Author + information
- Received September 1, 2004
- Revision received November 22, 2004
- Accepted December 14, 2004
- Published online April 5, 2005.
- William Whang, MD, MS⁎,⁎ (, )
- Christine M. Albert, MD, MPH⁎,†,
- Samuel F. Sears Jr, PhD‡,
- Rachel Lampert, MD§,
- Jamie B. Conti, MD‡,
- Paul J. Wang, MD∥,
- Jagmeet P. Singh, MD, DPhil⁎,
- Jeremy N. Ruskin, MD⁎,
- James E. Muller, MD⁎,
- Murray A. Mittleman, MD, DrPH¶,
- TOVA Study Investigators
- ↵⁎Reprint requests and correspondence:
Dr. William Whang, Cardiology Division, Mailstop GRB 109, Massachusetts General Hospital, 55 Fruit Street, Boston, Massachusetts 02114
Objectives We sought to examine the relationship between symptoms of depression and shock-treated ventricular arrhythmias among implantable cardioverter-defibrillator (ICD) patients.
Background Depression predicts mortality in patients with coronary artery disease (CAD), but whether this is via an increased risk of fatal ventricular arrhythmias is unclear.
Methods We prospectively analyzed data on symptoms of depression and risk of ventricular arrhythmia (ventricular tachycardia/ventricular fibrillation [VT/VF]) resulting in ICD discharge in the Triggers of Ventricular Arrhythmias (TOVA) study. Symptoms were assessed by the Center for Epidemiologic Studies-Depression (CES-D) scale. Scores of 16 to 26 and ≥27 represented mild and moderate/severe depression, respectively. The Cox and Anderson-Gill proportional hazards models were used to test for associations among all patients and patients with CAD.
Results Among 645 patients with baseline assessments, 90 (14%) were mildly depressed and 25 (3.9%) were moderately to severely depressed. Moderate/severe depression was associated with time to first shock for VT/VF (hazard ratio [HR] 3.2, 95% confidence interval [CI] 1.1 to 9.9) and all shocks for VT/VF including recurrent episodes (HR 3.2, 95% CI 1.2 to 8.6). Among the 476 CAD patients, the association with time to first shock (HR 6.4, 95% CI 1.9 to 21.1) and all shocks (HR 8.3, 95% CI 2.9 to 23.3) remained. The risk of shock for VT/VF was associated with depression severity in the total population (p for trend = 0.02) and among patients with CAD (p < 0.01), even after controlling for multiple confounders.
Conclusions More severe symptoms of depression predict shocks for VT/VF among ICD patients. The elevated risk of VT/VF among patients with CAD and depression suggests that arrhythmia may contribute significantly to total mortality in this subgroup.
Depression has been identified as a risk factor for the development of cardiovascular disease and for recurrent cardiac events and mortality among those with established coronary artery disease (CAD) (1,2). A number of studies have suggested that the poor cardiac prognosis associated with depression may be due to an arrhythmic mechanism (3,4), especially among patients with CAD (5–9). Among patients with implantable cardioverter-defibrillators (ICDs), who are at high risk for ventricular arrhythmias, there have been relatively few studies of depression as a predictor of arrhythmic outcomes (10,11).
We sought to determine whether depression was associated with an increased risk of shock-treated malignant ventricular arrhythmia in a large, prospectively collected sample of ICD patients. In addition, we sought to gain insight into the mode by which depression worsens prognosis among patients with CAD by analyzing the subgroup of patients with CAD separately.
The Triggers of Ventricular Arrhythmias (TOVA) study has been described previously (12). Briefly, it is a prospective cohort study of ICD patients conducted at 31 centers in the United States and designed to identify triggers of ventricular tachycardia/ventricular fibrillation (VT/VF). Patients met class I to IIb indications for ICD implant according to the 1998 American College of Cardiology/American Heart Association guidelines (13). Enrollment began in June 2000, and patients were identified at the time of implant or through the ICD clinics in participating institutions.
Depression symptom assessment
At entry into the study, information was collected on baseline clinical characteristics and information on symptoms of depression was collected on a subset of the patients. Of the 1,188 patients participating in TOVA, the last 645 patients enrolled were asked to complete the Centers for Epidemiologic Studies-Depression (CES-D) scale (14). The CES-D scale is a 20-item survey that measures mood, somatic symptoms, social interactions, and motor functioning according to a 4-point Likert scale. The survey asks about the frequency in the past week that one has felt or behaved a certain way. Items include: “I felt depressed,” “My sleep was restless,” “I lost interest in my usual activities,” “I felt like I was moving too slowly,” and “I did not like eating; my appetite was poor.” The CES-D scale score is a sum total over the 20 items and can range from 0 to 60, with higher scores indicating more severe symptoms of depression. The CES-D scale has been shown to have high internal consistency (Cronbach's alpha 0.85 in the general population) and satisfactory test-retest reliability over several weeks' time (14–16). For this analysis, we categorized patients with scores of 16 to 26 as mildly depressed (17) and those with scores of 27 or greater as moderately to severely depressed. The cutoff of 27 or greater has been shown to detect clinical depression with a specificity of 70% among medical patients (18).
Patients were asked to call the study and clinical staff at participating institutions within 72 h of an ICD discharge. All of the devices in the study had electrogram storage capability, and reported device discharges were evaluated by an electrophysiologist (P.J.W., C.M.A., J.P.S.) at the central core laboratory to identify episodes of sustained VT or VF. The primary end point in this analysis was first post-enrollment shock for VT or VF. A secondary end point was two or more appropriate ICD shocks within a 24-h period (19,20).
Means or proportions for baseline clinical variables were calculated according to the severity of depression. Chi-square tests were used to compare categoric variables. Visual inspection of histograms as well as skewness and kurtosis statistics were used to assess continuous variables for normality, and analysis of variance or non-parametric tests were used to compare continuous variables by depression category. Cumulative event proportions were calculated by the Kaplan-Meier method, and outcome differences of patients with depression versus those without were assessed with the log-rank test.
Cox proportional hazards models were used to evaluate associations between depression severity at enrollment and time to first ICD discharge for a confirmed episode of VT/VF. The proportional hazards assumption was tested with appropriate time interaction variables. In addition, Anderson-Gill proportional hazards models were used in order to incorporate recurrent ICD discharges in our analysis (21). All of the covariates included in the multivariable models varied significantly according to depression category or resulted in a change of >10% in the coefficient for either depression group in a model that already included age and gender. The relationship between depression severity and ICD discharge was assessed with a linear trend variable that assumed three possible values for no symptoms of depression, mild, or moderate-to-severe symptoms. The interaction between CAD and depression category was tested with multiplicative indicator variables among all patients, and multivariable proportional hazards analyses were performed in the subgroup of patients with a history of CAD at baseline. In a secondary analysis, we modeled the time to first episode of 2 or more appropriate ICD shocks within a 24-h time period using Cox proportional hazards models. All p values are two-tailed and all confidence intervals (CIs) computed at the 95% level (SAS Version 8.2, SAS, Cary, North Carolina).
Among 645 patients for whom baseline CES-D scale scores were available, 90 patients (14%) were categorized as mildly depressed (CES-D scale core 16 to 26) and 25 (3.9%) were categorized as moderately to severely depressed (CES-D ≥27). Depressed patients were younger and more often women compared with patients without depression (Table 1).The median time from ICD implant to enrollment was 16.4 (interquartile range 6.7 to 35.2) months among patients without symptoms of depression, 21.4 (interquartile range 8.0 to 38.9) months among patients with mild symptoms, and 23.4 (interquartile range 16.7 to 37) months among patients with moderate to severe symptoms (p = 0.36, Kruskal-Wallis test). Alcohol use was less frequent among depressed patients. Several baseline characteristics, including history of CAD, angina, congestive heart failure, and prior ICD discharge, differed significantly between patient groups but exhibited a non-linear relationship with CES-D scale score. These characteristics were more frequent in patients with mild depression compared with those without depression, but less frequent in patients with moderate to severe depression compared to those with mild depression. Selective serotonin reuptake inhibitor (SSRI) use was higher with worsening severity of depression, and beta-blocker use was not significantly different between groups.
CES-D scale score and risk of shock for VT/VF
During a median follow-up of 359 days (interquartile range 180 to 526), 59 patients experienced a total of 103 appropriate ICD shocks. Kaplan-Meier curves of the time to appropriate ICD discharge stratified according to the presence of depression (CES-D scale score ≥16) are shown in Figure 1(p = 0.02, log rank test). A Cox proportional hazards model that contained baseline depression category and adjusted for age and gender revealed positive relationships between both categories of depression and ICD discharges for VT or VF (Table 2).In a multivariable Cox proportional hazards model that adjusted for age, gender, number of prior ICD discharges (0, 1, 2 to 4, 5 or more), time from ICD implant to study enrollment, cardiac arrest as a device indication, CAD, angina class, congestive heart failure class, left ventricular ejection fraction, smoking, alcohol use, SSRI use, and use of angiotensin-converting enzyme (ACE) inhibitors or angiotensin receptor blocker (ARBs), moderate to severe depression (CES-D scale score ≥27) was associated with a statistically significant increased risk of appropriate ICD shocks (hazard ratio [HR] 3.2, 95% CI 1.1 to 9.9). The Andersen-Gill model, which included recurrent shocks in the dependent variable and which adjusted for the same covariates as the Cox model, also indicated an increased risk of shocks for patients with moderate to severe depression (hazard ratio 3.2, 95% CI 1.2 to 8.6). A test for trend was statistically significant for both the Cox and Anderson-Gill models (p ≤ 0.02), indicating that higher levels of depressive symptoms were associated with a higher risk of appropriate shock.
Among the subgroup of patients with CAD, there was a strong positive relationship between moderate to severe depression and appropriate shocks (Table 3).In the Cox and Anderson-Gill multivariable proportional hazards models, depression scores ≥27 conferred hazard ratios of 6.4 (95% CI 1.9 to 21.1) and 8.3 (95% CI 2.9 to 23.3), respectively. In models that included all patients, tests for an interaction between CAD and depression score were statistically significant for Anderson-Gill models (p = 0.04), but not for Cox models (p = 0.16), although our statistical power to examine this issue was quite limited.
In a secondary analysis of two or more appropriate ICD shocks within a 24-h time period, moderate to severe depression was associated with an elevated risk of such episodes in multivariable analyses, despite only 16 events (Table 4).We also estimated Cox proportional hazards models after removal of those who had had ICD discharges before study enrollment. The HR for moderate to severe depression was similar to our main analysis (HR 3.7, 95% CI 1.2 to 11.6), with a significant test for trend across categories (p = 0.04). Finally, a post-hoc Cox regression analysis suggested that the relationship between depression category and risk of shock was stronger among patients with ICD implant >6 months before enrollment (p for interaction = 0.09).
In this prospective multicenter study, symptoms of depression significantly predicted appropriate ICD shocks, even with adjustment for other important comorbidities, medications, and history of prior shocks. Few prior studies have evaluated the long-term risk conferred by depression in ICD patients. Dunbar et al. (10) studied 172 patients with newly implanted ICDs through a survey that described 6 mood states including depression-dejection and tension-anxiety, and found that 1-month total mood disturbance score predicted shock or anti-tachycardia pacing at 3 months. Hermann et al. (11) assessed 105 patients before ICD implantation with the Hospital Anxiety and Depression Scale. The odds of VT at one year were significantly higher among those with abnormal depression scores in multivariable analyses. Our analysis provides further prospective evidence for the long-term association between depression and ventricular arrhythmias in ICD patients and is strengthened by the size of the TOVA study sample and by the involvement of patients from multiple centers in the U.S.
In our study, depression conferred an elevated risk for ICD shocks among the subgroup of patients with a history of CAD. Previous studies have shown that depressed post-myocardial infarction patients have higher risks for mortality (2,9), mortality or cardiac arrest (6), and sudden death or sustained VT (7). Our analysis is also consistent with the study by Carney et al. (8) in which depression was associated with VT episodes detected by 24-h Holter monitoring among patients with CAD, even after adjustment for differences in beta-blocker use. In the context of these previous investigations, our results indicate that an arrhythmic mechanism may account for a significant part of the elevated mortality risk associated with depression among patients with CAD.
Although our analysis indicates that depression may predict the development of malignant ventricular arrhythmias, to determine whether depression is causative will require confirmatory studies, as well as studies of the mechanism of the arrhythmic risk. Reduced heart rate variability (HRV) is more common among depressed patients after acute myocardial infarction (22). A rodent model of depression, manifested as anhedonia in response to chronic mild stress, has revealed decreased HRV and shorter time to onset of ventricular arrhythmias in response to aconitine infusion (23). Depression has also been associated with reductions in red blood cell levels of omega-3 fatty acids (24), which are related to the risk of sudden death (25,26). In addition, depression has been linked to higher levels of platelet activation (27) and to enhanced activity of the clotting cascade (28).
Possible interventions to reduce the arrhythmic risk associated with depression could involve numerous approaches, from treatment of the “upstream” mood disorder to stabilization with anti-arrhythmic therapy. Treatment of patients with psychotropic medications has shown mixed results. An uncontrolled pilot study of paroxetine in the treatment of five ICD patients with panic disorder, agoraphobia, and depression found that the rate of ectopy and ICD therapies decreased after six months of treatment (29). The Sertraline Antidepressant Heart Attack Randomized Trial study did not detect a significant difference in ventricular arrhythmias by 24-h Holter monitor among 369 patients with acute coronary syndrome and depression randomized to sertraline or placebo, although the study was not powered for this outcome (30). An analysis of the Canadian Amiodarone Myocardial Infarction Arrhythmia Trial found that symptoms of depression were associated with sudden cardiac death among placebo-treated patients but not among amiodarone-treated patients, suggesting a possible protective effect from amiodarone (5). Understanding of the underlying pathways linking depression and arrhythmia may further guide the search for an effective treatment.
Our finding that episodes of two or more appropriate shocks within a 24-h period were more common among depressed patients raises possible implications for device therapy. A number of studies have shown that shocks worsen symptoms of depression and psychologic well-being in ICD patients (31–34). Therefore, depression may increase the susceptibility to ventricular arrhythmias that require ICD shocks, which in turn worsen depressive symptoms and lead to further malignant arrhythmias. If this were the case, anti-tachycardia pacing could interrupt this cycle by reducing the frequency of shocks. Recent data from the PainFREE Rx II Trial that support the use of anti-tachycardia pacing for fast VT events may be particularly relevant for shock prevention in depressed patients.
Finally, it is notable that about 18% of the patients in our sample were classified as mildly to severely depressed according to their CES-D score, a prevalence estimate that corresponds to other cardiovascular disease samples, yet only about 10% of these patients were treated with SSRIs. Although our estimate of the use of SSRIs may be subject to underreporting, it is likely that depression was undertreated in our sample. The relatively low number of ICD patients that received pharmacologic treatment for depression suggests that increased surveillance for depression in ICD patients may be needed. Primary care studies have suggested that inquiring about depressed mood and anhedonia have suitable sensitivity and specificity for detecting depression in medical settings (35). More aggressive surveillance and treatment of mood disorders with one or both of an SSRI and counseling may significantly affect the quality of life of ICD patients (36).
There are several limitations to our study. We did not assess symptoms of depression in the entire TOVA study cohort, because the CES-D scale was introduced to the study after enrollment had already begun. Although this limits the power of our study, we were able to detect significant relationships between appropriate ICD shocks and symptoms of depression in multivariable analyses. In addition, we used symptoms of depression assessed at baseline for our analysis; thus, we cannot evaluate whether changes in symptoms, which may have additional prognostic information (37), alter the risk of ICD shocks for VT/VF. Also, although we attempted to adjust for as many potential confounding factors as possible in our multivariable analyses, there may be unmeasured confounders that explain the relationship between depression and shocks in our analysis.
In summary, we found that symptoms of depression are predictive of shock-treated ventricular arrhythmias among a sample of ICD patients. These data support ventricular arrhythmia as a potential mechanism for the elevated mortality risk among depressed patients with CAD. Determining the optimal therapy to reduce this risk remains an important avenue of future investigation.
The authors thank Natalya Gomelskaya for expert statistical assistance. The authors are indebted to Beth Kelly, Jennifer Merlan, Jane Sherwood, and the patients who participated in the TOVA study for their outstanding commitment and cooperation, and to the study staff and investigators at the clinical sites.
For a list of the TOVA study investigators, please see the April 5, 2005, issue of JACCat www.onlinejacc.org.
This study was supported by funding from the National Heart, Lung, and Blood Institute (HL041016); Guidant Foundation; and Guidant Cardiac Rhythm Management.
- Abbreviations and acronyms
- angiotensin-converting enzyme
- angiotensin receptor blocker
- Centers for Epidemiologic Studies-Depression
- heart rate variability
- implantable cardioverter-defibrillator
- selective serotonin reuptake inhibitor
- ventricular tachycardia/ventricular fibrillation
- Received September 1, 2004.
- Revision received November 22, 2004.
- Accepted December 14, 2004.
- American College of Cardiology Foundation
- Kennedy G.J.,
- Hofer M.A.,
- Cohen D.,
- Cohen D.,
- Shindledecker R.,
- Fisher J.D.
- Luukinen H.,
- Laippala P.,
- Huikuri H.V.
- Irvine J.,
- Basinski A.,
- Baker B.,
- et al.
- Ladwig K.H.,
- Kieser M.,
- Konig J.,
- Breithardt G.,
- Borggrefe M.
- Frasure-Smith N.,
- Lesperance F.,
- Talajic M.
- Herrmann C.,
- Bergmann G.,
- Drinkmann A.,
- et al.
- Whang W.,
- Mittleman M.A.,
- Rich D.Q.,
- et al.
- Gregoratos G.,
- Cheitlin M.D.,
- Conill A.,
- et al.
- Raloff L.S.
- Weissman M.M.,
- Sholomskas D.,
- Pottenger M.,
- Prusoff B.A.,
- Locke B.Z.
- Kowey P.R.,
- Levine J.H.,
- Herre J.M.,
- et al.
- Scheinman M.M.,
- Levine J.H.,
- Cannom D.S.,
- et al.
- Carney R.M.,
- Blumenthal J.A.,
- Stein P.K.,
- et al.
- Grippo A.J.,
- Santos C.M.,
- Johnson R.F.,
- et al.
- Tofler G.H.,
- Massaro J.,
- Kelly-Hayes M.,
- et al.
- Schron E.B.,
- Exner D.V.,
- Yao Q.,
- et al.