- Email: [email protected]
The Effect of Comorbidity on the Competing Risk of Sudden and Nonsudden Death in an Ambulatory Heart Failure Population
Canadian Journal of Cardiology 27 (2011) 254 –261
Clinical Research
The Effect of Comorbidity on the Competing Risk of Sudden and Nonsudden Death in an Ambulatory Heart Failure Population Brian Clarke, MD,a Jonathan Howlett, MD,b John Sapp MD,a Pantelis Andreou, MSc, PhD,c and Ratika Parkash, MD, MSc, FRCPCa a
Department of Medicine, Division of Cardiology, Dalhousie University and the Queen Elizabeth II Health Science Centre, Halifax, Nova Scotia, Canada b c
Division of Cardiology, University of Calgary, Foothills Medical Centre, Calgary, Alberta, Canada
Department of Community Health and Epidemiology, Faculty of Medicine, Dalhousie University, Nova Scotia, Canada
ABSTRACT
RÉSUMÉ
Background: Sudden death (SD) and non–sudden cardiac death are responsible for the majority of deaths in patients with heart failure. We sought to identify the influence of comorbid illness (Charlson Comorbidity Index [CCI]) on competing modes of death in heart failure. Methods: A retrospective analysis of 824 patients followed in a tertiary care heart failure clinic was performed. We analyzed the cumulative incidence of sudden and nonsudden death. Competing risk regression was used to examine the association between medical comorbidities and mode of death. The outcomes of interest were overall mortality, SD, SD and/or appropriate implantable cardioverterdefibrillator therapy (ICD), and non-SD. Results: Mean age of the study population was 64.1 ⫾ 14.7 years, 68.6% were male, and mean ejection fraction was 32.8% ⫾ 13.5%. Over a mean follow-up of 4.4 years, 229 patients (27.8%) died. SD accounted for 33 deaths (14.4%), whereas SD/appropriate ICD therapy occurred in 56 patients (24.5%). The risk of non-SD and total mortality increased (P ⬍ .0001) as the CCI increased, whereas the risk of SD decreased (P ⫽ .03). The cumulative incidence of SD, SD and/or ventricular tachycardia/fibrillation, and non-SD at 5 years was 5.6%, 9.1%, and 27.8%, respectively. In multivariate competing risk analysis, advancing age, New York Heart Association class, and a CCI ⬎4 were significantly associated with non-SD.
Introduction : La mort subite (MS) et la mort cardiaque non subite sont responsables de la majorité des décès chez les patients ayant une insuffisance cardiaque. Nous avons cherché à déterminer l’influence de la comorbidité de la maladie (indice de comorbidité de Charlson [ICC]) sur les causes concurrentes de mortalité dans l’insuffisance cardiaque. Méthodes : Une analyse rétrospective a été réalisée auprès de 824 patients suivis dans une clinique de soins tertiaires pour une insuffisance cardiaque. Nous avons analysé l’incidence cumulative de la mort subite et non subite. La régression des risques concurrents a été utilisée pour examiner le lien entre les comorbidités médicales et les causes de décès. Les résultats recherchés étaient la mortalité globale, la MS, la MS et/ou le traitement par défibrillateur automatique implantable approprié (DAI), et la mort non subite. Résultats : L’âge moyen de la population étudiée était de 64,1 ⫾ 14,7 ans, dont 68,6 % était des hommes, et la fraction d’éjection moyenne était de 32,8 % ⫾ 13,5 %. Sur un suivi moyen de 4,4 ans, 229 patients (27,8 %) sont morts. La MS a représenté 33 décès (14,4 %), alors que la MS et le traitement avec un DAI approprié sont survenus chez 56 patients (24,5 %). Le risque de mort non subite et la mortalité totale a augmenté (P ⬍ 0,0001) comme le ICC augmentait, alors que le risque de mort subite a diminué (P ⫽ 0,03). L’incidence cumulative de la mort subite, la mort subite et/ou la tachycardie/fibrillation ventriculaire, et la mort
Heart failure is a major public health concern, with prevalence rates of 1%-2% in the general population under the age of 65 years and of ⬎10% of people over the age of 65. The incidence ranges
from 10-16 per 1000 population annually when adjusted for age.1 Mortality in the heart failure population is predominantly due to either sudden cardiac death (SD) or progressive heart failure, with SD accounting for 20%-50% of all deaths.2-5 Prophylactic implantable cardioverter-defibrillator therapy (ICD) therapy has been shown to reduce overall mortality via a reduction in arrhythmic deaths in patients with heart failure with severe systolic dysfunction.6,7 However, the accurate prediction of mode of death in heart failure using clinically relevant information beyond ejection fraction (EF) and New York Heart Association (NYHA) functional class has yet to be achieved.
Received for publication September 28, 2010. Accepted November 23, 2010. Corresponding author: Dr Ratika Parkash, Dalhousie University, Room XX, HI Site, QEII Health Sciences Centre, 1796 Summer St, Halifax, Nova Scotia, B3H 3A7, Canada. Tel.: ⫹1 902-473-4474; fax.: ⫹1 902-473-3158. E-mail: [email protected] See page 260 for disclosure information.
0828-282X/$ – see front matter © 2011 Canadian Cardiovascular Society. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.cjca.2010.12.053
Clarke et al. Comorbidity and SCD in Ambulatory HF Population
255
Conclusion: Patients with heart failure with significant comorbidities are much more likely to sustain non-SD. These findings may have implications in optimal selection of patients with heart failure for interventions such as prophylactic ICD therapy.
non subite après 5 ans, était de 5,6 %, 9,1 %, et 27,8 %, respectivement. Dans l’analyse du risque concurrent multifactorielle, l’âge avancé, la classification selon la New York Heart Association et un ICC ⬎ 4 étaient significativement associés à la mort non subite. Conclusion : Les patients ayant une insuffisance cardiaque avec des comorbidités importantes sont probablement plus à risque d’une mort non subite. Ces découvertes peuvent avoir des implications dans la sélection optimale des patients ayant une insuffisance cardiaque pour des interventions comme le traitement par DAI prophylactique.
There is significant interest in additional risk stratification of patients with heart failure to identify those who may derive the most benefit and, more important, to identify those who are least likely to derive benefit from prophylactic ICD therapy, avoiding the unnecessary risks and complications from device implantation in those whose risk of SD is far exceeded by other competing causes of death.8,9 We used competing risk methodology to estimate the probability of SD, SD and/or ventricular tachycardia/fibrillation (VT/VF), and non-SD in an ambulatory heart failure patient population. Furthermore, we aimed to examine the effect of medical comorbidity, as quantified by the Charlson Comorbidity Index (CCI), on mode of death.
obtained through an electronic database and detailed review of medical records (performed by B.C.). The baseline data were collected from information available as of the initial visit to the heart failure clinic. Deaths were documented in the database and mode of death was ascertained through extensive review of inpatient and outpatient medical records and classified as SD or non-SD by 2 independent physicians (B.C., J.H.), as defined later. In 14 cases where modality of death was uncertain, death certificates were obtained from Nova Scotia Vital Statistics in cooperation with the Population Health Research Unit of the Department of Health. Information from death certificates was corroborated with family physicians. None of these deaths were SDs. The prime determinant in classifying mode of death as sudden or nonsudden in these
Methods Table 2. Baseline characteristics of the study population
Study population This was a retrospective study of 824 consecutive patients followed at a tertiary care specialty ambulatory heart failure clinic in Halifax, NS, Canada, from December 1998 through December 2004. There were no prespecified criteria for referral to the heart failure clinic, other than at least one episode of clinical heart failure. Patients could be referred by family physicians, general internists, or other specialists. Data collection Approval from the institution’s research ethics board was obtained. Demographic, clinical, and laboratory information was Table 1. The Charlson comorbidity index Assigned weights for diseases 1
2
3 6
Conditions Myocardial infarct Congestive heart failure Peripheral vascular disease Cerebrovascular disease Dementia Chronic obstructive pulmonary disease Connective tissue disease Peptic ulcer disease Mild liver disease Diabetes Hemiplegia Moderate or severe renal disease (creatinine ⬎230 mmol/L) Diabetes with end organ damage Any tumor Leukemia Lymphoma Moderate or severe liver disease Metastatic solid tumor AIDS
Variable Age, y Male, n NYHA class, n I II III IV Ejection fraction, % QRS duration, ms ⱖ1 hospitalization for CHF Atrial fibrillation, n Ischemic heart disease, n Diabetes, n Diabetes with end organ damage, n Valvular disease, n Hypertension, n Dyslipidemia, n Peripheral vascular disease, n Cerebrovascular disease, n Chronic pulmonary disease, n Connective tissue disease, n Mild liver disease, n Moderate/severe liver disease, n Hemiplegia, n Moderate/Severe renal disease, n Any tumor, n Leukemia, n Lymphoma, n Metastatic solid tumor, n AIDS, n Creatinine, mmol/L ACE inhibitor, n Beta-blocker, n Diuretic, n Spironolactone, n Statin, n
Total population (N ⫽ 824) 64.2 ⫾ 14.7 565 (68.6%) 89 (10.8%) 237 (28.8%) 438 (53.1%) 60 (7.3%) 32.8 ⫾ 13.5 123.8 ⫾ 36.8 261 (31.7%) 259 (31.4%) 490 (59.5%) 272 (33.0%) 130 (47.8%) 404 (49.1%) 448 (54.4%) 436 (52.9%) 102 (12.4%) 129 (15.7%) 178 (21.6%) 13 (1.6%) 14 (1.7%) 3 (0.4%) 0 47 (5.7%) 77 (9.3%) 1 (⬍1%) 8 (1%) 6 (0.7%) 0 129.3 ⫾ 71.4 621 (75.9%) 532 (65.0%) 658 (80.4%) 192 (23.5%) 324 (39.6%)
ACE, Angiotensin-converting enzyme; CHF, congestive heart failure; NYHA, New York Heart Association.
256
Canadian Journal of Cardiology Volume 27 2011
Figure 1. Distribution of Charlson Comorbidity Index in the study population.
14 cases rested with clinical information provided by family physicians. Charlson Comorbidity Index (CCI) values were derived for each patient based on chart review and assigned according to baseline clinical information. The CCI is a weighted numeric scoring system designed to reflect the magnitude of comorbidity10 (Table 1). It was developed empirically, based on 1-year mortality from an inception cohort study in 1984, and has since been extensively validated in a variety of populations to predict survival.10-14 Deaths were classified as SD or non-SD. SD was defined as death within 1 hour of change of symptoms or during sleep or if death was unobserved, as has been previously defined.15. Non-SDs included all other deaths. Patients with ICDs were reviewed for number and types of therapies delivered. All ICD events were reviewed and adjudicated for appropriateness by 2 independent electrophysiologists. Appropriate therapies were defined as antitachycardia pacing (ATP) for VT or shocks delivered for VT or VF. Inappropriate therapies were defined as ATP or shocks delivered for rhythms other than VT/VF (ie, sinus tachycardia, atrial fibrillation). The outcomes included overall mortality, SD, a composite of SD or appropriate ICD therapy for VT/VF, and non-SD.
Figure 3. The association between mode of cardiac death and total mortality with the Charlson Comorbidity Index (CCI) in heart failure. White bars: Mortality from sudden death (SD). The P-value for trend was .03 as the CCI increased. Gray bars: Mortality from non-SD. The P-value for trend was ⬍.001 as the CCI increased. Black bars: Total mortality. The P-value for trend was ⬍.001 as the CCI increased.
Statistical analysis Baseline variables between groups were compared using Student’s t test for continuous variables and chi-square test for categorical variables. Univariate Cox proportional hazards models were constructed to identify those variables that significantly predicted overall mortality. The variables included in the multivariate Cox proportional hazards model were chosen from those variables that had a statistical significance of P ⬍ .10 on univariate analysis. The following covariates were included in the model: age, EF, NYHA class, dyslipidemia, the CCI, and the presence of an ICD. The proportional hazard assumption was examined for all covariates. We used cumulative incidence functions to display the proportion of patients with the event of interest or the competing event as time progressed.16 We used a multivariate competing risk model to analyze the effect of baseline predictors on the cumulative incidence function.17 A single analysis of the 2 cause-specific hazards was performed. The SAS macros, CumInc and CumIncv, developed by Rosthoj et al.18 were used for estimation of the cumulative incidence functions. These macros are based on multistate models and transition probabilities as developed by Anderson et al.19 All analyses were performed using SAS STAT software, version 9.2. A 2-sided
Table 3. Cox multivariate regression model for overall mortality
Figure 2. Distribution of mode of death (N ⫽ 229). CHF, Congestive heart failure; SD, sudden death.
Variable
HR
95% CI
P value
CCI NYHA III/IV Age Dyslipidemia ICD
1.26 2.50 1.04 1.39 1.86
1.19-1.35 1.80-3.46 1.03-1.05 1.06-1.81 1.01-3.42
⬍.0001 ⬍.0001 ⬍.0001 ⬍.016 ⬍.046
CCI, Charlson Comorbidity Index; CI, confidence interval; HR, hazard ratio; ICD, implantable cardioverter-defibrillator; NYHA, New York Heart Association.
Clarke et al. Comorbidity and SCD in Ambulatory HF Population
257
Figure 4. Cumulative incidence function for sudden death (SD), SD and/or ventricular tachycardia/fibrillation (VT/VF), and non-SD.
P value of ⬍.05 was taken as statistically significant for all analyses. Results Patient characteristics Baseline characteristics of the study cohort are shown in Table 2. Mean age was 64.1 ⫾ 14.7 years, 68.6% were male, mean EF was 32.8% ⫾ 13.5% (21% had EF ⱕ20%, 43% had EF 21%-35%, 26% had EF 36%-50%, 10% had EF ⬎50%), 60.4% were NYHA functional class III/IV, 59.5% had coronary artery disease, and 9.8% had an ICD. Patients were followed for a mean of 4.4 years; 31.7% had at least one rehospitalization for heart failure during the follow-up period. The distribution of the CCI for the study population is shown in Figure 1. Relationship between mode of death, ICD therapies, and the CCI Over a mean follow-up period of 4.4 years, 229 patients (27.8%) died. Of these, SD occurred in 33 patients (14.4%)
(Fig. 2). Twenty-three patients (28.4%) received at least one appropriate therapy for VT/VF. The risk of total mortality and non-SD was associated with an increase in the CCI (P ⬍ .0001), whereas the risk of SD decreased with increasing CCI (P ⫽ .03) (Fig. 3). In a Cox regression analysis, predictors of mortality included increasing CCI, age, NYHA class III/IV, dyslipidemia, and the absence of an ICD (Table 3). The 5-year cumulative incidence for SD, SD and/or VT/VF, and non-SD was 5.6%, 9.1%, and 27.8%, respectively (Fig. 4). A CCI cutoff of 4 was chosen based on the ROC curve. In multivariate competing risk analysis, advancing age (by 10-year increments), a CCI ⬎4, and advancing NYHA class were significant predictors of non-SD (Table 4). When appropriate ICD therapy for VT/VF was included in the analysis, there was a trend toward better freedom from SD or appropriate ICD therapy with increasing EF (HR 0.98, P ⫽ .057 for every 10% increase in EF; Table 5). No variables were found to significantly predict SD. The effect of the CCI on cumulative incidence is shown in Figure 5 and Table 6. Patients with a CCI ⬎4 had a significantly higher risk of non-SD with a 5-year cu-
Table 4. Multivariate competing risk regression: SD versus non-SD
Table 5. Multivariate competing risk regression: composite outcome of SD and VT/VF versus non-SD
Variable
Variable
Non-SD Age EF CCI ⬎ 4 NYHA class SD Age EF CCI ⬎ 4 NYHA class
HR
P value
95% CI
1.05 0.99 2.78 2.89
⬍.0001 .07 ⬍.0001 ⬍.0001
1.03-1.06 0.97-1.00 2.09-3.70 1.99-4.21
1.01 0.99 0.49 1.27
.68 .54 .15 .52
0.98-1.03 0.97-1.02 0.19-1.28 0.62-2.64
CCI, Charlson Comorbidity Index; CI, confidence interval; HR, hazard ratio; ICD, implantable cardioverter-defibrillator; NYHA, New York Heart Association; SD, sudden death.
Non-SD Age EF CCI ⬎ 4 NYHA class (I/II vs III/IV) SD/VT/VF Age EF CCI ⬎ 4 NYHA class (I/II vs III/IV)
HR
P value
95% CI
1.05 0.99 0.36 0.34
⬍.0001 .058 ⬍.0001 ⬍.0001
1.03-1.06 0.98-1.00 0.27-0.48 0.23-0.50
1.01 0.99 1.44 0.87
.30 .23 .28 .62
0.99-1.03 0.97-1.01 0.74-2.81 0.50-1.51
CCI, Charlson Comorbidity Index; CI, confidence interval; EF, ejection fraction; ICD, implantable cardioverter-defibrillator; NYHA, New York Heart Association; SD, sudden death; VT/VF, ventricular tachycardia/fibrillation.
258
Canadian Journal of Cardiology Volume 27 2011
Figure 5. Cumulative incidence function for sudden death (SD), SD and/or ventricular tachycardia/fibrillation (VT/VF), and non-SD by Charlson Comorbidity Index (CCI). A: CCI ⬎ 4. B: CCI ⱕ 4.
mulative incidence of 48.8% compared to 33.5% with a CCI ⱕ4 (P ⬍ .0001). Discussion In this study of ambulatory patients with heart failure, we report the relationship between underlying comorbid illness, as measured by the CCI, on modes of death. We used a competing risk analysis to evaluate the effect of comorbidity and other covariates on outcomes of SD, SD and/or VT/VF, and non-SD in patients with heart failure. Numerous studies have demonstrated the association between underlying medical comorbidities and mortality in a variety of patient populations using the CCI.10-14 In general, these
studies noted an increase in overall mortality with increasing comorbidity. This is the first study to use the CCI in the heart failure population. Our principal findings demonstrate the absolute risk of SD remains relatively constant with increasing CCI, while the risk of non-SD increases with increasing comorbidity burden. Thus, patients with multiple comorbidities have a significantly higher relative risk of non-SD. Results from recent clinical trials have established mortality benefit with ICD therapy in ischemic and nonischemic cardiomyopathy with reduced EF.6,7 However, these trials include a select group of patients with heart failure. Our study uses a real-world cohort of ambulatory patients
Clarke et al. Comorbidity and SCD in Ambulatory HF Population
259
Table 6. Cumulative incidence functions of competing risk model Competing risk model 5-Year cumulative incidence Variable CCI ⱕ4 ⬎4 NYHA class I/II III/IV EF class ⱕ30% ⬎30% Overall cumulative risk
SD
SD/VT death
Non-SD
0.0159 (0.00905, 0.0228) 0.0232 (0.0030, 0.0434)
0.0256 (0.0324, 0.0189) 0.0494 (0.0686, 0.0302)
0.335 (0.0703, 0.599) 0.488 (0.419, 0.557)
0.0124 (0.00519, 0.0197) 0.0457 (0.0243, 0.0671)
0.0214 (0.0277, 0.0152) 0.0732 (0.0921, 0.0544)
0.102 (0.0249, 0.178) 0.372 (0.323, 0.421)
0.0528 (0.0323, 0.0814) 0.0399 (0.0183, 0.0615) 0.0563
0.0922 (0.0692, 0.115) 0.0637 (0.0453, 0.0820) 0.0908
0.371 (0.1640, 0.578) 0.259 (0.211, 0.307) 0.278
CCI, Charlson Comorbidity Index; CI, confidence interval; EF, ejection fraction; NYHA, New York Heart Association; SD, sudden death.
with heart failure. Toma et al.20 applied SCD-HeFT criteria to a real-world population of patients with heart failure and found that just over half of all patients seen in a specialized heart failure clinic would qualify for ICDs. Many of these patients may not benefit due to competing risks of death, as seen in this study, identifying the need for additional stratification schemes beyond NYHA class and EF, when considering patients for ICD therapy for primary prevention of SD. ICD shocks (both appropriate and inappropriate) have recently been found to be associated with an increase in mortality, raising the possibility that ICD therapy may not be such a benign intervention.21-23 An analysis of SCDHeFT demonstrated that both appropriate and inappropriate shocks increased the risk of death.21 Of the 269 patients who received shocks in this study, 141 patients (52.4%) received at least one inappropriate shock and 87 patients (32.3%) received only inappropriate shocks. Furthermore, Koller et al.24 noted that 11%-23% of patients with heart failure with ICDs died without prior ICD therapy and 36% of patients remained alive and never used the device. Predicting mode of death is important as there are some therapies that modify the risk of SD. Reduction of non-SD may be achieved with aggressive conventional medical therapy (beta-blockers, angiotensin-converting enzyme [ACE] inhibitors/angiotensin II receptor blocker [ARB], spironolactone) and cardiac resynchronization therapy (CRT), whereas prevention of arrhythmic SD rests with ICD therapy in addition to conventional medical therapy. This therapy is invasive and has its inherent procedural and long-term risks including device advisories25 and complications at the time of implant as well as with replacement.26,27 Patients with higher comorbidity burden have higher rates of procedural complications, as demonstrated in the REPLACE registry, where the complication rate for ICD generator replacement was 13.1% for those patients with a CCI ⱖ2 versus 8.5% in those with a CCI ⬍2.26 The majority of the ambulatory heart failure population in this study (86%) are in this higher risk category. Recent evidence illustrates that much prognostic information can be gained from easily determined historical factors, such as age and other medical comorbidities.28 In an observational study of a large cohort of real-world ICD recipients, competing noncardiac comorbidities such as renal failure, chronic pulmonary disease, peripheral vascular disease, and diabetes were associated with increased mortality after ICD implantation.29 Braunstein et al.30 evaluated the effect of comor-
bid illness on rehospitalization rates and overall mortality in a cohort of U.S. Medicare beneficiaries with chronic heart failure and found that the risk of hospitalizations and potentially preventable hospitalizations strongly increased with the number of chronic conditions. Chronic obstructive pulmonary disease, renal failure, diabetes, depression, and lower respiratory diseases conferred a notably higher risk. Several observations from this study deserve special attention. The SD rate that was observed in this cohort of patients is much lower than reported in clinical trials; however, the observed overall mortality rate is similar to contemporary clinical trials.3 SD accounted for 64% and 59% of total mortality in NYHA II and III patients with heart failure, respectively, in MERIT-HF, in contrast to 14.4% of patients in this study. This may be due to several factors. Our cohort is representative of a mixed population of patients with heart failure, with a broader spectrum of EF values, more patients with NYHA class I and IV heart failure, and comorbidities that would have precluded entry into randomized trials. In addition, 81 patients had an ICD. When appropriate therapy for VT/VF was considered as a surrogate for SD, as has been done in other studies, 25% of patients were considered to have SD,31 recognizing, however, that appropriate ICD shocks may occur more frequently than SD and may overestimate the actuarial SD rate.32 In addition, MERIT-HF used a wider definition of SD, including patients who died within 28 days of resuscitated cardiac arrest. Patients were excluded if the EF was ⱖ40% and mean EF was 28%, in contrast to this study, where there were no exclusion criteria based on EF and mean EF was higher at 32.8%. There are some limitations to this retrospective study that deserve mention. We observed an ICD implantation rate of approximately 10% in this population. This can partially be accounted by the effective study period. The major clinical trials establishing the efficacy of prophylactic ICD therapy in heart failure were released in the early 2000s; since then, the indications have expanded significantly. This study included patients from 1998-2004, who at that time may not have been candidates for ICDs but now may qualify for ICD therapy according to current guidelines. Finally, the population studied was restricted to patients followed in an ambulatory heart failure clinic, which may limit applicability to the population as a whole. Whether this association holds in hospitalized patients with heart failure has yet to be determined.
260
We used standard definitions in classifying mode of death; however, there remains a risk of misclassification. To minimize this risk, deaths documented in medical records were independently reviewed by 2 physicians, and in cases where mode of death was uncertain, clinical details from primary care physicians were used to corroborate cause of death as listed on death certificate information. The clinical information used to calculate the CCI was the information that was available at the time of analysis. There is a small risk of underscoring the CCI if, for example, a patient with diabetes has unrecognized complications such as retinopathy. Our findings suggest a novel method in which to apply the CCI to ambulatory patients with heart failure to assist in determining the most appropriate medical and device-based therapies. The findings from this study could be applied prospectively to an ICD population to see if it predicted the occurrence of ventricular arrhythmia as a surrogate for SD, mindful of the fact that VT/VF is a surrogate for this. Conclusion The relationship between risks of SD and non-SD in the ambulatory heart failure is complex. Our findings demonstrate that the relative risk of each of these competing modes of death vary significantly with the CCI. The magnitude of benefit of invasive therapies to prevent SD, such as ICDs, may differ in ambulatory patients with heart failure based on the degree of comorbid illness. This has important implications for using this kind of invasive therapy, which carries a significant risk of complications in patients with a higher CCI. Patients with a high CCI and advanced heart failure may have very little benefit from an invasive therapy such as an ICD. This may have implications on how this resource should be allocated in the ever-expanding heart failure population. Acknowledgements The authors thank Kara Thompson for providing statistical support. Funding Sources This study was funded by a grant from the University of Internal Medicine Research Foundation. Disclosures The authors have no conflicts of interest to disclose. References 1. McMurray J, Stewart S: Heart failure: epidemiology, aetiology, and prognosis of heart failure. Heart 2000;83:596-602. 2. Rami T, Hih J: Update of implantable cardioverter/defibrillator and cardiac resynchronization therapy in heart failure. Curr Opin Cardiol 2004;19: 264-9. 3. MERIT-HF Study Group: Effect of metoprolol CR/XL in chronic heart failure: Metoprolol CR/XL Randomised Intervention Trial in Congestive Heart Failure (MERIT-HF). Lancet 1999;353:2001-7. 4. The Captopril-Digoxin Multicentre Research Group: Comparative effects of therapy with captopril and digoxin in mild to moderate heart failure. JAMA 1988;259:539-44. 5. Zheng ZJ, Croft JB, Giles WH, et al. Sudden cardiac death in the United States 1989 to 1998. Circulation 2001;104:2158-63.
Canadian Journal of Cardiology Volume 27 2011 6. Moss AJ, Zareba W, Hall WJ et al. Prophylactic implantation of defibrillator in patients with myocardial infarction and reduced left ventricular ejection fraction. N Engl J Med 2002;346:877-83. 7. Bardy GH, Lee KL, Mark DB, et al. Amiodarone or an implantable cardioverter-defibrillator for congestive heart failure. N Engl J Med 2005; 352:225-37. 8. Josephson M, Wellens H: Implantable defibrillators and sudden cardiac death. Circulation 2004;109:2685-91. 9. Warner Stevenson L and Desai A: Selecting patients for discussion of the ICD as primary prevention for sudden death in heart failure. J Card Fail 2006;12:407-12. 10. Charlson M, Pompei P, Ales K, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis 1987;40:373-83. 11. Rochon PA, Katz JN, Morrow LA, et al. Comorbid illness is associated with survivalandlengthofhospitalstayinpatientswithchronicdisability.Aprospective comparison of three comorbidity indices. Med Care 1996;34:1093-101. 12. West DW, Satariano WA, Ragaland DR, et al. Comorbidity and breast cancer survival: a comparison between black and white women. Ann Epidemiol 1996;6:413-9. 13. Skiest DJ, Rubinstien E, Carley N, et al. The importance of comorbidity in HIV-infected patients over 55: a retrospective case– control study. Am J Med 1996;101:605-11. 14. Poses RM, McClish DK, Smith WR, et al. Prediction of survival of critically ill patients by admission comorbidity. J Clin Epidemiol 1996;49:743-7. 15. Kearney MT, Fox KAA, Lee AJ, et al. Predicting death due to progressive heart failure in patients with mild-to-moderate chronic heart failure. J Am Coll Cardiol 2002;40:1801-8. 16. Putter H, Fiocco M, Geskus RB. Tutorial in biostatistics: competing risks and multi-state models. Stat Med 2007;26:2389-430. 17. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of acompeting risk. J Am Stat Assoc 1999;94:496-50. 18. Rosthej S, Andersen PK, Abildstrom SZ. SAS macros for estimation of the cumulative incidence functions based on a Cox regression model for competing risks survival data. Comp Meth Progr Biomed 2004;74:69-75. 19. Anderson PK, Bogan Ø, Gill RD, et al. Statistical Models Based on Counting Processes. New York: Springer, 1993. 20. Toma M, McAlister FA, Ezekowitz J, et al. Proportion of patients followed in a specialized heart failure clinic needing an implantable cardioverter defibrillator as determined by applying different trial eligibility criteria. Am J Cardiol 2006;97:882-5. 21. Poole JE, Johnson GW, Hellkamp AS, et al. Prognostic importance of defibrillator shocks in patients with heart failure. N Engl J Med 2008;359:1009-17. 22. Sweeney MO, Sherfesee L, DeGroot PJ, et al. Differences in effects of electrical therapy type for ventricular arrhythmias on mortality in implantable cardioverter-defibrillator patients. Heart Rhythm 2010;7:353-60. 23. El-Damaty A, Sapp J, Basta M, et al. Inappropriate ICD shocks: clinical risk factors and mortality. Heart Rhythm 2010;7:s186-7. 24. Koller MT, Schaer B, Wolbers M, et al. Death without prior appropriate implantable cardioverter-defibrillator therapy: a competing risk study. Circulation 2008;117:1918-26. 25. Gould PA, Krahn AD; Canadian Heart Rhythm Society Working Group on Device Advisories. Complications associated with implantable cardioverter-defibrillator replacement in response to device advisories. JAMA 2006;295:1944-6.
Clarke et al. Comorbidity and SCD in Ambulatory HF Population 26. Gleva MJ, Holcomb R, Uslan D, et al. Charlson Comorbidity Index and risk of cardiac rhythm device generator replacement: results from the REPLACE Registry. Circulation 2009;120:S637 Abstract 2435. 27. Ravindran BK, Poole JE, Johnson GW, et al. Timing of device-specific ICD complications: analysis of the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT). Heart Rhythm 2010;7:s9. 28. Buxton AE, Lee KL, Hafley GE, et al. Limitations of ejection fraction for prediction of sudden death risk in patients with coronary artery disease. J Am Coll Cardiol 2007;50:1150-7. 29. Lee DS, Tu JV, Austin PC, et al.: Effect of cardiac and noncardiac conditions on survival after defibrillator implantation. J Am Coll Cardiol 2007;49:2408-15.
261 30. 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;42:1226-33. 31. Gold MR, Ip JH, Costantini O, et al. Role of microvolt T-wave alternans in assessment of arrhythmia vulnerability among patients with heart failure and systolic dysfunction. Circulation 2008;118:2022-8. 32. Ellenbogen, KA, Levine JH, Berger RD, et al; for the Defibrillators in Non-Ischemic Cardiomyopathy Treatment Evaluation (DEFINITE) Investigators. Are implantable cardioverter defibrillator shocks a surrogate for sudden cardiac death in patients with nonischemic cardiomyopathy?. Circulation 2006;113:776-82.
Clinical Research
The Effect of Comorbidity on the Competing Risk of Sudden and Nonsudden Death in an Ambulatory Heart Failure Population Brian Clarke, MD,a Jonathan Howlett, MD,b John Sapp MD,a Pantelis Andreou, MSc, PhD,c and Ratika Parkash, MD, MSc, FRCPCa a
Department of Medicine, Division of Cardiology, Dalhousie University and the Queen Elizabeth II Health Science Centre, Halifax, Nova Scotia, Canada b c
Division of Cardiology, University of Calgary, Foothills Medical Centre, Calgary, Alberta, Canada
Department of Community Health and Epidemiology, Faculty of Medicine, Dalhousie University, Nova Scotia, Canada
ABSTRACT
RÉSUMÉ
Background: Sudden death (SD) and non–sudden cardiac death are responsible for the majority of deaths in patients with heart failure. We sought to identify the influence of comorbid illness (Charlson Comorbidity Index [CCI]) on competing modes of death in heart failure. Methods: A retrospective analysis of 824 patients followed in a tertiary care heart failure clinic was performed. We analyzed the cumulative incidence of sudden and nonsudden death. Competing risk regression was used to examine the association between medical comorbidities and mode of death. The outcomes of interest were overall mortality, SD, SD and/or appropriate implantable cardioverterdefibrillator therapy (ICD), and non-SD. Results: Mean age of the study population was 64.1 ⫾ 14.7 years, 68.6% were male, and mean ejection fraction was 32.8% ⫾ 13.5%. Over a mean follow-up of 4.4 years, 229 patients (27.8%) died. SD accounted for 33 deaths (14.4%), whereas SD/appropriate ICD therapy occurred in 56 patients (24.5%). The risk of non-SD and total mortality increased (P ⬍ .0001) as the CCI increased, whereas the risk of SD decreased (P ⫽ .03). The cumulative incidence of SD, SD and/or ventricular tachycardia/fibrillation, and non-SD at 5 years was 5.6%, 9.1%, and 27.8%, respectively. In multivariate competing risk analysis, advancing age, New York Heart Association class, and a CCI ⬎4 were significantly associated with non-SD.
Introduction : La mort subite (MS) et la mort cardiaque non subite sont responsables de la majorité des décès chez les patients ayant une insuffisance cardiaque. Nous avons cherché à déterminer l’influence de la comorbidité de la maladie (indice de comorbidité de Charlson [ICC]) sur les causes concurrentes de mortalité dans l’insuffisance cardiaque. Méthodes : Une analyse rétrospective a été réalisée auprès de 824 patients suivis dans une clinique de soins tertiaires pour une insuffisance cardiaque. Nous avons analysé l’incidence cumulative de la mort subite et non subite. La régression des risques concurrents a été utilisée pour examiner le lien entre les comorbidités médicales et les causes de décès. Les résultats recherchés étaient la mortalité globale, la MS, la MS et/ou le traitement par défibrillateur automatique implantable approprié (DAI), et la mort non subite. Résultats : L’âge moyen de la population étudiée était de 64,1 ⫾ 14,7 ans, dont 68,6 % était des hommes, et la fraction d’éjection moyenne était de 32,8 % ⫾ 13,5 %. Sur un suivi moyen de 4,4 ans, 229 patients (27,8 %) sont morts. La MS a représenté 33 décès (14,4 %), alors que la MS et le traitement avec un DAI approprié sont survenus chez 56 patients (24,5 %). Le risque de mort non subite et la mortalité totale a augmenté (P ⬍ 0,0001) comme le ICC augmentait, alors que le risque de mort subite a diminué (P ⫽ 0,03). L’incidence cumulative de la mort subite, la mort subite et/ou la tachycardie/fibrillation ventriculaire, et la mort
Heart failure is a major public health concern, with prevalence rates of 1%-2% in the general population under the age of 65 years and of ⬎10% of people over the age of 65. The incidence ranges
from 10-16 per 1000 population annually when adjusted for age.1 Mortality in the heart failure population is predominantly due to either sudden cardiac death (SD) or progressive heart failure, with SD accounting for 20%-50% of all deaths.2-5 Prophylactic implantable cardioverter-defibrillator therapy (ICD) therapy has been shown to reduce overall mortality via a reduction in arrhythmic deaths in patients with heart failure with severe systolic dysfunction.6,7 However, the accurate prediction of mode of death in heart failure using clinically relevant information beyond ejection fraction (EF) and New York Heart Association (NYHA) functional class has yet to be achieved.
Received for publication September 28, 2010. Accepted November 23, 2010. Corresponding author: Dr Ratika Parkash, Dalhousie University, Room XX, HI Site, QEII Health Sciences Centre, 1796 Summer St, Halifax, Nova Scotia, B3H 3A7, Canada. Tel.: ⫹1 902-473-4474; fax.: ⫹1 902-473-3158. E-mail: [email protected] See page 260 for disclosure information.
0828-282X/$ – see front matter © 2011 Canadian Cardiovascular Society. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.cjca.2010.12.053
Clarke et al. Comorbidity and SCD in Ambulatory HF Population
255
Conclusion: Patients with heart failure with significant comorbidities are much more likely to sustain non-SD. These findings may have implications in optimal selection of patients with heart failure for interventions such as prophylactic ICD therapy.
non subite après 5 ans, était de 5,6 %, 9,1 %, et 27,8 %, respectivement. Dans l’analyse du risque concurrent multifactorielle, l’âge avancé, la classification selon la New York Heart Association et un ICC ⬎ 4 étaient significativement associés à la mort non subite. Conclusion : Les patients ayant une insuffisance cardiaque avec des comorbidités importantes sont probablement plus à risque d’une mort non subite. Ces découvertes peuvent avoir des implications dans la sélection optimale des patients ayant une insuffisance cardiaque pour des interventions comme le traitement par DAI prophylactique.
There is significant interest in additional risk stratification of patients with heart failure to identify those who may derive the most benefit and, more important, to identify those who are least likely to derive benefit from prophylactic ICD therapy, avoiding the unnecessary risks and complications from device implantation in those whose risk of SD is far exceeded by other competing causes of death.8,9 We used competing risk methodology to estimate the probability of SD, SD and/or ventricular tachycardia/fibrillation (VT/VF), and non-SD in an ambulatory heart failure patient population. Furthermore, we aimed to examine the effect of medical comorbidity, as quantified by the Charlson Comorbidity Index (CCI), on mode of death.
obtained through an electronic database and detailed review of medical records (performed by B.C.). The baseline data were collected from information available as of the initial visit to the heart failure clinic. Deaths were documented in the database and mode of death was ascertained through extensive review of inpatient and outpatient medical records and classified as SD or non-SD by 2 independent physicians (B.C., J.H.), as defined later. In 14 cases where modality of death was uncertain, death certificates were obtained from Nova Scotia Vital Statistics in cooperation with the Population Health Research Unit of the Department of Health. Information from death certificates was corroborated with family physicians. None of these deaths were SDs. The prime determinant in classifying mode of death as sudden or nonsudden in these
Methods Table 2. Baseline characteristics of the study population
Study population This was a retrospective study of 824 consecutive patients followed at a tertiary care specialty ambulatory heart failure clinic in Halifax, NS, Canada, from December 1998 through December 2004. There were no prespecified criteria for referral to the heart failure clinic, other than at least one episode of clinical heart failure. Patients could be referred by family physicians, general internists, or other specialists. Data collection Approval from the institution’s research ethics board was obtained. Demographic, clinical, and laboratory information was Table 1. The Charlson comorbidity index Assigned weights for diseases 1
2
3 6
Conditions Myocardial infarct Congestive heart failure Peripheral vascular disease Cerebrovascular disease Dementia Chronic obstructive pulmonary disease Connective tissue disease Peptic ulcer disease Mild liver disease Diabetes Hemiplegia Moderate or severe renal disease (creatinine ⬎230 mmol/L) Diabetes with end organ damage Any tumor Leukemia Lymphoma Moderate or severe liver disease Metastatic solid tumor AIDS
Variable Age, y Male, n NYHA class, n I II III IV Ejection fraction, % QRS duration, ms ⱖ1 hospitalization for CHF Atrial fibrillation, n Ischemic heart disease, n Diabetes, n Diabetes with end organ damage, n Valvular disease, n Hypertension, n Dyslipidemia, n Peripheral vascular disease, n Cerebrovascular disease, n Chronic pulmonary disease, n Connective tissue disease, n Mild liver disease, n Moderate/severe liver disease, n Hemiplegia, n Moderate/Severe renal disease, n Any tumor, n Leukemia, n Lymphoma, n Metastatic solid tumor, n AIDS, n Creatinine, mmol/L ACE inhibitor, n Beta-blocker, n Diuretic, n Spironolactone, n Statin, n
Total population (N ⫽ 824) 64.2 ⫾ 14.7 565 (68.6%) 89 (10.8%) 237 (28.8%) 438 (53.1%) 60 (7.3%) 32.8 ⫾ 13.5 123.8 ⫾ 36.8 261 (31.7%) 259 (31.4%) 490 (59.5%) 272 (33.0%) 130 (47.8%) 404 (49.1%) 448 (54.4%) 436 (52.9%) 102 (12.4%) 129 (15.7%) 178 (21.6%) 13 (1.6%) 14 (1.7%) 3 (0.4%) 0 47 (5.7%) 77 (9.3%) 1 (⬍1%) 8 (1%) 6 (0.7%) 0 129.3 ⫾ 71.4 621 (75.9%) 532 (65.0%) 658 (80.4%) 192 (23.5%) 324 (39.6%)
ACE, Angiotensin-converting enzyme; CHF, congestive heart failure; NYHA, New York Heart Association.
256
Canadian Journal of Cardiology Volume 27 2011
Figure 1. Distribution of Charlson Comorbidity Index in the study population.
14 cases rested with clinical information provided by family physicians. Charlson Comorbidity Index (CCI) values were derived for each patient based on chart review and assigned according to baseline clinical information. The CCI is a weighted numeric scoring system designed to reflect the magnitude of comorbidity10 (Table 1). It was developed empirically, based on 1-year mortality from an inception cohort study in 1984, and has since been extensively validated in a variety of populations to predict survival.10-14 Deaths were classified as SD or non-SD. SD was defined as death within 1 hour of change of symptoms or during sleep or if death was unobserved, as has been previously defined.15. Non-SDs included all other deaths. Patients with ICDs were reviewed for number and types of therapies delivered. All ICD events were reviewed and adjudicated for appropriateness by 2 independent electrophysiologists. Appropriate therapies were defined as antitachycardia pacing (ATP) for VT or shocks delivered for VT or VF. Inappropriate therapies were defined as ATP or shocks delivered for rhythms other than VT/VF (ie, sinus tachycardia, atrial fibrillation). The outcomes included overall mortality, SD, a composite of SD or appropriate ICD therapy for VT/VF, and non-SD.
Figure 3. The association between mode of cardiac death and total mortality with the Charlson Comorbidity Index (CCI) in heart failure. White bars: Mortality from sudden death (SD). The P-value for trend was .03 as the CCI increased. Gray bars: Mortality from non-SD. The P-value for trend was ⬍.001 as the CCI increased. Black bars: Total mortality. The P-value for trend was ⬍.001 as the CCI increased.
Statistical analysis Baseline variables between groups were compared using Student’s t test for continuous variables and chi-square test for categorical variables. Univariate Cox proportional hazards models were constructed to identify those variables that significantly predicted overall mortality. The variables included in the multivariate Cox proportional hazards model were chosen from those variables that had a statistical significance of P ⬍ .10 on univariate analysis. The following covariates were included in the model: age, EF, NYHA class, dyslipidemia, the CCI, and the presence of an ICD. The proportional hazard assumption was examined for all covariates. We used cumulative incidence functions to display the proportion of patients with the event of interest or the competing event as time progressed.16 We used a multivariate competing risk model to analyze the effect of baseline predictors on the cumulative incidence function.17 A single analysis of the 2 cause-specific hazards was performed. The SAS macros, CumInc and CumIncv, developed by Rosthoj et al.18 were used for estimation of the cumulative incidence functions. These macros are based on multistate models and transition probabilities as developed by Anderson et al.19 All analyses were performed using SAS STAT software, version 9.2. A 2-sided
Table 3. Cox multivariate regression model for overall mortality
Figure 2. Distribution of mode of death (N ⫽ 229). CHF, Congestive heart failure; SD, sudden death.
Variable
HR
95% CI
P value
CCI NYHA III/IV Age Dyslipidemia ICD
1.26 2.50 1.04 1.39 1.86
1.19-1.35 1.80-3.46 1.03-1.05 1.06-1.81 1.01-3.42
⬍.0001 ⬍.0001 ⬍.0001 ⬍.016 ⬍.046
CCI, Charlson Comorbidity Index; CI, confidence interval; HR, hazard ratio; ICD, implantable cardioverter-defibrillator; NYHA, New York Heart Association.
Clarke et al. Comorbidity and SCD in Ambulatory HF Population
257
Figure 4. Cumulative incidence function for sudden death (SD), SD and/or ventricular tachycardia/fibrillation (VT/VF), and non-SD.
P value of ⬍.05 was taken as statistically significant for all analyses. Results Patient characteristics Baseline characteristics of the study cohort are shown in Table 2. Mean age was 64.1 ⫾ 14.7 years, 68.6% were male, mean EF was 32.8% ⫾ 13.5% (21% had EF ⱕ20%, 43% had EF 21%-35%, 26% had EF 36%-50%, 10% had EF ⬎50%), 60.4% were NYHA functional class III/IV, 59.5% had coronary artery disease, and 9.8% had an ICD. Patients were followed for a mean of 4.4 years; 31.7% had at least one rehospitalization for heart failure during the follow-up period. The distribution of the CCI for the study population is shown in Figure 1. Relationship between mode of death, ICD therapies, and the CCI Over a mean follow-up period of 4.4 years, 229 patients (27.8%) died. Of these, SD occurred in 33 patients (14.4%)
(Fig. 2). Twenty-three patients (28.4%) received at least one appropriate therapy for VT/VF. The risk of total mortality and non-SD was associated with an increase in the CCI (P ⬍ .0001), whereas the risk of SD decreased with increasing CCI (P ⫽ .03) (Fig. 3). In a Cox regression analysis, predictors of mortality included increasing CCI, age, NYHA class III/IV, dyslipidemia, and the absence of an ICD (Table 3). The 5-year cumulative incidence for SD, SD and/or VT/VF, and non-SD was 5.6%, 9.1%, and 27.8%, respectively (Fig. 4). A CCI cutoff of 4 was chosen based on the ROC curve. In multivariate competing risk analysis, advancing age (by 10-year increments), a CCI ⬎4, and advancing NYHA class were significant predictors of non-SD (Table 4). When appropriate ICD therapy for VT/VF was included in the analysis, there was a trend toward better freedom from SD or appropriate ICD therapy with increasing EF (HR 0.98, P ⫽ .057 for every 10% increase in EF; Table 5). No variables were found to significantly predict SD. The effect of the CCI on cumulative incidence is shown in Figure 5 and Table 6. Patients with a CCI ⬎4 had a significantly higher risk of non-SD with a 5-year cu-
Table 4. Multivariate competing risk regression: SD versus non-SD
Table 5. Multivariate competing risk regression: composite outcome of SD and VT/VF versus non-SD
Variable
Variable
Non-SD Age EF CCI ⬎ 4 NYHA class SD Age EF CCI ⬎ 4 NYHA class
HR
P value
95% CI
1.05 0.99 2.78 2.89
⬍.0001 .07 ⬍.0001 ⬍.0001
1.03-1.06 0.97-1.00 2.09-3.70 1.99-4.21
1.01 0.99 0.49 1.27
.68 .54 .15 .52
0.98-1.03 0.97-1.02 0.19-1.28 0.62-2.64
CCI, Charlson Comorbidity Index; CI, confidence interval; HR, hazard ratio; ICD, implantable cardioverter-defibrillator; NYHA, New York Heart Association; SD, sudden death.
Non-SD Age EF CCI ⬎ 4 NYHA class (I/II vs III/IV) SD/VT/VF Age EF CCI ⬎ 4 NYHA class (I/II vs III/IV)
HR
P value
95% CI
1.05 0.99 0.36 0.34
⬍.0001 .058 ⬍.0001 ⬍.0001
1.03-1.06 0.98-1.00 0.27-0.48 0.23-0.50
1.01 0.99 1.44 0.87
.30 .23 .28 .62
0.99-1.03 0.97-1.01 0.74-2.81 0.50-1.51
CCI, Charlson Comorbidity Index; CI, confidence interval; EF, ejection fraction; ICD, implantable cardioverter-defibrillator; NYHA, New York Heart Association; SD, sudden death; VT/VF, ventricular tachycardia/fibrillation.
258
Canadian Journal of Cardiology Volume 27 2011
Figure 5. Cumulative incidence function for sudden death (SD), SD and/or ventricular tachycardia/fibrillation (VT/VF), and non-SD by Charlson Comorbidity Index (CCI). A: CCI ⬎ 4. B: CCI ⱕ 4.
mulative incidence of 48.8% compared to 33.5% with a CCI ⱕ4 (P ⬍ .0001). Discussion In this study of ambulatory patients with heart failure, we report the relationship between underlying comorbid illness, as measured by the CCI, on modes of death. We used a competing risk analysis to evaluate the effect of comorbidity and other covariates on outcomes of SD, SD and/or VT/VF, and non-SD in patients with heart failure. Numerous studies have demonstrated the association between underlying medical comorbidities and mortality in a variety of patient populations using the CCI.10-14 In general, these
studies noted an increase in overall mortality with increasing comorbidity. This is the first study to use the CCI in the heart failure population. Our principal findings demonstrate the absolute risk of SD remains relatively constant with increasing CCI, while the risk of non-SD increases with increasing comorbidity burden. Thus, patients with multiple comorbidities have a significantly higher relative risk of non-SD. Results from recent clinical trials have established mortality benefit with ICD therapy in ischemic and nonischemic cardiomyopathy with reduced EF.6,7 However, these trials include a select group of patients with heart failure. Our study uses a real-world cohort of ambulatory patients
Clarke et al. Comorbidity and SCD in Ambulatory HF Population
259
Table 6. Cumulative incidence functions of competing risk model Competing risk model 5-Year cumulative incidence Variable CCI ⱕ4 ⬎4 NYHA class I/II III/IV EF class ⱕ30% ⬎30% Overall cumulative risk
SD
SD/VT death
Non-SD
0.0159 (0.00905, 0.0228) 0.0232 (0.0030, 0.0434)
0.0256 (0.0324, 0.0189) 0.0494 (0.0686, 0.0302)
0.335 (0.0703, 0.599) 0.488 (0.419, 0.557)
0.0124 (0.00519, 0.0197) 0.0457 (0.0243, 0.0671)
0.0214 (0.0277, 0.0152) 0.0732 (0.0921, 0.0544)
0.102 (0.0249, 0.178) 0.372 (0.323, 0.421)
0.0528 (0.0323, 0.0814) 0.0399 (0.0183, 0.0615) 0.0563
0.0922 (0.0692, 0.115) 0.0637 (0.0453, 0.0820) 0.0908
0.371 (0.1640, 0.578) 0.259 (0.211, 0.307) 0.278
CCI, Charlson Comorbidity Index; CI, confidence interval; EF, ejection fraction; NYHA, New York Heart Association; SD, sudden death.
with heart failure. Toma et al.20 applied SCD-HeFT criteria to a real-world population of patients with heart failure and found that just over half of all patients seen in a specialized heart failure clinic would qualify for ICDs. Many of these patients may not benefit due to competing risks of death, as seen in this study, identifying the need for additional stratification schemes beyond NYHA class and EF, when considering patients for ICD therapy for primary prevention of SD. ICD shocks (both appropriate and inappropriate) have recently been found to be associated with an increase in mortality, raising the possibility that ICD therapy may not be such a benign intervention.21-23 An analysis of SCDHeFT demonstrated that both appropriate and inappropriate shocks increased the risk of death.21 Of the 269 patients who received shocks in this study, 141 patients (52.4%) received at least one inappropriate shock and 87 patients (32.3%) received only inappropriate shocks. Furthermore, Koller et al.24 noted that 11%-23% of patients with heart failure with ICDs died without prior ICD therapy and 36% of patients remained alive and never used the device. Predicting mode of death is important as there are some therapies that modify the risk of SD. Reduction of non-SD may be achieved with aggressive conventional medical therapy (beta-blockers, angiotensin-converting enzyme [ACE] inhibitors/angiotensin II receptor blocker [ARB], spironolactone) and cardiac resynchronization therapy (CRT), whereas prevention of arrhythmic SD rests with ICD therapy in addition to conventional medical therapy. This therapy is invasive and has its inherent procedural and long-term risks including device advisories25 and complications at the time of implant as well as with replacement.26,27 Patients with higher comorbidity burden have higher rates of procedural complications, as demonstrated in the REPLACE registry, where the complication rate for ICD generator replacement was 13.1% for those patients with a CCI ⱖ2 versus 8.5% in those with a CCI ⬍2.26 The majority of the ambulatory heart failure population in this study (86%) are in this higher risk category. Recent evidence illustrates that much prognostic information can be gained from easily determined historical factors, such as age and other medical comorbidities.28 In an observational study of a large cohort of real-world ICD recipients, competing noncardiac comorbidities such as renal failure, chronic pulmonary disease, peripheral vascular disease, and diabetes were associated with increased mortality after ICD implantation.29 Braunstein et al.30 evaluated the effect of comor-
bid illness on rehospitalization rates and overall mortality in a cohort of U.S. Medicare beneficiaries with chronic heart failure and found that the risk of hospitalizations and potentially preventable hospitalizations strongly increased with the number of chronic conditions. Chronic obstructive pulmonary disease, renal failure, diabetes, depression, and lower respiratory diseases conferred a notably higher risk. Several observations from this study deserve special attention. The SD rate that was observed in this cohort of patients is much lower than reported in clinical trials; however, the observed overall mortality rate is similar to contemporary clinical trials.3 SD accounted for 64% and 59% of total mortality in NYHA II and III patients with heart failure, respectively, in MERIT-HF, in contrast to 14.4% of patients in this study. This may be due to several factors. Our cohort is representative of a mixed population of patients with heart failure, with a broader spectrum of EF values, more patients with NYHA class I and IV heart failure, and comorbidities that would have precluded entry into randomized trials. In addition, 81 patients had an ICD. When appropriate therapy for VT/VF was considered as a surrogate for SD, as has been done in other studies, 25% of patients were considered to have SD,31 recognizing, however, that appropriate ICD shocks may occur more frequently than SD and may overestimate the actuarial SD rate.32 In addition, MERIT-HF used a wider definition of SD, including patients who died within 28 days of resuscitated cardiac arrest. Patients were excluded if the EF was ⱖ40% and mean EF was 28%, in contrast to this study, where there were no exclusion criteria based on EF and mean EF was higher at 32.8%. There are some limitations to this retrospective study that deserve mention. We observed an ICD implantation rate of approximately 10% in this population. This can partially be accounted by the effective study period. The major clinical trials establishing the efficacy of prophylactic ICD therapy in heart failure were released in the early 2000s; since then, the indications have expanded significantly. This study included patients from 1998-2004, who at that time may not have been candidates for ICDs but now may qualify for ICD therapy according to current guidelines. Finally, the population studied was restricted to patients followed in an ambulatory heart failure clinic, which may limit applicability to the population as a whole. Whether this association holds in hospitalized patients with heart failure has yet to be determined.
260
We used standard definitions in classifying mode of death; however, there remains a risk of misclassification. To minimize this risk, deaths documented in medical records were independently reviewed by 2 physicians, and in cases where mode of death was uncertain, clinical details from primary care physicians were used to corroborate cause of death as listed on death certificate information. The clinical information used to calculate the CCI was the information that was available at the time of analysis. There is a small risk of underscoring the CCI if, for example, a patient with diabetes has unrecognized complications such as retinopathy. Our findings suggest a novel method in which to apply the CCI to ambulatory patients with heart failure to assist in determining the most appropriate medical and device-based therapies. The findings from this study could be applied prospectively to an ICD population to see if it predicted the occurrence of ventricular arrhythmia as a surrogate for SD, mindful of the fact that VT/VF is a surrogate for this. Conclusion The relationship between risks of SD and non-SD in the ambulatory heart failure is complex. Our findings demonstrate that the relative risk of each of these competing modes of death vary significantly with the CCI. The magnitude of benefit of invasive therapies to prevent SD, such as ICDs, may differ in ambulatory patients with heart failure based on the degree of comorbid illness. This has important implications for using this kind of invasive therapy, which carries a significant risk of complications in patients with a higher CCI. Patients with a high CCI and advanced heart failure may have very little benefit from an invasive therapy such as an ICD. This may have implications on how this resource should be allocated in the ever-expanding heart failure population. Acknowledgements The authors thank Kara Thompson for providing statistical support. Funding Sources This study was funded by a grant from the University of Internal Medicine Research Foundation. Disclosures The authors have no conflicts of interest to disclose. References 1. McMurray J, Stewart S: Heart failure: epidemiology, aetiology, and prognosis of heart failure. Heart 2000;83:596-602. 2. Rami T, Hih J: Update of implantable cardioverter/defibrillator and cardiac resynchronization therapy in heart failure. Curr Opin Cardiol 2004;19: 264-9. 3. MERIT-HF Study Group: Effect of metoprolol CR/XL in chronic heart failure: Metoprolol CR/XL Randomised Intervention Trial in Congestive Heart Failure (MERIT-HF). Lancet 1999;353:2001-7. 4. The Captopril-Digoxin Multicentre Research Group: Comparative effects of therapy with captopril and digoxin in mild to moderate heart failure. JAMA 1988;259:539-44. 5. Zheng ZJ, Croft JB, Giles WH, et al. Sudden cardiac death in the United States 1989 to 1998. Circulation 2001;104:2158-63.
Canadian Journal of Cardiology Volume 27 2011 6. Moss AJ, Zareba W, Hall WJ et al. Prophylactic implantation of defibrillator in patients with myocardial infarction and reduced left ventricular ejection fraction. N Engl J Med 2002;346:877-83. 7. Bardy GH, Lee KL, Mark DB, et al. Amiodarone or an implantable cardioverter-defibrillator for congestive heart failure. N Engl J Med 2005; 352:225-37. 8. Josephson M, Wellens H: Implantable defibrillators and sudden cardiac death. Circulation 2004;109:2685-91. 9. Warner Stevenson L and Desai A: Selecting patients for discussion of the ICD as primary prevention for sudden death in heart failure. J Card Fail 2006;12:407-12. 10. Charlson M, Pompei P, Ales K, et al. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chron Dis 1987;40:373-83. 11. Rochon PA, Katz JN, Morrow LA, et al. Comorbid illness is associated with survivalandlengthofhospitalstayinpatientswithchronicdisability.Aprospective comparison of three comorbidity indices. Med Care 1996;34:1093-101. 12. West DW, Satariano WA, Ragaland DR, et al. Comorbidity and breast cancer survival: a comparison between black and white women. Ann Epidemiol 1996;6:413-9. 13. Skiest DJ, Rubinstien E, Carley N, et al. The importance of comorbidity in HIV-infected patients over 55: a retrospective case– control study. Am J Med 1996;101:605-11. 14. Poses RM, McClish DK, Smith WR, et al. Prediction of survival of critically ill patients by admission comorbidity. J Clin Epidemiol 1996;49:743-7. 15. Kearney MT, Fox KAA, Lee AJ, et al. Predicting death due to progressive heart failure in patients with mild-to-moderate chronic heart failure. J Am Coll Cardiol 2002;40:1801-8. 16. Putter H, Fiocco M, Geskus RB. Tutorial in biostatistics: competing risks and multi-state models. Stat Med 2007;26:2389-430. 17. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of acompeting risk. J Am Stat Assoc 1999;94:496-50. 18. Rosthej S, Andersen PK, Abildstrom SZ. SAS macros for estimation of the cumulative incidence functions based on a Cox regression model for competing risks survival data. Comp Meth Progr Biomed 2004;74:69-75. 19. Anderson PK, Bogan Ø, Gill RD, et al. Statistical Models Based on Counting Processes. New York: Springer, 1993. 20. Toma M, McAlister FA, Ezekowitz J, et al. Proportion of patients followed in a specialized heart failure clinic needing an implantable cardioverter defibrillator as determined by applying different trial eligibility criteria. Am J Cardiol 2006;97:882-5. 21. Poole JE, Johnson GW, Hellkamp AS, et al. Prognostic importance of defibrillator shocks in patients with heart failure. N Engl J Med 2008;359:1009-17. 22. Sweeney MO, Sherfesee L, DeGroot PJ, et al. Differences in effects of electrical therapy type for ventricular arrhythmias on mortality in implantable cardioverter-defibrillator patients. Heart Rhythm 2010;7:353-60. 23. El-Damaty A, Sapp J, Basta M, et al. Inappropriate ICD shocks: clinical risk factors and mortality. Heart Rhythm 2010;7:s186-7. 24. Koller MT, Schaer B, Wolbers M, et al. Death without prior appropriate implantable cardioverter-defibrillator therapy: a competing risk study. Circulation 2008;117:1918-26. 25. Gould PA, Krahn AD; Canadian Heart Rhythm Society Working Group on Device Advisories. Complications associated with implantable cardioverter-defibrillator replacement in response to device advisories. JAMA 2006;295:1944-6.
Clarke et al. Comorbidity and SCD in Ambulatory HF Population 26. Gleva MJ, Holcomb R, Uslan D, et al. Charlson Comorbidity Index and risk of cardiac rhythm device generator replacement: results from the REPLACE Registry. Circulation 2009;120:S637 Abstract 2435. 27. Ravindran BK, Poole JE, Johnson GW, et al. Timing of device-specific ICD complications: analysis of the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT). Heart Rhythm 2010;7:s9. 28. Buxton AE, Lee KL, Hafley GE, et al. Limitations of ejection fraction for prediction of sudden death risk in patients with coronary artery disease. J Am Coll Cardiol 2007;50:1150-7. 29. Lee DS, Tu JV, Austin PC, et al.: Effect of cardiac and noncardiac conditions on survival after defibrillator implantation. J Am Coll Cardiol 2007;49:2408-15.
261 30. 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;42:1226-33. 31. Gold MR, Ip JH, Costantini O, et al. Role of microvolt T-wave alternans in assessment of arrhythmia vulnerability among patients with heart failure and systolic dysfunction. Circulation 2008;118:2022-8. 32. Ellenbogen, KA, Levine JH, Berger RD, et al; for the Defibrillators in Non-Ischemic Cardiomyopathy Treatment Evaluation (DEFINITE) Investigators. Are implantable cardioverter defibrillator shocks a surrogate for sudden cardiac death in patients with nonischemic cardiomyopathy?. Circulation 2006;113:776-82.