Author + information
- John S. Rumsfeld, MD, PhD, FACC*,* ( )(, )
- Edward Havranek, MD, FACC†,
- Frederick A. Masoudi, MD, MSPH, FACC†,
- Eric D. Peterson, MD, MPH, FACC‡,
- Philip Jones, MS§,
- Joseph F. Tooley, PharmD∥,
- Harlan M. Krumholz, MD, FACC¶,
- John A. Spertus, MD, MPH, FACC§,
- Cardiovascular Outcomes Research Consortium (CORC)
- ↵*Reprint requests and correspondence:
Dr. John S. Rumsfeld, Cardiology (111B), Denver VA Medical Center, 1055 Clermont Street, Denver, Colorado 80220, USA.
Objectives The purpose of this study was to assess whether depressive symptoms are independently associated with changes in heart failure (HF)-specific health status.
Background Depression is common in patients with HF, but the impact of depressive symptoms on the health status of these patients over time is unknown.
Methods We conducted a multicenter prospective cohort study of outpatients with HF. Data from 460 patients who completed a baseline Medical Outcomes Study-Depression Questionnaire and both a baseline and follow-up (6 ± 2 weeks) Kansas City Cardiomyopathy Questionnaire (KCCQ) were analyzed. The KCCQ measures HF-specific health status, including symptoms, physical and social function, and quality of life. Multivariable regression was used to evaluate depressive symptoms as a predictor of change in KCCQ scores, adjusting for baseline KCCQ scores and other patient variables. The primary outcome was change in KCCQ summary scores (range 0 to 100; higher scores indicate better health status; 5 points is a clinically meaningful change).
Results Approximately 30% (139/460) of the patients had significant depressive symptoms at baseline. Depressed patients had markedly lower baseline KCCQ summary scores (β = −19.6; p < 0.001). After adjustment for potential confounders, depressed patients were at risk for significant worsening of their HF symptoms, physical and social function, and quality of life (average change in KCCQ summary score = −7.1 points; p < 0.001). Depressive symptoms were the strongest predictor of decline in health status in the multivariable models.
Conclusions Depressive symptoms are a strong predictor of short-term worsening of HF-specific health status. The recognition and treatment of depression may be an important component of HF care.
Nearly five million Americans have heart failure (HF) and the prevalence continues to increase, leading some to conclude that there is an “epidemic of heart failure” in the U.S. (1,2). Patients with HF have impaired health status, including severe symptom burden (e.g., dyspnea), marked functional limitation, and diminished health-related quality of life (3–5). Optimizing the health status of patients with HF is an important goal of therapy, but little is known about the specific factors influencing the health status of these patients over time (6).
Depressive symptoms are common in patients with HF (5,7,8)and may be an important determinant of health status. However, the relationship between depression and the health status of patients with HF independent of traditional markers of disease severity and other patient characteristics has not been studied. If depression is a major predictor of HF-related health status, then improved recognition and treatment of depression may enhance outcomes for these patients.
The objectives of this study were to: 1) determine the frequency of depressive symptoms in a multicenter cohort of outpatients with HF; 2) evaluate the cross-sectional association between depressive symptoms and baseline HF-specific health status; and 3) evaluate whether depressive symptoms independently predict short-term changes in HF symptoms, functional status, and quality of life.
Patients were recruited through the Cardiovascular Outcomes Research Consortium, a collaborative of investigators from 14 North American centers (Appendix). Outpatients with the diagnosis of HF in their outpatient medical record or a hospitalization for decompensated HF within three years were screened for participation. Patients with documented left ventricular systolic dysfunction (left ventricular ejection fraction <0.40) and who were >30 years of age were asked to participate and to provide informed consent. There were no additional explicit exclusion criteria. Institutional review board approval was obtained at each of the study sites.
After enrollment, patients underwent a complete history and physical examination and a 6-min walk test (9). Study participants also completed baseline surveys, including the Kansas City Cardiomyopathy Questionnaire (KCCQ) and the Medical Outcomes Study-Depression (MOS-D) questionnaire (10,11). After 6 ± 2 weeks, patients returned to complete a follow-up KCCQ along with repeat physical examination, 6-min walk testing, and interim history. Of 546 patients enrolled, 460 (84.2%) completed both the baseline and follow-up measures and form the cohort for this study. Patients participating in follow-up were more likely to be female (25.3% vs. 13.2%; p = 0.03) and were less likely to have a history of angina (23.6% vs. 39.7%; p < 0.01) than those who did not participate in follow-up. There were no other significant differences in baseline characteristics between participants and non-participants.
The primary independent, or predictor, variable for this study was depressive symptoms as measured by the baseline MOS-D. This instrument is a validated, eight-item questionnaire for the screening of depressive symptoms (11–13). A score of ≥0.06 on the MOS-D is approximately 85% sensitive and 75% specific for major depression, which is comparable to other depression case finding instruments (12,13). This scale was chosen over other depressive symptom questionnaires because of minimal overlap in the somatic symptom questions with heart failure symptoms. Other independent variables included all of the demographic, cardiac, and comorbid variables listed in Table 1.
The primary dependent, or outcome, variable for this study was disease-specific health status, as measured by the KCCQ. The KCCQ is a 23-item, self-administered questionnaire with scales measuring the symptoms, physical function, social function, self-efficacy, and quality of life of patients with HF (10). These domains can be summarized in a single summary score reflecting overall health status for a given patient with HF. The scales are scored from 0 to 100 where higher scores represent fewer symptoms, better function, and/or better quality of life. The validity, reproducibility, and sensitivity of this measure to clinical changes have previously been established (10). A change of 5 points on KCCQ scale scores, either as a group mean or as an intra-individual change, is considered clinically important. The secondary outcome measure for this study was 6-min walk distance. Hospitalization for HF was also evaluated but was not an a priori outcome measure for this study.
Baseline characteristics of patients with substantial depressive symptoms (MOS-D score ≥0.06) were compared with those patients without substantial depressive symptoms (MOS-D score <0.06) using ttests for continuous variables and chi-square tests for categorical variables. Then, ttests were used to compare baseline KCCQ scores and 6-min walk distance between patients with and without depressive symptoms. Multiple linear regression was then used to define the independent, cross-sectional association between baseline depressive symptoms and baseline KCCQ summary scores, adjusting for the demographic, cardiac, comorbid, and HF treatment variables listed in Table 1. Depressive symptoms were dichotomized as present (MOS-D score ≥0.06) versus absent (MOS-D score <0.06) in the primary analyses. In secondary analyses, MOS-D scores were analyzed as a continuous variable (log-transformed), which did not significantly alter the results.
To assess changes in health status, one-way analysis of variance was used to compare changes in KCCQ scale scores (stratified by baseline KCCQ scores) and changes in 6-min walk distance (stratified by baseline 6-min walk quartiles) between patients with and without depressive symptoms. Then, a series of multiple linear regression models were developed to assess the independent relationship between depressive symptoms and change from baseline to follow-up in each of the outcome variables (change in KCCQ scores, change in 6-min walk distance), adjusting for the demographic, cardiac, comorbid, and heart failure treatment variables listed in Table 1. Models were built by applying forward stepwise selection to the variables in Table 1(p < 0.10 to enter the model; p < 0.05 to stay in the final model) and adding the depressive symptom variable to these models as a final step. Unselected regression models were also developed including all of the Table 1variables as well as depression to maximally control for confounding. Because the results of the selected and unselected models were similar, only the parameter estimates from the unselected models are presented. Finally, multiple logistic regression was used to model a decline of ≥5 points in KCCQ summary scores (i.e., a clinically important decline) versus no change or improvement in score between baseline and follow-up. Analyses of change in health status were stratified by baseline KCCQ scores or 6-min walk quartiles to account for baseline differences in health status between depressed and non-depressed patients. This approach also minimizes the chance that observed differences in health status changes were confounded by regression to the mean (14). Statistical analyses were performed using SAS version 8.0 (SAS Institute, Cary, North Carolina).
Of the 460 patients in the study, 139 (30.2%) had substantial depressive symptoms at baseline. Among patients classified as depressed, the median MOS-D score was 0.388 with a range of 0.060 to 0.930. Among the patients classified as non-depressed, the median MOS-D score was 0.002 with a range of 0.001 to 0.059. Depressed patients were younger, more likely to smoke, more likely to have a history of alcohol/substance abuse, and more likely to have had previous depression requiring treatment compared with non-depressed patients (Table 1). However, the two groups of patients were similar with regard to most baseline demographic, cardiac, and non-cardiac factors. For example, in both groups the mean ejection fraction was approximately 0.25, about one-half the patients had an ischemic etiology of their HF, and about one-third of the patients had diabetes.
There were no HF treatment differences between depressed and non-depressed patients, and a high proportion of patients were being treated with beta-blockers and angiotensin-converting enzyme inhibitor or angiotensin receptor blocker agents in both groups. Among patients with substantial depressive symptoms, 15.8% (22/139) were on anti-depressant medical therapy.
Baseline health status
At baseline, depressed patients had significantly lower KCCQ summary, symptom, physical function, social function, and quality of life scores than non-depressed patients (Table 2). In the multivariable model adjusting for potential confounders, depressive symptoms were the strongest correlate of lower baseline KCCQ summary scores (β = −19.6 ± standard error 2.2; p < 0.001).
Hospitalization rates during the follow-up period were 10.1% (14/139) for the depressed patients and 11.0% (36/327) for the non-depressed patients (p = 0.8).
Changes in health status
Figures 1 to 3⇓⇓⇓demonstrate the unadjusted relationship between baseline depressive symptoms and changes in KCCQ summary scores, symptom scores, and 6-min walk distance, respectively. Similar relationships were found for the associations between baseline depressive symptoms and changes in KCCQ physical limitation, social limitation, and quality of life scales (not shown). In each case, depressed patients with fair to good baseline health status (i.e., upper two quartiles) were much more likely to suffer significant declines in health status over the follow-up period compared with non-depressed patients (p < 0.001). Furthermore, depressed patients with poor baseline health status (i.e., lower two quartiles) had significantly less improvement over the follow-up period compared with non-depressed patients.
In multiple linear regression analyses, depressive symptoms were significantly associated with declines in HF-related health status (Table 3). Depressed patients were at risk for statistically and clinically significant declines (i.e., ≥5 points) on the KCCQ summary, symptom, physical function, social function, and quality of life scales. Depressive symptoms were also independently associated with declines in 6-min walk distance. In the multivariable models of change in KCCQ summary scores, symptom scores, social function, and quality of life, depressive symptoms had the largest magnitude of association with the outcome.
In multiple logistic regression analysis, depressive symptoms were predictive of clinically significant (≥5 points) declines in KCCQ summary scores (odds ratio [OR] 2.5, 95% confidence interval [CI] 1.5 to 4.2). Other significant predictors of decline in KCCQ scores included a history of alcohol/substance abuse (OR 2.4, 95% CI 1.4 to 4.3), diabetes (OR 1.7, 95% CI 1.1 to 2.7), older age (OR 1.2, 95% CI 1.0 to 1.5, per 10-year increment), and baseline KCCQ scores (OR 1.2, 95% CI 1.1 to 1.3, per 10-point increment).
The primary goal of this study was to evaluate the strength of association between depressive symptoms and six-week changes in HF-related health status. The study population was composed of outpatients with HF, and there were no differences in HF treatment between depressed and non-depressed patients at baseline. Nevertheless, depressive symptoms were common in this population, and depressed patients had markedly worse baseline health status than non-depressed patients. After adjusting for baseline health status and over 20 patient variables, depressed patients were at risk for significant worsening of their HF symptoms, physical and social functional status, and quality of life. In fact, depressive symptoms were the strongest predictor of decline in health status in the multivariable models. Therefore, HF outpatients with depression are not only at risk for worse baseline health status, but also for short-term declines in their HF symptoms, function, and quality of life.
To our knowledge, this is the first study to demonstrate the impact of depressive symptoms on HF-specific health status over time. The results of this study expand the scope of previous studies that demonstrated an independent association between depressive symptoms and adverse outcomes in patients with HF, including increased mortality, higher rates of hospitalization, and declines in activities in daily living (5,15,16). Although this study did not find a significant difference in hospitalization rates between depressed and non-depressed patients, it was not powered for this outcome. Nonetheless, we were able to demonstrate meaningful differences in health status outcomes between depressed and non-depressed patients. In fact, a focus on mortality and hospitalization would have obscured the burden of depressive symptoms on these HF patients.
Furthermore, a focus on health status outcomes is consistent with the Institute of Medicine's promotion of patient-centered care as one of the six key strategies to improving the quality of care in the U.S. (17). Factors such as the aging population, increasing prevalence of HF, and evidence that many older persons prefer quality of life to quantity of life further underscore the importance of evaluating patient-centered outcomes in this population (1,2,18). The results of this study suggest that patient-centered outcomes for HF outpatients may be improved through better recognition and treatment of depression.
To date, efforts to improve the quality of care for patients with HF have focused on processes of care specific to HF (e.g., angiotensin-converting enzyme inhibitors for left ventricular systolic dysfunction) (19), largely ignoring the impact of prevalent comorbid conditions. In our study, nearly one-third of the patients in this study had significant depressive symptoms, and yet, less than one-fifth of the depressed patients were receiving anti-depressant medication. These findings are consistent with previous studies reporting that depression is common in HF populations (5,7,8)and that the majority of depressed patients with HF are not being treated for depression (20). Because depression is common, is associated with adverse outcomes, and appears to be under-recognized and under-treated, it deserves increased attention from clinicians and investigators.
Previous studies in post-myocardial infarction populations have not shown that depression screening and treatment reduces mortality (21,22). However, these trials did show that depression detection and treatment strategies are effective in alleviating depression and demonstrated that there are safe and effective treatments available for depression in patients with cardiovascular disease, including selective serotonin reuptake inhibitors. Unfortunately, these studies did not focus on health status outcomes. Furthermore, the efficacy of depression screening and treatment has not been evaluated in HF populations. One may anticipate more transient, or situational, depression in post-myocardial infarction patients compared with HF patients, who are faced with a chronic condition characterized by exacerbations and clinical decline and, therefore, may remain depressed for prolonged periods of time (20,23). Future studies are warranted to investigate the impact of improved depression management in patients with HF on a broad range of outcomes. In the meantime, however, there are inherent benefits to the recognition and treatment of depression.
An important barrier to the incorporation of depression management into the care of patients with HF may be that non-psychiatrist clinicians do not feel capable of identifying and treating depression. However, it has been shown that significant depressive symptoms can be identified in primary care settings using two simple screening questions and that the vast majority of patients with depressive symptoms can be treated effectively without psychiatric referral (12,24). The recognition and management of depression may be enhanced through the use of multidisciplinary teams or disease management programs that have been successfully applied to other aspects of HF care (25–28).
The reasons for the high prevalence of depressive symptoms and the strong association between depressive symptoms and adverse outcomes in patients with HF are unknown. Depression may be a risk factor for the development of HF (29,30). Conversely, chronic illnesses such as HF may precipitate depression. In either case, there are both behavioral and pathophysiologic mechanisms by which depressive symptoms may lead to adverse outcomes in patients with HF. For example, depressed patients are at risk for medical noncompliance, and there is evidence that depression is associated with high sympathetic tone, hypercortisolemia, increased platelet aggregability, and elevated inflammatory markers (31–34).
Several potential limitations of this study should be addressed. First, this study enrolled a convenience sample of outpatients with HF and, therefore, may have limited generalizability. However, one may expect fewer depressed patients to participate in such a study, and therefore our sample may underestimate the true prevalence of depressive symptoms in outpatients with HF. Second, the MOS-D measures depressive symptoms and is not a diagnostic tool for major depressive disorder as defined by the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition. However, the diagnostic accuracy of the MOS-D for major depression is relatively high and is comparable to other depression screening instruments (12,13). Furthermore, multiple studies in cardiac populations have demonstrated the prognostic importance of depressive symptoms, irrespective of a diagnosis of major depression (34–36). Third, we only assessed patients at two points in time within a relatively short interval. However, we were able to demonstrate a strong association between depressive symptoms and declines in health status even within this time period. Future studies should examine the relationship between depressive symptoms and longer-term outcomes in outpatients with HF.
In conclusion, depressive symptoms are common in patients with HF, are associated with markedly worse baseline health status, and are a strong predictor of worsening of HF symptoms, functional status, and quality of life over time. While additional studies should explicitly evaluate whether the improved recognition and treatment of depression will improve outcomes, clinicians should be aware of the strong link between depressive symptoms and adverse health status outcomes in patients with HF. Appendix.
☆ Funding for this study was provided by Pharmacia Corporation, Project Grant 4501096072. Dr. Rumsfeld is supported by a VA Health Services Research Advanced Career Development Award (RCD 98341-2). Dr. Masoudi is supported by an NIA Career Development Award (KO8-AG01011).
- confidence interval
- heart failure
- Kansas City Cardiomyopathy Questionnaire
- Medical Outcomes Study-Depression Questionnaire
- odds ratio
- American College of Cardiology Foundation
- ↵American Heart Association. Heart Disease and Stroke Statistics—2003 Update. Dallas, TX: American Heart Association, 2002
- McCullough P.A.,
- Philbin E.F.,
- Spertus J.A.,
- Kaatz S.,
- Sandberg K.R.,
- Weaver W.D.
- Jeunger J.,
- Schellberg D.,
- Kraemer S.,
- et al.
- Steptoe A.,
- Mohabir A.,
- Mahon N.G.,
- McKenna W.J.
- Vaccarino V.,
- Kasl S.V.,
- Abramson J.,
- Krumholz H.M.
- Guyatt G.H.,
- Sullivan M.J.,
- Thompson P.J.,
- et al.
- Green C.P.,
- Porter C.B.,
- Bresnahan D.R.,
- Spertus J.A.
- Gehlbach S.H.
- Murberg T.A.,
- Bru E.,
- Svebak S.,
- Tveteras R.,
- Aarsland T.
- Institute of Medicine, Committee on Quality Health Care in America
- Kasper E.K.,
- Gerstenblith G.,
- Hefter G.,
- et al.
- Williams S.A.,
- Kasl S.V.,
- Heiat A.,
- Abramson J.L.,
- Krumholz H.M.,
- Vaccarino V.
- Rozanski A.,
- Blumenthal J.A.,
- Kaplan J.