Author + information
- Received November 1, 2000
- Revision received March 15, 2001
- Accepted March 26, 2001
- Published online July 1, 2001.
- Viola Vaccarino, MD, PhD∗,* (, )
- Stanislav V Kasl, PhD∗,
- Jerome Abramson, PhD∗ and
- Harlan M Krumholz, MD, FACC∗,†,‡,§
- ↵*Reprint requests and correspondence:
Dr. Viola Vaccarino, Department of Medicine, Division of Cardiology, Emory University School of Medicine, 1256 Briarcliff Road, Suite 1 North, Atlanta, Georgia 30306
We sought to examine whether depressive symptoms are associated with poorer prognosis in patients with heart failure.
Depression is an established risk factor for poor outcome in patients with coronary heart disease (CHD). Little is known of its role in patients with heart failure.
We prospectively followed 391 patients ≥50 years of age who met criteria for decompensated heart failure on hospital admission. The outcome of the study was death or decline in activities of daily living (ADL) at six months, relative to baseline. Depressive symptoms were measured at baseline by means of the Geriatric Depression Scale, Short-Form, with 6 to 7 symptoms, 8 to 10 symptoms and ≥11 symptoms indicating mild, moderate and severe levels of depressive symptoms, respectively.
There was a strong and graded association between the severity of depressive symptoms at baseline and the rate of the combined end point of either functional decline or death at six months. After adjustment for demographic factors, medical history, baseline functional status and clinical severity, patients with ≥11 depressive symptoms, compared with those with <6 depressive symptoms, had an 82% higher risk of either functional decline or death, whereas the intermediate levels of depressive symptoms showed intermediate risk (p = 0.003 for trend). A similar graded association was found for functional decline and death separately; however, after multivariate analysis, the association with mortality was less strong and no longer statistically significant.
An increasing number of depressive symptoms is a negative prognostic factor for patients with heart failure, just as it is for patients with CHD.
Depression is a relatively common condition among individuals with heart failure, with rates reported from 24% (1)to >40% (2)in patients with stable ambulatory heart failure. In patients with coronary heart disease (CHD), depression is associated with poor outcomes, including recurrent cardiac events (3,4), mortality (5–8)and low functional status or quality of life (9,10). These associations exist both when the risk factor is a major depressive episode (4,7)and when a high level of depressive symptoms occur in the absence of a clinical diagnosis of depression (5,8,11).
Despite the established role of depression as a prognostic factor in patients with CHD, little is known about whether depression plays a similar role in patients with heart failure. A link between depressive symptoms and the course of heart failure may be expected, because depressed individuals have elevated sympathoadrenal activation (12–15), a response that may be deleterious for patients with heart failure (16,17). Accordingly, the present study examined whether depressive symptoms increased the risk of death and disability in a consecutive sample of patients meeting clinical criteria for heart failure on hospital admission.
Study sample and general procedures
We studied consecutive patients ≥50 years of age admitted to Yale-New Haven Hospital in a 2.5-year period (March 1996 to September 1998) who met clinical criteria for the presence of heart failure on hospital admission. Yale-New Haven Hospital is a 900-bed hospital affiliated with Yale University and is the largest community hospital serving New Haven, Connecticut. To be eligible for the study, patients were identified with either a hospital admission diagnosis of heart failure or radiologic signs of heart failure on the admission chest X-ray film. Patients who met either of these conditions had their medical records reviewed to verify a set of additional criteria of symptoms and signs (18,19). At least one of the following symptoms had to be present on admission or had to have precipitated the patient’s hospital admission: new onset or worsening of symptoms of respiratory distress (e.g., dyspnea or shortness of breath at rest or during exertion, orthopnea, paroxysmal nocturnal dyspnea), chest pain consistent with myocardial ischemia or nonspecific symptoms that may be manifestations of heart failure (e.g., fatigue, confusion or disorientation, loss of consciousness, syncope). At least one of the following signs had to be present on the admission physical examination: S3gallop, bilateral pulmonary rales or crackles, hypotension or cardiogenic shock, cardiac arrest, respiratory rate ≥24 breaths/min, peripheral edema, cardiac tachyarrhythmias or conduction disorders (e.g., atrioventricular block of second degree or higher, left bundle branch block, bifascicular block). Although some of the qualifying symptoms and signs may appear nonspecific, they apply to patients who have already passed the first phase of screening (i.e., patients who have either signs of heart failure on the admission chest X-ray film or an admitting diagnosis of heart failure). Excluded from the study were patients transferred from other hospitals, patients admitted from nursing homes, patients with heart failure secondary to high-output states or noncardiac diseases and patients with a terminal illness in addition to heart failure. The Institutional Review Board approved the study, and all patients provided informed consent. Within three days of hospital admission, patients who met criteria for heart failure were interviewed face to face to assess study variables, as described subsequently. Medical charts were reviewed after hospital discharge.
Assessment of depressive symptoms
Depressive symptoms were assessed by means of the Geriatric Depression Scale (GDS), Short-Form, which is a validated 15-item, self-report depressive symptom scale designed to detect the presence of current depression in older adults (20). The GDS is one of the most commonly used instruments for depression screening among older adults (21,22), and it is well suited to measure depressive symptoms in severely ill patients in the hospital, because it focuses on the nonsomatic symptoms of depression (23). Using a cut-off point of six or more symptoms, this scale has a sensitivity between 88% to 92% and a specificity between 62% to 81%, as compared with a structured clinical interview for depression (21,24). Consistent with previous studies (20,22), we defined the presence of clinically relevant depressive symptoms as ≥6 symptoms, with 6 to 7 symptoms indicating a mild level of depressive symptoms, 8 to 10 symptoms indicating a moderate level of depressive symptoms and ≥11 symptoms indicating a severe level of depressive symptoms.
Baseline health status, physical function and clinical variables
Patients were asked about their level of physical function in the month before hospital admission, using self-assessments suitable for severely ill in-patients. We measured the patients’ functional capability in activities of daily living (ADL), given the impact of a decline in ADL on the quality of life of patients with heart failure (25,26). Limitations in ADL were assessed by means of the ADL scale by Katz et al. (27), which measures with high accuracy (28–30)the self-reported ability to perform one or more of the following activities without the help of another person: moving from bed to chair, using the toilet, bathing, dressing, eating and walking across a small room. We also assessed physical limitations due to dyspnea by means of the Dyspnea Index (31,32). Dyspnea at rest was defined as the presence of shortness of breath at rest, while sitting or lying down.
Patients were asked about their history of heart failure through questions on shortness of breath, orthopnea and paroxysmal nocturnal dyspnea, similar in structure to the Rose questionnaire on chronic conditions (33). They were also asked whether they had been previously admitted to the hospital for congestive heart failure or shortness of breath. Information on other medical history, suspected factors that precipitated the hospital admission, clinical characteristics on admission, treatment with angiotensin-converting enzyme inhibitors at discharge and left ventricular ejection fraction (LVEF) were abstracted from the medical records. If LVEF was not obtained during the index hospital period, previous values were recorded, if available. In our sample, 71% of the patients had LVEF assessed during the index admission; 13% had a previous value obtained within the past six months before admission and 17% had a value obtained earlier than six months before admission.
The main outcome of the study was a combined end point of either death or decline in ADL function at six months after the index hospital admission. The rationale for the use of this combined outcome is that death can be considered as the worst functional status (34); therefore, exclusion of patients who died before being able to provide information on functional status at follow-up could bias the results (35). Death and functional decline among survivors at six months, however, were also examined separately. Follow-up assessments were done in a blinded fashion with respect to depression status and GDS scores. At one and six months, patients were contacted by phone to ascertain their vital status and functional status by means of Katz’s ADL scale, the same as in the baseline interview. At six months, vital status was known for all the patients and functional status was known for >95% of the surviving patients. Decline in ADL function was defined as a greater number of ADL limitations at six months than at baseline. Five patients who had a functional decline by one month after discharge and for whom information on ADL function at six months was not available were classified as having a decline in ADL function.
First, we compared mean GDS depressive symptom scores according to the levels of baseline demographic factors, medical history, other baseline variables and LVEF by using the Student ttest or analysis of variance. Next, we conducted multivariable regression analyses that modeled the relative risk of study outcomes (i.e., death or decline in ADL function at six months) according to the levels of depressive symptoms at baseline, as defined earlier, after adjusting for other baseline factors (i.e., demographic characteristics, medical history, baseline functional status, clinical characteristics on presentation and LVEF). This was accomplished by means of a generalized linear model with a binomial error distribution and a logarithmic link (36).
We fitted three sequential models. In the first model, we adjusted for demographic factors only. In the second model, we added medical history and baseline functional status. In the third model, we added clinical severity features (i.e., clinical characteristics on presentation and LVEF). At each step, variables considered for inclusion were those associated with the GDS depressive symptom score, at p < 0.20 in the bivariate analyses (37).
All tests for statistical significance were two-tailed with an alpha level of 0.05. All analyses were conducted using SAS software, release 6.12 (Cary, North Carolina).
In the study period, 1,148 consecutively admitted patients were screened for study eligibility. A total of 520 met the eligibility criteria, and of these, 426 patients (81.9%) were enrolled. We excluded patients with a missing GDS score (n = 13), missing baseline functional status (n = 3) or missing follow-up functional status, unless the patients had died or had already declined in ADL function by one month (n = 16). Three additional patients were excluded because their medical records could not be located, which meant missing information on all clinical variables. These exclusions yielded a final sample of 391 patients.
Lower education and non-black race were associated with higher depressive symptom scores (Table 1). Women also tended to have higher depressive symptom scores compared with men, but there was no association with age.
With respect to medical history, previous myocardial infarctions and previous hospital admissions for heart failure were significantly associated with higher depressive symptom scores. A history of diabetes also showed a borderline association with higher GDS scores. Baseline functional status, including ADL limitations and dyspnea at rest, were strongly related to depressive symptoms.
Depressive symptoms were commonly noted: of all patients, 35% scored in the mild range, 33.5% in the moderate range and 9% in the severe range. There was no association between depressive symptom scores and suspected factors precipitating the current heart failure episode, including nonadherence to medications, uncontrolled hypertension, tachyarrhythmia and myocardial ischemia. Similarly, clinical presentation characteristics showed little relationship with depressive symptoms, with the exception of systolic blood pressure, which was inversely associated with depressive symptoms. There was no association between GDS score and clinical management, such as the likelihood of obtaining an ejection fraction measurement during the index admission or angiotensin-converting enzyme inhibitor treatment on discharge (data not shown).
When the rates for the combined end point of a decline in ADL activities or death at six months were examined according to depressive symptom severity at baseline, a strong and graded association was found (Table 2). The higher the level of depressive symptoms, the higher the rate of either functional decline or death at six months. For the combined end point, the rates increased from 30.7% in the absence of clinically relevant depressive symptoms to 60% in the presence of severe depressive symptoms (p = 0.001 for trend). When the rates of functional decline and death were examined separately, a similar graded association was found between depressive symptoms and each of these two outcomes (Table 2).
To determine whether the association between depressive symptoms and the study outcomes was independent of demographic factors, medical history, baseline functional status and clinical severity (i.e., clinical characteristics on presentation and LVEF), we adjusted for these covariables in a series of sequential models (Table 3). After adjusting for all of these factors, patients with severe depressive symptoms, compared with those without clinically relevant depressive symptoms, had an 82% higher risk of either functional decline or death, whereas the intermediate levels of depressive symptoms showed intermediate risk. Adjustment for baseline health status and clinical severity had a greater impact on the association between depressive symptoms and mortality than on the association between depressive symptoms and functional decline (Table 3). After adjusting for all of these factors, increasing levels of depressive symptoms retained a strong and significant trend toward a higher decline in functional status, whereas the trend for higher mortality was no longer statistically significant.
Effect of depression on functional decline and death in patients with heart failure
Depressive symptoms occurred commonly in this population, as noted by others (1,2,38). The prevalence of depressive symptoms was higher in our study than in previous reports, which might be due to the fact that our group consisted of severely ill in-patients, whereas other studies typically focused on patients with stable, ambulatory heart failure (1,2).
We found that, among patients admitted to the hospital for heart failure, an increasing number of depressive symptoms was associated with a significantly higher risk of the combined end point of functional decline or death. The association was strong and graded, with higher risks associated with higher depressive symptom scores, and was not explained by sociodemographic factors, baseline functional status, medical history or other baseline characteristics.
Virtually no previous study has addressed the link between depression and functional outcomes in heart failure. An earlier cross-sectional study reported higher levels of disability among patients with heart failure with increasing levels of depression (39). Using a prospective design, our study demonstrated a strong relationship between the level of depressive symptoms at baseline and subsequent functional decline in patients with heart failure.
When we examined mortality as a separate outcome, a clear, albeit nonsignificant trend was found between increasing depressive symptom levels and higher mortality rates at six months. The latter is consistent with previous studies (38,40,41), which showed mild to moderate associations between depression and mortality in patients with heart failure, although the difference was statistically significant in only in one of the studies (40). In our study, after adjusting for baseline characteristics, the effect of depressive symptoms on mortality was weaker than its effect on functional decline.
A number of potential explanations are possible for the association described earlier. First, higher depressive symptom levels at baseline may have been secondary to a greater clinical severity of heart failure. Such a circumstance could make it falsely appear as if depressive symptoms increased the risk of adverse outcomes, when, in fact, the higher risk was simply a function of disease severity. However, this explanation seems unlikely, because we adjusted for a number of medical history and clinical factors indicative of disease severity, including baseline functional status, dyspnea at rest, previous hospital admissions for heart failure and LVEF.
Second, depressed persons may become increasingly depressed over time and may therefore be more likely to report more disabilities over time, even when their functional status has actually remained stable (42,43). This explanation is also unlikely, because, as described earlier, we also found a trend for depressive symptoms to increase the risk of mortality, an end point that is not susceptible to this type of bias.
Third, the association we observed may have been mediated by unhealthy behaviors, such as smoking and excessive alcohol consumption. In our data, however, depressive symptoms were not significantly associated with smoking or a history of alcohol abuse; therefore, it is unlikely that these factors play a major role. Nonetheless, other behaviors could have mediated the association we observed. For example, there is evidence that depression reduces the likelihood that one will take medication properly, follow dietary advice or adhere to other aspects of medical regimens (44,45). In our study, however, we failed to find a baseline association between nonadherence to medications and depressive symptoms. This suggests that noncompliance with medical regimens should not play a major role in our findings. Yet, we did not have reliable information on adherence to medical regimens during the follow-up; therefore, we cannot completely measure the role of this factor in our results.
Fourth, depression may worsen the prognosis of heart failure through direct physiologic mechanisms. Depression has been associated with excessive activation of the sympathetic nervous system. Hypersecretion of norepinephrine in depressed patients has been documented by the presence of elevated norepinephrine and its metabolites in plasma and urine (12–15), and treatment with tricyclic antidepressants decreases plasma and urinary levels of norepinephrine and its metabolites (46,47). Heart failure is accompanied by heightened activation of the sympathetic nervous system. This initially aids the failing heart, but eventually is thought to play a major role in the progression of heart failure (16,17). Neurohormonal activation may exacerbate the hemodynamic abnormalities of heart failure by increasing the heart rate and inducing vasoconstriction and sodium retention. In addition, increased activity of the sympathoadrenal system may predispose patients to arrhythmias and sudden death (48). Therefore, by further increasing the level of sympathetic activity, which is already elevated in heart failure, depression may foster disease progression. Depression is also associated with impaired vagal tone and reduced heart rate variability (49,50), which is also a risk factor for cardiovascular morbidity and mortality (51,52). Because we did not have information on variables of sympathetic and parasympathetic nervous system activation, these mechanisms remain speculative. Nevertheless, such explanations are biologically plausible.
This study had some potential limitations. First, although the GDS is a valid measure of depressive symptoms among older persons, it is not synonymous with a clinical diagnosis of depression. However, we found no evidence of a threshold effect. On the contrary, there was a graded effect, with an increasing level of depressive symptoms being associated with a risk of higher outcomes. Therefore, a symptom checklist assessing the amount of depressive symptomatology may be more useful than a dichotomous diagnosis. In addition, from a practical standpoint, the GDS is easier to administer than a structured interview for clinical diagnosis of depression, and it does not require a mental health specialist. Therefore, it may provide a more feasible means of identifying depressed patients with heart failure who are in need of special attention.
A second weakness of our study was the lack of information on adherence to medical regimens during follow-up. Thus, we were not able to determine whether adherence after hospital discharge may have explained the relationship between depressive symptoms and risk of disability and death. However, the lack of association between adherence and depressive symptoms at baseline suggests that adherence may not have played an important role in our study. Finally, although our sample was larger than previous studies, its size did not provide adequate power to examine mortality as a separate end point.
The present study provides strong evidence that depressive symptoms represent a negative prognostic factor in patients with heart failure, just as they do for patients with CHD. Early identification and treatment of depressive symptoms and clinical depression should provide an important tool for decreasing the morbidity and mortality of heart failure.
☆ This study was supported by grant no. 95-094 from the Donaghue Medical Research Foundation, Hartford, Connecticut, and by grant no. P60 AG 104 from the Claude D. Pepper Older Americans Independence Center 69.
- activities of daily living
- coronary heart disease
- Geriatric Depression Scale
- left ventricular ejection fraction
- Received November 1, 2000.
- Revision received March 15, 2001.
- Accepted March 26, 2001.
- American College of Cardiology
- Denollet J,
- Brutsaert D.L
- Carney R.M,
- Rich M.W,
- Freedland K.E,
- et al.
- Frasure-Smith N,
- Lesperance F,
- Juneau M,
- Talajic M,
- Bourassa M.G
- Mayou R.A,
- Gill D,
- Thompson D.R,
- et al.
- Frasure-Smith N,
- Lesperance F,
- Talajic M
- Packer M
- Schocken D.D,
- Arrieta M.I,
- Leaverton P.E,
- Ross E.A
- Katz S,
- Downs T.D,
- Cash H.R,
- Grotz R.C
- Rose G.A,
- Blackburn H
- Little R.J.A,
- Rubin D.B
- Watcholder S
- Mickey R.M,
- Greenland S
- Murberg T.A,
- Bru E,
- Aarsland T,
- Svebak S
- Murberg T.A,
- Bru E,
- Svebak S,
- Tveteras R,
- Aarsland T
- Freedland K.E,
- Carney R.M,
- Rich M.W,
- et al.
- Kempen G.I,
- Steverink N,
- Ormel J,
- Deeg D.J
- Krittayaphong R,
- Cascio W,
- Light K,
- et al.
- Hordsten M,
- Ericson M,
- Perski A,
- Wamala S,
- Schenk-Gustafsson K,
- Orth-Gomer K
- Huikuri H.V,
- Makikallio T.H,
- Airaksinen J,
- et al.
- Liao D,
- Cai J,
- Rosamond W.D,
- et al.