Depression as a risk factor for ischaemic heart disease in men: population based case-control studyBMJ 1998; 316 doi: http://dx.doi.org/10.1136/bmj.316.7146.1714 (Published 06 June 1998) Cite this as: BMJ 1998;316:1714
- Julia Hippisley-Cox, lecturer in general practice ()a,
- Katherine Fielding, lecturer in medical statisticsb,
- Mike Pringle, professor of general practicea
- aDepartment of General Practice, The Medical School, Queen's Medical Centre, Nottingham NG7 2UH
- bTrent Institute for Health Services Research, The Medical School, Queen's Medical Centre
- Correspondence to: Dr Hippisley-Cox
- Accepted 2 February 1998
Objective: To determine the relation between depression, anxiety, and use of antidepressants and the onset of ischaemic heart disease.
Design: Population based case-control study.
Setting: All 5623 patients registered with one general practice.
Subjects:188 male cases with ischaemic heart disease matched by age to 485 male controls without ischaemic heart disease; 139 female cases with ischaemic heart disease matched by age to 412 female controls.
Main outcome measure: Adjusted odds ratios calculated by conditional logistic regression.
Results: The risk of ischaemic heart disease was three times higher among men with a recorded diagnosis of depression than among controls of the same age (odds ratio 3.09; 95% confidence interval 1.33 to 7.21; P=0.009). This association persisted when smoking status, diabetes, hypertension, and underprivileged area (UPA(8)) score were included in a multivariate model (adjusted 2.75; 1.13 to 6.69; P=0.03). Men with depression within the preceding 10 years were three times more likely to develop ischaemic heart disease than were the controls (3.13; 1.27 to 7.70; P=0.01). Men with ischaemic heart disease had a higher risk of subsequent ischaemic heart disease than men without ischaemic heart disease (adjusted 2.34; 1.34 to 4.10; P=0.003). Depression was not a risk factor for ischaemic heart disease in women on multivariate analysis (adjusted 1.34; 0.70 to 2.56; P=0.38). Anxiety and subsequent ischaemic heart disease were not significantly associated in men or women.
Conclusion: Depression may be an independent risk factor for ischaemic heart disease in men, but not in women.
So far, research into whether depression precedes myocardial infarction has been limited
This case-control study examined the relation between ischaemic heart disease and depression and the differences in this relation between men and women
Depression may be a risk factor for ischaemic heart disease in men but not women
This is independent of diabetes, hypertension, deprivation score, and smoking status
Depression is present in over 45% of patients admitted to hospital after a myocardial infarction1 and is an independent risk factor for increased mortality2–4 and increased morbidity 5 6 after myocardial infarction. Depression may precede myocardial infarction,7 although this is not certain. 8 9 Research in this area has been limited to studies of small numbers of highly selected hospital patients, often without any control group.10
Furthermore, the overall relation between depression, ischaemic heart disease, and cholesterol concentration is unclear. Some evidence shows that low cholesterol concentration may be related to depression11 and increased risk of suicide.12–15 Other evidence shows that no relation exists between low and declining cholesterol concentration and depression16–18 or suicide. 19 20 If ischaemic heart disease is associated with hyperlipidaemia, and depression is associated with low cholesterol concentration, then a lower prevalence of depression in patients who subsequently develop ischaemic heart disease would be expected.
We aimed to determine whether (a) an association exists between ischaemic heart disease and depression, (b) depression occurs before or after the onset of ischaemic heart disease, and (c) the relation between depression and ischaemic heart disease differs between men and women. We included diagnoses of anxiety as well as of depression, as the two conditions often coexist.
Selection of cases and controls
We conducted this case-control study in a rural dispensing training practice with 5623 patients on the borders of Nottinghamshire and Lincolnshire. Cases were male and female patients who have or have had ischaemic heart disease. Cases were identified from the practice computer; we selected those who either had a recorded diagnosis of ischaemic heart disease (including angina, myocardial infarction, coronary artery surgery) or were receiving repeat prescriptions of nitrates. The written records of all cases were reviewed to confirm the diagnosis, the date of onset, the first presenting illness (angina or myocardial infarction), and the results of supporting diagnostic investigations—that is, resting and exercise electrocardiography, and angiography.
We needed 299 matched case-control sets (one case to two controls) to show a relative risk of 2 for the onset of depression before the onset of ischaemic heart disease. This is based on a 20% prevalence of prior depression in cases compared with a 10% prevalence in controls. This sample size would give 95% power at the 5% significance level. Altogether, 327 patients with ischaemic heart disease were registered with the practice on 1 January 1996. There were insufficient patients aged over 80 for us to match two controls per case in that age group. To maintain the power of the study, therefore, we allocated between one and four age matched and sex matched controls according to the number of patients in each 10 year age band (four controls to each case under 60 years, three to each case aged 60-69, two to each case aged 70-79, and one to each case aged over 80 years). The controls were selected from an alphabetical list of patients currently registered with the practice. The next patient of exactly the same age in years, but with a different surname (to avoid family members), was chosen. Patients who had died were not included in the analysis as their manual records were no longer available.
Every control was given a “pseudodiagnosis” date, the date on which he or she was the same age as the matching case was at diagnosis of ischaemic heart disease. For cases and controls, Read codes that related to depression or anxiety were identified from the computer database, and the dates of first diagnoses of depression and the first diagnoses of anxiety were recorded. Diagnoses of postnatal depression or manic depression were excluded. Computerised data for the use of antidepressants—that is, tricyclic drugs, monoamine oxidase inhibitors, and selective serotonin reuptake inhibitors—during five years preceding the diagnosis or pseudodiagnosis date were also recorded, together with the date of first prescription. We searched a random sample of 30 manual records to validate the computer data and found no discrepancy. We also collected the following data on all subjects: age; sex; body mass index; underprivileged area (UPA(8)) score, which is a measure of deprivation on the basis of the subjects' postcode; date of onset of ischaemic heart disease; age at diagnosis of ischaemic heart disease; and most recently recorded smoking status (current or former smoker or non-smoker). The presence or absence of prior diabetes mellitus and hypertension were recorded.
The statistical analyses generally used were conditional multiple logistic regression analysis for individually matched case-control studies. The statistical package used was STATA (version 5.0). The dependent variable was the presence of ischaemic heart disease, and the principal variable was depression before the diagnosis or pseudodiagnosis date. When depression occurred in the same year as the onset of ischaemic heart disease, it was assumed to have occurred after onset. This was done because only the year of onset was recorded for some cases and controls. Such an assumption would tend to underestimate rather than overestimate the odds ratio. The univariate and multivariate associations for body mass index, deprivation score, anxiety, depression, use of antidepressants, diabetes, hypertension, and smoking status were determined. A case-control set was excluded if the information either for the case or for all the controls was not known for the variable in question. The multivariate models presented here comprise smoking status, hypertension, diabetes, and deprivation score. Body mass index was not included in the final model owing to the number of missing data points that would have greatly reduced the eventual sample size and therefore the power of the study.
Characteristics of study population
Of the 5623 patients registered with the practice, 327 patients had ischaemic heart disease; of these, 205 first presented with angina, 122 first presented with a myocardial infarction, and 23 had had previous coronary artery surgery. Altogether, 188 male cases (105 with angina, 83 with a myocardial infarction) were age matched to 485 male controls; 139 female cases (100 with angina, 39 with a myocardial infarction) were age matched to 412 female controls. Table 1 shows the baseline characteristics, and table 2 shows the Read codes used for the diagnoses of depression.
Depression as risk factor for ischaemic heart disease
Table 3 shows the results of the univariate and multivariate analysis for men with and without ischaemic heart disease. On univariate analysis, men with a recorded diagnosis of depression were three times more likely than controls of the same age to develop ischaemic heart disease (odds ratio 3.09; 95% confidence interval 1.33 to 7.21; P=0.009). The risk of ischaemic heart disease persisted when smoking status, diabetes, hypertension, and deprivation score were included in the calculations (adjusted 2.75; 1.13 to 6.69; P=0.03).
The data were reanalysed by comparing the mean values for each group of controls with the value for their respective case by using the Wilcoxon signed rank test. Men with ischaemic heart disease had a higher score for depression than the controls did (P=0.01). This is consistent with the results of the conditional logistic regression analysis.
Duration of depression
Duration of depression before ischaemic heart disease was categorised as “no depression,” “duration ≥10 years,” or “duration >10 years.” These categories were used because of the distribution of the data. Duration of depression was associated with risk of heart disease on univariate analysis (table 3). Men who had a recorded diagnosis of depression within the preceding 10 years had a risk of ischaemic heart disease three times as high as both the controls (3.13; 1.27 to 7.70; P=0.01) and the patients who had depression for more than 10 years before the onset of ischaemic heart disease. When smoking status, hypertension, diabetes, and underprivileged score were included in the calculations, the risk was essentially unaltered (adjusted 3.12; 1.23 to 7.93; P=0.02).
Tricyclic antidepressants before ischaemic heart disease
Only six male cases and six male controls were taking tricyclic antidepressants before the onset of ischaemic heart disease. Of these, three cases were taking dothiepin and three amitriptyline; five controls were taking dothiepin and one amitriptyline. The mean dose for cases was 68.8 (SD 24.7) mg and for controls was 54.2 (SD 40.1) mg. The median duration of use for cases was 3 (range 1-7) months and for controls was 10 (2-16) months.
The results in table 3 suggest that men who had been prescribed tricyclic antidepressants in the recent past have a risk of ischaemic heart disease three times as high as controls but with a wide confidence interval owing to small numbers (adjusted 3.55; 0.89 to 14.21; P=0.07). When the doses of tricyclic antidepressants were included as a categorical variable in the logistic regression model, increased doses seemed to be associated with increased risk of heart disease (table 3). However, these results were also not significant owing to the small sample size. The study was not designed to determine the particular effect of tricyclic antidepressants on risk of heart disease.
Depression after onset of ischaemic heart disease
Men with ischaemic heart disease are twice as likely to have a recorded diagnosis of depression after the onset of ischaemic heart disease as men without ischaemic heart disease (2.20; 1.28 to 3.79; P=0.005). When smoking, deprivation score, hypertension, and diabetes were included the increased risk of depression persisted (adjusted 2.34; 1.33 to 4.10; P=0.003).
Depression after ischaemic heart disease: effect of prior depression
The risk of any subsequent depression was related more to prior ischaemic heart disease (adjusted 2.42; 1.39 to 4.21; P=0.002) than to prior depression (adjusted 0.70; 0.15 to 3.16, P=0.64).
Depression and risk of ischaemic heart disease
Table 4 shows the results for the univariate and multivariate analysis for women. Depression was not associated with an increased risk of subsequent ischaemic heart disease on either univariate or multivariate analysis. When the use, dose, and duration of tricyclic antidepressants were examined in women, no significant associations were found (table 4). The odds ratios for antidepressants, however, were in the opposite direction to that found for men.
Depression after onset of ischaemic heart disease
On multivariate analysis women with ischaemic heart disease had twice the risk of having a recorded diagnosis of depression compared with women of the same age without ischaemic heart disease (adjusted 1.86; 1.10 to 3.16; P=0.02).
Anxiety before and after onset of ischaemic heart disease—Anxiety was not found to be a risk factor for ischaemic heart disease for men or women, on either univariate or multivariate analysis. Similarly, men and women with ischaemic heart disease were not at increased risk of having a recorded diagnosis of anxiety.
Although dead patients were not formally included in the case-control study, we had identified on the database 69 dead patients who had had ischaemic heart disease. These dead patients were no more likely than the 327 study patients with ischaemic heart diseases to have been depressed before or after the diagnosis of ischaemic heart disease. The two groups were similar for the baseline characteristics.
To our knowledge this is the first controlled study to show that depression is likely to be an independent risk factor for ischaemic heart disease for men in primary care. This risk persists regardless of smoking status, deprivation score, and presence of diabetes or hypertension. Previous studies have shown that certain personality traits predict increased cardiovascular risk—for example, type A personality21 and hostility. 20 22
Strengths and weaknesses of this study
This study used a larger sample of cases and controls than previous studies, and subjects were selected from the community. As such, the population is more representative of the population with ischaemic heart disease than the populations in studies that selected cases and controls from secondary care. This study has used routinely collected data from a general practice database which has been validated and found to have a high standard of data completeness and accuracy.23 General practice databases do not seem to have undue bias in epidemiological studies of patient morbidity.24 As this study is a case-control study, any minor limitations of the routinely collected data will apply to both cases and controls and are therefore unlikely to cause significant bias. Although this study has been conducted on a single practice population, we have no reason to believe that the patients studied are different from any other practice population.23
Validity of diagnosis
For the past year the practice has had a protocol for diagnosing and treating depression. This specifies diagnostic criteria and suggests which Read codes and antidepressant drugs to use. Only the doctors enter diagnoses of depression on the computer. When new patients register with the practice, the general practitioner reviews all their past records in order and enters diagnoses of depression and ischaemic heart disease (and dates of onset) on the computer. However, as the diagnoses had been made over a 30 year period, the protocol was not operational for most of the study period. We used diagnosis of depression rather than a numerical rating score. We think that depression of sufficient character and severity to warrant assessment by a general practitioner probably has greater validity than a score that assesses mood on one occasion. The pragmatic nature of this study is likely to have increased the generalisability of its results, particularly as most depressed patients are managed entirely in general practice. The diagnoses of depression in general practice are consistent with psychiatric criteria, although the disorder tends to be less severe. 25 26
Depression, ischaemic heart disease, and cholesterol
Although we did not include cholesterol concentrations, our results provide indirect evidence supporting other studies that have found no association between low or declining concentration of cholesterol and depression.16–18 If low cholesterol concentration was related to low mood then we would have expected that a population with a high predicted cholesterol concentration—that is, patients with ischaemic heart disease—would have a lower prevalence of depression. This is clearly not the case for men with ischaemic heart disease.
Plausibility of depression as cause for ischaemic heart disease
At least six possible explanations exist for why depression could be an aetiological factor for ischaemic heart disease. Firstly, depression may lead to coronary events directly or indirectly via poorer health behaviours, such as increased smoking or decreased activity.5 This has been shown in patients who are depressed after myocardial infarction,6 and the same mechanism could operate before infarction. Such behaviour changes may lead to a poorer cardiovascular risk profile—for example, higher cholesterol concentration or blood pressure. Secondly, the association between depression and risk of heart disease may be due to an effect of tricyclic antidepressant drugs. Our study did not have adequate power to detect the risk associated specifically with antidepressants. Thirdly, depression has been shown to be proarrhythmogenic in patients with established ischaemic heart disease. 2 3 This is thought to be due to changes in the balance between sympathetic and parasympathetic nerve activity—for example, an increase in sympathetic nerve activity or a decrease in parasympathetic nerve activity, or both of these. This mechanism might operate in depressed patients without established ischaemic heart disease, increasing their risk of developing it or accelerating its onset. Fourthly, depression might result in an unfavourable lipid profile resulting from an interaction between the catecholamine and steroid axes.3Fifthly, depression might be confused with “vital exhaustion”—the prodromal symptoms of tiredness, apathetic mood, and sadness—which can occur immediately before a myocardial infarction. We do not think that this explains our findings as we took the year of onset of both depression and ischaemic heart disease. When both years were identical, we assumed that depression occurred after ischaemic heart disease. Finally, there may be a separate, and yet unidentified, aetiological factor that causes both depression and ischaemic heart disease in men.
Men versus women
None of the above factors has so far explained why depressed men seem to be at a higher risk of ischaemic heart disease than women, but several possible explanations exist. Firstly, men, who are at higher absolute risk of ischaemic heart disease, are more susceptible than women to changes in autonomic nerve activity or changes in the operation of the catecholamine and steroid axes. Secondly, depression may lead to an increase in smoking and a decrease in physical activity that is more pronounced in men than women. Thirdly, the discrepancy in risk might be due to the variation in prevalence of both diseases in men and women. Fourthly, men's higher risk might result from a difference in general practitioners' ability or opportunity to make diagnoses of depression in men and women. It might reflect differences in severity of depression and illness behaviour between the sexes—for example, men may be diagnosed with depression only if it is of a certain severity. If depression is a risk factor for cardiovascular disease then there could be a “dose response” relation whereby patients with severe depression have a higher coronary risk. If men with a recorded diagnosis of depression have a more severe illness than women, then we would expect men to have a higher coronary risk. We had intended to use secondary care referral as a marker for the severity of depression, although data were insufficient to allow such an analysis.
We thank Collingham Medical Centre for providing us with access to the data for this study, Lynne Wright for extracting some of the computerised data, and Professor Clair Chilvers for her helpful comments on this manuscript.
Contributors: JH-C conceived the idea, contributed to the study design, reviewed the literature, collected the data, did the statistical analysis, interpreted the results, drafted the paper, and is the guarantor for the paper. KF designed the study and advised on the statistical analysis. MP advised on the design and on the interpretation of the results and contributed to the writing of the paper. Dr Hutton and partners, of Collingham Medical Centre, allowed access to the centre's high quality database. Ms Lynne Wright assisted with some of the data collection. Ms April McCambridge collected the references for the literature review. Professor Clair Chilvers commented on a final draft of the manuscript.
Conflict of interest: None.