Worry affects the circulation, the heart, the glands, the whole nervous system. I have never known a man who died from overwork, but many who died from doubt.
—Charles Horace Mayo (1865—1939)
The heart has its reasons which reason does not know.
—Blaise Pascal (1623—1662)
Cardiovascular disease (CVD) is the leading cause of death in the United States, claiming more lives than the next five leading causes of death combined—among them cancer, accidents, and diabetes. One in five men and women in the United States have some form of CVD, and in 2001, one of every 2.6 deaths was attributable to CVD. For 2004, the direct and indirect cost of CVD was estimated at $368.4 billion (1). Inasmuch as anxiety disorders affect 20 million Americans annually and depression 17 million (2, 3), it is not surprising that there is extensive comorbidity of psychiatric disorders and cardiovascular disease. Yet there appears to be more to the relationship of mental disorders with heart disease than a coincidental occurrence. Mood and anxiety problems may be risk factors for heart disease, and the occurrence of heart disease both increases the risk and complicates the treatment of psychiatric illness.
Evidence for the association of depression with cardiovascular disease first emerged in studies of institutionalized psychiatric patients more than 70 years ago (4, 5). Anecdotal reports of anger- and anxiety-induced cardiac events also provided a basis for speculation about mind-body relationships in heart disease, an issue addressed further in Cannon’s work on voodoo death (6). In recent years, there has been a dramatic increase in research about the relationship between psychiatric symptoms and disorders and cardiovascular disease. Such research has been spurred by several factors:
This article focuses primarily on depression and anxiety, which have been the most studied psychiatric conditions in relation to cardiovascular disease. We will also discuss research on personality traits, acute stress, and other psychiatric symptoms or conditions that have been studied in relation to coronary heart disease and heart failure.
Cardiac disease has an enormous impact on an individual’s overall health and daily functioning. Cardiac events often result in disability and a change in social role function and affect the individual’s perception of his or her mortality. Hence it is not surprising that depression appears to be the most common psychiatric disorder in patients with coronary artery disease (CAD) (4). Estimates of the prevalence of depression in post—myocardial infarction (MI) patients in studies in the 1960s and 1970s ranged from 10% to 87%. The breadth of this range is probably attributable to wide variations in diagnostic criteria, in the time windows after MI in which depression was measured, and in the studies’ inclusion and exclusion criteria (5). More recent studies consistently indicate a prevalence of post-MI depression in the range of 16%—27% (7—11). The Montreal Heart Institute study found the 1-year cumulative incidence of depression in 218 post-MI patients to be 31% (12), a result that was recently reproduced in an expanded sample (13). A study of 200 patients who had suffered a first MI found a 1-year cumulative incidence of depression of 25% (14).
In inpatients as well as outpatients with CAD, studies have placed the point prevalence of major depressive disorder within the range of 15%—24% (15—19). In the first few weeks or months following coronary artery bypass graft (CABG) surgery, the point prevalence of depression is in the range of 20%—30% (20—22). In patients with congestive heart failure (CHF), three studies have found the prevalence of depression to be approximately 20% (23—25), and one study found a 35% point prevalence of depression in 374 CHF patients admitted to an inpatient cardiology service (24).
The epidemiology of anxiety disorders in CAD is not as well studied as that of depression, even though symptoms of anxiety are relatively common in cardiac patients. The incidence of anxiety symptoms in patients with acute coronary disease in cardiac care units is approximately 50% (26, 27). Anxiety symptoms have been reported to be elevated in 5%—10% of patients with chronic heart disease (28). A recent study of 100 stable outpatients with coronary heart disease (CHD) using structured interviews for DSM-IV diagnosis demonstrated that multiple psychiatric diagnoses were common. Generalized anxiety disorder and posttraumatic stress disorder (PTSD) were identified in more than 20% of patients. Alcohol abuse, recurrent major depression, past episodes of depression, and current dysthymia were also prevalent, but single-episode current major depression was rare (29). Of 61 post-MI patients referred to a cardiac rehabilitation program, 4.9% were diagnosed with social phobia and 3.3% with agoraphobia without panic attacks (30). Lane and colleagues, in a study of 288 patients hospitalized for MI, found a 26% incidence of high levels of state anxiety (STAI≥40) (31). In a study of CABG patients, 55% had high levels of state anxiety preoperatively, and 32% had clinically significant levels of anxiety 3 months after surgery (32).
There are few data on the prevalence of anxiety disorders in patients with CHF (33). In a review of data from the Studies of Left Ventricular Dysfunction (SOLVD) trials, women were found to have significantly worse anxiety than women in several comparison groups, including normal subjects, geriatric subjects, hypertension patients, and cancer patients (34). High levels of anxiety symptoms were also reported in subjects in a survey of ambulatory outpatients with dilated cardiomyopathy (35).
Patients with implanted cardioverter-defibrillators (ICDs) are a special subgroup with a high risk of anxiety problems. ICDs, approved for use by the Food and Drug Administration in 1985, automatically deliver electrical shocks to the heart to interrupt abnormal rhythms; the device improves survival in patients at risk of developing lethal arrhythmias (36). The shock from the ICD is strong and unpleasant, a sensation sometimes described as a "kick in the chest." Discharges may occur without any antecedent subjective symptoms, and the presence of the device is itself a stressor for many patients, especially those who have already experienced a discharge. After experiencing shocks, symptoms of PTSD, such as avoidance, hypervigilance, and reexperiencing, are common. This is especially true if patients experience multiple sequential shocks while conscious, which they may endure with a sense of helplessness. Some patients with ICDs meet criteria for full PTSD (37).
In an early study of psychiatric morbidity in patients with ICDs, Morris and colleagues found a 50% prevalence of psychiatric disorders (adjustment disorders, major depression, and panic disorder) in 20 patients examined 3 to 21 months after the ICD implant (38). Chevalier and colleagues reported a 20% prevalence of anxiety disorders in 30 patients (39). Although one study found that no new anxiety disorders developed in ICD patients by 9—18 months after implantation (40), Bourke and associates found that over 6 years, 6 of 35 patients demonstrated "florid" psychological problems, with anxiety symptoms, including two cases of generalized anxiety disorder, two cases of agoraphobia with panic, and two cases of major depression (41). Similarly, in a study of 72 ICD patients evaluated 1 to 6 years after implantation, Godemann et al. found that 14 patients (19.4%) had agoraphobia and/or panic disorder, 11 of whom developed the anxiety disorder only after ICD placement (42).
Recent large-scale outcome studies of ICD therapy for ventricular arrhythmias confirm that discharges that are repetitive or frequent or occur soon after implantation are associated with reduced mental well-being, reduced physical function, and increased anxiety (43, 44). One recent trial of defibrillator-delivered pacing and shock therapy for atrial fibrillation found no significant difference in perceived quality of life between those who had and those who had not received shocks (45). Similarly, Duru and colleagues found no differences in symptoms of anxiety and depression between pacemaker patients and ICD patients, regardless of whether or not they had received shocks (46).
Depression and known cardiac risk factors
Depression is linked to a large number of major risk factors for CAD or cardiac-related mortality, including cigarette smoking, diabetes, and obesity. The statistical overlap between depression and having one or more of these risk factors has long been known. Prospective studies indicate that depression is an independent risk factor in the development of these behaviors or conditions.
Cigarette smoking is clearly associated with an increased risk of CVD. Current smokers have almost three times the rate of acute MI when compared with those who have never smoked (47). Depression is independently linked to daily smoking and nicotine dependence (48). Researchers have found that people with a history of major depression have a threefold elevation in risk of becoming a smoker (49). Depressed smokers are less likely to quit successfully and are more likely to have withdrawal symptoms during attempts to quit (50—52). Smokers with a history of depression have been shown to have an exaggerated belief in the positive effects of smoking (53). They also have less confidence in their ability to refrain from smoking (54).
Diabetes is associated with a three- to fourfold increase in risk of developing CVD and cardiovascular mortality (55). Depression is a risk factor for developing diabetes even after other cardiac risk factors are controlled for. An 8-year prospective study of 2,764 male employees found that subjects who had a moderate or severe level of depressive symptoms had 2.3 times the risk of developing type 2 diabetes compared with those who had no depressive symptoms (56). Two large prospective studies have recently been published that further support this finding. The Atherosclerosis Risk in Communities (ARIC) study included 11,615 initially nondiabetic adults 48—67 years of age who were followed for 6 years for the development of type 2 diabetes. After adjustment for age, race, sex, and education, the risk of individuals in the highest quartile of depressive symptoms was 1.6 times that of those in the lowest quartile. Even after adjustment for lifestyle factors such as smoking and metabolic factors such as weight and blood pressure, the risk was still 1.4 times that of the nondepressed subjects (57). A prospective study of 72,178 female nurses 45—72 years of age who were followed for 4 years found that the age-adjusted risk of developing type 2 diabetes among women with depressive symptoms was 1.55. After adjustment for body mass index (BMI) the risk was 1.36, a finding consistent with the ARIC study results (58).
Depression is also linked to the development of obesity, another CVD risk factor. One prospective study of patients 6—17 years of age found that after controlling for other risk factors, including age, cigarette use, and socioeconomic class, there was a significant correlation between childhood depression and BMI 10—15 years later. Those who had major depression had a BMI±SD of 26.1±5.2, while those without depression had a BMI 24.2±4.1 (59). A prospective cohort study of 9,374 adolescents in grades 7 through 12 assessed via in-home interviews and self-reported height and weight found at 1-year follow-up that depressed mood at baseline independently raised the risk of obesity twofold, even after potential confounders were controlled for. Baseline obesity did not predict follow-up depression even though depression did predict obesity (60).
Depression as an independent coronary disease risk factor
Even after the effects of depression on health behaviors, smoking, diabetes, and weight are taken into account, there is ample evidence that depressive symptoms and a history of depression are strong independent risk factors for the development of cardiovascular disease, of acute coronary events, and of mortality from cardiac illness. In general, studies show that the relative risk of incident cardiac disease in healthy individuals with depression or symptoms of depression is about 1.5 to 2, depending on which cardiac endpoint is used. The risk tends to be graduated, increasing with the level of depression (5, 61, 62). A full clinical diagnosis of depression yields a higher risk of developing CHD. The Mini-Finland Health Survey, which followed subjects for an average of 6.6 years, demonstrated a relative risk of 3.36 for the development of CHD in those with major depression (63). A follow-up of the Baltimore Epidemiologic Catchment Area (ECA) study that looked at 1,551 subjects without heart disease over a 13-year period found that those with a history of a major depressive episode had an odds ratio for having a nonfatal MI of 4.54 when compared with those without a history of major depression or dysphoria (64).
The impact of depression on the prognosis of patients with preexisting cardiac disease has been extensively studied. In the relatively few studies of patients with verifiable, established CAD (e.g., diagnosed by angiography) but without any recent cardiac event, depression has been found to be predictive of future cardiac mortality and morbidity (65—67). Depression has also been found to be a risk factor for mortality after CABG, in follow-up periods ranging from 2 to 12 years (20, 68, 69).
In post-MI patients, probably the most widely cited study is that of Frasure-Smith and colleagues (10, 70). In 222 patients, they demonstrated that the risk of death in the first 6 months after MI for patients with major depression was five times that of the nondepressed patients (odds ratio, 4.29) (9). While patients with depressive symptoms (Beck Depression Inventory score ≥ 10) but without major depressive disorder did not have an elevated risk at 6 months, by 18 months they had an odds ratio of 7.82 compared with those with a Beck Depression Inventory score < 10. For patients with Beck Depression Inventory ≥ 10 and more than nine premature ventricular contractions per hour, the odds ratio for mortality was 29.17 compared with other patients (70). Lesperance and colleagues followed 896 post-MI patients for 5 years, evaluating depression during admission and again 1 year after MI, and found a significant dose-response relationship between initial in-hospital depressive symptoms and cardiac mortality. Depressive symptoms at 1 year were not independently correlated with cardiac mortality. Patients with mild to moderate depression symptoms at baseline who had reduced depression symptoms at 1-year follow-up went on to demonstrate lower cardiac mortality, while patients with high depression symptoms at baseline did not. This raises the question of whether there are distinct subtypes of depression with different prognostic implications (71).
Not all studies have had such clearly positive results. Irvine and colleagues found that the apparent twofold increase in risk of mortality in a 2-year follow-up of post-MI patients was reduced by 30% and no longer statistically significant after controlling for dyspnea and fatigue (72). Strik and colleagues looked at 318 men who completed questionnaires on anxiety, depression, and hostility after MI and found that while depression seemed to be associated with a higher risk of cardiac events over an average of 3.4 years of follow-up, anxiety symptoms alone were the independent predictor of cardiac events (73). The seemingly positive results of two other studies, both with low statistical power, also did not survive risk factor adjustment (10, 74).
In patients with CHF, studies have shown that depression is associated with a higher risk of mortality. Murberg and colleagues evaluated 119 stable outpatients with heart failure and found that depressed mood doubled the risk of mortality (25% vs. 11.3%), with a significant difference remaining even after other factors were controlled for (75). In 374 inpatients with CHF, a Beck Depression Inventory ≥ 10 combined with a diagnosis of major depression on interview significantly increased mortality at 1-year follow-up (24). After other risk factors were controlled for, only increased readmission rates at 1 year persisted as a significant difference (24). A 5-year follow-up assessment of 396 inpatients with heart failure due to dilated cardiomyopathy found that the patients with clinical depression had a threefold greater risk of mortality (25). Major depression was associated with increased mortality at 1 year in 374 inpatients with CHF (76). Low emotional support was associated with worse 1-year cardiac outcome in a sample of 168 female elderly patients admitted with heart failure, although no such association was noted in the 124 male patients (77). Another study, however, failed to find an effect of depression on survival in hospitalized CHF patients after adjustment for illness severity (78).
Among patients who have depression after a cardiac event, those with a history of prior depression appear to have the greatest risk of mortality. Lesperance et al. (12) studied patients with depression after MI and found that at 1 year, 40% of those with a prior history of depression had died, compared with only 10% of those who experienced their first episode of depression after the cardiac event.
Three community-based studies have shown a significant relationship between anxiety disorders and sudden cardiac death. One study of 1,457 male subjects followed over 6 years found the risk of cardiac-related death elevated almost fourfold in those with phobic anxiety (79). In a study of 33,999 middle-aged male health professionals followed over 2 years, a dose-dependent relationship was observed between degree of phobic anxiety and risk of cardiac-related death. Highly anxious subjects had a sixfold elevation in risk of sudden cardiac death (80). In a study of 2,271 men followed over 32 years, anxiety was associated with a 1.9-fold increase in risk of fatal cardiac disease and a 4.7-fold increase in risk of sudden cardiac death, in a dose-dependent fashion (81). All three studies showed no link between anxiety and MI. Thus, anxiety-related cardiac mortality risk seems to be confined to sudden cardiac death, at least in males. This suggests that the predominant mechanism by which anxiety disorders increase risk in nonclinical samples is by inducing ventricular arrhythmias and not by increasing the risk of atherosclerosis.
Patients with preexisting cardiac disease, however, are in a different risk category than the general population. The presence of anxiety at a baseline cardiac event significantly predicts future MIs and cardiac events, not just sudden cardiac death. Frasure-Smith and colleagues (82) followed 222 post-MI patients for 12 months and found that elevated levels of self-reported state and trait anxiety at baseline were significantly and independently linked both to the occurrence of another MI and the development of any new cardiac event. Moser and Dracup (26) looked at anxiety in 86 new-MI patients within 48 hours of admission to the hospital, using a brief self-rating instrument. In-hospital complications occurred in 19.6% of those with high levels of anxiety symptoms, compared with 6% in those with low levels. Even after other risk factors were controlled for, patients with high anxiety still had almost a fivefold elevation in risk of cardiac complications or death compared with those with little or no anxiety. Denollet and Brutsaert studied 87 middle-aged post-MI patients with a left ventricular ejection fraction of 50% or less who were psychologically evaluated at baseline and then followed up an average of 8 years later. A high level of anxiety symptoms was not an independent risk factor for recurrent events after adjustment for other risk factors, but the combination of high anxiety and a high level of social inhibition was associated with an independent relative risk of 4.7 for having a new cardiac event (83).
Acutely stressful events dramatically increase cardiac morbidity and mortality. In a prospective study of 95,647 Finnish men and women, the risk of all-cause mortality in the first week following spousal death was two times normal. The risk of any ischemic heart disease event in that period was elevated 2.3-fold in men and 3.5-fold in women, an effect independent of age; the risk diminished dramatically by the end of the first month (84). The Northridge, California, earthquake of 1994 was followed by a 35% rise in hospitalizations for acute MI (85) and an almost fivefold increase in incidence of sudden cardiac death (86). A study on cardiac events in the New York City area following the September 11, 2001, attack on the World Trade Center showed a 2.3-fold increase in the rate of tachyarrhythmias in patients with ICDs during the month after the attack (87), and another study showed a 68% increase in the frequency of defibrillator shocks for ventricular arrhythmias in ICD patients in Florida during the same month (88). Events less severe than the terrorist attacks also appear to increase cardiac events. In a 5-year retrospective study in England, mortality rates from myocardial infarction or stroke were increased by 1.28 times in men (but not in women) on days when the local professional soccer team lost (89). A study of over 200,000 Chinese and Japanese U.S. residents—ethnic groups that consider the number 4 to be particularly unlucky—found that cardiac mortality peaks on the fourth day of the month (90).
Chronic stress, such as job strain, also increases the risk of CHD. The Whitehall II study (91), a prospective cohort study of 6,895 male and 3,413 female civil servants 35—55 years of age followed over a period of 11 years on average, looked at rates of coronary death, angina, and nonfatal MI. Even after adjustment for age, sex, pay grade, and coronary risk factors, the employees with low decision latitude and high demands (high job strain) had an increased relative risk of CHD events compared with those without high job strain. The effect was most correlated with high job demands and less so with low decision latitude. Interestingly, the effect was more pronounced in lower age groups, and there was no effect difference related to employment grade or social support at work. The INTERHEART study, a recent very large multinational study of risk factors in myocardial infarction patients, revealed that stress levels were higher among individuals with new MI compared with age-, sex-, and site-matched control subjects. The population attributable risk for psychological stress, that is, the proportion of MI cases attributable to stress, was 12%—33% (92).
Anger, hostility, and type A behavior
Early research exploring the relationship of behavior patterns with cardiovascular disease focused on the effect of the type A behavior pattern (TABP)—time urgency, hostility, and achievement-striving-competitiveness—described in a seminal article by Friedman and Rosenman in 1959 (93). In 1975, the Western Collaborative Group Study of more than 3,000 middle-aged men found that over an 8-year period, men with type A behavior were twice as likely as men without those type A traits to develop coronary heart disease (94). Support of this theory was significantly weakened in 1985 by the publication of the MRFIT Behavior Pattern Study (95), a prospective study of 3,110 men with no history of a cardiac event. In the 7 years of follow-up there was no effect of TABP on the occurrence of MIs or coronary death. The Aspirin Myocardial Infarction Study prospectively followed 2,314 male and female heart attack survivors for at least 3 years. No correlation was found between the patients’ self-reported measures of type A behavior and the risk of MI or coronary death (96). A 1987 meta-analysis documented the weakening of the association of type A behavior with coronary heart disease outcomes in larger and newer studies (97). As a result, much of the subsequent research shifted to focus on specific traits within the type A construct, particularly hostility.
In 1980, Williams and colleagues studied 424 patients undergoing coronary angiography for suspected CHD, using both the TABP interview and the Cook-Medley Hostility Scale measurement, derived from the MMPI. Both TABP and Hostility Scale score were independently linked to the presence of atherosclerosis, but the Hostility Scale score was more strongly associated (98). A 25-year follow-up study of 255 physicians found that the Hostility Scale score was predictive of both CAD incidence and total mortality (99). Hostility Scale score has also been found to be associated with increased coronary artery calcification, a measure of atherosclerotic plaque development, in young asymptomatic men (100); with reduced high-frequency heart period variability, a risk factor for coronary and all-cause mortality, in young adults (101); and with increased platelet activation (102). These data stand in contrast to the finding that out of 14 population-based prospective cohort studies with at least 500 subjects published from 1974 to 1997, only six studies showed any link between TABP or hostility and CHD. If studies that include angina in their endpoint of cardiac events are excluded, only three of the 14 studies show any significant risk of cardiac events, with the relative risk ranging from 1.47 to 2.95 (103).
In patients with preexisting CHD, there is some evidence that hostility is associated with a worse prognosis. In 250 CHD patients and 500 control subjects who were followed over 8.5 years in the Western Collaborative Group Study, hostility was the only component independently linked to CHD incidence, with a relative risk of 1.93 (104). Some studies have found that in patients with existing CHD, hostility is associated with a higher risk of cardiac events, CHD mortality, and total mortality (105, 106). Hostility is also associated with more severe exercise-induced myocardial ischemia (107) and a higher risk of restenosis after percutaneous transluminal coronary angioplasty (PTCA) (108). In these studies there was significant variability in how hostility was measured.
There have been negative studies as well. A 33-year follow-up study of 1,399 men who completed the MMPI as freshmen at the University of Minnesota found that high Hostility Scale score did not predict CHD mortality or morbidity (109). In the Normative Aging Study, anger was found to predict the incidence of a combined endpoint (cardiac mortality, nonfatal MI, and angina) but did not significantly predict either cardiac mortality or nonfatal MI (110). Kaufmann and colleagues studied 331 post-MI patients and found that Hostility Scale score did not predict mortality at 6 or 12 months (10). One review identified five prospective cohort studies with 100 or more CHD patients that showed no prognostic role for type A or hostility measurements (103). Another study compared multiple questionnaire measures of anger and hostility in patients with CAD, in patients with valvular heart disease, and in normal control men. Only anger expression had a significant independent correlation with CAD severity (111). In summary, there may be a significant relationship between some aspect of anger and hostility and coronary disease incidence and prognosis, but it is clear that there is a lack of consensus on the best measure of anger and hostility as a cardiovascular risk factor. It is possible that self-report measures of anger and hostility are inevitably confounded by social desirability bias or lack of insight. Provocative data have been presented demonstrating that spouse ratings of men’s negative affects are superior to self-ratings of negative affects by male patients in their correlation with the men’s CAD risk (112).
In European studies, vital exhaustion, defined as a state of diminished energy, increased irritability, and demoralization, has been identified as a coronary disease risk factor and a risk factor for coronary events and sudden cardiac arrest in patients with established CAD (113, 114). The overlap between vital exhaustion and depression has not been definitively determined, and few studies of the issue have been conducted in the United States. This concept requires further study.
A number of biological mechanisms may explain why patients with anxiety or depression have an elevated risk of CVD. Stress and depression cause dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis. Under stress, the hypothalamus produces corticotropin-releasing factor (CRF), which stimulates the anterior pituitary to release corticotropin, signaling the adrenal glands to release cortisol. There is ample evidence of HPA hyperactivity in depressed patients, such as the presence of elevated levels of CRF in spinal fluid (115, 116) and nonsuppression of cortisol following administration of dexamethasone (117). Elevated levels of cortisol have been linked to the development of atherosclerosis and hypertension, although whether this is a direct or an indirect effect is unclear (118, 119). HPA hyperactivity is frequently accompanied by hyperactivity of the sympathoadrenal system. Chronic HPA and sympathoadrenal hyperactivity has been linked to atherosclerosis (120). Acute stress or anger initiates excessive release of epinephrine and norepinephrine, which elevates heart rate, increases blood pressure, and increases vascular resistance. Sympathetic hyperresponsiveness is also linked to the development of ischemia during exercise or mental stress (121, 122). Vagal withdrawal, that is, decreased parasympathetic tone, has been linked to mental stress, which can increase both heart rate and ventricular irritability.
Heart rate variability, which is controlled centrally by the hypothalamus, limbic system, and brainstem as well as peripherally by the vagus nerve, can be construed as an index of the ability to maintain cardiovascular homeostasis. In patients with cardiac disease, heart rate variability is decreased and is a predictor of cardiac mortality and morbidity (123—125). Depression, hostility, and chronic worry are also linked to decreased heart rate variability (126—129). This effect may partly explain the cardiovascular risks of depression and anxiety.
Anxiety disorders and stress cause ventricular instability, which would explain the link between anxiety disorders and sudden cardiac death rather than MI. Animals conditioned to anticipate unpleasant restraint will experience increased ventricular ectopy and ventricular tachycardia-fibrillation when exposed to the threat of restraint. This effect depends on the presence of current or past myocardial ischemia; stress alone does not induce events. In acute ischemia or vulnerable myocardium (e.g., previous infarction), stress induces ventricular tachycardia. The effect can be disrupted by the use of intracerebral β-adrenergic blockade (130—136).
Depression (137—141) and emotional stress (138, 142) are also linked to impairments in platelet function, such as increased reactivity and the secretion of platelet products, including platelet factor 4, β-thromboglobulin, and serotonin, which cause platelet aggregation. Platelet dysfunction can cause the formation of thrombi and also contribute to damage of vascular endothelial cells. Since selective serotonin reuptake inhibitors (SSRIs) reduce the ability of platelets to store serotonin, they may reduce the risk of thrombus formation (see the section Treatment Issues for Cardiac Patients). Chronic stress is also linked to alterations in the fibrinolytic system (143).
The link between chronic infection and inflammatory response and heart disease has been another area of interest in the past decade (144). Inflammatory responses such as increased leukocyte adhesiveness (145), T-cell activation (146, 147), and proinflammatory cytokines (148, 149) have been linked to depression and "burnout." Ford and Erlinger (150) recently found that a history of depression was linked in a graduated fashion—based on temporal proximity of depression to the index measurement and number of episodes—to elevation of C-reactive protein in men.
Can treatment of psychiatric conditions associated with CVD reduce the risk of developing CVD or improve the prognosis of patients who already have CVD? As the epidemiological findings of an association between psychiatric factors and heart disease have become more robust, these questions have taken on profound public health significance. Despite recent studies addressing the treatment of depression in patients with coronary disease, these questions remain largely unanswered (151, 152).
Musselman and associates (153) studied the impact of paroxetine treatment on platelet reactivity in depressed patients. In this small nonrandomized study of 15 depressed and 12 nondepressed patients, none of whom had ischemic heart disease, the depressed patients had greater platelet activity and increased plasma concentrations of platelet factor 4. Treatment with paroxetine for 6 weeks in the depressed patients reduced all parameters of increased platelet activity. Confounding these results, 10 of the 15 depressed patients had one or more risk factors for ischemic heart disease (hypertension, obesity, and so on). None of the 12 control patients had any ischemic heart disease risk factors.
Sauer, Berlin, and Kimmel (154) conducted a case-control study of first MI in 653 smokers 30 to 65 years of age, compared with 2,990 community subjects as controls. After confounding variables were controlled for (including CVD risk factors, age, education, and physical activity), the odds ratio for MI among current subjects who used SSRIs within the index week was 0.35. There was a nonsignificant reduction in risk for those taking non-SSRI antidepressants, which was a very small group.
A large double-blind randomized controlled trial by Glassman and colleagues (152), the Sertraline Antidepressant Heart Attack Randomized Trial (SADHART), studied the safety and efficacy of sertraline treatment for 24 weeks in 369 patients initially hospitalized for acute MI or unstable angina. They reported a decreased incidence of "severe" cardiac events (as rated by the patients’ treating physicians) in the group treated with sertraline (14.5% compared with 22.4% on placebo), although the difference was not statistically significant (the study was not designed to have sufficient statistical power to test the significance of this effect). A follow-up study on the SADHART group data (155) showed that in a subset of 64 patients who participated in platelet studies, treatment with sertraline was associated with significantly decreased markers of platelet activation compared with placebo, an effect independent of the use of antiplatelet regimens, including aspirin, clopidogrel, or other antiplatelet agents and/or oral anticoagulants.
Sauer, Berlin, and Kimmel (156) conducted a case-control study of 1,080 first-MI patients 40 to 75 years of age and 4,256 control subjects who were followed over 3 years to determine the relationship of antidepressant serotonin transporter affinity to MI protection. After controlling for multiple CVD risk factors, SSRI users as a whole in this study did not have a significantly reduced risk of MI compared with nonusers. SSRI users taking high-affinity SSRIs (paroxetine, sertraline, and fluoxetine) did have a significantly reduced risk of MI compared with nonusers, with an odds ratio of 0.59. If confirmed by other studies, this result would suggest that high-affinity SSRIs might be of special benefit for patients at high risk of MI.
The effect of ECT on cardiac morbidity and mortality has not yet been adequately assessed. Schultz and colleagues examined nine patients with depression to test the hypothesis that treatment of depression with ECT would increase heart rate variability and potentially reduce the risk of cardiac mortality. Despite an improvement in patients’ depressive symptoms (mean Hamilton Depression Rating Scale scores decreased from 34.4 to 11.2), a decrease was noted in heart rate variability (157). In elderly patients with major depressive disorder, ECT responders have an increase in heart rate variability after treatment (158, 159).
Psychotherapy and supportive interventions
The Recurrent Coronary Prevention Project looked at 1,013 survivors of acute MI to see if group counseling to reduce type A behavior could reduce the risks of future MI and cardiac mortality (160). The patients were randomly assigned to a control condition, in which group cardiac counseling only was provided; an experimental condition, in which both group cardiac counseling and type A behavior counseling were provided; or a comparison group, which received no group counseling. The type A behavior counseling (primarily cognitive behavior psychotherapy techniques) was significantly more effective in reducing type A behavior than ordinary group counseling. The cumulative cardiac recurrence rate over 4.5 years was 12.9%, approximately half that in the group-cardiac-counseling-only group (21.2%) and in or the comparison group (28.2%).
Two large trials have failed to show any benefit of other specific psychosocial interventions on post-MI survival rates. In the Montreal Heart Attack Readjustment Trial (M-HART), 1,376 patients had monthly screenings for depression and anxiety as well as follow-up home nursing interventions for those who were "distressed." Not only did the intervention have only a small impact on depression and anxiety, it had no impact on cardiac or all-cause mortality in men and showed a trend toward increasing cardiac and all-cause mortality in women (161). A multicenter study of 2,328 patients in England assessing a 7-week outpatient group program of psychological counseling, therapy, relaxation training, and stress management also found minimal impact on depression and anxiety and no impact on clinical sequelae or mortality (162). A 1996 meta-analysis of 23 randomized controlled trials evaluating the effects of psychosocial interventions on post-MI patients found that psychosocial treatment did reduce mortality and cardiac event recurrence rates over 2 years (163). Another meta-analysis of 37 psychoeducational programs (health education and stress management) for CHD patients, published in 1999, found a 34% decrease in cardiac mortality and a 29% decrease in recurrent MI (164).
The Enhancing Recovery in Coronary Heart Disease (ENRICHD) trial studied 2,481 patients from eight clinical centers within 28 days of their admission to the hospital for MI (151). Patients diagnosed with depression and/or low perceived social support were randomly assigned to a usual-care control group or to an intervention group. The intervention group received up to 6 months of individual cognitive behavior therapy sessions and group therapy for up to 3 additional months. After the initial treatment period, patients in either group could also take antidepressants if needed. No difference was seen in mortality or recurrent MI, even in the depressed subgroup of patients. Initially a modest but significantly greater improvement in depression and social support was observed in the treatment group, but by 42 months no difference could be seen. Notably, in patients on any antidepressant, and especially in the subgroup taking SSRIs, there was a trend toward decreased mortality but no difference in nonfatal MI. It should be noted that use of antidepressants was not randomized.
The safe psychopharmacological treatment of cardiac patients requires consideration of three main issues: (1) the cardioactive effects of psychiatric medications; (2) the interaction of psychotropic medications with other medications the patient is taking or is likely to be prescribed; (3) the impact of any comorbid health problems. All of these factors must be taken into account when deciding on which psychotropic medications to use in patients with cardiac disease.
Except in the case of severe CHF causing reduced cardiac output, hepatic congestion, and renal impairment, the absorption, metabolism, and elimination of psychotropics other than lithium is generally not substantially impaired. Therefore, antidepressants must be used in therapeutically effective doses and not reduced unnecessarily for mild CHF.
Tricyclic antidepressants have many properties that make them relatively less desirable for use in cardiac patients. They cause cardiac conduction delays, including bundle-branch block or complete atrioventricular nodal block, and in overdose they may cause ventricular arrhythmias. Tricyclics are class IA antiarrhythmic agents, which prolong atrial and ventricular depolarization, causing increased P-R, QRS, and QT intervals. Studies have shown that QTc intervals of over 440 ms, and especially over 500 msec, are associated with an increased risk of sudden death (165, 166). Class IA antiarrhythmic agents have been proven to increase mortality in post-MI patients with premature ventricular contractions (167, 168).
Tricyclics can also cause significant orthostatic hypotension and tachycardia due to α1-adrenergic blockade. Patients with CHF are often on other medications that cause orthostatic hypotension, such as diuretics and vasodilators, and the risk of falling and sustaining fractures can be significant for elderly patients. Tachycardia increases cardiac demand in general and reduces left ventricular filling time, worsening diastolic function. Nortriptyline and desipramine tend to have fewer anticholinergic side effects and are better tolerated by cardiac patients than tertiary-amine tricyclics such as amitriptyline or doxepin (169—171).
Given their effects, tricyclic antidepressants should be avoided whenever possible in patients with recent MI and should not be considered first-line agents for patients with ischemic heart disease or preexisting intraventricular conduction delays. In certain patients, the benefits may outweigh the risks, so consideration should be given to the entire clinical situation (172). When tricyclics are prescribed for patients with cardiac disease, orthostatic blood pressure measurements should be obtained at baseline and during treatment. A baseline ECG should also be obtained, and follow-up ECGs should be obtained when a therapeutic level of the drug has been reached to evaluate the P-R, QRS, and QTc intervals and to monitor for bundle-branch block or complete atrioventricular block (173).
Selective serotonin reuptake inhibitors
SSRIs have few cardiac effects in healthy patients. They can cause slowing of heart rate by a few beats per minute, which is usually clinically insignificant (174). A number of studies have confirmed that, overall, SSRIs carry little or no risk of increasing the QTc interval or causing other ECG changes (175—177). Nonetheless, a few case reports have been published of significant sinus bradycardia, dysrhythmias, syncope (178), and QTc prolongation (179) in patients taking SSRIs.
In studies of patients who have preexisting cardiac disease, SSRIs have been found to have minimal negative effects on blood pressure or cardiac conduction. Until recently no randomized controlled trials with SSRIs and a placebo or an alternative treatment group with more than 81 total patients had been published (180, 181). The SADHART study, with 369 patients, was the first large randomized controlled trial to examine the effects of an SSRI, sertraline, in patients with depression after an acute coronary event. The study found that there was no effect of sertraline on heart rate, blood pressure, arrhythmias, ejection fraction, or cardiac conduction (152). Sertraline was found to be an effective treatment for depression, demonstrating a weak response in the overall sample but a very good response in the groups with severe and with recurrent major depressive disorder. Another small study showed a statistically significant 7% improvement in ejection fraction in patients with preexisting cardiac disease taking fluoxetine (182).
Cardiac effects of SSRIs may be dose dependent and vary from drug to drug. In a review of 6,000 ECGs of 1,789 patients treated with citalopram in clinical trials, no evidence was found of QTc prolongation (174). Studies of patients who have taken overdoses of various SSRIs indicate that in overdose citalopram causes significantly greater QTc prolongation than other SSRIs (183, 184). This may be related to the finding that citalopram inhibits cardiac sodium and calcium channels in animal studies (185). While very large doses (400 mg or more) were required to produce QTc prolongation in healthy patients, these data would suggest that, in general, extremely high doses of citalopram should be used with careful monitoring in cardiac patients, particularly those who are also taking inhibitors of cytochrome P450 enzyme 2C19 or 3A4.
Other antidepressants have not been as well studied with regard to cardiac effects. In a randomized double-blind crossover study of 10 patients with impaired left ventricular ejection fraction taking imipramine or bupropion, bupropion had no significant cardiac effects (171). In a study of 36 inpatients with cardiac disease, bupropion caused an increase in supine blood pressure, although it did not cause significant orthostatic hypotension, conduction disturbances, or ventricular arrhythmias. Bupropion did cause an exacerbation of baseline hypertension in two patients (186). Venlafaxine has not been specifically studied in cardiac patients, but it may cause a dose-dependent increase in diastolic blood pressure starting at 150 mg/day, with significant increases at doses of 300 mg/day or above (187). A 1-week randomized controlled trial of 20 patients—10 taking mirtazapine and 10 taking imipramine—found that mirtazapine caused a significant increase in heart rate and decrease in heart rate variability and had no effect on blood pressure or blood pressure variability (188).
Nefazodone is rarely used in cardiac patients because of multiple drug interactions. One small open-label study in CHF patients showed that it was effective in treating major depression, with a significant reduction in heart rate and no changes in heart rate variability. The QT interval was increased, which is consistent with a reduced heart rate, but the QTc interval did not change (189).
Monoamine oxidase inhibitors are almost never used in cardiac patients because of drug interactions, orthostatic hypotension, and the risk of hypertensive crises.
A number of SSRIs inhibit cytochrome P450 pathways, which has an impact on treatment decisions. Fluoxetine, paroxetine, and duloxetine inhibit 2D6. Nefazodone, fluoxetine, and, to a lesser extent, sertraline, are 3A4 inhibitors. Other medications, such as ketoconazole and erythromycin, are potent 3A4 inhibitors. Amiodarone and quinidine are 2D6 inhibitors and can elevate blood levels of fluoxetine, risperidone, and several tricyclics.
Tables 1T1, 2T2, and 3T3 summarize interactions between psychotropic, cardioactive, and cardiovascular drugs.
Antipsychotic medications may be used in small doses for short periods to treat delirium in cardiac patients who are in acute cardiac care settings or inpatient units. Longer-term and higher-dose therapy may be needed for cardiac patients who have a chronic psychotic disorder. The two situations are distinct, and some thought must be given to both the short-term and long-term risks and benefits of the various antipsychotic medications.
A number of antipsychotics have been linked to torsade de pointes and sudden death, among them pimozide, sertindole, droperidol, haloperidol, and thioridazine (190). The greatest risk is associated with thioridazine. In a large retrospective case-control study of all sudden deaths in psychiatric wards of five hospitals in England over 11 years, the only antipsychotic that was found to be an independent risk factor for sudden death was thioridazine (191). Of six major antipsychotics (thioridazine, ziprasidone, quetiapine, risperidone, olanzapine, and haloperidol), thioridazine produces the largest QTc prolongation, with a mean change of 35.6 ms (190, 192, 193). Several excellent reviews have examined the relationship of antipsychotics to QTc prolongation (191, 193—198).
Among the atypical antipsychotics, none has been linked to torsade de pointes, even though most have a larger impact on QTc than haloperidol, which has been linked to torsade de pointes. Risperidone has been linked to one fatality due to pulseless electrical activity, and it has been reported to cause QTc and QRS prolongation in two cases of overdose and in 8 of 380 patients in a double-blind study by the manufacturer (199). The fatality may have been due to factors other than QTc prolongation (198). In two reported cases of ziprasidone overdose, neither patient developed torsade de pointes, though one had QRS prolongation and the other had QTc prolongation (198). Aripiprazole, olanzapine, and quetiapine have not been linked to torsade de pointes.
In considering the use of drugs that may prolong the QT interval, factors to be reviewed include a family or personal history of long QT syndrome, a history of sudden cardiac arrest, syncope or unexplained seizure, arrhythmias, hypertension, valvular heart disease, bradycardia, and use of other medications that may prolong the QT interval or interfere with the metabolism of QT-prolonging agents. Class IA and class III antiarrhythmic drugs, dolasetron, droperidol, tacrolimus, levomethadyl acetate, other antipsychotic agents, many antibiotics (the "floxacins"), and antifungal agents may increase the risk of torsade. Magnesium and potassium levels should be monitored, as abnormalities may also increase the vulnerability to developing torsade (190, 200, 201).
In addition to the potential proarrhythmic effects of the antipsychotics, the decision of which medication to select for patients with cardiac disease should take into account the drug’s potential to cause orthostatic hypotension, glucose intolerance, and hyperlipidemia. Orthostatic hypotension, which is related to the α1-adrenergic receptor blocking properties of antipsychotics, is seen most frequently with low-potency antipsychotics, such as chlorpromazine. Some atypical antipsychotics may also cause orthostatic hypotension. Hyperlipidemia, glucose intolerance, and diabetes mellitus, with or without weight gain, have been linked to most atypical antipsychotics, including clozapine, olanzapine, quetiapine, and risperidone. According to their manufacturers, aripiprazole and ziprasidone are less likely to cause these effects. In the initial phase of treatment, and periodically thereafter, blood sugar and lipid levels should be monitored.
Overall, benzodiazepines and buspirone are clinically safe and effective in patients with cardiac disease. Benzodiazepines have been shown to increase heart rate acutely, but they also reduce vagal tone and heart period variability, most likely by potentiating γ-aminobutyric acid (202, 203). Since they reduce anxiety and sympathetic nervous system activation, they decrease the heart rate and pressor responses to stress. On the other hand, buspirone, in one study, has been linked to decreased baseline heart rate, and it may actually enhance increased heart rate in response to stress (204).
Stimulants are particularly useful for treating apathy, fatigue, and psychomotor slowing in the medically ill, and they offer the advantage of working within a few days instead of several weeks. Dextroamphetamine and methylphenidate in doses of 5—30 mg/day are well tolerated by cardiac patients and have no significant effects on heart rate or blood pressure (205). They can also be used with good efficacy in poststroke patients, with no evidence of adverse cardiac effects (206).
At therapeutic doses, lithium can cause sinus node dysfunction that is generally reversible when the medication is discontinued, although there have been rare reports of sinus arrest, atrioventricular block, and aggravation of ventricular arrhythmias (207—209). Lithium toxicity can cause sinoatrial block, atrioventricular block or dissociation, bradyarrhythmias, and ventricular tachycardia or fibrillation (208). Generally, lithium can be used safely in cardiac patients even with reduced cardiac output by decreasing the dosage; the dosage can be decreased even further in patients with impaired renal function in advanced heart failure. Greater caution is necessary for patients taking diuretics, particularly thiazides, and those on salt-restricted diets. Lithium may also increase the risk of cardiac arrhythmias in patients taking angiotensin-converting enzyme inhibitors (194). In patients with acute congestive heart failure exacerbations or acute coronary syndromes, the use of lithium should generally be avoided because of rapid electrolyte and fluid balance shifts.
Valproate and lamotrigine have no apparent cardiovascular effects. Carbamazepine has a tricyclic-like class IA antiarrhythmic effect and should be used only with cautions similar to those discussed above. It also induces cytochrome P450 enzyme 3A4, which can increase the metabolism of a number of anticoagulant and cardiovascular medications.