Depression, mortality, and medical morbidity in patients with coronary heart disease

Abstract

There is substantial evidence that depression is a risk factor for cardiac morbidity and mortality, both for patients without clinical evidence of coronary heart disease at index examination and for patients with established coronary disease. The relationship is most apparent for patients with a recent acute myocardial infarction. Many questions about the impact of depression on heart disease remain unresolved.

Introduction

The study of the relationship between depression and heart disease has a long history. Scientific evidence of an association between these conditions dates back at least to the 1930s, when several published studies found a higher incidence of coronary heart disease (CHD)–related death among depressed psychiatric patients than in control groups Fuller 1935, Malzberg 1937. Subsequent studies that were larger and better controlled found a relationship between depression and CHD-related mortality in both bipolar and unipolar depression (Rabins et al 1985; Tsuang et al 1980; Week et al 1987).

More recent studies have found that a history of major depression Cohen et al 2001, Pratt et al 1996, current depression symptoms Anda et al 1993, Ariyo et al 2000, Barefoot et al 1996, Ferketich et al 2000, Mendes de Leon et al 1998, Sesso et al 1998, Simonsick et al 1995, Whooley and Browner 1998, a diagnosis of clinical depression Aromaa et al 1994, Ford et al 1998, Hippisley-Cox et al 1998, Penninx et al 1998, Pratt et al 1996, and an increase in depressive symptoms with time Penninx et al 1998, Wassertheil-Smoller et al 1996 predict incident CHD or cardiac mortality in cohorts without any clinical evidence of CHD when depression was initially assessed. The participants in these studies were randomly selected from specific geographic areas or institutions Barefoot et al 1996, Penninx et al 2001, Pratt et al 1996 or were subjects in studies of other CHD risk factors Ariyo et al 2000, Ferketich et al 2000, Wassertheil-Smoller et al 1996. They ranged in age from the early 20s (Ford et al 1998) to older than 65 years (Penninx et al 1998). The adequacy of adjustment for potential confounders has varied considerably across studies. Although a few reports have failed to find that depression predicts incident CHD (Thomas et al 1992), most studies have found significant effects of depression in men, women, or both.

The consistency of these findings is impressive, given the variety of assessment tools that have been used and the diversity of the populations that have been studied. The assessment instruments have ranged from nonstandardized questionnaires to structured psychiatric interviews. Many of the studies were not originally designed to investigate depression as a risk factor for CHD, but brief depression questionnaires were included in the baseline assessment battery for other reasons. As interest in the role of depression in CHD grew during the 1980s and 1990s, researchers began to assess the role of depression as a risk factor for CHD end points in these archival data sets. The results suggest that depression, even when assessed with simple self-report measures, can predict CHD end points occurring many years later. One of the possible reasons for this is that depression often follows a chronic or recurrent course. Unfortunately, most of these studies only assessed depression on one or two occasions. Thus the relationship between the chronicity of depression and mortality risk remains unclear.

A clinical diagnosis of depression seems to carry a greater risk of incident CHD than do subclinical symptoms of depression. For example, Aromaa and colleagues (Aromaa et al 1994) found a relative risk of 3.36 for development of CHD among subjects with major depression, and Pratt and colleagues (Pratt et al 1996) reported a relative risk of 4.16 for incident CHD among community residents with major depression. Studies that have relied on questionnaires to assess symptoms of depression have generally reported relative risks of 2.0 or less (Barefoot et al 1996). Furthermore, all of the studies to date that have failed to find a relationship between depression and CHD end points relied on depression questionnaires rather than on structured interviews and clinical diagnoses. In summary, numerous studies that have found that depressed individuals, even otherwise healthy persons in their 20s or30s, are more likely to acquire or die from CHD than are nondepressed persons with the same age and risk factor profile.

Depression also predicts cardiac morbidity and mortality among patients with established CHD. Patients with positive results of coronary angiography (Barefoot et al 1996; Carney et al 1988) or exercise stress testing (Herrmann et al 2000) and patients who have recently undergone bypass surgery (Burg et al 2001; Connerney et al 2000) are at greater risk for nonfatal cardiac events and death than are comparable nondepressed patients.

Depression is a particularly significant risk factor for CHD morbidity and mortality after acute myocardial infarction (MI), and patients after MI have received the most study Ahern et al 1990, Bush et al 2001, Denollet et al 1995, Frasure-Smith et al 1993, Frasure-Smith et al 1995, Kaufmann et al 1999, Irvine et al 1999, Ladwig et al 1991. One of the first of these studies found that the adjusted risk of death during the first 6 months after an acute MI is approximately four times higher in depressed than nondepressed patients (Frasure-Smith et al 1993). The findings of subsequent studies have been reasonably consistent, despite differences in the methods used to assess depression.

In most of these studies, the relationship between depression and mortality has survived adjustment for medical and demographic risk factors Bush et al 2001, Denollet et al 1995, Frasure-Smith et al 1993, Frasure-Smith et al 1995, Irvine et al 1999; however, some studies have found only nonsignificant trends after risk factor adjustment Kaufmann et al 1999, Ladwig et al 1991. In the study by Ladwig and colleagues (Ladwig et al 1991) the relationship was not significant (p < .07), despite an odds ratio of 4.9 for mortality among patients with high levels of depression after covariate adjustment. The study was limited by a small sample size and by the fact that only 2% of the participants died during follow-up. In the study by Kaufmann and colleagues (Kaufmann et al 1999), the unadjusted odds ratio was 2.3, but no adjusted odds ratio was reported. Once again, low statistical power may have been responsible for the nonsignificant findings.

At least two studies have failed to find even a univariate relationship between depression and mortality Lane et al 2001, Mayou et al 2000. Mayou and colleagues (Mayou et al 2000) reported a nonsignificant relative risk of death of 1.6 with the Hospital Anxiety and Depression Scale. The Hospital Anxiety and Depression Scale has generally not been as predictive of CHD-related death as have other measures of depression (Hermann et al 2000), and it may be relatively insensitive to depression symptoms in cardiac patients (Hermann et al 1993). Lane and colleagues (Lane et al 2001) used the Beck Depression Inventory (BDI), which (unlike the Hospital Anxiety and Depression Scale) has been shown to predict post-MI death. Only about two in three eligible patients agreed to participate in this study, and the mean BDI score in the “depressed” group was in the “mildly depressed” range (15.2). This suggests the possibility that patients with relatively severely depression may have declined to participate, which could have reduced the association between depression and CHD mortality. In other respects, the study was well designed and carefully conducted.

These two reports are among the most recent studies of post-MI depression and mortality. It is possible that advances in the treatment of acute MI and in the prevention of recurrent events have begun to suppress or delay the effect of depression on mortality. The effects of depression may be minimal for the first 6 months after the acute MI, while patients are receiving particularly aggressive cardiologic care, and then become more prominent later, when the protective effects of the acute and early post-MI interventions begin to abate. This may make it more difficult to detect early effects of depression than it was in the 1990s, and it suggests that patients will have to be followed up for at least 1 or 2 years after the index MI if an effect is to be detected in future studies.

The declining long-term mortality rate after MI (Centers for Disease Control and Prevention 1999) will also make it more difficult to detect the effects of depression. The odds of survival after acute MI have improved substantially in recent years, partly as a result of advances in thrombolytic therapy, early revascularization, statins, aspirin, and angiotensin-converting enzyme inhibitors (Centers for Disease Control and Prevention 1999). This may have contributed to a reduction in the relative risk of death associated with depression. As the mortality rate declines, larger samples are needed to study risk factors with adequate statistical power (Lespérance and Frasure-Smith 1999). In summary, larger studies and longer follow-ups will be needed to document any association between depression and mortality and other cardiac events after MI.

There has been only one study of the effects of depression on patients with unstable angina (Lespérance et al 2000). More than 40% of the 430 participants in this study were found to be depressed, and these patients were more likely to die or have a nonfatal MI during the follow-up period than were the nondepressed patients, with an adjusted odds ratio of 6.73. These results are quite dramatic. More studies are needed to further document and clarify this relationship.

Both the prevalence and prognostic importance of a risk factor must be considered when judging its public health significance. Depression is common among patients with CHD. Approximately one in five patients undergoing diagnostic angiography has major depression Carney et al 1987, Gonzalez et al 1996, Hance et al 1995, and comparable numbers have minor depression (Hance et al 1995). Similarly, about one in five patients has major depression after an acute MI Carney et al 1990, Forrester et al 1992, Frasure-Smith et al 1993, Schleifer et al 1989, and another one in five has minor depression (Schleifer et al 1989).

There are 1.5 million MIs per year in the United States, and approximately 10% of patients die within 1 year of an initial MI (American Heart Association 1995). Approximately two fifths of these patients (about 600,000 patients annually) meet the diagnostic criteria for major or minor depression within a few weeks after MI Carney et al 1990, Forrester et al 1992, Schleifer et al 1989. If we assume the adjusted relative risk of post-MI death associated with depression to be 3.0 for patients with major depression and 2.0 for those with minor depression, more than 90,000 deaths per year may be at least partially attributable to post-MI depression. Furthermore, this estimate does not address the medical morbidity and diminished quality of life associated with depression after MI, nor does it consider the mortality and medical morbidity occurring at other times in the course of the cardiac illness.

Few studies have directly compared the prognostic values of alternative methods of measuring depression. Of all of the measures that have been used, standardized diagnostic interviews and the BDI have shown the most consistent relationship with medical morbidity and mortality. Endorsing even a few depression symptoms on the BDI identifies patients at risk for death Bush et al 2001, Lespérance et al 2002; however, the relative risks tend to be higher for clinically significant depression than for subsyndromal depressive symptoms. Thus there may be a dose-response relationship that extends from subsyndromal depressive symptoms to severe major depressive disorder.

Clinical diagnostic interviews should be used to help ensure that depression is truly what is being assessed. The BDI is not highly specific for depression; scores tend to be elevated not only depressed patients but also in nondepressed patients with other psychiatric disorders and in psychiatrically well individuals during with transient adjustment reactions. Consequently, some authors have argued that the risk factor identified in these studies should be labeled “distress” rather than depression (Denollet et al 1998).