- Email: [email protected]
Depression Predicts Repeated Heart Failure Hospitalizations
Journal of Cardiac Failure Vol. 18 No. 3 2012
Depression Predicts Repeated Heart Failure Hospitalizations TRICIA J. JOHNSON, PhD,1 SANJIB BASU, PhD,1,2 BARBARA A. PISANI, DO,1 ELIZABETH F. AVERY, MS,1 JOSE C. MENDEZ, MD,1 JAMES E. CALVIN JR, MD,1 AND LYNDA H. POWELL, PhD1 Chicago, Illinois
ABSTRACT Objective: Management of depression, if it is independently associated with repeated hospitalizations for heart failure (HF), offers promise as a viable and cost-effective strategy to improve health outcomes and reduce health care costs for HF. The objective of this study was to assess the association between depression and the number of HF-related hospitalizations in patients with low-to-moderate systolic or diastolic dysfunction, after controlling for illness severity, socioeconomic factors, physician adherence to evidencebased medications, patient adherence to HF drug therapy, and patient adherence to salt restrictions. Methods and Results: The Heart Failure Adherence and Retention Trial (HART) was a randomized behavioral trial to evaluate whether patient self-management skills coupled with HF education improved patient outcomes. Depression was measured at baseline with the Geriatric Depression Scale (GDS). The number of hospitalizations was analyzed with a negative binomial regression model that included an offset term to account for the differential duration of follow-up for individual subjects. The average unadjusted number of hospitalizations per year was 0.40 in the depressed group (GDS $10) and 0.33 in the nondepressed group (GDS !10). Depression was a strong predictor (incident rate ratio 1.45; P 5 .006) after adjusting for physician adherence to evidence-based medication use, patient adherence to HF drug therapy, patient adherence to salt restriction, illness severity, HF severity (6-minute walk !620 feet), and socioeconomic factors. Conclusions: Depression is a strong psychosocial predictor of repeated hospitalizations for HF. Compared with nondepressed individuals, those with depression were hospitalized for HF 1.45 times more often, even after controlling for physician adherence to evidence-based medications and patient adherence to HF drug therapy and salt restrictions. This finding suggests that clinicians should screen for depression early in the course of HF management. (J Cardiac Fail 2012;18:246e252) Key Words: Hospitalizations, depression, heart failure, adherence to evidence-based medications, adherence to drug therapy, adherence to salt restrictions.
per year.1 Unnecessary and preventable hospitalizations for HF are expensive, with the average hospitalization for HF costing $9,400. The aggregate cost of HF-related hospitalizations was $10.4 billion in 2004.2 It is clear that more research is needed to identify modifiable risk factors for repeated hospitalizations that extend beyond clinical severity and adherence to drug therapy and lifestyle modifications and to understand why readmission is so common. Depression is more common in people with HF than in the general population.3 It has been linked to several adverse outcomes in this population, beginning with an increased incidence of the disease in people with coronary artery disease4 as well as poorer adherence to medical treatment5 and increased risk of mortality in people with HF.6,7 In a meta-analysis of 8 studies on depression and mortality, Rutledge et al6 found that people with HF who had symptoms of depression were more than twice as likely to die as those without depressive symptoms. There is scant
Despite tremendous progress in understanding the clinical progression of heart failure (HF) and consistent evidence that supports the effectiveness of lifestyle modifications and medications to treat the disease, rates of hospitalization for HF are high. Approximately 17% of people aged $65 years with HF are hospitalized at least once per year for an HF-related condition, and 65% of those with HF are hospitalized for a related or unrelated reason From the 1Rush University Medical Center, Chicago, Illinois and Northern Illinois University, Chicago, Illinois. Manuscript received December 3, 2009; revised manuscript received December 8, 2011; revised manuscript accepted December 13, 2011. Reprint requests: Tricia J. Johnson, PhD, Department of Health Systems Management, Rush University/Rush University Medical Center, 1700 West Van Buren Street, TOB Suite 126B, Chicago, IL 60612 USA. Tel: 312942-7107; Fax: 312-942-4957. E-mail: [email protected] See page 251 for disclosure information. 1071-9164/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.cardfail.2011.12.005 2
246
Depression and Repeated Hospitalizations
and contradictory evidence, however, about whether depression also predicts repeated hospitalizations for HF. Although several smaller studies have failed to find a relationship between depression and hospitalizations,8e12 one study found that symptoms of depression significantly increased the risk of a hospitalization or death.7 Furthermore, a large population-based study of Medicare beneficiaries found that diagnosed depression increased the likelihood of an HF hospitalization in 1 year by 11%.1 These studies are limited, however, by their failure to account for patient adherence to drug therapy and recommended HF medical treatments. Two studies that examined rehospitalization for HF included baseline but not longitudinal control for HF drug therapy.9,12 Some research has shown that depression affects an individual’s adherence to HF treatment,5 which may have an indirect effect on rehospitalization, but it is not known whether depression exerts an independent effect on hospitalization after controlling for adherence to HF drug therapy and lifestyle modifications. Identification of novel modifiable risk factors extending beyond lifestyle modifications and drug therapy may help to stem the high rate of repeated hospitalizations. Depression is one potential modifiable factor that could be diagnosed and treated early in the course of HF management and might provide a relatively low-cost intervention in the community setting. The primary objective of the present study was to examine whether depression is independently predictive of repeated HF hospitalizations (ie, total number of HF hospitalizations) after controlling for severity of illness, socioeconomic factors, patient adherence to HF drug therapy, patient adherence to salt restrictions, and physician adherence to evidence-based medications. Methods Description of HART The Heart Failure and Retention Trial (HART) was a randomized controlled trial to evaluate the benefit of patient selfmanagement skills training in combination with HF education over HF education alone. Powell et al13 provide a complete description of HART. The HART cohort included 902 participants who were randomized to a self-management intervention or education control and received 18 treatment contacts over 1 year, annual follow-ups, and 3-month phone calls to assess primary end points. The primary end points measured in HART included death, hospitalizations for HF, and all-cause hospitalizations. Subjects were eligible for HART if they had HF for at least the preceding 3 months, defined as either left ventricular ejection fraction #40% by echocardiography, radiographic ventriculography, or radionuclide ventriculograpy or diuretic therapy for at least 3 months and 1 previous hospitalization for HF. Exclusion criteria were the following: 1) patients whose 12-month prognosis was uncertain (ie, New York Heart Association (NYHA) functional class IV, likelihood of cardiac transplant over the next year, symptomatic or sustained ventricular tachycardia not controlled by therapy within the preceding 3 months, or other illnesses that limit 12month survival); 2) patients classified as NYHA functional class
Johnson et al
247
I, because they were unlikely to die or be hospitalized in the next year; 3) patients who were unlikely to undergo or benefit from the behavioral treatment (ie, presence of cognitive dysfunction or psychologic comorbidity); 4) patients whose symptoms may be eliminated by surgery; 5) patients with logistical issues, such as being enrolled in a conflicting protocol or not English speaking; 6) patients whose physicians refused access; 7) patients who indicated that they were unwilling to make lifestyle changes; and 8) patients who had unstable angina, myocardial infarction, coronary artery bypass grafting, or percutaneous transluminal coronary angiography within the preceding month. For purposes of the present analysis, we further limited the sample to include only subjects who, within the first 60 days after randomization into HART, did not die, withdraw, or become lost to follow-up. Subjects were followed for up to 1,095 days. Description of the Variables
Heart Failure-Related Hospitalizations. Ascertainment of each HF-related hospitalization was based on information from the full medical record, dated chest x-ray, electrocardiogram, serum enzymes of CK, CK-MB, and troponins, and all consultation and progress notes. HF hospitalizations were adjudicated blindly by a team of cardiologists based on the presence of either shortness of breath, peripheral edema, or chest x-ray evidence of pulmonary edema without evidence of another disease process accounting for the symptoms or signs. The hospitalizations were confirmed if the patient responded to antifailure therapy or had a documented decrease in left ventricular function. Emergency department visits that did not result in an inpatient hospital stay were not counted as an HF hospitalization. A hospitalization was defined as an inpatient hospital stay lasting more than 24 hours. A hospital admission that occurred on the same day as a previous hospital discharge was combined with the previous hospitalization. We included inpatient hospital stays with a primary diagnosis of one of the following: HF, ventricular tachycardia/ventricular fibrillation, placement of an automatic implantable cardioverterdefibrillator, atrial arrhythmia, or myocardial infarction. We used this more comprehensive definition of HF-related hospitalizations to more completely capture all hospital stays that were related to the patient’s HF. Depression. Depression was measured with the Geriatric Depression Scale (GDS-30),14 a 30-item self-administered questionnaire where each item is answered in a yes/no format. ‘‘No’’ responses are summed to calculate a total depression score. Research has demonstrated that the GDS-30 is an internally consistent, reliable, and valid measure of depression in older adults14 that was designed to be easy to administer by physicians in screening for depression. The items were summed to yield a score from 0 to 30, with a higher score indicating more depression. A cutoff point of 10 is a sensitive and specific screen for depressive symptoms.15 Our models used this cutoff point to create a dichotomous variable (depressed or not). Self-Management Skills Training. We included a dichotomous variable indicating whether the subject received self-management skills training with HF education or HF education alone. The self-management skills training included 18 2-hour group sessions, with each session addressing a different American Heart Association tip sheet (eg, sudden weight gain, salt restrictions). Each session addressed the health education tips with the use of 5 self-management skills: self-monitoring, environmental restructuring of home and work, social support,
248 Journal of Cardiac Failure Vol. 18 No. 3 March 2012 cognitive restructuring to refocus thoughts to be stress reducing, and deep breathing. The attention control group received the same 18 American Heart Association tip sheets on the same schedule as the selfmanagement skills training group sessions. The education materials were mailed to the patient’s home, and a study coordinator followed up with the subject 2e3 days after receipt to ensure that the materials had been read and to address any questions. These telephone calls only addressed information in the tip sheets. The subject was referred back to his or her provider for any questions that were not related to the tip sheets.
Physician Adherence to Evidence-Based Medications. Evidence-based medication use is a central component of effective HF management.16 Medication management of HF following evidence-based guidelines was defined using the most current class I American College of Cardiology/American Heart Association guidelines for the management of HF.16 Patients with systolic dysfunction were classified as receiving evidence-based medication if they were on a regimen of angiotensin-converting enzyme (ACE) inhibitor or angiotensin II receptor blocker (ARB), betablocker, and statin (if the patient had coronary artery disease). The algorithm was adjusted to exclude a drug if there was a contraindication. Patients with diastolic dysfunction were classified as receiving evidence-based medication if there was adherence to blood pressure control (!130/80 mm Hg), use of diuretics, beta-blockers, and aspirin (if the patient had coronary artery disease).16 Drug use was assessed by having patients bring all medications that they were currently taking to the baseline exam. Physician adherence to evidence-based medication was coded as a dichotomous variable in our analysis. Adherence to Drug Therapy. Physician adherence to evidence-based medications is only 1 element of effective drug therapy. In addition, patients must also adhere to the prescribed drug therapy. Adherence to drug therapy was measured by a Medication Event Monitoring System (MEMS) electronic pill cap container (MEMS V Trackcap; Aardex, Zug, Switzerland), which tracked a single medication, typically an ACE inhibitor, that served as a marker of overall adherence to drug therapy. The electronic pill cap only registered use if the cap was off of the bottle for $15 seconds. Incidents where the cap was off the bottle for ! 15 seconds were not counted. If the subject was not taking an ACE inhibitor, the following sequence was used to select the medication to use in the MEMS electronic pill cap: 1) ARB; 2) beta-blocker; and 3) diuretic. Adherence to drug therapy was monitored for 1 month after the baseline exam and was measured as the proportion of pills taken relative to the prescribed amount each day. Subjects could not receive credit for taking more than the prescribed amount of pills per day. We dichotomized medication adherence at the clinically significant cutoff point of 80%. Adherence to Salt Restriction. Adherence to salt restrictions is another critical component of effective HF management and facilitates lower necessary dosages of diuretics.16 Salt intake was assessed for 1 week with the use of the Stanford Heart Failure Food Intake Checklist,17 which was developed to assess dietary items that are the main sources of sodium in the American diet. Brief questionnaires to evaluate sodium intake based on food consumption have been shown to be valid and internally consistent measures.18 The checklist produces an estimated total sodium intake in mg/d. It was administered at baseline as a self-reported questionnaire that asked about the frequency of eating various food items over the preceding week. We dichotomized sodium intake as
whether the subject exceeded the clinically significant cutoff point of 2,400 milligrams of sodium per day to minimize the sensitivity of the models to extreme values. Although the recommended sodium intake is 2,000 mg/d,16 very few subjects achieved this recommendation, so we had to use a higher, but still clinically meaningful, threshold of 2,400 mg/d, as recommended by the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure.19 Severity of Illness. HF severity was measured by the distance walked in 6 minutes. The 6-minute walk test has been shown to be associated with an increased risk of hospitalization and death in other studies of individuals with HF.20e23 Because there is no clinically meaningful cutoff point for the 6-minute walk test, and distance varies by the population it is used for, we dichotomized distance at the lowest tertile: #620 feet and O620 feet. Medical comorbidity was measured by the self-reported number of comorbidities. The number of self-reported comorbidities could range from 0 to 13 and included previous myocardial infarction, high blood pressure, diabetes, cancer, stroke, renal disease, arthritis, lung disease, liver disease, asthma, sleep apnea, and Parkinson disease. Overall frailty was measured by a binary variable for whether the subject died during the study. Socioeconomic Status. Socioeconomic status was assessed by age, sex, education, and income. Education was measured with a binary variable indicating whether the individual had more than a high school education. Income was measured with a binary variable indicating whether annual household income was $$50,000. Statistical Analyses We used a series of simple negative binomial regression models with an offset for the differential follow-up duration of the individual subjects, which was equal to the natural logarithm of days in the study, to test the univariate relationship between each of the independent variables and number of hospitalizations. We used a negative binomial regression model with an offset equal to the natural logarithm of days in the study to estimate the number of HF-related hospitalizations as a function of depression, controlling for physician adherence to evidence-based medications, patient adherence to drug therapy, patient adherence to salt restriction, severity of illness, and socioeconomic status. Because 28% of the patients had missing data for either adherence to drug therapy or adherence to salt restrictions, we fit a regression model that excluded these 2 variables, and a second model with both variables included, to check the sensitivity of the results to the missing data.
Results Of the 902 subjects in HART, complete data were available for 784 subjects at baseline, with 14 subjects excluded because they died or were lost to follow-up within the first 60 days after randomization, 71 excluded because of missing 6-minute walk information, and 32 excluded because of missing data on age or income. Table 1 reports the characteristics of the 784 subjects at baseline. The sample was 53% male with an overall average age of 63.1 years (SD 13.2). The majority had an average income of !$50,000 (75%) and more than a high school education (57%). On average, the subjects had 3.2 (SD 1.7) comorbidities.
Depression and Repeated Hospitalizations Table 1. Description of Characteristics at Baseline (n 5 784) Percentage or Mean (SD) Overall Survival status Survived 81.9 Died 18.1 Self-management skills training No 50.5 Yes 49.5 Sex Female 47.5 Male 52.6 Age (y) 63.14 (13.19) Income !$50,000 75.4 $$50,000 24.6 Education High school or less 43.5 More than high school 56.5 Six-minute walk #620 ft 42.9 O620 ft 57.1 Six-minute walk 831.96 (461.66) distance (ft) No. of comorbidities 3.19 (1.71) Adherence to salt restrictiony Adherence 28.4 Nonadherence 71.6 Salt intake per day (mg) 3,698.81 (1,575.84) Physician adherence to evidence-based medications Adherence 48.3 Nonadherence 51.7 Adherence to heart failure drug therapy Adherence 66.8 Nonadherence 33.2 Depression Not depressed 71.7 Depressed 28.3 Geriatric Depression 7.45 (5.98) Scale
Mean no. of Hospitalizations per Year 0.35 (0.70) 0.21 (0.44) 1.00 (1.15) 0.32 (0.66) 0.38 (0.74) 0.39 (0.78) 0.30 (0.60) * 0.26 (0.62) 0.38 (0.72) 0.40 (0.76) 0.31 (0.65) 0.48 (0.86) 0.25 (0.53) * * 0.37 (0.74) 0.34 (0.66) * 0.32 (0.65) 0.38 (0.75) 0.31 (0.64) 0.39 (0.78) 0.33 (0.68) 0.40 (0.74) *
*Mean number of hospitalizations per year not reported for continuous variables. y n 5 680.
Adherence to a low-sodium diet was low, with 28% consuming #2,400 mg/d sodium. At baseline, 48% were on evidence-based medications, and two-thirds were adherent to HF drug therapy. Twenty-eight percent were classified as depressed (GDS $ 10). Overall, the average number of hospitalizations was 0.35 (SD 0.70) per year, with 0.40 (SD 0.74) in the depressed group and 0.33 (SD 0.68) in the nondepressed group (GDS !10). Of the 784 subjects, 496 (63%) had no hospitalizations and 29 (4%) had $4 hospitalizations (Table 2). Table 3 reports the predictors of repeated hospitalizations and the incident rate ratio (IRR) of hospitalization for each characteristic from the univariate analyses. The IRR of hospitalization was significantly higher for income !$50,000 (IRR 1.53, 95% confidence interval [CI] 1.12e2.10; P 5 .008). Both a 6-minute walk of #620 feet (IRR 1.79, 95% CI 1.38e2.32; P ! .001) and the number of comorbidities (IRR 1.09, 95% CI 1.01e1.17; P 5 .029) were significant predictors of hospitalizations in the simple
Johnson et al
249
Table 2. Distribution of the Number of Hospitalizations (n 5 784) n (%) 0 1 2 3 $4
496 158 70 31 29
(63%) (20%) (9%) (4%) (4%)
regression analyses. Physician adherence to evidence-based medications and patient adherence to drug therapy and salt restriction were insignificant. The IRR of hospitalization was significantly higher for those who were depressed compared with not depressed (IRR 1.35, 95% CI 1.01e1.79; P 5 .040). Table 4 reports the results from 3 negative binomial regression models that analyzed the association between number of hospitalizations and depression. Model 1 controlled for physician adherence to evidence-based medications, sociodemographic characteristics, severity of illness, and self-management skills training. Depression was associated with a 1.41 times greater rate of hospitalization (95% CI 1.08e1.86). Model 2 controlled for the additional factors of patient adherence to drug therapy and patient adherence to salt restrictions. In this model, depression was associated with a 1.45 times greater rate of hospitalization (95% CI 1.05e2.00). Because death censors the number of hospitalizations, model 3 considered an analysis on the subgroup of those who were surviving at the end of follow-up. This subgroup had a median follow-up of 1,095 days, and model 3 included both physician and patient adherence measures. In this model, depression was associated with an even larger rate of hospitalization (IRR 1.69; 95% CI 1.12e2.57). Discussion Depression has been shown to increase the incidence of HF, decrease adherence to prescribed medical treatment, and increase the risk of mortality.4e7 The present study provides further evidence that depression also increases the risk of costly repeated hospitalizations. Earlier studies on depression and hospitalizations for people with HF primarily examined whether subjects were hospitalized at least once or examined the time to hospitalization, rather than the number of times subjects were hospitalized during a particular time frame.1,7e10,24,25 Two studies specifically examined the number of HF hospitalizations. Sullivan et al26 found that depression increased overall health care costs for people with HF; however, the outpatient cost differences between those with and without depression were larger than the inpatient differences. In a claims-based study of Medicare beneficiaries hospitalized for HF, Sayers et al27 found that depression increased the number of hospitalizations by 0.39 per year (30%) annually. Both studies used retrospective data collected for cost or billing
250 Journal of Cardiac Failure Vol. 18 No. 3 March 2012 Table 3. Univariable Predictors of Repeated Hospitalizations (n 5 784) Incident Rate Ratio (95% CI) of Hospitalization
Variable Self-management skills training Female Age Income !$50,000 High school education or less Six-minute walk #620 ft No. of comorbidities Adherence to salt restrictionx Physician adherence to evidencebased medications Adherence to heart failure drug therapy{ Depression
1.15 0.80* 1.01 1.53z 1.29* 1.79z 1.09y 0.94 0.79*
(0.89e1.50) (0.61e1.04) (1.00e1.02) (1.12e2.10) (0.99e1.68) (1.38e2.32) (1.01e1.17) (0.43e0.32) (0.61e1.03)
1.16 (0.85e1.59) 1.35y (1.01e1.79)
*Significant at the #.10 level. y Significant at the #.05 level. z Significant at the #.01 level. x n 5 644; {n 5 679.
purposes, and were unable to account for HF severity or adherence to salt restrictions or medications. Our study extends earlier findings by controlling for adherence and HF severity. We found, in a moderately severe communitybased sample, that depression increases the number of hospitalizations by 50%, even after controlling for physician adherence to evidence-based medications and HF severity. These results are consistent with Sayers et al’s findings in a population of older patients with more severe depression, given that their definition of depression was based on the presence of diagnosis codes from claims data. Depression is common in people with HF, with the prevalence ranging from 12% in those with mild HF (NYHA functional class II) to 40% for those with the most severe HF (NYHA functional class IV).28 Cardiologists and general practitioners caring for patients with HF need a simple
tool to screen individuals for depression. Depression is often undiagnosed by clinicians, and symptoms are instead attributed to limitations in functional capacity. In a study of changes in depression for people with HF, 26% of people with minor depression and 45% of those with major depression were taking antidepressant drugs and psychotherapy was rare, with 6% of those with minor depression and 7% of those with major depression receiving psychotherapy.29 There are a number of inventories for clinicians to use in diagnosing depression, in addition to the criteria in the Diagnostic and Statistical Manual of Mental Disorders, such as the GDS, the Center for Epidemiologic Studies Depression Scale, the Beck Depression Inventory, the Hamilton Depression Rating Scale, and the Patient Health Questionnaire 2 (PHQ-2).30,31 Short measures, such as the PHQ-2, may be particularly useful in the clinical setting for depression screening, because the questions can easily be integrated into a routine visit. The use of the GDS may have misclassified some people regarding the presence or absence of depression. In a recent meta-analysis of the use of the GDS in primary care, Mitchell et al32 found that the 30-question GDS had a sensitivity of 77.4% and specificity of 65.4%, indicating a moderate fraction correct of 71.2%. The 30-question GDS was better than primary care physicians without the use of other diagnostic aids in correctly identifying patients with depression (sensitivity 56.3%), but it did not perform as well in correctly identifying those without depression (specificity 73.6%). Our results should, therefore, be viewed as conservative estimates of the impact of depression on the number of hospitalizations, because we may have misclassified some depressed people as not having depression. More work is needed to explore the underlying reasons for the lack of significant relationship between the number of hospitalizations and physician adherence to evidencebased medications, patient adherence to drug therapy, and
Table 4. Multivariable Predictors of the Number of Heart FailureeRelated Hospitalizations (Full Results) Incident Rate Ratio (95% CI) of Hospitalization Variable Depression Physician adherence to evidence-based medications Adherence to salt restriction Adherence to HF drug therapy Self-management skills training Female Age Income !$50,000 High school education or less 6-minute walk !620 ft No. of comorbidities Death
Model 1 (n 5 784) y
Model 2 (n 5 564)
Model 3 (Survivors Only) (n 5 458)
1.41 (1.08e1.86) 0.83 (0.64e1.08)
1.45 (1.05e2.00) 0.90 (0.67e1.22)
1.69y (1.12e2.57) 0.90 (0.61e1.32)
1.29y 0.74y 1.00 1.32* 1.04 1.59z 0.99 4.39z
1.01 0.93 1.22 0.73y 0.99 1.60y 1.06 1.35* 1.00 4.31z
1.03 0.96 1.28 0.68* 0.99 1.54* 1.08 1.46* 1.04
(1.01e1.65) (0.57e0.95) (0.99e1.01) (0.96e1.81) (0.81e1.35) (1.22e2.08) (0.92e1.07) (3.27e5.89)
y
(0.70e1.47) (0.68e1.29) (0.91e1.62) (0.54e0.98) (0.98e1.01) (1.09e2.34) (0.78e1.44) (0.99e1.84) (0.92e1.09) (3.08e6.04)
(0.63e1.70) (0.63e1.46) (0.89e1.86) (0.46e1.01) (0.98e1.01) (0.95e2.50) (0.73e1.58) (0.97e2.20) (0.93e1.17)
All models control for sex, age, income !$50,000, high school education or less, 6-minute walk #620 ft, and number of comorbidities. Models 1 and 2 also control for death. *Significant at the #.10 level. y Significant at the #.05 level. z Significant at the #.01 level.
Depression and Repeated Hospitalizations
patient adherence to salt restrictions. One plausible explanation is that that dichotomizing these variables resulted in a loss of information that explained variation in the number of hospitalizations. We tested whether continuous measures of these variables were significant in the multivariable models and found no relationship. Our measure of adherence to salt restrictions was based on patient self-report using a food frequency questionnaire, which is less detailed than a food diary or food recall. Our food frequency questionnaire asked about frequency of consumption during the past week but did not ask about quantities consumed, which may have resulted in some measurement error. Although food frequency questionnaires do not include the same level of detail about the quantity or types of food, they are both less costly to administer and process and less burdensome for subjects to complete than dietary recalls and food records, which likely results in higher response rates.33 Because of the extensive battery of questionnaires administered at baseline, it was not feasible to administer a more comprehensive dietary assessment measure. In addition, the electronic pill caps counted medication use when the pill caps were opened; they could not differentiate between pills removed from the bottle and taken from instances when the pill bottle was opened and pills were not removed or the pills were not taken by the individual. There is no evidence, however, that pills are frequently removed from the bottle and not taken by the individual. In addition, the pill caps registered use only when the pill cap was removed for $15 seconds and counted pill use only up to the maximum dosage per day and for the bottle. The present results should be interpreted with caution when generalizing to people with HF. Our sample included people with NYHA functional class II and III HF, purposefully excluding those with the least and most severe HF from the study. Furthermore, people with other psychologic comorbidities and people who indicated that they were unwilling to make lifestyle changes were excluded. Our results extend the findings of other studies that have examined depression and HF hospitalizations. In a nationally representative study of Medicare beneficiaries hospitalized for HF, Sayers et al27 found that depression increased hospitalizations by 30% (by 0.39 hospitalizations per year) and hospital charges by 25%. Our results, from a sample of younger and healthier people which did not rely upon diagnosis-based information to classify depression, were similar, with depression increasing the number of HF hospitalizations by 45% (0.15 hospitalizations per year). Further work is needed to understand how best to treat depression in the presence of HF, because antidepressive drugs have some known cardiovascular side effects.30 Seven percent of our sample reported antidepressive drug use at baseline, with similar proportions classified as depressed and not depressed (3.7% and 3.1%, respectively). However, we did not have information about the length of time that depressive symptoms were present nor duration of antidepressant use before the baseline data collection, and because of this, antidepressant use was not examined in this study. Nevertheless, our results suggest that it is
Johnson et al
251
critical to diagnose major depression in people with HF, owing to its impact on health care utilization as well as quality of life, despite an uncertainty whether depression treatment decreases the time to remission of depressive symptoms or the risk of mortality.29,34 Reducing the number of avoidable hospitalizations related to HF has the potential to reduce expenditures for Medicare, Medicaid, and private health insurers. Depression substantially increases the average health care costs for people with HF. Sullivan et al26 found that people with HF who were on antidepressant drugs had 29% higher health care costs than those who were not taking antidepressant drugs, with increases in both inpatient and outpatient medical care driving these differences. In our sample of 784 individuals, 28% were classified as depressed. If early treatment could eliminate the depressive symptoms, 35 hospitalizations per year could potentially have been avoided in our sample. Disclosures None.
References 1. Braunstein JB, Anderson GF, Gerstenblith G, et al. Noncardiac comorbidity increases preventable hospitalizations and mortality among Medicare beneficiaries with chronic heart failure. J Am Coll Cardiol 2003;42:1226e33. 2. Russo CA, Ho K, Elixhauser A. Hospital stays for circulatory diseases, 2004. H-CUP statistical brief #26. Rockville, MD: Agency for Healthcare Research and Quality; 2007. 3. MacMahon KMA, Lip GYH. Psychological factors in heart failure: A review of the literature. Arch Intern Med 2002;162:509e16. 4. May HT, Horne BD, Carlquist JF, Sheng X, Joy E, Catinella AP. Depression after coronary heart disease is associated with heart failure. J Am Coll Cardiol 2009;53:1440e7. 5. DiMatteo MR, Lepper HS, Croghan TW. Depression is a risk factor for noncompliance with medical treatment. Arch Intern Med 2000; 160:2101e7. 6. Rutledge T, Reis VA, Linke SE, Greenberg BH, Mills PJ. Depression in heart failure: a meta-analytic review of prevalence, intervention effects, and associations with clinical outcomes. J Am Coll Cardiol 2006;48:1527e37. 7. Sherwood A, Blumenthal JA, Trivdei R, et al. Relationship of depression to death or hospitalization in patients with heart failure. Arch Intern Med 2007;167:367e73. 8. Krumholz HM, Butler J, Miller J, et al. Prognostic importance of emotional support for elderly patients hospitalized with heart failure. Circulation 1998;97:958e64. 9. Jiang W, Alexander J, Christopher E, et al. Relationship of depression to increased risk of mortality and rehospitalization in patients with congestive heart failure. Arch Intern Med 2001;161:1849e56. 10. Koenig HG. Depression in hospitalized older patients with congestive heart failure. General Hospital Psychiatry 1998;20:29e43. 11. Macchia A, Monte S, Pellegrini F, et al. Depression worsens outcomes in elderly patients with heart failure: an analysis of 48,117 patients in a community setting. Eur J Heart Fail 2008;10:714e21. 12. Rodriguez-Artalejo F, Guallar-Castillon P, Herrera MC, et al. Social network as a predictor of hospital readmission and mortality among older patients with heart failure. J Card Fail 2006;12:621e7.
252 Journal of Cardiac Failure Vol. 18 No. 3 March 2012 13. Powell LH, Calvin JE, Mendes de Leon CF, et al. The Heart Failure Adherence and Retention Trial (HART): design and rationale. Am Heart J 2008;156:452e60. 14. Yesavage J, Brink T, Rose T, et al. Development and validation of a geriatric depression screening scale: a preliminary report. J Psychiatric Res 1983;17:37e49. 15. Lyness JM, Noel TK, Cox C, King DA, Conwell Y, Caine ED. Screening for depression in elderly primary care patients: a comparison of the Center for Epidemiologic StudieseDepression Scale and the Geriatric Depression Scale. Arch Intern Med 1997;157: 449e54. 16. Hunt S. ACC/AHA 2005 guideline update for the diagnosis and management of chronic heart failure in the adult: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Update the 2001 Guidelines for the Evaluation and Management of Heart Failure). J Am Coll Cardiol 2005;46:E1e82. 17. DeBusk R. A new approach to risk factor modification. Cardiol Clin 1996;14:143e57. 18. Charlton KE, Steyn K, Levitt NS, Jonathan D, Zulu JV, Nel JH. Development and validation of a short questionnaire to assess sodium intake. Public Health Nutr 2008;11:83e94. 19. Chobanian AV, Bakris GL, Black HR, et al. Seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Hypertension 2003;42:1206e52. 20. Reibis RK, Treszl A, Wegscheider K, Ehrlich B, Dissmann R, V€ oller H. Exercise capacity is the most powerful predictor of 2-year mortality in patients with left ventricular systolic dysfunction. Herz 2010;35:104e10. 21. Shah MR, Hasselblad V, Gheorghiade M, et al. Prognostic usefulness of the six-minute walk in patients with advanced congestive heart failure secondary to ischemic or nonischemic cardiomyopathy. Am J Cardiol 2001;88:987e93. 22. Boxer R, Kleppinger A, Ahmad A, Annis K, Hager D, Kenny A. The 6-minute walk is associated with frailty and predicts mortality in older adults with heart failure. Congest Heart Fail 2010;16:208e13.
23. Bittner V, Weiner DH, Yusuf S, et al. Prediction of mortality and morbidity with a 6-minute walk test in patients with left ventricular dysfunction. JAMA 1993;270:1702e7. 24. Kato N, Kinugawa K, Yao A, Hatano M, Shiga T, Kazuma K. Relationship of depressive symptoms with hospitalization and death in Japanese patients with heart failure. J Card Fail 2009;15:912e9. 25. Lesman-Leegte I, van Veldhuisen DJ, Hillege HL, Moser D, Sanderman R, Jaarsma T. Depressive symptoms and outcomes in patients with heart failure: data from the COACH Study. Eur J Heart Fail 2009;11:1202e7. 26. Sullivan M, Simon G, Spertus J, Russo J. Depression-related costs in heart failure care. Arch Intern Med 2002;162:1860e6. 27. Sayers SL, Hanrahan N, Kutney A, Clarke SP, Reis BF, Riegel B. Psychiatric comorbidity and greater hospitalization risk, longer length of stay, and higher hospitalization costs in older adults with heart failure. J Am Geriatr Soc 2007;55:1585e91. 28. Freedland K, Rich M, Skala J, Carney R, Davila-Roman V, Jaffe A. Prevalence of depression in hospitalised patients with congestive heart failure. Psychosom Med 2003;65:119e28. 29. Koenig HG. Depressive outcome in inpatients with congestive heart failure. Arch Intern Med 2006;166:991e6. 30. Tousoulis D, Antonopoulos AS, Charalambos A, Chrysa Saldari, et al. Role of depression in heart failuredchoosing the right antidepressive treatment. Int J Cardiol 2010;140:12e8. 31. Kroenke K, Spitzer RL, Williams JBW. The Patient Health Questionnaire-2: validity of a two-item depression screener. Med Care 2003;41:1284e92. 32. Mitchell AJ, Bird V, Rizzo M, Meader N. Diagnostic validity and added value of the Geriatric Depression Scale for depression in primary care: a meta-analysis of GDS(30) and GDS(15). JAffect Disord 2010;125:10e7. 33. Thompson FE, Subar AF. Dietary assessment methodology. In: Coulston AM, Boushey CJ, editors. Nutrition in the prevention and treatment of disease. 2nd ed. Amsterdam: Academic Press; 2008. p. 3e39. 34. O’Connor CM, Jiang W, Kuchibhatla M, et al. Antidepressant use, depression and survival in patients with heart failure. Arch Intern Med 2008;168:2232e7.
Depression Predicts Repeated Heart Failure Hospitalizations TRICIA J. JOHNSON, PhD,1 SANJIB BASU, PhD,1,2 BARBARA A. PISANI, DO,1 ELIZABETH F. AVERY, MS,1 JOSE C. MENDEZ, MD,1 JAMES E. CALVIN JR, MD,1 AND LYNDA H. POWELL, PhD1 Chicago, Illinois
ABSTRACT Objective: Management of depression, if it is independently associated with repeated hospitalizations for heart failure (HF), offers promise as a viable and cost-effective strategy to improve health outcomes and reduce health care costs for HF. The objective of this study was to assess the association between depression and the number of HF-related hospitalizations in patients with low-to-moderate systolic or diastolic dysfunction, after controlling for illness severity, socioeconomic factors, physician adherence to evidencebased medications, patient adherence to HF drug therapy, and patient adherence to salt restrictions. Methods and Results: The Heart Failure Adherence and Retention Trial (HART) was a randomized behavioral trial to evaluate whether patient self-management skills coupled with HF education improved patient outcomes. Depression was measured at baseline with the Geriatric Depression Scale (GDS). The number of hospitalizations was analyzed with a negative binomial regression model that included an offset term to account for the differential duration of follow-up for individual subjects. The average unadjusted number of hospitalizations per year was 0.40 in the depressed group (GDS $10) and 0.33 in the nondepressed group (GDS !10). Depression was a strong predictor (incident rate ratio 1.45; P 5 .006) after adjusting for physician adherence to evidence-based medication use, patient adherence to HF drug therapy, patient adherence to salt restriction, illness severity, HF severity (6-minute walk !620 feet), and socioeconomic factors. Conclusions: Depression is a strong psychosocial predictor of repeated hospitalizations for HF. Compared with nondepressed individuals, those with depression were hospitalized for HF 1.45 times more often, even after controlling for physician adherence to evidence-based medications and patient adherence to HF drug therapy and salt restrictions. This finding suggests that clinicians should screen for depression early in the course of HF management. (J Cardiac Fail 2012;18:246e252) Key Words: Hospitalizations, depression, heart failure, adherence to evidence-based medications, adherence to drug therapy, adherence to salt restrictions.
per year.1 Unnecessary and preventable hospitalizations for HF are expensive, with the average hospitalization for HF costing $9,400. The aggregate cost of HF-related hospitalizations was $10.4 billion in 2004.2 It is clear that more research is needed to identify modifiable risk factors for repeated hospitalizations that extend beyond clinical severity and adherence to drug therapy and lifestyle modifications and to understand why readmission is so common. Depression is more common in people with HF than in the general population.3 It has been linked to several adverse outcomes in this population, beginning with an increased incidence of the disease in people with coronary artery disease4 as well as poorer adherence to medical treatment5 and increased risk of mortality in people with HF.6,7 In a meta-analysis of 8 studies on depression and mortality, Rutledge et al6 found that people with HF who had symptoms of depression were more than twice as likely to die as those without depressive symptoms. There is scant
Despite tremendous progress in understanding the clinical progression of heart failure (HF) and consistent evidence that supports the effectiveness of lifestyle modifications and medications to treat the disease, rates of hospitalization for HF are high. Approximately 17% of people aged $65 years with HF are hospitalized at least once per year for an HF-related condition, and 65% of those with HF are hospitalized for a related or unrelated reason From the 1Rush University Medical Center, Chicago, Illinois and Northern Illinois University, Chicago, Illinois. Manuscript received December 3, 2009; revised manuscript received December 8, 2011; revised manuscript accepted December 13, 2011. Reprint requests: Tricia J. Johnson, PhD, Department of Health Systems Management, Rush University/Rush University Medical Center, 1700 West Van Buren Street, TOB Suite 126B, Chicago, IL 60612 USA. Tel: 312942-7107; Fax: 312-942-4957. E-mail: [email protected] See page 251 for disclosure information. 1071-9164/$ - see front matter Ó 2012 Elsevier Inc. All rights reserved. doi:10.1016/j.cardfail.2011.12.005 2
246
Depression and Repeated Hospitalizations
and contradictory evidence, however, about whether depression also predicts repeated hospitalizations for HF. Although several smaller studies have failed to find a relationship between depression and hospitalizations,8e12 one study found that symptoms of depression significantly increased the risk of a hospitalization or death.7 Furthermore, a large population-based study of Medicare beneficiaries found that diagnosed depression increased the likelihood of an HF hospitalization in 1 year by 11%.1 These studies are limited, however, by their failure to account for patient adherence to drug therapy and recommended HF medical treatments. Two studies that examined rehospitalization for HF included baseline but not longitudinal control for HF drug therapy.9,12 Some research has shown that depression affects an individual’s adherence to HF treatment,5 which may have an indirect effect on rehospitalization, but it is not known whether depression exerts an independent effect on hospitalization after controlling for adherence to HF drug therapy and lifestyle modifications. Identification of novel modifiable risk factors extending beyond lifestyle modifications and drug therapy may help to stem the high rate of repeated hospitalizations. Depression is one potential modifiable factor that could be diagnosed and treated early in the course of HF management and might provide a relatively low-cost intervention in the community setting. The primary objective of the present study was to examine whether depression is independently predictive of repeated HF hospitalizations (ie, total number of HF hospitalizations) after controlling for severity of illness, socioeconomic factors, patient adherence to HF drug therapy, patient adherence to salt restrictions, and physician adherence to evidence-based medications. Methods Description of HART The Heart Failure and Retention Trial (HART) was a randomized controlled trial to evaluate the benefit of patient selfmanagement skills training in combination with HF education over HF education alone. Powell et al13 provide a complete description of HART. The HART cohort included 902 participants who were randomized to a self-management intervention or education control and received 18 treatment contacts over 1 year, annual follow-ups, and 3-month phone calls to assess primary end points. The primary end points measured in HART included death, hospitalizations for HF, and all-cause hospitalizations. Subjects were eligible for HART if they had HF for at least the preceding 3 months, defined as either left ventricular ejection fraction #40% by echocardiography, radiographic ventriculography, or radionuclide ventriculograpy or diuretic therapy for at least 3 months and 1 previous hospitalization for HF. Exclusion criteria were the following: 1) patients whose 12-month prognosis was uncertain (ie, New York Heart Association (NYHA) functional class IV, likelihood of cardiac transplant over the next year, symptomatic or sustained ventricular tachycardia not controlled by therapy within the preceding 3 months, or other illnesses that limit 12month survival); 2) patients classified as NYHA functional class
Johnson et al
247
I, because they were unlikely to die or be hospitalized in the next year; 3) patients who were unlikely to undergo or benefit from the behavioral treatment (ie, presence of cognitive dysfunction or psychologic comorbidity); 4) patients whose symptoms may be eliminated by surgery; 5) patients with logistical issues, such as being enrolled in a conflicting protocol or not English speaking; 6) patients whose physicians refused access; 7) patients who indicated that they were unwilling to make lifestyle changes; and 8) patients who had unstable angina, myocardial infarction, coronary artery bypass grafting, or percutaneous transluminal coronary angiography within the preceding month. For purposes of the present analysis, we further limited the sample to include only subjects who, within the first 60 days after randomization into HART, did not die, withdraw, or become lost to follow-up. Subjects were followed for up to 1,095 days. Description of the Variables
Heart Failure-Related Hospitalizations. Ascertainment of each HF-related hospitalization was based on information from the full medical record, dated chest x-ray, electrocardiogram, serum enzymes of CK, CK-MB, and troponins, and all consultation and progress notes. HF hospitalizations were adjudicated blindly by a team of cardiologists based on the presence of either shortness of breath, peripheral edema, or chest x-ray evidence of pulmonary edema without evidence of another disease process accounting for the symptoms or signs. The hospitalizations were confirmed if the patient responded to antifailure therapy or had a documented decrease in left ventricular function. Emergency department visits that did not result in an inpatient hospital stay were not counted as an HF hospitalization. A hospitalization was defined as an inpatient hospital stay lasting more than 24 hours. A hospital admission that occurred on the same day as a previous hospital discharge was combined with the previous hospitalization. We included inpatient hospital stays with a primary diagnosis of one of the following: HF, ventricular tachycardia/ventricular fibrillation, placement of an automatic implantable cardioverterdefibrillator, atrial arrhythmia, or myocardial infarction. We used this more comprehensive definition of HF-related hospitalizations to more completely capture all hospital stays that were related to the patient’s HF. Depression. Depression was measured with the Geriatric Depression Scale (GDS-30),14 a 30-item self-administered questionnaire where each item is answered in a yes/no format. ‘‘No’’ responses are summed to calculate a total depression score. Research has demonstrated that the GDS-30 is an internally consistent, reliable, and valid measure of depression in older adults14 that was designed to be easy to administer by physicians in screening for depression. The items were summed to yield a score from 0 to 30, with a higher score indicating more depression. A cutoff point of 10 is a sensitive and specific screen for depressive symptoms.15 Our models used this cutoff point to create a dichotomous variable (depressed or not). Self-Management Skills Training. We included a dichotomous variable indicating whether the subject received self-management skills training with HF education or HF education alone. The self-management skills training included 18 2-hour group sessions, with each session addressing a different American Heart Association tip sheet (eg, sudden weight gain, salt restrictions). Each session addressed the health education tips with the use of 5 self-management skills: self-monitoring, environmental restructuring of home and work, social support,
248 Journal of Cardiac Failure Vol. 18 No. 3 March 2012 cognitive restructuring to refocus thoughts to be stress reducing, and deep breathing. The attention control group received the same 18 American Heart Association tip sheets on the same schedule as the selfmanagement skills training group sessions. The education materials were mailed to the patient’s home, and a study coordinator followed up with the subject 2e3 days after receipt to ensure that the materials had been read and to address any questions. These telephone calls only addressed information in the tip sheets. The subject was referred back to his or her provider for any questions that were not related to the tip sheets.
Physician Adherence to Evidence-Based Medications. Evidence-based medication use is a central component of effective HF management.16 Medication management of HF following evidence-based guidelines was defined using the most current class I American College of Cardiology/American Heart Association guidelines for the management of HF.16 Patients with systolic dysfunction were classified as receiving evidence-based medication if they were on a regimen of angiotensin-converting enzyme (ACE) inhibitor or angiotensin II receptor blocker (ARB), betablocker, and statin (if the patient had coronary artery disease). The algorithm was adjusted to exclude a drug if there was a contraindication. Patients with diastolic dysfunction were classified as receiving evidence-based medication if there was adherence to blood pressure control (!130/80 mm Hg), use of diuretics, beta-blockers, and aspirin (if the patient had coronary artery disease).16 Drug use was assessed by having patients bring all medications that they were currently taking to the baseline exam. Physician adherence to evidence-based medication was coded as a dichotomous variable in our analysis. Adherence to Drug Therapy. Physician adherence to evidence-based medications is only 1 element of effective drug therapy. In addition, patients must also adhere to the prescribed drug therapy. Adherence to drug therapy was measured by a Medication Event Monitoring System (MEMS) electronic pill cap container (MEMS V Trackcap; Aardex, Zug, Switzerland), which tracked a single medication, typically an ACE inhibitor, that served as a marker of overall adherence to drug therapy. The electronic pill cap only registered use if the cap was off of the bottle for $15 seconds. Incidents where the cap was off the bottle for ! 15 seconds were not counted. If the subject was not taking an ACE inhibitor, the following sequence was used to select the medication to use in the MEMS electronic pill cap: 1) ARB; 2) beta-blocker; and 3) diuretic. Adherence to drug therapy was monitored for 1 month after the baseline exam and was measured as the proportion of pills taken relative to the prescribed amount each day. Subjects could not receive credit for taking more than the prescribed amount of pills per day. We dichotomized medication adherence at the clinically significant cutoff point of 80%. Adherence to Salt Restriction. Adherence to salt restrictions is another critical component of effective HF management and facilitates lower necessary dosages of diuretics.16 Salt intake was assessed for 1 week with the use of the Stanford Heart Failure Food Intake Checklist,17 which was developed to assess dietary items that are the main sources of sodium in the American diet. Brief questionnaires to evaluate sodium intake based on food consumption have been shown to be valid and internally consistent measures.18 The checklist produces an estimated total sodium intake in mg/d. It was administered at baseline as a self-reported questionnaire that asked about the frequency of eating various food items over the preceding week. We dichotomized sodium intake as
whether the subject exceeded the clinically significant cutoff point of 2,400 milligrams of sodium per day to minimize the sensitivity of the models to extreme values. Although the recommended sodium intake is 2,000 mg/d,16 very few subjects achieved this recommendation, so we had to use a higher, but still clinically meaningful, threshold of 2,400 mg/d, as recommended by the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure.19 Severity of Illness. HF severity was measured by the distance walked in 6 minutes. The 6-minute walk test has been shown to be associated with an increased risk of hospitalization and death in other studies of individuals with HF.20e23 Because there is no clinically meaningful cutoff point for the 6-minute walk test, and distance varies by the population it is used for, we dichotomized distance at the lowest tertile: #620 feet and O620 feet. Medical comorbidity was measured by the self-reported number of comorbidities. The number of self-reported comorbidities could range from 0 to 13 and included previous myocardial infarction, high blood pressure, diabetes, cancer, stroke, renal disease, arthritis, lung disease, liver disease, asthma, sleep apnea, and Parkinson disease. Overall frailty was measured by a binary variable for whether the subject died during the study. Socioeconomic Status. Socioeconomic status was assessed by age, sex, education, and income. Education was measured with a binary variable indicating whether the individual had more than a high school education. Income was measured with a binary variable indicating whether annual household income was $$50,000. Statistical Analyses We used a series of simple negative binomial regression models with an offset for the differential follow-up duration of the individual subjects, which was equal to the natural logarithm of days in the study, to test the univariate relationship between each of the independent variables and number of hospitalizations. We used a negative binomial regression model with an offset equal to the natural logarithm of days in the study to estimate the number of HF-related hospitalizations as a function of depression, controlling for physician adherence to evidence-based medications, patient adherence to drug therapy, patient adherence to salt restriction, severity of illness, and socioeconomic status. Because 28% of the patients had missing data for either adherence to drug therapy or adherence to salt restrictions, we fit a regression model that excluded these 2 variables, and a second model with both variables included, to check the sensitivity of the results to the missing data.
Results Of the 902 subjects in HART, complete data were available for 784 subjects at baseline, with 14 subjects excluded because they died or were lost to follow-up within the first 60 days after randomization, 71 excluded because of missing 6-minute walk information, and 32 excluded because of missing data on age or income. Table 1 reports the characteristics of the 784 subjects at baseline. The sample was 53% male with an overall average age of 63.1 years (SD 13.2). The majority had an average income of !$50,000 (75%) and more than a high school education (57%). On average, the subjects had 3.2 (SD 1.7) comorbidities.
Depression and Repeated Hospitalizations Table 1. Description of Characteristics at Baseline (n 5 784) Percentage or Mean (SD) Overall Survival status Survived 81.9 Died 18.1 Self-management skills training No 50.5 Yes 49.5 Sex Female 47.5 Male 52.6 Age (y) 63.14 (13.19) Income !$50,000 75.4 $$50,000 24.6 Education High school or less 43.5 More than high school 56.5 Six-minute walk #620 ft 42.9 O620 ft 57.1 Six-minute walk 831.96 (461.66) distance (ft) No. of comorbidities 3.19 (1.71) Adherence to salt restrictiony Adherence 28.4 Nonadherence 71.6 Salt intake per day (mg) 3,698.81 (1,575.84) Physician adherence to evidence-based medications Adherence 48.3 Nonadherence 51.7 Adherence to heart failure drug therapy Adherence 66.8 Nonadherence 33.2 Depression Not depressed 71.7 Depressed 28.3 Geriatric Depression 7.45 (5.98) Scale
Mean no. of Hospitalizations per Year 0.35 (0.70) 0.21 (0.44) 1.00 (1.15) 0.32 (0.66) 0.38 (0.74) 0.39 (0.78) 0.30 (0.60) * 0.26 (0.62) 0.38 (0.72) 0.40 (0.76) 0.31 (0.65) 0.48 (0.86) 0.25 (0.53) * * 0.37 (0.74) 0.34 (0.66) * 0.32 (0.65) 0.38 (0.75) 0.31 (0.64) 0.39 (0.78) 0.33 (0.68) 0.40 (0.74) *
*Mean number of hospitalizations per year not reported for continuous variables. y n 5 680.
Adherence to a low-sodium diet was low, with 28% consuming #2,400 mg/d sodium. At baseline, 48% were on evidence-based medications, and two-thirds were adherent to HF drug therapy. Twenty-eight percent were classified as depressed (GDS $ 10). Overall, the average number of hospitalizations was 0.35 (SD 0.70) per year, with 0.40 (SD 0.74) in the depressed group and 0.33 (SD 0.68) in the nondepressed group (GDS !10). Of the 784 subjects, 496 (63%) had no hospitalizations and 29 (4%) had $4 hospitalizations (Table 2). Table 3 reports the predictors of repeated hospitalizations and the incident rate ratio (IRR) of hospitalization for each characteristic from the univariate analyses. The IRR of hospitalization was significantly higher for income !$50,000 (IRR 1.53, 95% confidence interval [CI] 1.12e2.10; P 5 .008). Both a 6-minute walk of #620 feet (IRR 1.79, 95% CI 1.38e2.32; P ! .001) and the number of comorbidities (IRR 1.09, 95% CI 1.01e1.17; P 5 .029) were significant predictors of hospitalizations in the simple
Johnson et al
249
Table 2. Distribution of the Number of Hospitalizations (n 5 784) n (%) 0 1 2 3 $4
496 158 70 31 29
(63%) (20%) (9%) (4%) (4%)
regression analyses. Physician adherence to evidence-based medications and patient adherence to drug therapy and salt restriction were insignificant. The IRR of hospitalization was significantly higher for those who were depressed compared with not depressed (IRR 1.35, 95% CI 1.01e1.79; P 5 .040). Table 4 reports the results from 3 negative binomial regression models that analyzed the association between number of hospitalizations and depression. Model 1 controlled for physician adherence to evidence-based medications, sociodemographic characteristics, severity of illness, and self-management skills training. Depression was associated with a 1.41 times greater rate of hospitalization (95% CI 1.08e1.86). Model 2 controlled for the additional factors of patient adherence to drug therapy and patient adherence to salt restrictions. In this model, depression was associated with a 1.45 times greater rate of hospitalization (95% CI 1.05e2.00). Because death censors the number of hospitalizations, model 3 considered an analysis on the subgroup of those who were surviving at the end of follow-up. This subgroup had a median follow-up of 1,095 days, and model 3 included both physician and patient adherence measures. In this model, depression was associated with an even larger rate of hospitalization (IRR 1.69; 95% CI 1.12e2.57). Discussion Depression has been shown to increase the incidence of HF, decrease adherence to prescribed medical treatment, and increase the risk of mortality.4e7 The present study provides further evidence that depression also increases the risk of costly repeated hospitalizations. Earlier studies on depression and hospitalizations for people with HF primarily examined whether subjects were hospitalized at least once or examined the time to hospitalization, rather than the number of times subjects were hospitalized during a particular time frame.1,7e10,24,25 Two studies specifically examined the number of HF hospitalizations. Sullivan et al26 found that depression increased overall health care costs for people with HF; however, the outpatient cost differences between those with and without depression were larger than the inpatient differences. In a claims-based study of Medicare beneficiaries hospitalized for HF, Sayers et al27 found that depression increased the number of hospitalizations by 0.39 per year (30%) annually. Both studies used retrospective data collected for cost or billing
250 Journal of Cardiac Failure Vol. 18 No. 3 March 2012 Table 3. Univariable Predictors of Repeated Hospitalizations (n 5 784) Incident Rate Ratio (95% CI) of Hospitalization
Variable Self-management skills training Female Age Income !$50,000 High school education or less Six-minute walk #620 ft No. of comorbidities Adherence to salt restrictionx Physician adherence to evidencebased medications Adherence to heart failure drug therapy{ Depression
1.15 0.80* 1.01 1.53z 1.29* 1.79z 1.09y 0.94 0.79*
(0.89e1.50) (0.61e1.04) (1.00e1.02) (1.12e2.10) (0.99e1.68) (1.38e2.32) (1.01e1.17) (0.43e0.32) (0.61e1.03)
1.16 (0.85e1.59) 1.35y (1.01e1.79)
*Significant at the #.10 level. y Significant at the #.05 level. z Significant at the #.01 level. x n 5 644; {n 5 679.
purposes, and were unable to account for HF severity or adherence to salt restrictions or medications. Our study extends earlier findings by controlling for adherence and HF severity. We found, in a moderately severe communitybased sample, that depression increases the number of hospitalizations by 50%, even after controlling for physician adherence to evidence-based medications and HF severity. These results are consistent with Sayers et al’s findings in a population of older patients with more severe depression, given that their definition of depression was based on the presence of diagnosis codes from claims data. Depression is common in people with HF, with the prevalence ranging from 12% in those with mild HF (NYHA functional class II) to 40% for those with the most severe HF (NYHA functional class IV).28 Cardiologists and general practitioners caring for patients with HF need a simple
tool to screen individuals for depression. Depression is often undiagnosed by clinicians, and symptoms are instead attributed to limitations in functional capacity. In a study of changes in depression for people with HF, 26% of people with minor depression and 45% of those with major depression were taking antidepressant drugs and psychotherapy was rare, with 6% of those with minor depression and 7% of those with major depression receiving psychotherapy.29 There are a number of inventories for clinicians to use in diagnosing depression, in addition to the criteria in the Diagnostic and Statistical Manual of Mental Disorders, such as the GDS, the Center for Epidemiologic Studies Depression Scale, the Beck Depression Inventory, the Hamilton Depression Rating Scale, and the Patient Health Questionnaire 2 (PHQ-2).30,31 Short measures, such as the PHQ-2, may be particularly useful in the clinical setting for depression screening, because the questions can easily be integrated into a routine visit. The use of the GDS may have misclassified some people regarding the presence or absence of depression. In a recent meta-analysis of the use of the GDS in primary care, Mitchell et al32 found that the 30-question GDS had a sensitivity of 77.4% and specificity of 65.4%, indicating a moderate fraction correct of 71.2%. The 30-question GDS was better than primary care physicians without the use of other diagnostic aids in correctly identifying patients with depression (sensitivity 56.3%), but it did not perform as well in correctly identifying those without depression (specificity 73.6%). Our results should, therefore, be viewed as conservative estimates of the impact of depression on the number of hospitalizations, because we may have misclassified some depressed people as not having depression. More work is needed to explore the underlying reasons for the lack of significant relationship between the number of hospitalizations and physician adherence to evidencebased medications, patient adherence to drug therapy, and
Table 4. Multivariable Predictors of the Number of Heart FailureeRelated Hospitalizations (Full Results) Incident Rate Ratio (95% CI) of Hospitalization Variable Depression Physician adherence to evidence-based medications Adherence to salt restriction Adherence to HF drug therapy Self-management skills training Female Age Income !$50,000 High school education or less 6-minute walk !620 ft No. of comorbidities Death
Model 1 (n 5 784) y
Model 2 (n 5 564)
Model 3 (Survivors Only) (n 5 458)
1.41 (1.08e1.86) 0.83 (0.64e1.08)
1.45 (1.05e2.00) 0.90 (0.67e1.22)
1.69y (1.12e2.57) 0.90 (0.61e1.32)
1.29y 0.74y 1.00 1.32* 1.04 1.59z 0.99 4.39z
1.01 0.93 1.22 0.73y 0.99 1.60y 1.06 1.35* 1.00 4.31z
1.03 0.96 1.28 0.68* 0.99 1.54* 1.08 1.46* 1.04
(1.01e1.65) (0.57e0.95) (0.99e1.01) (0.96e1.81) (0.81e1.35) (1.22e2.08) (0.92e1.07) (3.27e5.89)
y
(0.70e1.47) (0.68e1.29) (0.91e1.62) (0.54e0.98) (0.98e1.01) (1.09e2.34) (0.78e1.44) (0.99e1.84) (0.92e1.09) (3.08e6.04)
(0.63e1.70) (0.63e1.46) (0.89e1.86) (0.46e1.01) (0.98e1.01) (0.95e2.50) (0.73e1.58) (0.97e2.20) (0.93e1.17)
All models control for sex, age, income !$50,000, high school education or less, 6-minute walk #620 ft, and number of comorbidities. Models 1 and 2 also control for death. *Significant at the #.10 level. y Significant at the #.05 level. z Significant at the #.01 level.
Depression and Repeated Hospitalizations
patient adherence to salt restrictions. One plausible explanation is that that dichotomizing these variables resulted in a loss of information that explained variation in the number of hospitalizations. We tested whether continuous measures of these variables were significant in the multivariable models and found no relationship. Our measure of adherence to salt restrictions was based on patient self-report using a food frequency questionnaire, which is less detailed than a food diary or food recall. Our food frequency questionnaire asked about frequency of consumption during the past week but did not ask about quantities consumed, which may have resulted in some measurement error. Although food frequency questionnaires do not include the same level of detail about the quantity or types of food, they are both less costly to administer and process and less burdensome for subjects to complete than dietary recalls and food records, which likely results in higher response rates.33 Because of the extensive battery of questionnaires administered at baseline, it was not feasible to administer a more comprehensive dietary assessment measure. In addition, the electronic pill caps counted medication use when the pill caps were opened; they could not differentiate between pills removed from the bottle and taken from instances when the pill bottle was opened and pills were not removed or the pills were not taken by the individual. There is no evidence, however, that pills are frequently removed from the bottle and not taken by the individual. In addition, the pill caps registered use only when the pill cap was removed for $15 seconds and counted pill use only up to the maximum dosage per day and for the bottle. The present results should be interpreted with caution when generalizing to people with HF. Our sample included people with NYHA functional class II and III HF, purposefully excluding those with the least and most severe HF from the study. Furthermore, people with other psychologic comorbidities and people who indicated that they were unwilling to make lifestyle changes were excluded. Our results extend the findings of other studies that have examined depression and HF hospitalizations. In a nationally representative study of Medicare beneficiaries hospitalized for HF, Sayers et al27 found that depression increased hospitalizations by 30% (by 0.39 hospitalizations per year) and hospital charges by 25%. Our results, from a sample of younger and healthier people which did not rely upon diagnosis-based information to classify depression, were similar, with depression increasing the number of HF hospitalizations by 45% (0.15 hospitalizations per year). Further work is needed to understand how best to treat depression in the presence of HF, because antidepressive drugs have some known cardiovascular side effects.30 Seven percent of our sample reported antidepressive drug use at baseline, with similar proportions classified as depressed and not depressed (3.7% and 3.1%, respectively). However, we did not have information about the length of time that depressive symptoms were present nor duration of antidepressant use before the baseline data collection, and because of this, antidepressant use was not examined in this study. Nevertheless, our results suggest that it is
Johnson et al
251
critical to diagnose major depression in people with HF, owing to its impact on health care utilization as well as quality of life, despite an uncertainty whether depression treatment decreases the time to remission of depressive symptoms or the risk of mortality.29,34 Reducing the number of avoidable hospitalizations related to HF has the potential to reduce expenditures for Medicare, Medicaid, and private health insurers. Depression substantially increases the average health care costs for people with HF. Sullivan et al26 found that people with HF who were on antidepressant drugs had 29% higher health care costs than those who were not taking antidepressant drugs, with increases in both inpatient and outpatient medical care driving these differences. In our sample of 784 individuals, 28% were classified as depressed. If early treatment could eliminate the depressive symptoms, 35 hospitalizations per year could potentially have been avoided in our sample. Disclosures None.
References 1. Braunstein JB, Anderson GF, Gerstenblith G, et al. Noncardiac comorbidity increases preventable hospitalizations and mortality among Medicare beneficiaries with chronic heart failure. J Am Coll Cardiol 2003;42:1226e33. 2. Russo CA, Ho K, Elixhauser A. Hospital stays for circulatory diseases, 2004. H-CUP statistical brief #26. Rockville, MD: Agency for Healthcare Research and Quality; 2007. 3. MacMahon KMA, Lip GYH. Psychological factors in heart failure: A review of the literature. Arch Intern Med 2002;162:509e16. 4. May HT, Horne BD, Carlquist JF, Sheng X, Joy E, Catinella AP. Depression after coronary heart disease is associated with heart failure. J Am Coll Cardiol 2009;53:1440e7. 5. DiMatteo MR, Lepper HS, Croghan TW. Depression is a risk factor for noncompliance with medical treatment. Arch Intern Med 2000; 160:2101e7. 6. Rutledge T, Reis VA, Linke SE, Greenberg BH, Mills PJ. Depression in heart failure: a meta-analytic review of prevalence, intervention effects, and associations with clinical outcomes. J Am Coll Cardiol 2006;48:1527e37. 7. Sherwood A, Blumenthal JA, Trivdei R, et al. Relationship of depression to death or hospitalization in patients with heart failure. Arch Intern Med 2007;167:367e73. 8. Krumholz HM, Butler J, Miller J, et al. Prognostic importance of emotional support for elderly patients hospitalized with heart failure. Circulation 1998;97:958e64. 9. Jiang W, Alexander J, Christopher E, et al. Relationship of depression to increased risk of mortality and rehospitalization in patients with congestive heart failure. Arch Intern Med 2001;161:1849e56. 10. Koenig HG. Depression in hospitalized older patients with congestive heart failure. General Hospital Psychiatry 1998;20:29e43. 11. Macchia A, Monte S, Pellegrini F, et al. Depression worsens outcomes in elderly patients with heart failure: an analysis of 48,117 patients in a community setting. Eur J Heart Fail 2008;10:714e21. 12. Rodriguez-Artalejo F, Guallar-Castillon P, Herrera MC, et al. Social network as a predictor of hospital readmission and mortality among older patients with heart failure. J Card Fail 2006;12:621e7.
252 Journal of Cardiac Failure Vol. 18 No. 3 March 2012 13. Powell LH, Calvin JE, Mendes de Leon CF, et al. The Heart Failure Adherence and Retention Trial (HART): design and rationale. Am Heart J 2008;156:452e60. 14. Yesavage J, Brink T, Rose T, et al. Development and validation of a geriatric depression screening scale: a preliminary report. J Psychiatric Res 1983;17:37e49. 15. Lyness JM, Noel TK, Cox C, King DA, Conwell Y, Caine ED. Screening for depression in elderly primary care patients: a comparison of the Center for Epidemiologic StudieseDepression Scale and the Geriatric Depression Scale. Arch Intern Med 1997;157: 449e54. 16. Hunt S. ACC/AHA 2005 guideline update for the diagnosis and management of chronic heart failure in the adult: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Update the 2001 Guidelines for the Evaluation and Management of Heart Failure). J Am Coll Cardiol 2005;46:E1e82. 17. DeBusk R. A new approach to risk factor modification. Cardiol Clin 1996;14:143e57. 18. Charlton KE, Steyn K, Levitt NS, Jonathan D, Zulu JV, Nel JH. Development and validation of a short questionnaire to assess sodium intake. Public Health Nutr 2008;11:83e94. 19. Chobanian AV, Bakris GL, Black HR, et al. Seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure. Hypertension 2003;42:1206e52. 20. Reibis RK, Treszl A, Wegscheider K, Ehrlich B, Dissmann R, V€ oller H. Exercise capacity is the most powerful predictor of 2-year mortality in patients with left ventricular systolic dysfunction. Herz 2010;35:104e10. 21. Shah MR, Hasselblad V, Gheorghiade M, et al. Prognostic usefulness of the six-minute walk in patients with advanced congestive heart failure secondary to ischemic or nonischemic cardiomyopathy. Am J Cardiol 2001;88:987e93. 22. Boxer R, Kleppinger A, Ahmad A, Annis K, Hager D, Kenny A. The 6-minute walk is associated with frailty and predicts mortality in older adults with heart failure. Congest Heart Fail 2010;16:208e13.
23. Bittner V, Weiner DH, Yusuf S, et al. Prediction of mortality and morbidity with a 6-minute walk test in patients with left ventricular dysfunction. JAMA 1993;270:1702e7. 24. Kato N, Kinugawa K, Yao A, Hatano M, Shiga T, Kazuma K. Relationship of depressive symptoms with hospitalization and death in Japanese patients with heart failure. J Card Fail 2009;15:912e9. 25. Lesman-Leegte I, van Veldhuisen DJ, Hillege HL, Moser D, Sanderman R, Jaarsma T. Depressive symptoms and outcomes in patients with heart failure: data from the COACH Study. Eur J Heart Fail 2009;11:1202e7. 26. Sullivan M, Simon G, Spertus J, Russo J. Depression-related costs in heart failure care. Arch Intern Med 2002;162:1860e6. 27. Sayers SL, Hanrahan N, Kutney A, Clarke SP, Reis BF, Riegel B. Psychiatric comorbidity and greater hospitalization risk, longer length of stay, and higher hospitalization costs in older adults with heart failure. J Am Geriatr Soc 2007;55:1585e91. 28. Freedland K, Rich M, Skala J, Carney R, Davila-Roman V, Jaffe A. Prevalence of depression in hospitalised patients with congestive heart failure. Psychosom Med 2003;65:119e28. 29. Koenig HG. Depressive outcome in inpatients with congestive heart failure. Arch Intern Med 2006;166:991e6. 30. Tousoulis D, Antonopoulos AS, Charalambos A, Chrysa Saldari, et al. Role of depression in heart failuredchoosing the right antidepressive treatment. Int J Cardiol 2010;140:12e8. 31. Kroenke K, Spitzer RL, Williams JBW. The Patient Health Questionnaire-2: validity of a two-item depression screener. Med Care 2003;41:1284e92. 32. Mitchell AJ, Bird V, Rizzo M, Meader N. Diagnostic validity and added value of the Geriatric Depression Scale for depression in primary care: a meta-analysis of GDS(30) and GDS(15). JAffect Disord 2010;125:10e7. 33. Thompson FE, Subar AF. Dietary assessment methodology. In: Coulston AM, Boushey CJ, editors. Nutrition in the prevention and treatment of disease. 2nd ed. Amsterdam: Academic Press; 2008. p. 3e39. 34. O’Connor CM, Jiang W, Kuchibhatla M, et al. Antidepressant use, depression and survival in patients with heart failure. Arch Intern Med 2008;168:2232e7.