Given their advanced age and frequent co-morbidities, it is unclear whether octogenarians and nonagenarians with decreased left ventricular ejection fraction (LVEF) derive a survival benefit from implantable cardioverter–defibrillators (ICDs) in the primary prevention setting. The purpose of this study was to examine the effect of ICDs, age, and multiple co-morbidities on survival in elderly patients who otherwise meet implantation criteria for primary prevention of sudden cardiac death. Patients ≥80 years of age who received an ICD for LVEF ≤35% at our institution from 2001 through 2008 (n = 99) were compared to a cohort of patients ≥80 years of age with similarly low LVEF who did not receive an ICD (n = 53). Co-morbid conditions were assessed using the Charlson co-morbidity index (CCI). The overall cohort (n = 152, 84 ± 4 years old, 72% men, 87% with ischemic cardiomyopathy, LVEF 25 ± 7%, CCI 5.9 ± 3.2) was followed for 2.3 ± 2.0 years. Patients with ICDs were younger (82 ± 3 vs 86 ± 4 years, p <0.001) and had fewer co-morbidities reflected in a lower CCI (5.3 ± 3.1 vs 6.7 ± 3.1, p = 0.021). Patients with ICDs also had a trend toward lower LVEF (24 ± 6% vs 27 ± 7%, p = 0.06). During follow-up 93 (61%) patients died, 58 (59%) in the ICD group and 35 (66%) in the no-ICD group. ICD recipients had better 1-year survival compared to patients with no ICD (72% vs 52%, p = 0.014). However, after adjusting for age, LVEF, glomerular filtration rate (GFR), and CCI using multivariate Cox models, an ICD did not confer any survival benefit (hazard ratio 0.71, 95% confidence interval 0.42 to 1.20, p = 0.20), whereas age (p = 0.043) and GFR (p = 0.006) were the only independent predictors of survival. In conclusion, age and GFR are the main determinant of survival in octogenarians and nonagenarians with LV dysfunction. After correcting for these parameters, an ICD does not appear to confer a survival benefit.
The purpose of the present study was to evaluate the effect of implantable cardioverter–defibrillators (ICDs) on mortality in octogenarians and nonagenarians with left ventricular (LV) dysfunction and account for the presence of other life-limiting co-morbid conditions.
Methods
We identified all patients who received an ICD at our institution from January 2000 through December 2008 and were ≥80 years of age at the time of their ICD implantation (n = 485). The study was limited to this period to assure contemporary and homogenous cardiac care, namely as it relates to indications for ICD implantations. From this group patients were excluded if the ICD was implanted for secondary prevention (n = 150), if the original implantation was before 2000 but a device change out was performed in the period of interest (n = 56), if their LV ejection fraction (EF) was >35% (n = 100), or if they did not have follow-up at our institution (n = 80). Ninety-nine patients were therefore included in the final analysis. Baseline characteristics were similar between the study group and patients excluded for lack of follow-up at our institution. A group of patients ≥80 years of age who had an LVEF ≤35% but did not have an ICD served as a control group (n = 53). ICD and no-ICD groups were not matched by any clinical criteria. Patients were followed for a mean of 2.3 ± 2.0 years. Retrospective data collected on all patients including gender, age, LVEF, presence of ischemic heart disease, previous coronary artery bypass surgery, hypertension, diabetes, New York Heart Association (NYHA) class before ICD implantation, history of cancer, congestive heart failure, chronic kidney disease, medical therapy (including antiarrhythmic agents, β-adrenergic antagonists, angiotension-converting enzyme inhibitors or angiotension receptor blockers), and implanted device type. Patients were excluded if they received their postimplantation care at another institution or if they had no documented follow-up.
The primary end point for the study was all-cause mortality. Time of death was determined through medical record review, outpatient clinic follow-up visits, and a query of the United States Social Security Death Index. The beginning of follow-up was defined as the date of ICD implantation for ICD recipients and the date of first documented low LVEF for patients with no ICD. Surviving patients were censored at the time of their latest identifiable follow-up visit.
For the purposes of this study, primary prevention was defined as implantation in patients with (1) ischemic or nonischemic cardiomyopathy and EF ≤35% with NYHA class II or III heart failure or (2) patients with EF ≤30%, previous myocardial infarction with NHYA class I. Estimated glomerular filtration rates (GFRs) were determined using the Cockcroft–Gault equation.
Comorbid conditions in the ICD and no-ICD groups were assessed using the Charlson co-morbidity index (CCI), which incorporates 19 medical conditions that affect longevity. Table 1 lists variables included in the CCI. The CCI was calculated by assigning a weight score of 1, 2, 3, or 6 depending on the risk of dying associated with this condition. Overall co-morbidity score, which predicts mortality for each patient, is the arithmetic sum of values assigned to each identified co-morbid condition.
| Score | Condition |
|---|---|
| 1 | myocardial infarction |
| 1 | congestive heart failure |
| 1 | peripheral vascular disease |
| 1 | cerebrovascular disease |
| 1 | dementia |
| 1 | chronic pulmonary disease |
| 1 | connective tissue disease |
| 1 | peptic ulcer disease |
| 1 | mild liver disease |
| 1 | diabetes |
| 2 | hemiplegia |
| 2 | moderate or severe renal disease |
| 2 | diabetes with end-organ damage |
| 2 | any tumor |
| 2 | leukemia |
| 2 | lymphoma |
| 3 | moderate or severe liver disease |
| 6 | metastatic solid tumor |
| 6 | acquired immunodeficiency syndrome |
Continuous variables were expressed as mean ± SD and compared using Student’s t test. Categorical variables were presented as percentages and compared using chi-square test. We used the Kaplan–Meier method to compare survival between patient groups using the log-rank test. For this purpose cutoffs for age and GFR were chosen as the mean for the overall population (ICD and no-ICD groups). Multivariate analyses of the independent effect of an ICD on time to death were performed using Cox proportional hazards model. Inclusion of age and CCI in multivariate models was predetermined. Other variables (GFR and LVEF) were included because they predicted death in univariate analyses (p <0.1) or because they are established in the literature as predictors of mortality. Multiple models were therefore analyzed adjusting for the following confounding variables: (1) age only; (2) age and CCI; (3) age, CCI, and LVEF; (4) age, CCI, and GFR; and (5) age, CCI, LVEF, and GFR. Given the known influence of heart failure class on survival, a sixth model was analyzed including this covariate. A 2-tailed p value <0.05 was considered statistically significant. All analyses were performed using SPSS 11.0 (SPSS, Inc., Chicago, Illinois).
Results
Baseline characteristics of the study population are presented in Table 2 . The overall cohort (n = 152) was 84 ± 4 years old (range 80 to 96) and consisted primarily of men (72%) with ischemic cardiomyopathy (87%). There were 140 octogenarians and 12 nonagenarians in this cohort. Mean LVEF of the overall study population was 25 ± 7%, and CCI was 5.9 ± 3.2. Patients implanted with an ICD were younger (82 ± 3 vs 86 ± 4 years, p <0.001) and had fewer co-morbidities reflected in a lower CCI (5.3 ± 3.1 vs 6.7 ± 3.1, p = 0.021). Patients with ICDs also had better GFR (42 ± 20 vs 35 ± 18 ml/min, p = 0.034) and a trend toward lower LVEF (24 ± 6% vs 27 ± 7%, p = 0.06). The ICD and no-ICD groups were otherwise similar.
| Variable | ICD | No ICD | p Value |
|---|---|---|---|
| (n = 99) | (n = 53) | ||
| Women/men | 23%/77% | 33%/67% | 1.000 |
| Age (years) | 82 ± 3 | 86 ± 4 | 0.001 |
| Ejection fraction (%) | 24 ± 6 | 27 ± 7 | 0.06 |
| Ischemic cardiomyopathy | 87% | 93% | 0.359 |
| Previous coronary bypass | 53% | 39% | 0.184 |
| Diabetes mellitus | 34% | 63% | 0.004 |
| Hypertension | 96% | 98% | 1.000 |
| Cancer | 17% | 26% | 0.295 |
| Heart failure class III or IV | 44% | 72% | 0.001 |
| Chronic kidney disease | 61% | 80% | 0.182 |
| Glomerular filtration rate (ml/min) | 42 ± 20 | 35 ± 18 | 0.034 |
| β Blocker | 70% | 61% | 0.310 |
| Angiotensin-converting enzyme-inhibitor | 77% | 65% | 0.162 |
| Antiarrhythmics drugs (classes 1 and 3) | 9% | 22% | 0.047 |
| Charlson co-morbidity index | 5.3 ± 3.1 | 6.7 ± 3.1 | 0.021 |
During 2.3 ± 2.0 years of follow-up 93 (61%) patients died, 58 (59%) in the ICD group and 35 (66%) in the no-ICD group. Median survival for the overall cohort was 1.8 years. Compared to survivors patients who died had lower LVEF (24 ± 7% vs 27 ± 7%, p = 0.042) and lower GFR (38 ± 19 vs 54 ± 19 ml/min, p <0.001) but comparable age at the beginning of follow-up (83 ± 3 vs 84 ± 4 years, p = 0.242) and CCI (5.6 ± 3.0 vs 5.7 ± 3.0, p = 0.776). Time to death was significantly shorter in patients with lower GFR (1-year mortality 47% vs 24%, p = 0.013; Figure 1 ) and older age (1-year mortality 43% vs 35%, p = 0.023; Figure 1 ).
ICD recipients had better 1-year survival compared to patients with no ICD (1-year survival 72% vs 52%, p = 0.014). However, ICDs did not confer survival benefit after adjusting for age, CCI, LVEF, and GFR in various combinations using a multivariate Cox model (hazard ratio 0.78, 95% confidence interval [CI] 0.44 to 1.30, p = 0.312). In this model, only age (hazard ratio 1.07, 95% CI 1.00 to 1.14, p = 0.043; Table 3 ) and GFR (hazard ratio 0.98, 95% CI 0.97 to 0.99, p = 0.006; Table 3 ) were independent predictors of survival.
