf) status
GHQ 'Cases' (%)
Odds ratio adjusted for employment status
Type ' A ' Total
25(31-3) 80
1-53(0-85-2-75)
37 (67-3) 55
5-9(3-0- 11-5)
Type'B' Tolal
11(17-4) 63
204(0-95-4-39)
16(33-3) 48
3-6(1-8 7-4)
Table 4. Psychiatric disorder on the General Health Questionnaire and past history of myocardial infarction Myocardial infarction No
Electrocardiographic evidence of cardiac ischaemia and psychiatric disorder
Yes
Psychiatric disorder
N
(%)
N
(%)
Not a 'case' (GHQ (M) Case' (GHQ £ 5)
1590
(79-4)
123
(61 2)
infarction (38-8 %) than in men without such a history (20-6 %) (Table 4). This association was unchanged by stratification for social class, employment status, age and Framingham Type ' A ' behaviour.
(5 levels), age (3 levels), social contacts (3 levels) and Type ' A ' Behaviour (2 levels); and (b) simultaneous stratification on social class, employment status age, and Framingham Type' A' behaviour. However, there was a fall of the odds ratio from 7-19 to 5-04 comparing 'no chest pain' to Angina Grade II after controlling for employment status (Table 2). Moreover, among Type ' B' men, particularly on the Framingham scale, there was less psychiatric morbidity in those with angina, than in Type ' A ' men with angina (Table 3). However, there is likely to be confounding between Type ' A ' behaviour on the Framingham scale and the GHQ.
In view of the powerful association between symptomatic measures of ischaemic heart disease, angina and myocardial infarction and psychiatric disorder there is a natural progression to examining the association between 'objective' measures of ischaemic heart disease, namely the electrocardiogram and psychiatric disorder. In Table 5 there is a small non-significant gradient in rates of possible psychiatric caseness by cardiac ischaemia. This suggests that the association between reported chest pain and psychiatric disorder cannot be explained by ' objective' evidence of coronary vascular disease as measured by ECG. Men with ECG abnormalities would include both those who were aware of their evidence of ischaemic heart disease and men who were not aware of it; and about one-third of all men with ECG ischaemia had either angina or a past history of severe chest pain.
History of myocardial infarction and psychiatric disorder
Chest pain, electrocardiographic findings and psychiatric disorder
An acknowledgement on questionnaire of severe chest pain lasting half an hour or more was confirmed as myocardial infarction with the general practitioner. There were statistically significantly more possible psychiatric cases among men with a past history of myocardial
The association between angina (categorized as 'no chest pain', and Angina Grade I, Angina Grade II) and psychiatric morbidity on the GHQ remained statistically highly significant after controlling for electro-cardiographic evidence of ischaemia (Cochran-Mantel-Haenszel
Total
412 2002
(20-6) (100)
78 201
(38-8) (100)
2203
X' = 35-l, d f = 1, P< 0-0001.
Ischaemic heart disease and psychiatric disorder
943
Table 5. Psychiatric disorder on the General Health Questionnaire and electrocardiographic {ECG) evidence of ischaemic heart disease ECG evidence of ischaemia None Psychiatric disorder
Possible
N 1451 407
Not a 'case' (GHQ 0-4) •Case'(GHQ^S)
202 61
(781) (21 9)
1858
Total (2203)
N
(%)
(100)
263
Probable
(%)
N
(%)
(76-8) (23-2)
60 22
(73-2) (26-8)
82
(100)
(100)
Mantel-Haenzel x' Irend = 114, df = 1, /> = 0-28.
Table 6. Association of chest pain and psychiatric disorder stratified by electrocardiographic evidence of ischaemic heart disease No chest pain
Non-specific chest pain
Angina grade I
Angina grade II
Psychiatric disorder
/V
(%)
N
(%)
N
(%)
N
(%)
ECG negative ' C a s e ' G H Q J: 5 Total (1852)
96 756
(12-7)
252 927
(27-2)
26 102
(25-5)
31 67
(46-3)
ECG positive •Case1 G H Q & 5 Total (345)
17 108
(15-7)
34 158
(21-5)
10 41
(24-3)
22 38
(57-9)
X' for ECG negative = 76-7, d f - 3, P < 0-0001. X* for ECG positive = 28-5, df = 3, P < 00001.
Table 7. Psychiatric disorder by past history of myocardial infarction {MI) stratified by current electrocardiographic evidence of ischaemic heart disease Past history of MI No Electrocardiogram Negative
Psychiatric disorder
N
(%)
N
(%)
'Non-case' •Case'
1371 352 1723
(79-6) (20-4)
65 45 110
(59-1) (40-9)
'Non-case' 'Case'
168 42 210
(80-0) (200)
33 19 52
(635) (36-5)
'Non-case' 'Case'
35 9 44
(79-6) (20-4)
24 13 37
(64-9) (351)
Total Possible Total Probable
Yes
Total
For ECG negative / = 25-6, df = 1, P < 00001. For ECG possible x' = 6-4, df = 1, P < 005. For ECG probable x' = 2-2, df = 1, P < 0139.
/ test for trend = 29-2, df = 1, P < 00001). The percentage of possible psychiatric cases among those with non-specific chest pain was lower among those with probable ischaemia (21-5%) than among those without evidence of ischaemia
(27-2%) (Table 6), suggesting that men with non-specific chest pain were more likely to have psychiatric morbidity if their ECGs were normal. Conversely, however, there was a gradient of increasing psychiatric morbidity among men
944
Type A
S. A. Stansfeld and others
E 00
c
1
S3 u.
, ,
irder
Type
co r
1992 Cambridge University Press
Printed in Great Britain
A population survey of ischaemic heart disease and minor psychiatric disorder in men S. A. STANSFELD, 1 D. S. SHARP, J. E. J. GALLACHER AND J. W. G. YARNELL From the Academic Department of Psychiatry, University College and Middlesex Hospital School of Medicine; and MRC Epidemiology Unit (South Wales), Penarth
Associations between ischaemic heart disease and psychiatric morbidity in hospital recruited samples may be confounded by differential referral of patients with co-morbidity. Associations of angina, past history of myocardial infarction, blood pressure, and electrocardiographic evidence of ischaemia with psychiatric disorder can best be examined in community samples as reported here in 2204 middle-aged men from the Caerphilly Collaborative Study. There was a strong association between past history of myocardial infarction, non-specific chest pain, Angina Grade II and psychiatric disorder measured by the 30-item General Health Questionnaire. Electrocardiographic evidence of ischaemia alone was not significantly associated with psychiatric disorder. It is suggested that non-specific chest pain is a symptom of psychiatric disorder; conversely in severe angina psychiatric disorder is secondary to the pain, restricted activity and threat to life which angina implies. SYNOPSIS
INTRODUCTION The finding of psychiatric morbidity among patients attending a cardiac clinic or complaining of chest pain attributed to the heart has long been established (Lewis, 1917; Wood, 1941; Mayou, 1989). These studies have been refined by the aid of coronary angiography providing an objective measure of the extent of coronary atherosclerosis. Such studies have demonstrated an association between established ischaemic heart disease and psychological disorder (Murrell et al. 1983; Carney et al. 1987; Young et al. 1988), but have demonstrated even stronger associations in patients with chest pain and normal coronary angiography (Bass & Wade, 1984; Katon et al. 1988; Carney et al. 1990). Investigations using detailed measures of psychiatric disorder have demonstrated associations between ischaemic heart disease and panic disorder (Bass & Wade, 1984; Beitman et al. 1987 a) as well as depressive disorder (Beitman et al. 1987/?). 1 Address for correspondence: Dr S. A. Stansfeld, Academic Department of Psychiatry, University College and Middlesex Hospital, School of Medicine, Wolfson Building, Riding House Street, London WIN 8AA.
Although there have been studies in primary care settings (Goldberg & Bridges, 1988), the size and nature of associations between cardiac disease and psychiatric disorder has not been set in the context of the general population (Chambers & Bass, 1990). Associations between angina, myocardial infarction and psychiatric disorder in studies of hospital patients may be confounded by differential referral to hospital for those with co-morbidity. It may be that highly anxious or depressed patients with chest pain are more likely to be referred for angiography by their general practitioners than patients without psychological symptoms, thus, potentially inflating the association between chest pain and psychological morbidity. This argument of referral bias has been used to cast doubt upon the association between chest pain and psychiatric disorder in those with normal coronary angiograms and alternatively to postulate a pathogenic mechanism in terms of abnormal function of the coronary microcirculation (microvascular angina) (Cannon, 1988; Wielgosz, 1988). Ischaemic heart disease is a major chronic health problem in industrialized nations, and the delineation of the size of the associations between the different mani-
939
940
S. A. Stansfeld and others
festations of ischaemic heart disease and psychological morbidity has important public health implications for planning rational psychological management of ischaemic heart disease in the community. This paper presents the cross-sectional findings from the Caerphilly Collaborative Heart Disease Study examining the association of psychological morbidity and ischaemic heart disease in men. It aims to clarify the nature of this association in a population sample and provide initial clues to the mechanisms underlying this association unbiased by referral selection factors. METHOD The Caerphilly Collaborative Heart Disease Study includes cross-sectional and prospective studies of risk factors for ischaemic heart disease and their determinants for Caerphilly and Speedwell Collaborative Group, 1984). Sample All men between the ages of 45 and 64 years, living in Caerphilly, South Wales, and its environs were invited to attend a screening clinic where physiological measurements were obtained and questionnaires completed. The electoral roll and general practitioner records were used as the sampling frame. Morning clinics from 6 a.m. and evening clinics were used to encourage working men to attend. All nonattenders were followed up by a home visit. Instruments All men completed a battery of questionnaires which included the 30-item General Health Questionnaire (GHQ) (Goldberg, 1972), the Spielberger Trait Anxiety Inventory (Spielberger et al. 1970) and brief social support measures (Marmot, 1983). The 30-item General Health Questionnaire was validated in a subsample of 97 men using the Clinical Interview Schedule (Goldberg & Blackwell, 1970). As a result of Receiver Operating Characteristic analysis respondents scoring 0-4 on the GHQ, were considered non-cases and those scoring 5 + were considered possible cases of minor psychiatric disorder (Stansfeld et al. 1991). Angina was measured by the London School of Hygiene Chest Pain Questionnaire (LSHTM) using the original author's definition of angina
(Rose, 1962; Rose et al. 1977). Angina was subdivided into Grade I (chest pain on walking uphill only) and Grade II (chest pain on walking on the level). Type ' A ' behaviour was measured by the Framingham Type A Scale (Haynes et al. 1978 a), the 21-item Jenkins Activity Survey (Jenkins et al. 1971) and 14-item Bortner Scale (Bortner, 1969). Past history of myocardial infarction was taken from the single question from the LSHTM chest pain questionnaire ' Have you ever had a severe pain across the front of your chest lasting for half an hour or more?' As at least 50% of those answering 'yes' to the single question had no evidence of myocardial infarction, additional criteria needed to be fulfilled for qualification for' past history of myocardial infarction'. These were that the respondent had (a) seen a doctor because of the pain, (b) was admitted to hospital and (c) had been told a diagnosis of myocardial infarction had been made (Yarnell et al. 1988). Demographic data collected included age, marital status, employment status, and socioeconomic status classified according to the Registrar General Social class index. Physiological measurements Blood pressure was measured by a medical observer using either a Hawksley random zero sphygmomanometer or a regular mercury sphygmomanometer after a short rest period in a sitting position. A 12-lead electrocardiogram was recorded and coded by two experienced coders using the Minnesota code. Two classes of ECG ischaemia were adopted, as in the Whitehall Study (Reid et al. 1974); possible ischaemia 1-3 (minor Q waves), 4-1, 4-2, 4-3, < M ( S T wave changes), 5-1, 5-2, 5 -3, (T wave changes) and 7-1 (left bundle branch block) and probable ischaemia 1-1, 1-2 (major Q waves). RESULTS The response rate for the study was 89%. All those who did not attend clinic were seen at home and 50% of these men agreed to an electrocardiogram. There seemed more physical illness among the 11 % of men not attending clinic but apart from this there was nothing to suggest their GHQ scores would differ from other participants. In total, 2204 men attending clinics were given General Health Question-
941
Ischaemic heart disease and psychiatric disorder
Table 1. Psychiatric disorder on the general health questionnaire and non-specific chest pain and angina No chest pain Psychiatric disorder Nol a 'case' (GHQ 0 4) 'Case' (GHQ & 5) Total (2198)
Non-specific chest pain
Angina Grade II
Angina Grade I
(%)
N
(%)
N
(%)
N
(%)
751
(86-9)
799
(73-6)
107
(74-8)
52
(49-5)
113
(131)
287
(26-4)
36
(25-2)
53
(50-5)
N
864
(100)
1086
(100)
143
(100)
105
(100)
'= 1020, df = 3, P< 00001.
Table 2. Odds ratio (95% CI) for psychiatric disorder for three types of chest pain relative to no chest pain (N = 2178)
No adjustment for covariates Adjustment for age Social class Employment status Framingham Type ' A ' status All 4 covariates
Non-specific chest pain
Angina Grade I
Angina Grade II
2-42 (1-9-3-1) 2-39 (1-9-3-0) 2-39 (1-9-3-0) 2-22 (1-7-2-8) 216 (1-7-2-8) 1-87 (1-5-2-4)
2-28 (1-5-3-5) 2-31 (1-5-3-5) 2-43 (1-6-3-7) 1-90 (1-2-3-0) 202 (1-3-3-1) 1 89 (1-1-3-2)
719 (4-9-10-6) 7-45 (5-0-11-0) 7-50 (5-0-11-2) 504 (3-1-8-1) 7-25 (4-8-11-0) 4-64 (2-7-8-0)
naires, 8'1 % (N = 194) of questionnaires were possible psychiatric cases among men with nonnot completed and 0-8 % and 0-1 % missed 1 or specific chest pain (defined as a positive response 2 items, respectively (0-7% of 'low' scorers; to: 'Have you ever had any pain or discomfort 2-2% of'high' scorers missed 1 or 2 items). All in the chest?' If ' n o ' - ' a n y pressure or heavisubjects who scored on 28 items or more on the ness?') with a comparable proportion (25%) GHQ were included. Those with 28 or 29 items among men with Angina Grade I. If non-specific had their raw score adjusted by a multiplication chest pain was incorporated into the 'no chest factor of either 30/28, or 30/29, respectively. pain' group the level of possible psychiatric Eighty-eight per cent of the total sample were caseness rose from 13-1% to 20-5% but the married, 5 % were single, 3 % widowed, and 4 % difference between 'no chest pain' and angina divorced. Fifty-three per cent were employed or still remained highly significant (x2 = 52-5, self employed, 15% unemployed, and 32% d f = 2 , P > 0-0001). retired. Thirty-five per cent were between 45 and In order to test for possible confounding by 55 years, 32 % 55 and 60 years and 33 % 60 and socio-demographic variables, odds ratios for 64 years. each of the three successive levels of chest pain relative to 'no chest pain' were examined both Chest pain and psychiatric disorder separately and in combination with relevant There was a highly significant association be- covariates (Table 2). The odds ratios were not tween chest pain and minor psychiatric disorder significantly different upon examining the Bres(Table 1). Over 50% of men with Angina Grade low-Day tests for homogeneity of the odds II were identified as possible psychiatric cases. ratios after; (a) separate stratification for social There was also a notable proportion (26%) of class (non-manual, manual), employment status
S1. A. Stansfeld and others
942
Table 3. Psychiatric disorder and angina stratified by Type 'A' behaviour: for psychiatric disorder for angina adjusted by employment
Angina Grade II
Angina Grade I Odds ratio adjusted for employment status
GHQ 'Cases' (%)
Framingham scale
odds ratios ( 9 5 % Cf) status
GHQ 'Cases' (%)
Odds ratio adjusted for employment status
Type ' A ' Total
25(31-3) 80
1-53(0-85-2-75)
37 (67-3) 55
5-9(3-0- 11-5)
Type'B' Tolal
11(17-4) 63
204(0-95-4-39)
16(33-3) 48
3-6(1-8 7-4)
Table 4. Psychiatric disorder on the General Health Questionnaire and past history of myocardial infarction Myocardial infarction No
Electrocardiographic evidence of cardiac ischaemia and psychiatric disorder
Yes
Psychiatric disorder
N
(%)
N
(%)
Not a 'case' (GHQ (M) Case' (GHQ £ 5)
1590
(79-4)
123
(61 2)
infarction (38-8 %) than in men without such a history (20-6 %) (Table 4). This association was unchanged by stratification for social class, employment status, age and Framingham Type ' A ' behaviour.
(5 levels), age (3 levels), social contacts (3 levels) and Type ' A ' Behaviour (2 levels); and (b) simultaneous stratification on social class, employment status age, and Framingham Type' A' behaviour. However, there was a fall of the odds ratio from 7-19 to 5-04 comparing 'no chest pain' to Angina Grade II after controlling for employment status (Table 2). Moreover, among Type ' B' men, particularly on the Framingham scale, there was less psychiatric morbidity in those with angina, than in Type ' A ' men with angina (Table 3). However, there is likely to be confounding between Type ' A ' behaviour on the Framingham scale and the GHQ.
In view of the powerful association between symptomatic measures of ischaemic heart disease, angina and myocardial infarction and psychiatric disorder there is a natural progression to examining the association between 'objective' measures of ischaemic heart disease, namely the electrocardiogram and psychiatric disorder. In Table 5 there is a small non-significant gradient in rates of possible psychiatric caseness by cardiac ischaemia. This suggests that the association between reported chest pain and psychiatric disorder cannot be explained by ' objective' evidence of coronary vascular disease as measured by ECG. Men with ECG abnormalities would include both those who were aware of their evidence of ischaemic heart disease and men who were not aware of it; and about one-third of all men with ECG ischaemia had either angina or a past history of severe chest pain.
History of myocardial infarction and psychiatric disorder
Chest pain, electrocardiographic findings and psychiatric disorder
An acknowledgement on questionnaire of severe chest pain lasting half an hour or more was confirmed as myocardial infarction with the general practitioner. There were statistically significantly more possible psychiatric cases among men with a past history of myocardial
The association between angina (categorized as 'no chest pain', and Angina Grade I, Angina Grade II) and psychiatric morbidity on the GHQ remained statistically highly significant after controlling for electro-cardiographic evidence of ischaemia (Cochran-Mantel-Haenszel
Total
412 2002
(20-6) (100)
78 201
(38-8) (100)
2203
X' = 35-l, d f = 1, P< 0-0001.
Ischaemic heart disease and psychiatric disorder
943
Table 5. Psychiatric disorder on the General Health Questionnaire and electrocardiographic {ECG) evidence of ischaemic heart disease ECG evidence of ischaemia None Psychiatric disorder
Possible
N 1451 407
Not a 'case' (GHQ 0-4) •Case'(GHQ^S)
202 61
(781) (21 9)
1858
Total (2203)
N
(%)
(100)
263
Probable
(%)
N
(%)
(76-8) (23-2)
60 22
(73-2) (26-8)
82
(100)
(100)
Mantel-Haenzel x' Irend = 114, df = 1, /> = 0-28.
Table 6. Association of chest pain and psychiatric disorder stratified by electrocardiographic evidence of ischaemic heart disease No chest pain
Non-specific chest pain
Angina grade I
Angina grade II
Psychiatric disorder
/V
(%)
N
(%)
N
(%)
N
(%)
ECG negative ' C a s e ' G H Q J: 5 Total (1852)
96 756
(12-7)
252 927
(27-2)
26 102
(25-5)
31 67
(46-3)
ECG positive •Case1 G H Q & 5 Total (345)
17 108
(15-7)
34 158
(21-5)
10 41
(24-3)
22 38
(57-9)
X' for ECG negative = 76-7, d f - 3, P < 0-0001. X* for ECG positive = 28-5, df = 3, P < 00001.
Table 7. Psychiatric disorder by past history of myocardial infarction {MI) stratified by current electrocardiographic evidence of ischaemic heart disease Past history of MI No Electrocardiogram Negative
Psychiatric disorder
N
(%)
N
(%)
'Non-case' •Case'
1371 352 1723
(79-6) (20-4)
65 45 110
(59-1) (40-9)
'Non-case' 'Case'
168 42 210
(80-0) (200)
33 19 52
(635) (36-5)
'Non-case' 'Case'
35 9 44
(79-6) (20-4)
24 13 37
(64-9) (351)
Total Possible Total Probable
Yes
Total
For ECG negative / = 25-6, df = 1, P < 00001. For ECG possible x' = 6-4, df = 1, P < 005. For ECG probable x' = 2-2, df = 1, P < 0139.
/ test for trend = 29-2, df = 1, P < 00001). The percentage of possible psychiatric cases among those with non-specific chest pain was lower among those with probable ischaemia (21-5%) than among those without evidence of ischaemia
(27-2%) (Table 6), suggesting that men with non-specific chest pain were more likely to have psychiatric morbidity if their ECGs were normal. Conversely, however, there was a gradient of increasing psychiatric morbidity among men
944
Type A
S. A. Stansfeld and others
E 00
c
1
S3 u.
, ,
irder
Type
co r
