Abstract
Objective:
There is little evidence that antidepressants are efficacious for depression in patients with heart failure (HF), and equivocal evidence that they are safe. This study identified characteristics that are associated with antidepressant use in hospitalized patients with HF.
Method:
Logistic regression models were used to identify independent correlates of antidepressant use in 400 patients hospitalized with HF between 2014 and 2016. The measure of depression in the primary analysis was a DSM-5 diagnosis based on a structured interview; this was replaced by a PHQ-9 depression score in a secondary analysis.
Results:
In the primary analysis, there were positive associations between antidepressant use and white race, younger age, unemployment, non-ischemic HF, number of other prescribed medications, current minor depression, history of major depression, and functional impairment. In the secondary analysis, there were positive associations with white race, unemployment, number of other prescribed medications, and functional impairment; the effect of current severity of depression differed between patients with vs. without a history of major depression.
Conclusions:
Current depression is only one of several factors that influence the use of antidepressant medications in patients with HF. Further research is needed to ensure that these agents are being used appropriately in this patient population.
Keywords: Antidepressive agents, Depression, Depressive Disorder, Drug Utilization, Heart Failure
1. Introduction
Antidepressant use has increased substantially over the past two decades [1], including in patients with heart failure (HF). For example, only 13% of a cohort of 662 patients hospitalized with HF between 1994 and 1999 were on an antidepressant [2], compared to 29% of patients with HF enrolled in the CASA trial between 2012 and 2015 [3].
This increase has occurred despite studies that have raised questions about the benefits and safety of antidepressants for patients with HF. The two largest randomized controlled trials (RCTs) of antidepressants for patients with HF and comorbid major depression found that sertraline [4] and escitalopram [5] are safe but not superior to placebo with respect to depression outcomes.
The evidence regarding the effects of antidepressants on medical outcomes in HF is equivocal. Sertraline had no effect on cardiovascular events in the SADHART-CHF trial [4] and escitalopram had no effect on all-cause mortality or hospitalization in the MOOD-HF trial [5]. Antidepressant use also had no effect on event-free survival in an observational study of 209 outpatients with HF [6], hospitalization-free survival in 147 outpatients with HF [7], or mortality in 1,017 outpatients with HF [8], although fluoxetine use predicted mortality in the latter study. In 1,006 patients with HF with reduced ejection fraction (HFrEF), antidepressant use predicted mortality in an unadjusted analysis but not when depression was added to the model [9].
Observational studies have also examined whether antidepressants may have adverse effects in patients with HF. In 121,252 patients with HF in the Danish National Health Registry, antidepressant use predicted all-cause and cardiovascular mortality regardless of depression status [10]. Tricyclic antidepressants (TCAs) and selective serotonin reuptake inhibitors (SSRIs) predicted all-cause and cardiovascular death in a separate study of 99,335 Danish National Health Service patients with HF [11], although uncontrolled confounding by indication (depression) may help to explain these findings. Antidepressant use also predicted mortality in 3,346 patients in the Danish Heart Failure Clinics Network registry [12], but uncontrolled confounding by indication raises questions about these findings as well. The limitations of these studies have left us with ambiguous evidence about the safety of antidepressants for comorbid depression in HF. On the other hand, there is little evidence to suggest that antidepressant use improves survival or other medical outcomes in patients with HF.
Depression is a prevalent comorbidity in HF [13–15]. It is associated with poor quality of life [16] and is an independent predictor of rehospitalization [2, 17, 18] and mortality [9, 17–24]. Antidepressant medications are still the first-line treatment despite the lack of evidence that they are efficacious for comorbid depression in HF and the equivocal evidence for their safety. This is consistent with recent evidence that some medications that may be contraindicated, including certain antidepressants, are often continued or started following hospitalization for HF [25].
Given the uncertainty surrounding the risks of antidepressants for patients with HF, research is needed on factors that are associated with their use of these medications. Unfortunately, the existing studies have yielded inconsistent results. In 121,252 Danish National Health Service patients with HF, antidepressant use was independently associated with female gender; older age; higher socioeconomic status; more comorbidities; higher use of statins, spironolactone, and aspirin; lower use of beta-blockers and ACE-inhibitors; more severe HF; and a clinical diagnosis of depression [10]. In 209 outpatients in the American HF Quality of Life registry, antidepressant use was associated with female gender and depression but not with other patient characteristics [6]. In a Spanish study of 1,017 outpatients with HF, depressive symptoms, female gender, higher New York Heart Association (NYHA) class, higher body mass index (BMI), and worse functional impairment was associated with antidepressant use [8]. In 1,006 patients in the Duke University Medical Center HF registry, those who were prescribed antidepressants were more likely to be white and married than other patients; other patient characteristics were not associated with antidepressant use [9]. In 864 patients (21% of whom had HF) evaluated at the Diagnostic Cardiac Catheterization Laboratory at Duke University, antidepressant use was significantly higher in whites (21%) than in African Americans (12%), despite no differences in the prevalence or severity of depression [26]. Finally, in the Danish HF Clinics Network study, female gender and higher NYHA class were the only variables associated with antidepressant use [12].
The most consistent finding across these studies is that in patients with HF, women are more likely than men to use antidepressants, even after taking depression into account. NYHA class was associated with antidepressant use in two of the studies, but none of the other associations have been replicated. Also, none of the studies were based on DSM-5 [27] diagnoses derived from structured interviews; depression status was based on medical records or screening questionnaires.
Because of the inconsistencies among these studies, further research is needed on factors that may influence clinician’s decisions to prescribe antidepressant medications for patients with HF. Thus, the purpose of the present study was to identify factors that are independently associated with antidepressant use in patients hospitalized with HF. We examined most of the same variables that have been reported in previous studies to determine which ones could be replicated, along with additional candidates. Unlike the previous studies, we also examined antidepressant use in relation to the presence or absence of DSM-5 depression diagnoses based on structured interviews.
2. Methods
2.1. Ethics
This report utilizes data from a study of the relationship of depression and other psychosocial factors to hospitalization and mortality in patients with HF. The study was approved by the Institutional Review Board of Washington University School of Medicine. Written informed consent was obtained from all participants before enrollment.
2.2. Participants
Between July 1, 2014 and December 31, 2016, a total of 1,440 patients with a clinical diagnosis of HF were screened for study eligibility while hospitalized at Barnes-Jewish Hospital alarge, urban, tertiary care teaching hospital at Washington University Medical Center in St. Louiswith a catchment area that includes nearby urban, suburban, and rural counties of Missouri and Illinois.A daily query of the hospital’s electronic medical record (EMR) system was run to identify all patients with a clinical diagnosis of HF, including patients with either reduced (HFrEF) or preserved (HFpEF) ejection fraction, who were admitted within the past 24 hours. Most of these patients were admitted to cardiac wards but some were admitted to other wards. The study nurse-recruiters applied the European Society of Cardiology (ESC) HF criteria [28] to confirm the HF diagnosis. The ESC criteria for HFrEF require at least two symptoms (e.g., dyspnea, fatigue, etc.), two signs (e.g., elevated jugular venous pressure, hepatojugular reflux, etc.), and reduced left ventricular ejection fraction (LVEF). The criteria for HFpEF require two symptoms, two signs, and evidence of structural heart disease (e.g., left ventricular hypertrophy) and/or diastolic dysfunction.
The study required participants to have chronic HF and to be able to complete interviews and questionnaires, both at bedside during hospitalization and during follow-up assessments. Consequently, patients who met any of the following criteria were excluded: 1) isolated right HF or reversible HF due to valve disease with impending surgical correction; 2) dementia or severe cognitive impairment; 3) severe medical comorbidities with a poor one-year prognosis; 4) age less than 18 years or over 89 years; 5) active substance abuse, alcoholism; 6) bipolar disorder, schizophrenia, or other psychotic disorder; 7) patient or attending physician refusal. A total of 400 (17%) of the patients with HF met the eligibility criteria and were enrolled in the study. Eligibility screening was discontinued when the 400th patient was enrolled. Figure 1 displays the CONSORT flow diagram for the study.
Fig. 1.
CONSORT diagram of reasons for exclusion from the study.
2.3. Sociodemographic and medical data
Sociodemographic, medical history, and current medical status data were obtained from the patient’s electronic medical record (EMR). Left ventricular ejection fraction (LVEF) was obtained from the most recent echocardiogram; only echocardiograms obtained within one year of enrollment were used in the analysis. NYHA class during the two weeks prior to hospital admission was obtained by reviewing the EMR and interviewing the patient.
A list of the patient’s current medications was also obtained from the EMR. Medications that are typically classified as antidepressants were coded as antidepressants for the purpose of this analysis if and only if the prescribed dosage was at or above the minimum level indicated for treatment of depression. This rule was used to eliminate low-dose prescriptions for indications other than depression, such as low-dose trazodone for insomnia.
2.4. Depression and psychosocial data
A trained interviewer administered the Depression Interview and Structured Hamilton (DISH) [29] to identify current and past major depressive episodes according to the DSM-5 criteria [27], as well as current minor depressive episodes (Unspecified Depressive Disorder in DSM-5). The interviewer also administered several questionnaires including 1) the Patient Health Questionnaire (PHQ-9) [30]; 2) the Generalized Anxiety Disorder questionnaire (GAD-7) [31]; 3) the Perceived Stress Scale (PSS) [32]; 4) the ENRICHD Social Support Instrument (ESSI) [33]; 5) the Kansas City Cardiomyopathy Questionnaire [34]; and 5) the Duke Activity Status Index (DASI) [35].
2.5. Hypotheses
The primary analysis evaluated relationships between depressive disorders, a priori covariables, and antidepressant use. Based on the studies discussed above, we hypothesized positive associations between antidepressant use and 1) current depression diagnosis, 2) history of depression, 3) female gender, 4) older age, 5) white race, 6) HF severity, 7) multimorbidity, 8) higher socioeconomic status, and 9) anxiety. We had intended to include health insurance status in the model but did not do so because 95% of the patients had some form of insurance (e.g., Medicare, Medicaid, private insurance, etc.) Based on studies that have found a high prevalence of depression in patients who have had an acute myocardial infarction [36], we hypothesized an association between ischemic etiology of HF and antidepressant use. Finally, because the potential for drug-drug interactions and poor adherence increases along with the number of medications, we also hypothesized a negative association between antidepressant use and polypharmacy.
In a secondary analysis, we removed current depression diagnosis from the model and entered the patient’s current PHQ-9 score instead. In both the primary and secondary analysis, we entered the sociodemographic, medical, and psychiatric variables in a single set, and then added self-report measures of perceived stress, social support, and physical impairment in a second set for exploratory purposes. We predicted that after adjustment for the variables in the first set, patients with a higher level of perceived stress, lower level of perceived social support, or more impairment of physical functioning would be more likely to be prescribed antidepressant medications.
2.6. Statistical analysis
The target sample size (n=400) was based on assumptions of a 30% prevalence of major and minor depression combined, 30% attrition attributable primarily to post-discharge mortality, moderate overdispersion in the readmission process, and 90% power to detect a 30% increase in readmission rates. (The rates are not included in this report). One-way analysis of variance (ANOVA) and chi-square tests were used to compare the sociodemographic, medical, psychiatric, and psychosocial characteristics of patients who were taking an antidepressant at enrollment to those who were not. Due to missing information for some hypothesized correlates, the data were imputed under the assumption that the data were plausibly missing at random (MAR). Imputer’s models consisted of the incomplete correlates and any auxiliary variables that were at least moderately correlated (r ≥ 0.30) with them, and the other potential correlates that were relevant to the primary and secondary hypotheses. Separate imputer’s models were created and 50 imputed datasets were generated for each statistical model. Multiple imputation (MI) increases the limiting sample size for all statistical models which provides greater statistical power to detect association(s) between any hypothesized correlate and antidepressant use as well as reduces the risk of overfitting the statistical models.
Multivariable logistic regression models were used to test the hypotheses; specifically, the log-odds of antidepressant use was regressed on a linear combination of the hypothesized correlates described previously. Statistical inference was based on estimation of the odds ratio (OR), which quantifies the effect that a given correlate has on the odds of antidepressant use. Model assumptions, fit (Cox-Snell R2), and discriminatory power (c-statistic) were evaluated to provide valid inference and to assess the adequacy of the models. All statistical hypotheses were two-tailed with a Type I error rate of 0.05. SAS software version 9.3 (SAS Institute, Cary, NC) was used to create the imputation models and to fit the statistical models.
3. Results
The average age was 58.4 + 13.1 years, 51% were women and 51% were members of racial or ethnic minorities; all but three minority patients were African American. At enrollment, 111 (28%) of the patients were on an antidepressant. However, 11 of these patients were taking <150 mg/day of trazadone for insomnia, so they were classified as not being on an antidepressant at a therapeutic dosage. Thus, 100 (25%) of the patients had been prescribed an antidepressant medication.
The percentage of patients who were on an antidepressant medication was higher in those with current major (39%) or minor (34%) depression than in the nondepressed subgroup (17%). The most common antidepressants were citalopram (22% of the 100 patients who were on an antidepressant), sertraline (14%), escitalopram (12%), fluoxetine (12%), bupropion (11%), amitriptyline (9%), and duloxetine (9%). As shown in Table 1, patients who were on an antidepressant medication were significantly more likely than other patients to be white, unemployed, in NYHA class III/IV, currently depressed per DSM-5 criteria, and to have a history of major depression. They also had more medical comorbidities, were taking more medications (other than antidepressants), and scored higher on measures of depression, anxiety, and perceived stress, and lower on measures of functional capacity and health-related quality of life.
Table 1.
Participant characteristics of the subgroups defined by current use of antidepressant medication during hospitalization with heart failure (n=400).
| Characteristic | Antidepressant Use | P | |
|---|---|---|---|
| No (n=300) | Yes (n=100) | ||
| Female (n = 202) | 144 (71.3) | 58 (28.7) | |
| Non-White (n = 204) | 172 (84.3) | 32 (15.7) | |
| Age (years) | 58.6 ± 13.4 | 58.0 ± 12.1 | .72 |
| +12 years (n = 234) | 174 (74.4) | 60 (25.6) | |
| Yes (n= 158) | 111 (70.3) | 47 (29.7) | |
| Yes (n = 95) | 72 (75.8) | 23 (24.2) | |
| Yes (n = 79) | 68 (86.1) | 11 (13.9) | |
| ≥$30,000 (n=198) | 146 (73.7) | 52 (26.3) | |
| Body mass index (kg/m2) | 34.4 ± 10.5 | 34.1 ± 10.8 | .78 |
| ≥45% (n = 145) | 103 (71.0) | 42 (29.0) | |
| III-IV (n = 214) | 150 (70.1) | 64 (29.9) | |
| Yes (n= 171) | 123 (71.9) | 48 (28.1) | |
| Multimorbidity count (max. 14) | 5.5 ± 2.3 | 6.3 ± 2.3 | .003 |
| Number of other prescribed medications | 8.6 ± 2.9 | 10.3 ± 2.6 | <.0001 |
| Major depression (n = 94) | 57 (60.6) | 37 (39.4) | |
| Yes (n= 150) | 91 (60.7) | 59 (39.3) | |
| PHQ-9 score (depression) | 8.5 ± 6.1 | 11.6 ± 5.9 | <.0001 |
| GAD-7 score (anxiety) | 4.8 ± 5.1 | 7.9 ± 5.9 | <.0001 |
| PSS score (perceived stress) | 14.1 ± 8.5 | 19.5 ± 8.7 | <.0001 |
| ESSI score (social support) | 22.3 ± 5.7 | 21.1 ± 6.6 | .07 |
| DASI score (functional capacity) | 20.1 ± 14.1 | 12.8 ± 10.7 | <.0001 |
| KCCQ score (quality of life) | 44.5 ± 22.7 | 34.2 ± 20.8 | <.0001 |
Note. Continuous variables are presented as mean ± standard deviation. Categorical variables are presented as number (row percentage).
The primary results are shown in Table 2. In Model 1, contrary to our hypothesis, patients with major depression were no more likely to be on an antidepressant than were patients who were not depressed. However, patients with current minor depression or a history of depression were more likely to be on an antidepressant. The interaction between history of depression and current depression diagnosis was not significant, i.e., the relationship between current depression diagnosis and antidepressant use did not differ by depression history. The other independent correlates of antidepressant use were younger age, white race, unemployment, non-ischemic HF, and a higher medication count. When the PSS, ESSI, DASI, and KCCQ were added to Model 2, lower functional capacity was also independently related to antidepressant use. No violations of model assumptions were found for the final multivariable logistic model, and it had relatively strong predictive (Cox-Snell R2 = 0.26) and discriminatory (c-statistic = 0.84) power.
Table 2.
Primary model: relationships between depression diagnosis, covariables, and antidepressant use.
| Variable | Model 1 | Model 2 | ||||||
|---|---|---|---|---|---|---|---|---|
| Odds Ratio | 95% CI | t | P | Odds Ratio | 95% CI | t | P | |
| Female | 1.68 | 0.93, 3.02 | 1.73 | .09 | 1.34 | 0.72, 2.50 | 0.93 | .35 |
| White | 4.57 | 2.36, 8.85 | 4.51 | <.0001 | 4.38 | 2.22, 8.65 | 4.26 | <.0001 |
| Age (per 5 years) | 0.87 | 0.77, 0.99 | −2.16 | .03 | 0.87 | 0.76, 0.99 | −2.15 | .03 |
| Education ≤ 12 years | 0.96 | 0.55, 1.69 | 0.89 | .59 | 0.84 | 0.46, 1.53 | −0.58 | .56 |
| Married or partnered | 1.46 | 0.80, 2.66 | 1.24 | .21 | 1.30 | 0.68, 2.48 | 0.78 | .44 |
| Employed | 0.38 | 0.17, 0.87 | −2.29 | .02 | 0.40 | 0.17, 0.94 | −2.11 | .04 |
| Income <$30,000 per year | 0.77 | 0.38, 1.56 | −0.74 | .46 | 0.73 | 0.35, 1.53 | −0.83 | .41 |
| Body mass index (kg/m) | 0.98 | 0.95, 1.01 | −1.52 | .13 | 0.97 | 0.94, 1.01 | −1.58 | .11 |
| LVEF (%) | 1.01 | 1.00, 1.03 | 1.46 | .15 | 1.01 | 1.00, 1.03 | 1.45 | .15 |
| NYHA class (ordinal) | 1.23 | 0.83, 1.83 | 1.05 | .29 | 1.01 | 0.65, 1.56 | 0.03 | .98 |
| Ischemic heart failure | 0.50 | 0.25, 0.97 | −2.06 | .04 | 0.48 | 0.24, 0.95 | −2.10 | .04 |
| Multimorbidity count | 1.00 | 0.84, 1.18 | −0.05 | .96 | 0.97 | 0.81, 1.16 | −0.38 | .71 |
| Number of other medications | 1.30 | 1.16, 1.46 | 4.43 | <.0001 | 1.33 | 1.18, 1.50 | 4.63 | <.0001 |
| Minor depression vs. none | 2.88 | 1.33, 6.26 | 2.68 | .007 | 2.36 | 1.05, 5.30 | 2.09 | .04 |
| Major depression vs. none | 2.08 | 0.95, 4.57 | 1.83 | .07 | 1.80 | 0.78, 4.15 | 1.37 | .17 |
| History of major depression | 2.14 | 1.22, 3.73 | 2.66 | .008 | 2.04 | 1.15, 3.63 | 2.43 | .02 |
| GAD-7 (anxiety) | 1.06 | 0.99, 1.12 | 1.77 | .08 | 1.02 | 0.95, 1.10 | 0.59 | .56 |
| PSS (perceived stress) | 1.04 | 1.00, 1.09 | 1.77 | .08 | ||||
| ESSI (social support) | 1.00 | 0.94, 1.06 | −0.04 | .97 | ||||
| DASI (functional capacity) | 0.96 | 0.93, 0.99 | −2.49 | .01 | ||||
| KCCQ (quality of life) | 1.00 | 0.99, 1.02 | 0.33 | .74 | ||||
Table 3 displays the results of the secondary analysis in which the PHQ-9 depression score replaced the depression diagnosis. Age and ischemic HF did not have significant effects in Model 1, but white race, unemployment, and a higher medication count were independently associated with antidepressant use. The interaction between history of major depression and the current PHQ-9 score was statistically significant, indicating that the association between current depressive symptoms and antidepressant use depends on whether there is a history of majority depression. Table 4 quantifies the relationship between history of depression and antidepressant use at various levels of current depression severity. It shows that history of depression was associated with antidepressant use only among patients with low PHQ scores. When the PSS, ESSI, DASI, and KCCQ were added to Model 2 in Table 3, lower functional capacity was independently associated with antidepressant use. There were no violations of model assumptions for the final multivariable model; it had strong predictive (Cox-Snell R2 = 0.27) and discriminatory (c-statistic = 0.84) power.
Table 3.
Secondary model: Relationships between PHQ-9 score, covariables, and antidepressant use.
| Variable | Model 1 | Model 2 | ||||||
|---|---|---|---|---|---|---|---|---|
| Odds Ratio | 95% CI | t | P | Odds Ratio | 95% CI | t | P | |
| Female | 1.69 | 0.93, 3.06 | 1.73 | .08 | 1.35 | 0.72, 2.54 | 0.94 | .35 |
| White | 4.41 | 2.30, 8.45 | 4.46 | <.0001 | 4.23 | 2.17, 8.25 | 4.23 | <.0001 |
| Age (per 5 years) | 0.89 | 0.78, 1.01 | −1.79 | .07 | 0.88 | 0.77, 1.01 | −1.85 | .06 |
| Education ≤ 12 years | 1.01 | 0.57, 1.80 | 0.04 | .97 | 0.88 | 0.48, 1.64 | −0.39 | .70 |
| Married or partnered | 1.25 | 0.69, 2.26 | 0.72 | .47 | 1.15 | 0.60, 2.20 | 0.42 | .68 |
| Employed | 0.39 | 0.17, 0.89 | −2.23 | .03 | 0.40 | 0.17, 0.94 | −2.10 | .04 |
| Income <$30,000 per year | 0.69 | 0.33, 1.44 | −1.00 | .32 | 0.66 | 0.30, 1.43 | −1.06 | .29 |
| Body mass index (kg/m2) | 0.98 | 0.95, 1.01 | −1.41 | .16 | 0.98 | 0.95, 1.01 | −1.50 | .13 |
| LVEF (%) | 1.01 | 0.99, 1.03 | 1.23 | .22 | 1.01 | 0.99, 1.03 | 1.31 | .19 |
| NYHA (ordinal) | 1.35 | 0.90, 2.03 | 1.47 | .14 | 1.10 | 0.70, 1.72 | 0.41 | .68 |
| Ischemic heart failure | 0.58 | 0.30, 1.11 | −1.64 | .10 | 0.53 | 0.27, 1.05 | −1.83 | .07 |
| Multimorbidity count | 0.95 | 0.80, 1.12 | −0.62 | .53 | 0.93 | 0.77, 1.11 | −0.83 | .41 |
| No. prescribed medications | 1.30 | 1.16, 1.46 | 4.41 | <.0001 | 1.33 | 1.18, 1.50 | 4.62 | <.0001 |
| PHQ-9 score | 1.12 | 1.04, 1.21 | 2.96 | .003 | 1.09 | 1.00, 1.18 | 2.04 | .04 |
| History of major depression | 8.14 | 2.54, 26.1 | 3.53 | .0004 | 6.31 | 1.91, 20.8 | 3.02 | .003 |
| PHQ-9 x history of depression | 0.89 | 0.81, 0.97 | −2.55 | .01 | 0.91 | 0.83, 0.99 | −2.09 | .04 |
| GAD-7 (anxiety) | 1.05 | 0.98, 1.12 | 1.35 | .18 | 1.02 | 0.94, 1.10 | 0.46 | .65 |
| PSS (perceived stress) | 1.05 | 1.00, 1.10 | 1.85 | .07 | ||||
| ESSI (social support) | 0.99 | 0.94, 1.05 | −0.25 | .80 | ||||
| DASI (functional capacity) | 0.96 | 0.93, 0.99 | −2.40 | .02 | ||||
| KCCQ (quality of life) | 1.00 | 0.99, 1.02 | 0.37 | .71 | ||||
Table 4.
Odds ratios of antidepressant use for history of depression at selected PHQ-9 scores to illustrate the significant interaction between PHQ-9 score and history of major depression.
| Variable | PHQ-9 Score | Model 1 | Model 2 | ||||
|---|---|---|---|---|---|---|---|
| Odds Ratio | 95% CI | P | Odds Ratio | 95% CI | P | ||
| History of depression | 0 | 8.14 | 2.54, 26.1 | .0004 | 6.31 | 1.91, 20.8 | .003 |
| 5 | 4.52 | 2.04, 10.0 | .0002 | 3.85 | 1.69, 8.76 | .001 | |
| 10 | 2.51 | 1.42, 4.43 | .002 | 2.35 | 1.30, 4.24 | .004 | |
| 15 | 1.39 | 0.73, 2.67 | .32 | 1.44 | 0.74, 2.80 | .29 | |
| 20 | 0.77 | 0.30, 2.03 | .60 | 0.88 | 0.33, 2.35 | .80 | |
| 25 | 0.43 | 0.11, 1.67 | .22 | 0.54 | 0.13, 2.15 | .38 | |
4. Discussion
This study found that hospitalized patients with HF and comorbid major depression were more likely than nondepressed patients to be on an antidepressant. However, the percentages of patients who were on antidepressants differed very little between the subgroups with current major and minor depression, and approximately one out of five nondepressed patients were also on an antidepressant. This is not the pattern that one would expect if antidepressants were prescribed only for patients with current major depression and if these medications were discontinued when partial or full remission occurs. However, antidepressants are often continued long after depression has improved in order to reduce the risk of relapse or recurrence [37–40]. They are also being used in some cases to treat conditions other than depression, such as generalized anxiety disorder [41].
Since this was a cross-sectional analysis, it was not possible to determine when patients started or stopped taking an antidepressant medication, or whether (or how severely) they were depressed at the time. In some cases, the patient’s depressive symptoms probably improved after the initiation of antidepressant therapy, although not necessarily to the point of complete remission. Thus, some of the patients who were not depressed at baseline or who had current minor depression might have met the criteria for major depression sometime before enrollment in the study. This possibility is strengthened by the relatively strong relationship between history of major depression and antidepressant use. It is also possible, however, that some patients with HF who do not have comorbid major depression are prescribed antidepressant medications anyway.
It is difficult to differentiate on an initial interview between current minor depression and major depression in partial remission because many patients have difficulty recalling and describing the timing of changes in depression symptoms. Similarly, some of the patients who did not meet the criteria for a current depressive episode might have had a depressive disorder that was currently in remission. To examine whether this might have inflated the depression scores of the currently nondepressed group or the group with current minor depression, we stratified the sample by history of depression; within these strata, we then compared the PHQ-9 depression scores of subgroups defined by depression diagnosis and antidepressant use. In the currently nondepressed patients, and in those with current minor depression, PHQ-9 scores differed little by history of depression (results not displayed). These data make it less plausible that many patients with a major depressive episode in remission were misclassified by the interview as being nondepressed or as having minor depression. However, this possibility cannot be eliminated altogether.
Our findings raise the possibility that antidepressants are being prescribed for at least two off-label indications in patients with HF, i.e., minor depression and difficulty coping with stressful circumstances such as unemployment or functional impairment. The results of both the primary and secondary analyses are consistent with these possibilities in that minor depression, unemployment, and functional impairment were independently associated with antidepressant use.
The results of this study replicate previous evidence of racial disparities in antidepressant use [9]; a higher percentage of white than African American patients were on an antidepressant. The difference between these groups held after adjustment for current depression, history of depression, and socioeconomic factors. This finding is consistent with evidence that antidepressant medications tend to be less acceptable to African American and Hispanic than to white patients [42]. There may also be other reasons why white patients with HF are more likely than African American patients to be treated with antidepressants, but this study did not reveal them.
Several previous studies have found that women are more likely than men to use antidepressants even after adjusting for differences in depression [6, 8–10, 12]. Although we found that a slightly higher proportion of women than men were on an antidepressant, the difference was not significant. In addition, female gender was not independently associated with antidepressant use in either the primary or the secondary analysis. We also failed to replicate previous findings of associations between antidepressant use and older age [10], higher socioeconomic status [10], more severe HF [8, 12], multimorbidity [10], obesity [8], or marital status [9]. Contrary to previous findings, our primary model suggests that younger patients with HF are more likely than older patients to take antidepressants.
We also identified two independent correlates of antidepressant use that have not been found in previous studies of patients with HF, i.e., non-ischemic HF and polypharmacy. Both findings are contrary to our hypotheses. We expected that patients with a greater burden of physical illness, whether due to having more severe HF, more comorbidities, or a history of myocardial infarction would be more likely than other patients to be on an antidepressant, but our findings suggest otherwise. Physicians may be reluctant to prescribe antidepressant medications to patients with ischemic HF, although they do not seem to be deterred by a low LVEF.
We hypothesized an inverse relationship between antidepressant use and polypharmacy on the assumption that the potential for drug-drug interactions might compel physicians to withhold antidepressants in some cases in order to prevent interactions with high-priority medications. Contrary to our hypothesis, we found a positive association between antidepressant use and the total number of other medications prescribed. Although we did not obtain data on the duration of antidepressant use, these findings raise the possibility that antidepressant medications are not being discontinued even when a patient’s worsening medical status and growing list of medications might warrant a reevaluation of the risks and benefits of continuing antidepressant therapy. If so, this could be an example of clinical inertia [43]. This phenomenon pertains both to situations in which initiation or intensification of a medication is indicated but no action is taken, and to those in which a medication should be tapered or discontinued but it is continued instead. Studies in various patient populations have found associations between clinical inertia and old age, low life expectancy, multimorbidity, and psychiatric comorbidity [44]. On the other hand, antidepressant continuation and maintenance therapy has been shown to prevent relapse or recurrence of major depression [37, 45]. Continuation or maintenance antidepressant therapy may be helpful for patients with HF who are high risk for a relapse or recurrence of major depression while their medical status is worsening.
Differences in sample characteristics may help to explain differences in the correlates of antidepressant use between this study and previous investigations. There are multiple differences in sample characteristics among these studies, but one of the most salient is that approximately 50% of our participants were African American; the percentages of African American participants were substantially lower in previous studies [6, 8–10, 12, 26]. Utilization of antidepressants among African Americans in this study was less than half that of white participants, and that may have attenuated some of the associations that were hypothesized. Also, this study’s eligibility criteria were designed to screen out patients who were too medically ill or cognitively impaired to provide reliable responses to the research interviews and questionnaires. The eligibility criteria for the earlier population-based and registry studies were less restrictive, and this may also help to explain differences between the present findings and the results of those studies.
The strengths of this study include the use of a structured clinical interview to evaluate major and minor depression according to the DSM-5 criteria and the collection of extensive clinical data on HF and medical comorbidities from the hospital’s electronic medical record. The most significant limitation is the study’s cross-sectional design. Because the participants were enrolled during a hospital stay and the patient’s current medication list was used to determine antidepressant use, it was not possible to document indication(s) for the antidepressant prescription, the duration of use, or the patient’s adherence to the antidepressant regimen. Similarly, we were able to retrospectively determine whether there was a history of major depression and whether a current major or minor depressive episode was present, but we were not able to obtain reliable data on the duration of the current episode. This may have resulted in underestimation of the prevalence of antidepressant use in this patient population, and it prevent us from correlating antidepressant use with changes in mood state. In addition, the interview was designed to be as brief as possible in order to reduce the respondent burden on the hospitalized participants. Consequently, it did not evaluate disorders other than depression that may have been treated with antidepressant medications, including anxiety disorders, sleep disorders, and chronic pain.
The cumulative evidence from the present results and previous studies can help clinicians identify patients with HF who may be depressed but who are relatively unlikely to receive any treatment for depression as part of their routine care. Previous findings have shown that male patients with HF are less likely to receive treatment for depression than otherwise similar female patients. Our findings suggest that there are also disparities in depression care between African American and other patients, as well as between patients who are functionally intact or only mildly functionally impaired compared to more severely impaired patients. Some of these patients may decline antidepressants, and these agents may be ineffective for some patients who do accept them. Some patients who do not accept or respond to antidepressants may respond to nonpharmacological interventions such as cognitive behavior therapy [46]. This highlights the need to integrate mental health care services into HF care.
5. Conclusion
This study found independent, positive associations between antidepressant use and current minor depression, history of major depression, white race, younger age, unemployment, non-ischemic HF, functional incapacity, and polypharmacy. Because of the limitations imposed by the cross-sectional analysis, inconsistencies between the depression diagnosis and PHQ models, and inconsistencies with previous studies, the findings raise more questions than answers about antidepressant use in patients with HF. Prospective studies of well-characterized patients are needed to clarify the determinants of antidepressant use in patients with HF. This research could help to pave the way for safer and more effective approaches to the management of comorbid major depression in heart failure. Nevertheless, the cumulative evidence from this and previous studies suggest that many male and African American patients with HF, as well as patients without very salient functional impairment, may not be receiving depression care when they need it. Clinicians should discuss unmet needs for depression care with these patients.
Funding
This work was support by the National Heart, Lung, and Blood Institute, Bethesda, Maryland, USA [grant number 5R01HL119286].
List of Abbreviations
- BMI
Body mass index
- DASI
Duke Activity Status Index
- DISH
Depression Interview and Structured Hamilton
- DSM-5
American Psychiatric Association Diagnostic and Statistical Manual, 5th edition
- EMR
Electronic medical record
- ESSI
ENRICHD Social Support Instrument
- GAD-7
Generalized Anxiety Disorder questionnaire
- HF
Heart failure
- HFpEF
Heart failure with preserved ejection fraction
- HFrEF
Heart failure with reduced ejection fraction
- KCCQ
Kansas City Cardiomyopathy Questionnaire
- LVEF
Left ventricular ejection fraction
- NYHA
New York Heart Association
- PHQ-9
Patient Health Questionnaire
- PSS
Perceived Stress Scale
- RCT
Randomized controlled trial
- SSRI
Selective serotonin reuptake inhibitor
- TCA
Tricyclic antidepressant
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References
- [1].Pratt LA, Brody DJ and Gu Q. Antidepressant use among persons aged 12 and over: United States, 2011–2014. NCHS Data Brief 2017:1–8. [PubMed] [Google Scholar]
- [2].Freedland KE, Carney RM, Rich MW, Steinmeyer BC, Skala JA and Davila-Roman VG. Depression and multiple rehospitalizations in patients with heart failure. Clinical cardiology 2016;39:257–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Bekelman DB, Allen LA, McBryde CF, Hattler B, Fairclough DL, Havranek EP, et al. Effect of a Collaborative Care Intervention vs Usual Care on Health Status of Patients With Chronic Heart Failure: The CASA Randomized Clinical Trial. JAMA Intern Med 2018;178:511–519. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].O’Connor CM, Jiang W, Kuchibhatla M, Silva SG, Cuffe MS, Callwood DD, et al. Safety and efficacy of sertraline for depression in patients with heart failure: results of the SADHART-CHF (Sertraline Against Depression and Heart Disease in Chronic Heart Failure) trial. J Am Coll Cardiol 2010;56:692–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Angermann CE, Gelbrich G, Stork S, Gunold H, Edelmann F, Wachter R, et al. Effect of Escitalopram on All-Cause Mortality and Hospitalization in Patients With Heart Failure and Depression: The MOOD-HF Randomized Clinical Trial. Jama 2016;315:2683–93. [DOI] [PubMed] [Google Scholar]
- [6].Chung ML, Dekker RL, Lennie TA and Moser DK. Antidepressants do not improve event-free survival in patients with heart failure when depressive symptoms remain. Heart & lung : the journal of critical care 2013;42:85–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Sherwood A, Blumenthal JA, Hinderliter AL, Koch GG, Adams KF Jr., Dupree CS, et al. Worsening depressive symptoms are associated with adverse clinical outcomes in patients with heart failure. J Am Coll Cardiol 2011;57:418–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Diez-Quevedo C, Lupon J, Gonzalez B, Urrutia A, Cano L, Cabanes R, et al. Depression, antidepressants, and long-term mortality in heart failure. Int J Cardiol 2013;167:1217–25. [DOI] [PubMed] [Google Scholar]
- [9].O’Connor CM, Jiang W, Kuchibhatla M, Mehta RH, Clary GL, Cuffe MS, et al. Antidepressant use, depression, and survival in patients with heart failure. Arch Intern Med 2008;168:2232–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Brouwers C, Christensen SB, Damen NL, Denollet J, Torp-Pedersen C, Gislason GH, et al. Antidepressant use and risk for mortality in 121,252 heart failure patients with or without a diagnosis of clinical depression. Int J Cardiol 2016;203:867–73. [DOI] [PubMed] [Google Scholar]
- [11].Fosbol EL, Gislason GH, Poulsen HE, Hansen ML, Folke F, Schramm TK, et al. Prognosis in heart failure and the value of {beta}-blockers are altered by the use of antidepressants and depend on the type of antidepressants used. Circulation Heart failure 2009;2:582–90. [DOI] [PubMed] [Google Scholar]
- [12].Veien KT, Videbaek L, Schou M, Gustafsson F, Hald-Steffensen F and Hildebrandt PR. High mortality among heart failure patients treated with antidepressants. Int J Cardiol 2011;146:64–7. [DOI] [PubMed] [Google Scholar]
- [13].Freedland KE, Rich MW, Skala JA, Carney RM, Davila-Roman VG and Jaffe AS. Prevalence of depression in hospitalized patients with congestive heart failure. Psychosom Med 2003;65:119–28. [DOI] [PubMed] [Google Scholar]
- [14].Koenig HG. Depression in hospitalized older patients with congestive heart failure. Gen Hosp Psychiatry 1998;20:29–43. [DOI] [PubMed] [Google Scholar]
- [15].Rutledge T, Reis VA, Linke SE, Greenberg BH and Mills PJ. Depression in heart failure a meta-analytic review of prevalence, intervention effects, and associations with clinical outcomes. J Am Coll Cardiol 2006;48:1527–37. [DOI] [PubMed] [Google Scholar]
- [16].Hallas CN, Wray J, Andreou P and Banner NR. Depression and perceptions about heart failure predict quality of life in patients with advanced heart failure. Heart & lung : the journal of critical care 2011;40:111–21. [DOI] [PubMed] [Google Scholar]
- [17].Alhurani AS, Dekker RL, Abed MA, Khalil A, Al Zaghal MH, Lee KS, et al. The association of co-morbid symptoms of depression and anxiety with all-cause mortality and cardiac rehospitalization in patients with heart failure. Psychosomatics 2015;56:371–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Jiang W, Alexander J, Christopher E, Kuchibhatla M, Gaulden LH, Cuffe MS, et al. Relationship of depression to increased risk of mortality and rehospitalization in patients with congestive heart failure. Arch Intern Med 2001;161:1849–56. [DOI] [PubMed] [Google Scholar]
- [19].Adams J, Kuchibhatla M, Christopher EJ, Alexander JD, Clary GL, Cuffe MS, et al. Association of depression and survival in patients with chronic heart failure over 12 Years. Psychosomatics 2012;53:339–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Freedland KE, Hesseler MJ, Carney RM, Steinmeyer BC, Skala JA, Davila-Roman VG, et al. Major Depression and Long-Term Survival of Patients With Heart Failure. Psychosom Med 2016;78:896–903. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Adelborg K, Schmidt M, Sundboll J, Pedersen L, Videbech P, Botker HE, et al. Mortality Risk Among Heart Failure Patients With Depression: A Nationwide Population-Based Cohort Study. J Am Heart Assoc 2016;5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Deveney TK, Belnap BH, Mazumdar S and Rollman BL. The prognostic impact and optimal timing of the Patient Health Questionnaire depression screen on 4-year mortality among hospitalized patients with systolic heart failure. Gen Hosp Psychiatry 2016;42:9–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].Frasure-Smith N, Lesperance F, Habra M, Talajic M, Khairy P, Dorian P, et al. Elevated depression symptoms predict long-term cardiovascular mortality in patients with atrial fibrillation and heart failure. Circulation 2009;120:134–40, 3p following 140. [DOI] [PubMed] [Google Scholar]
- [24].Gathright EC, Goldstein CM, Josephson RA and Hughes JW. Depression increases the risk of mortality in patients with heart failure: A meta-analysis. J Psychosom Res 2017;94:82–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Goyal P, Kneifati-Hayek J, Archambault A, Mehta K, Levitan EB, Chen L, et al. Prescribing patterns of heart failure-exacerbating medications following a heart failure hospitalization. JACC Heart failure 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Waldman SV, Blumenthal JA, Babyak MA, Sherwood A, Sketch M, Davidson J, et al. Ethnic differences in the treatment of depression in patients with ischemic heart disease. Am Heart J 2009;157:77–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].American Psychiatric Association. and American Psychiatric Association. DSM-5 Task Force. Diagnostic and statistical manual of mental disorders : DSM-5. Washington, D.C.: American Psychiatric Association, 2013. [Google Scholar]
- [28].Ponikowski P, Voors AA, Anker SD, Bueno H, Cleland JGF, Coats AJS, et al. 2016 ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure: The Task Force for the diagnosis and treatment of acute and chronic heart failure of the European Society of Cardiology (ESC)Developed with the special contribution of the Heart Failure Association (HFA) of the ESC. Eur Heart J 2016;37:2129–2200. [DOI] [PubMed] [Google Scholar]
- [29].Freedland KE, Skala JA, Carney RM, Raczynski JM, Taylor CB, Mendes de Leon CF, et al. The Depression Interview and Structured Hamilton (DISH): rationale, development, characteristics, and clinical validity. Psychosom Med 2002;64:897–905. [DOI] [PubMed] [Google Scholar]
- [30].Kroenke K, Spitzer RL and Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med 2001;16:606–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Spitzer RL, Kroenke K, Williams JB and Lowe B. A brief measure for assessing generalized anxiety disorder: the GAD-7. Arch Intern Med 2006;166:1092–7. [DOI] [PubMed] [Google Scholar]
- [32].Cohen S, Kamarck T and Mermelstein R. A global measure of perceived stress. J Health Soc Behav 1983;24:385–96. [PubMed] [Google Scholar]
- [33].Mitchell PH, Powell L, Blumenthal J, Norten J, Ironson G, Pitula CR, et al. A short social support measure for patients recovering from myocardial infarction: the ENRICHD Social Support Inventory. J Cardiopulm Rehabil 2003;23:398–403. [DOI] [PubMed] [Google Scholar]
- [34].Green CP, Porter CB, Bresnahan DR and Spertus JA. Development and evaluation of the Kansas City Cardiomyopathy Questionnaire: a new health status measure for heart failure. J Am Coll Cardiol 2000;35:1245–55. [DOI] [PubMed] [Google Scholar]
- [35].Hlatky MA, Boineau RE, Higginbotham MB, Lee KL, Mark DB, Califf RM, et al. A brief self-administered questionnaire to determine functional capacity (the Duke Activity Status Index). Am J Cardiol 1989;64:651–4. [DOI] [PubMed] [Google Scholar]
- [36].Feng L, Li L, Liu W, Yang J, Wang Q, Shi L, et al. Prevalence of depression in myocardial infarction: A PRISMA-compliant meta-analysis. Medicine 2019;98:e14596. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [37].Bockting CLH, Klein NS, Elgersma HJ, van Rijsbergen GD, Slofstra C, Ormel J, et al. Effectiveness of preventive cognitive therapy while tapering antidepressants versus maintenance antidepressant treatment versus their combination in prevention of depressive relapse or recurrence (DRD study): a three-group, multicentre, randomised controlled trial. The lancet Psychiatry 2018;5:401–410. [DOI] [PubMed] [Google Scholar]
- [38].Machmutow K, Meister R, Jansen A, Kriston L, Watzke B, Harter MC, et al. Comparative effectiveness of continuation and maintenance treatments for persistent depressive disorder in adults. The Cochrane database of systematic reviews 2019;5:Cd012855. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [39].Riihimaki K, Vuorilehto M and Isometsa E. Maintenance pharmacotherapy for recurrent major depressive disorder in primary care: A 5-year follow-up study. Eur Psychiatry 2017;41:111–114. [DOI] [PubMed] [Google Scholar]
- [40].Huijbregts KM, Hoogendoorn A, Slottje P, van Balkom A and Batelaan NM. Long-Term and Short-Term Antidepressant Use in General Practice: Data from a Large Cohort in the Netherlands. Psychother Psychosom 2017;86:362–369. [DOI] [PubMed] [Google Scholar]
- [41].Wetherell JL, Petkus AJ, White KS, Nguyen H, Kornblith S, Andreescu C, et al. Antidepressant medication augmented with cognitive-behavioral therapy for generalized anxiety disorder in older adults. Am J Psychiatry 2013;170:782–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [42].Cooper LA, Gonzales JJ, Gallo JJ, Rost KM, Meredith LS, Rubenstein LV, et al. The acceptability of treatment for depression among African-American, Hispanic, and white primary care patients. Medical care 2003;41:479–89. [DOI] [PubMed] [Google Scholar]
- [43].Phillips LS, Branch WT, Cook CB, Doyle JP, El-Kebbi IM, Gallina DL, et al. Clinical inertia. Ann Intern Med 2001;135:825–34. [DOI] [PubMed] [Google Scholar]
- [44].Milman T, Joundi RA, Alotaibi NM and Saposnik G. Clinical inertia in the pharmacological management of hypertension: A systematic review and meta-analysis. Medicine 2018;97:e11121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [45].Hansen R, Gaynes B, Thieda P, Gartlehner G, Deveaugh-Geiss A, Krebs E, et al. Meta-analysis of major depressive disorder relapse and recurrence with second-generation antidepressants. Psychiatr Serv 2008;59:1121–30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [46].Freedland KE, Carney RM, Rich MW, Steinmeyer BC and Rubin EH. Cognitive Behavior Therapy for Depression and Self-Care in Heart Failure Patients: A Randomized Clinical Trial. JAMA Intern Med 2015;175:1773–82. [DOI] [PMC free article] [PubMed] [Google Scholar]