Depression increases the risk of mortality in patients with heart failure (HF). Less is known about whether depression predicts multiple readmissions or whether multiple hospitalizations worsen depression in patients with HF. This study aimed to test the hypotheses that depression predicts multiple readmissions in patients hospitalized with HF, and conversely that multiple readmissions predict persistent or worsening depression. All-cause readmissions were ascertained over a 2-year follow-up of a cohort of 400 patients hospitalized with HF. The Patient Health Questionnaire-9 was used to assess depression at index and 3-month intervals. At enrollment in the study, 21% of the patients were mildly depressed and 22% were severely depressed. Higher Patient Health Questionnaire-9 depression scores predicted a higher rate of readmissions (adjusted hazard ratio 1.02, 95% confidence interval 1.00 to 1.04, p = 0.03). The readmission rate was higher in those who were severely depressed than in those without depression (p = 0.0003), but it did not differ between patients who were mildly depressed and patients without depression. Multiple readmissions did not predict persistent or worsening depression, but younger patients in higher New York Heart Association classes were more depressed than other patients. Depression is an independent risk factor for multiple all-cause readmissions in patients hospitalized with HF. Severe depression is a treatable psychiatric co-morbidity that warrants ongoing clinical attention in patients with HF.
Most studies of hospital readmissions of patients with heart failure (HF) have focused on 30-day readmissions, as have most efforts to reduce readmissions. However, many patients are readmitted multiple times over longer intervals. Risk factors such as depression that can persist throughout HF may increase the rate of readmissions over periods that extend far beyond the first 30 days after hospitalization. There have been few prospective investigations of risk factors for multiple readmissions in patients with HF. Depression is a common co-morbidity that typically follows a chronic or recurrent course and is associated with an increased risk of morbidity and mortality in HF. It was identified as a risk factor for multiple all-cause readmissions in a cohort of 662 patients who were hospitalized with HF during the 1990s. However, changes in HF care that have occurred since then raise questions about the generalizability of these findings to current patient populations. The primary aim of this study was to determine, in a contemporary cohort, whether depression is a risk factor for multiple readmissions of patients with HF. Hospitalization-related stress may increase the risk of depression. A unique secondary aim of this study was to determine whether multiple readmissions contribute to the persistence or worsening of depression, thereby creating a vicious cycle of hospital readmissions and depression.
Methods
This prospective cohort study was approved by the Institutional Review Board of Washington University School of Medicine. All participants provided written informed consent. Patients with a clinical diagnosis of HF were screened for study eligibility while hospitalized between July 1, 2014, and December 31, 2016, at Barnes-Jewish Hospital (BJH), a university teaching hospital in St. Louis. The European Society of Cardiology criteria were used to confirm the diagnosis of HF. Eligibility required both the presence of chronic HF and the ability to cooperate with interviews, questionnaires, and follow-up assessments; consequently, patients were excluded for (1) isolated right HF or reversible HF because of valve disease with impending surgical correction; (2) dementia; (3) medical co-morbidities with a poor 1-year prognosis; (4) age <18 years or >89 years; (5) active substance abuse or alcoholism; (6) bipolar disorder or schizophrenia; and (7) refusal by the patient or attending physician. Of the 1,159 patients who met the inclusion criteria for HF, 400 patients (35%) also met all other eligibility criteria and were enrolled in the study. Figure 1 displays a CONSORT (Consolidated Standards of Reporting Trials) diagram of the screened, excluded and enrolled samples.
Data on demographic factors, medical history, co-morbidities, and medications were obtained from the BJH electronic medical record (EMR) system. Left ventricular ejection fraction values were obtained from 397 echocardiograms; 245 (62%) were performed during the index admission, and 152 (38%) were performed a median of 85 days before enrollment. New York Heart Association (NYHA) class during the 2 weeks before the index admission was estimated by reviewing the EMR and interviewing the patient.
The Patient Health Questionnaire-9 (PHQ-9) was used to assess depression at baseline, and at 3-month intervals for 24 months. The EMR was monitored to identify all readmissions to BJH or affiliated hospitals. Patients were interviewed at 6-month intervals to identify admissions to other hospitals, and medical records were obtained to document these admissions.
The sample size (n = 400) was based on assumptions of a 30% prevalence of depression, 30% attrition primarily because of mortality after discharge, and 90% power to test the primary hypothesis. Depression scores and covariates were assumed to be missing at random and were imputed sequentially. A total of 50 multiply imputed datasets with all independent variables and any auxiliary variables that were moderately correlated ( r ≥0.30) with the outcome were generated for each model. Outcome data were not imputed. Supplementary Table 1 lists the number of imputed data points per variable.
Analysis of variance and chi-square tests were used to compare patient characteristics across 3 levels defined by standard PHQ-9 score ranges at baseline (patients without depression 0 to 9, mildly depressed 10 to 14, severely depressed 15 to 27). A marginal proportional rates model was used to test the primary hypothesis that PHQ-9 depression scores at baseline and follow-up independently predict the number of all-cause readmissions over 2 years. Deaths are incorporated into this model as terminating events in the readmission process. A joint modeling strategy was used to test the secondary hypothesis that having more readmissions predicts a worse longitudinal course of depression over 2 years. Cox frailty and random coefficients models were fitted to the longitudinal depression and terminal event processes, respectively.
Factors that were associated with HF outcomes in previous studies and extractible from the EMR were included as covariates and retained regardless of statistical significance. Interaction terms were added to test whether race or antidepressant therapy moderate the relation between depression and all-cause readmissions. Statistical tests were 2-tailed with a type I error rate of 0.05. SAS version 9.3 (SAS Institute, Cary, North Carolina) was used for imputation and the proportional rates model. The R statistical package (The R Foundation for Statistical Computing, Vienna, Austria) was used to generate an event, mean cumulative function, and Kaplan-Meier plots to fit the longitudinal joint model through the R package frailtypack . The R code for the joint model is presented in the Supplementary Methods section of the Supplementary Materials, along with plots to assess model assumptions and goodness of fit ( Supplementary Figures 1 to 3 ).
Results
The demographic and baseline characteristics of the sample are presented in Table 1 . Patients with depression tended to be younger and in higher NYHA classes, were more likely to have chronic obstructive pulmonary disease (COPD) and obstructive sleep apnea, and were more likely to be taking an antidepressant compared with patients without depression.
| Characteristic | All participants (n = 400) | PHQ-9 baseline score category | P | ||
|---|---|---|---|---|---|
| 0 to 9 (n = 227) | 10 to 14 (n = 85) | 15 to 27 (n = 88) | |||
| Age (years) | 58.4 ± 13.1 | 60.0 ± 13.3 | 56.1 ± 12.7 | 56.6 ± 12.3 | .02 |
| Men | 198 (50%) | 112 (49%) | 46 (54%) | 40 (46%) | .52 |
| Minority race | 204 (51%) | 106 (47%) | 50 (59%) | 48 (55%) | .12 |
| LVEF | |||||
| % | 38.1 ± 19.1 | 37.9 ± 19.6 | 35.9 ± 18.1 | 40.6 ± 19.2 | .25 |
| <45% | 256 (64%) | 149 (66%) | 58 (68%) | 49 (56%) | .17 |
| NYHA class | |||||
| Ordinal | 2.5 ± 0.8 | 2.4 ± 0.8 | 2.6 ± 0.9 | 2.8 ± 0.8 | .0002 |
| I-II | 187 (47%) | 124 (55%) | 38 (45%) | 25 (28%) | .0001 |
| Body mass index (kg/m 2 ) | 34.3 ± 10.5 | 33.4 ± 9.4 | 35.4 ± 11.2 | 35.8 ± 12.4 | .10 |
| Coronary heart disease | 171 (43%) | 93 (41%) | 46 (54%) | 32 (36%) | .04 |
| Diabetes mellitus | 202 (51%) | 112 (49%) | 39 (46%) | 51 (58%) | .25 |
| COPD | 114 (29%) | 57 (25%) | 23 (27%) | 34 (39%) | .05 |
| OSA | 162 (41%) | 78 (34%) | 38 (45%) | 46 (52%) | .001 |
| Atrial fibrillation/flutter | 159 (40%) | 99 (44%) | 30 (35%) | 30 (34%) | .19 |
| Hypertension | 356 (89%) | 200 (88%) | 75 (88%) | 81 (92%) | .59 |
| eGFR MDRD category | |||||
| On dialysis | 26 (7%) | 15 (7%) | 3 (4%) | 8 (9%) | .19 |
| <60 mL/min/1.73 m 2 | 186 (47%) | 114 (50%) | 34 (40%) | 38 (43%) | |
| ≥60 mL/min/1.73 m 2 | 188 (47%) | 98 (43%) | 48 (57%) | 42 (48%) | |
| BUN (mg/dl) | 28.2 ± 18.4 | 28.8 ± 18.4 | 27.3 ± 20.7 | 27.3 ± 16.0 | .71 |
| Hemoglobin (non-A1c) | 12.4 ± 6.8 | 12.5 ± 7.7 | 12.9 ± 7.6 | 11.8 ± 2.2+ | .55 |
| Beta-blocker | 320 (80%) | 181 (80%) | 73 (86%) | 66 (75%) | .20 |
| ACE Inhibitor or ARB | 254 (64%) | 139 (61%) | 59 (69%) | 56 (64%) | .41 |
| Nitrate | 136 (34%) | 77 (34%) | 34 (40%) | 25 (28%) | .27 |
| MRA | 138 (35%) | 78 (34%) | 33 (39%) | 27 (31%) | .53 |
| Antidepressant | 100 (25%) | 45 (20%) | 22 (26%) | 33 (38%) | .005 |
| History of depression | 150 (39%) | 70 (31%) | 31 (37%) | 49 (56%) | .0002 |
| PHQ-9 score | 9.3 ± 6.2 | 4.8 ± 2.7 | 11.8 ± 1.3 | 18.5 ± 2.8 | <.0001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
