Author + information
- Received February 25, 2016
- Revision received October 18, 2016
- Accepted October 18, 2016
- Published online January 16, 2017.
- Jarrod K. Betz, MDa,
- David F. Katz, MDb,c,
- Pamela N. Peterson, MD, MSPHb,c,d,
- Ryan T. Borne, MDb,c,
- Sana M. Al-Khatib, MD, MHSe,
- Yongfei Wang, MSf,g,
- Carolina Malta Hansen, MDe,
- David D. McManus, MD, ScMh,
- Jehu S. Mathew, MDb,c and
- Frederick A. Masoudi, MD, MSPHb,c,∗ ()
- aDepartment of Medicine, University of Colorado, Denver, Colorado
- bDivision of Cardiology, University of Colorado, Denver, Colorado
- cColorado Cardiovascular Outcomes Research Group, Denver, Colorado
- dDenver Health Medical Center, Denver, Colorado
- eDivision of Cardiology, Duke University, Durham, North Carolina
- fDepartment of Internal Medicine, Yale University School of Medicine, New Haven, Connecticut
- gCenter for Outcomes Research and Evaluation, Yale-New Haven Hospital, New Haven, Connecticut
- hDivision of Cardiology, University of Massachusetts Medical School, Worcester, Massachusetts
- ↵∗Reprint requests and correspondence:
Dr. Frederick A. Masoudi, Division of Cardiology, University of Colorado, 12401 East 17th Avenue, Room 522, Aurora, Colorado 80045.
Background Clinical trials of implantable cardioverter-defibrillators (ICDs) for secondary prevention of sudden cardiac death were conducted nearly 2 decades ago and enrolled few older patients.
Objectives This study assessed morbidity and mortality of older patients receiving ICDs for secondary prevention in contemporary clinical practice.
Methods We identified 12,420 Medicare beneficiaries from the National Cardiovascular Data Registry ICD Registry undergoing first-time secondary prevention ICD implantation between 2006 and 2009 in 956 U.S. hospitals. Risks of death, hospitalization, and admission to a skilled nursing facility (SNF) were assessed over 2 years in age strata (65 to 69, 70 to 74, 75 to 79, and ≥80 years of age) using Medicare claims. The adjusted association between age and outcomes was evaluated using multivariable models.
Results The mean age was 75 years at the time of implantation; 25.3% were <70 years of age and 25.7% were ≥80 years of age. Overall, the risk of death at 2 years was 21.8%, ranging from 14.7% among those <70 years of age to 28.9% among those ≥80 years of age (adjusted risk ratio [aRR]: 2.01; 95% confidence interval [CI]: 1.85 to 2.33; p for trend <0.001). The cumulative incidence of hospitalizations was 65.4%, ranging from 60.5% in those <70 years of age to 71.5% in those ≥80 years of age (aRR: 1.27; 95% CI: 1.19 to 1.36; p for trend <0.001). The cumulative incidence of admission to a SNF ranged from 13.1% among those <70 years of age to 31.9% among those ≥80 years of age (aRR: 2.67; 95% CI: 2.37 to 3.01; p for trend <0.001); SNF admission risk was highest in the first 30 days.
Conclusions Almost 4 in 5 older patients receiving a secondary prevention ICD survives at least 2 years. High hospitalization and SNF admission rates, particularly among the oldest patients, identify substantial care needs after device implantation.
Implantable cardioverter-defibrillators (ICDs) were initially used in the early 1980s to treat individuals who had been successfully resuscitated from cardiac arrest (i.e., secondary prevention) (1). Although the indications for ICDs have since expanded to include high-risk individuals who have not experienced lethal ventricular arrhythmias (i.e., primary prevention), patients undergoing secondary prevention ICD implantation still account for approximately one-quarter of all procedures entered in the National Cardiovascular Data Registry (NCDR) (2). Although the outcomes in patients receiving an ICD for primary prevention have been characterized in detail (3), those for patients receiving secondary prevention ICDs are substantially more limited.
For several reasons, existing clinical trial data on secondary prevention ICDs may not apply to contemporary clinical practice, especially to older patients (4–7). First, the few randomized controlled trials in this context were performed nearly 2 decades ago, and were generally restricted to younger patients with a history of documented ventricular tachycardia (VT) and ventricular fibrillation. Older patients surviving cardiac arrest may have a higher burden of coexisting illnesses that may influence outcomes after ICD implantation. Second, therapies for underlying structural heart disease, including those for left ventricular systolic dysfunction and coronary artery disease, have evolved substantially. Finally, guideline recommendations for secondary prevention ICD therapy expand beyond the enrollment criteria of the randomized trials, and are generally predicated upon the assumption that patients considered for therapy have reasonable prospects for a life expectancy of at least 1 year (8,9). In the 15 years since the publication of these randomized trials, the outcomes of older patients receiving an ICD for secondary prevention in clinical practice have not been well characterized.
Accordingly, we analyzed data from the NCDR ICD Registry to assess rates of death, rehospitalization, and skilled nursing facility (SNF) admission among older persons undergoing secondary prevention ICD implantation. These data are intended to provide patients and clinicians with contemporary, representative estimates of the risks of adverse outcomes after ICD implantation to inform decision making and understand the resource needs of this population to support health policy.
Patients assessed in this study were enrolled in the NCDR ICD Registry (10,11). The registry includes data on patients receiving implantable devices in the United States across hospital and payer types. As a condition of reimbursement from the Centers for Medicare and Medicaid Services, all Medicare beneficiaries receiving a primary prevention ICD must be included in the ICD Registry. Although this requirement does not apply to patients receiving an ICD designated as secondary prevention, 91% (1,320 of 1,465) of participating sites have submitted data on patients receiving ICDs for secondary prevention indications. Clinical, demographic, and procedural data are collected using standardized definitions. Data are submitted by participating hospitals using certified software and are examined using a formal Data Quality Reporting and audit process (12). Medicare claims data were used to ascertain outcomes through linkage with NCDR data. Using an established validated method, eligible subjects were matched to Medicare claims data on the basis of indirect identifiers, including age, sex, admission or procedure date, and hospital Medicare provider number (13). Analyses of the NCDR ICD Registry are performed under an institutional review board approval by Yale University, with a waiver of informed consent because of the study design.
Medicare beneficiaries in the NCDR ICD registry ≥65 years of age were included. This study group was limited to those undergoing initial implantation of a secondary prevention ICD between 2006 and 2009. The implanting physician determined the designation of secondary prevention. The cohort was further limited to patients with a prior episode of sudden cardiac arrest, defined as having any of the following: 1) tachycardic arrest; 2) sustained, monomorphic VT; or 3) sustained polymorphic VT. Single-chamber, dual-chamber, and cardiac resynchronization therapy-defibrillator devices were included. Patients meeting the clinical eligibility criteria who could be linked to Medicare data formed the study cohort.
Clinical and demographic information on patients were obtained from the NCDR, including age, sex, and clinical characteristics. The primary predictor variable was age. The study patients were stratified into groups on the basis of age at time of implantation (<70, 70 to 74, 75 to 79, or ≥80 years of age). Other covariates considered included patient, clinician, and hospital characteristics, all ascertained from the NCDR. Patient-level characteristics included demographics (sex, race, and insurance payer), comorbidities and risk factors including cardiac arrest, history of sustained or nonsustained VT, syncope, family history of sudden death, history of heart failure, admission New York Heart Association functional classification, atrial fibrillation or flutter, presence of ischemic or nonischemic cardiomyopathy, myocardial infarction, coronary artery bypass graft surgery, percutaneous coronary intervention, valve surgery, cerebrovascular disease, chronic lung disease, diabetes, hypertension, chronic kidney disease (glomerular filtration rate <60 ml/min/1.73 m2), and renal failure requiring hemodialysis. Diagnostic information included ejection fraction; whether an electrophysiology study was performed, and if so, whether a sustaining ventricular arrhythmia was induced; serum creatinine; serum blood urea nitrogen; serum sodium level; and systolic blood pressure. The training of the implanting operator was examined, as were hospital characteristics, including hospital type (private or community, academic, and government), hospital size (as a continuous variable defined by number of beds), geographic location (by region in the United States), and geographic setting (rural vs. suburban vs. urban).
The primary outcomes of interest were rates of death from any cause, hospitalizations (all cause and heart failure, determined by diagnosis-related groups) and SNF admission at 2 years. Risks of these outcomes at 30 days and 1 year were also calculated. Medicare claims data were used to ascertain outcomes.
Patient, physician, and hospital characteristics were examined overall, and were compared among the different age strata using the chi-square test for categorical variables and the F test in analysis of variance for continuous variables.
Kaplan-Meier survival curves were used to evaluate unadjusted survival in each of the age strata in the overall cohort. Factors significantly associated with time to death within 2 years were identified using Cox proportional hazard regression analyses in the overall cohort. Independent relationships among the age groups and time to events were evaluated, adjusting for other patient characteristics using Cox regression models, with the youngest age group as the referent. To account for the competing risk of death in the models for hospitalization and SNF admission, the cumulative incidence function was calculated. Gray tests were used to assess the significance of the differences between groups (14). The assumption of proportionality for covariates were tested for and met. Missing variables were rare (<1%); imputation of missing values was performed using the study group median for continuous variables and the most common value for categorical variables.
In the NCDR ICD registry, 38,305 patients were identified as Medicare beneficiaries receiving an ICD for secondary prevention, as designated by the implanting physician between 2006 and 2009. Of this group, patients were included if they had a prior episode of: 1) tachycardic arrest; 2) sustained, monomorphic VT; or 3) sustained polymorphic VT. Patients were excluded if they had either: 1) documentation of only a bradycardic arrest (n = 655); or 2) a previous ICD (n = 16,899). This resulted in 12,420 Medicare patients ≥65 years of age with available outcomes data as the primary study cohort.
Patient characteristics stratified by age are presented in Table 1. Each of the age groups consisted of an approximately equal number of patients. In the total cohort, the majority was white (90.9%), and ischemic heart disease was common (75.1%). Over one-third (40.5%) had a left ventricular ejection fraction >35%. There was a significantly higher prevalence of a history of heart failure and a lower prevalence of both chronic kidney disease and diabetes mellitus with increasing age (for all conditions, p for trend <0.001).
The implanting operator training and hospital characteristics are presented in Table 2. In the total cohort, board-certified electrophysiology physicians performed 68.6% of ICD implantations, which was similar in all strata. The majority (86.2%) of implantations were performed in a private or community hospital, with the remaining implantations performed in a university (12.5%) or government (1.3%) hospital setting. Slightly more than one-half (55.3%) of implantations were performed at institutions designated as teaching hospitals.
Rates of death increased with increasing age (Central Illustration). The rate at 30 days in the youngest group (65 to 69 years of age) was 1.5% compared with 3.0% in those at least 80 years of age (p for trend <0.001) (Table 3). At 1 year, the overall death rate was 14.3%, ranging from 9.9% in the youngest patients to 18.9% in the oldest (p for trend <0.001). At 2 years, the overall death rate was 21.8%, increasing progressively across age groups from 14.7% in the youngest group to 28.9% in the oldest (Table 3). In multivariable models with 65 to 69 years of age as the referent group, the risks of death at 2 years increased significantly with age (Table 4); the adjusted risk ratio for those at least 80 years of age was 2.01 (95% confidence interval [CI]: 1.85 to 2.33; p for trend <0.001).
The cumulative incidence of hospitalization for all causes within 30 days of ICD implantation was 16.2%, and increased with increasing age (Central Illustration), ranging from 14.6% in the youngest group to 18.5% in the oldest (p for trend <0.001) (Table 3). At 2 years, hospitalization occurred in almost two-thirds of the entire cohort (65.4%), with a range of 60.5% for 65 to 69 years of age to 71.5% for those at least 80 years of age; the adjusted risk ratio for the oldest age group was 1.27 compared with the youngest group (95% CI: 1.19 to 1.36; p for trend <0.001) (Table 4).
Rates of hospitalization for heart failure within 30 days occurred in 3.1% of the entire cohort, and similarly increased significantly with increasing age (Central Illustration), ranging from 2.4% in the youngest group to 3.6% in the oldest (p for trend <0.001) (Table 3). At 2 years, the cumulative incidence of hospitalization for heart failure was 18.8% overall, with a range of 14.7% for 65 to 69 years of age to 23.5% for those at least 80 years of age, corresponding to an adjusted risk ratio of 1.50 (95% CI: 1.33 to 1.69; p for trend <0.001) (Table 4).
The cumulative incidence of admission to a SNF also increased progressively with advancing age, and was greatest in the first 30 days following device implantation (Central Illustration). At 30 days, the cumulative incidence of SNF admission overall was 8.8%, and varied from 4.6% in the youngest patients to 14.3% in the oldest group (Table 3). At 2 years, the cumulative incidence of SNF admission was 13.1% in the 65- to 69-year-old age group versus 31.9% of those who were 80+ years of age, corresponding to an adjusted hazard ratio of 2.67 (95% CI: 2.37 to 3.01; p for trend <0.001) (Table 4).
This study provides insights into a range of health outcomes in a large contemporary study group of older persons in the United States receiving an ICD for secondary prevention of sudden cardiac death. Although nearly 4 in 5 survived more than 2 years, almost two-thirds were hospitalized and more than 1 in 5 were admitted to a SNF over that time. The risks for these outcomes increased significantly across the age spectrum, especially for death and SNF admission. The risk for SNF admission was particularly high within the first month after implantation in all age groups. After accounting for other factors, those at least 80 years of age were at twice the risk for death and admission to a SNF, and also had significantly higher risks of hospitalizations for all causes and for heart failure compared with patients 65 to 69 years of age. Hospitalizations for heart failure constituted a minority of all hospitalizations. These data provide a perspective on the magnitude of the expected outcomes of older patients receiving secondary prevention ICD therapy, and identify substantial care needs after device implantation in this population.
The risks of death among older patients following implantation of an ICD for secondary prevention indications in contemporary practice are not well characterized. The randomized trials that assessed the efficacy of secondary prevention ICDs enrolled few older patients (4–7): in a pooled analysis of these trials, only 252 patients (14%) of those enrolled were at least 75 years of age (15). The applicability of the survival patterns of treated patients from these trials published 2 decades ago is further limited by the subsequent evolution in device technology, patient selection, and medical therapy for the underlying conditions that predispose to cardiac arrest, such as left ventricular systolic dysfunction. The risk of death for patients at least 75 years of age in the randomized trials comparing defibrillator therapy with antiarrhythmic therapy was approximately 35% among those receiving an ICD, which is substantially higher than that of the older patients in this study, even for those older than 80 years of age. Thus, the present study provides a different perspective on survival than that available from the trials.
Observational data on outcomes in the elderly after secondary prevention ICD placement are also limited. A cohort study from Ontario included 1,102 patients at least 60 years of age (of whom only 621 were at least 70 years of age) treated with an ICD for secondary prevention in Ontario (16). In this cohort, the proportion of deaths at 2 years was lower than in the present study (18% vs. 29% among those at least 80 years of age). This difference may in part reflect a higher comorbidity profile in the NCDR study group in the United States, which had higher rates of coexisting illnesses, including chronic lung disease and renal insufficiency. Our study also found greater risks for hospitalizations, both for all causes and for heart failure with increasing age. Moreover, the cumulative incidence of hospitalization for all causes was more than 3 times that for heart failure, suggesting a broad range of care needs in this patient population.
Outcomes of patients receiving ICDs for primary prevention provide an additional frame of reference. In a study from the NCDR of an older cohort receiving a primary prevention device between 2006 and 2010 (mean age 75 years), rates of death and hospitalization for any cause at 6 months were 6.8% and 35.3%, respectively, which are similar to the outcomes in the present study (3). In a Danish national study of patients receiving an ICD between 2000 and 2012, only one-third was older than 70 years of age. Death rates were comparable between those receiving primary and secondary prevention devices (5% at 1 year), but the characteristics of patients were not stratified by indication, limiting comparisons (17).
This is the first study to assess admission to a SNF, an outcome of particular relevance to the older population. The reasons for the use of a SNF undoubtedly vary widely and may initially represent short stays following hospitalization, but the increasing rate over time may reflect the loss of functional independence. The highest risk of SNF admission occurred within the first 30 days of device implantation, and the cumulative incidence of admission within 30 days was nearly 15% in the oldest patients. However, the risk of SNF admission also increased over time, with a cumulative incidence of more than 20% in the entire cohort at 2 years and more than 30% in the oldest patients; these high rates of SNF use are indicative of patients with advancing frailty and substantial health care needs who receive a therapy principally aimed at prolonging life. Although these data provide a useful perspective on what patients receiving this therapy might expect, an understanding of their physical function and quality of life would further support shared decision making for patients considering ICD therapy in the context of secondary prevention.
The data from this cohort are pertinent to the clinical guidelines for device-based therapy, which recommend only considering an ICD for patients with “a reasonable expectation of survival with a good functional status for more than 1 year” (8). More than 85% of the cohort survived at least a year, including more than 80% of those older than 80 years of age at the time of implantation. Given the challenges of ascertaining the risks of death prospectively, these rates suggest that the clinicians involved in the care of the patients represented in this cohort are relatively circumspect in their selection of candidates for secondary prevention ICD therapy in the older population. Although admission to SNF cannot necessarily be equated with a loss of physical function, the frequency of such admissions suggests that a better understanding of the factors leading to a loss of function among older persons may be useful in guiding the use of ICD technology.
Because this study did not include a comparison group of patients who did not receive an ICD, it is not possible to make inferences about the effectiveness of ICD therapy. However, the data provide an important perspective on outcomes among older patients who receive this therapy in contemporary U.S. practice. Given the large number of ICDs placed for secondary prevention (2), these findings are useful, particularly in light of the sparse published data otherwise available. Although this study expands the published reports, patient-reported health status outcomes and functional status were not collected. Device therapy data were also not available for the cohort. The cohort only included patients receiving ICD therapy; thus the study cannot assess the impact of ICDs on survival or describe the characteristics of older survivors of sudden cardiac death who did not receive this therapy. Finally, although the NCDR collects many cardiac and noncardiac variables that were included in the multivariable models, we cannot exclude the impact of confounding by unmeasured variables that could influence the observed relationship between age and outcomes.
In this large, nationally representative study group of older patients treated with ICD therapy for secondary prevention in a national U.S. registry, almost 8 in 10 patients survived 2 years, although the risk of death increased significantly with age. Rates of admissions to hospital and SNFs were high, providing a perspective of the substantial health care needs of this population after receiving an ICD for a secondary prevention indication.
COMPETENCY IN MEDICAL KNOWLEDGE: Patients surviving cardiac arrest are often evaluated for an ICD for the secondary prevention of recurrent sudden cardiac death. In current U.S. practice, a large proportion of these patients is elderly and may have multiple coexisting illnesses. Understanding the potential outcomes is necessary to inform shared decision making with such patients.
TRANSLATIONAL OUTLOOK: Additional research is needed to more precisely define which particular coexisting conditions have the greatest impact on a variety of outcomes including death, hospitalization, and loss of physical function in patients receiving a secondary prevention ICD.
Dr. McManus is supported by National Institutes of Health grants 1U01HL105268-01, 1UH2TR000921-02, 1R15HL121761-01A1, and KL2TR000160. Dr. McManus owns equity in ATRIA, Inc. and Mobile Sense, Inc.; has received grant funding from Philips, Biotronik, the Department of Defense, the National Institutes of Health, and the National Science Foundation; and has served as a consultant for Bristol-Myers Squibb. Dr. Hansen has received research grants from TrygFonden, Helsefonden, and The Laerdal Foundation. Dr. Masoudi has a contract with the American College of Cardiology as the Chief Science Officer of the National Cardiovascular Data Registries. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- confidence interval
- implantable cardioverter-defibrillator
- National Cardiovascular Data Registry
- skilled nursing facility
- ventricular tachycardia
- Received February 25, 2016.
- Revision received October 18, 2016.
- Accepted October 18, 2016.
- American College of Cardiology Foundation
- Masoudi F.A.,
- Ponirakis A.,
- Yeh R.W.,
- et al.
- Borne R.T.,
- Peterson P.N.,
- Greenlee R.,
- et al.
- Connolly S.J.,
- Gent M.,
- Roberts R.S.,
- et al.
- Kuck K.H.,
- Cappato R.,
- Siebels J.,
- et al.
- Connolly S.J.,
- Hallstrom A.P.,
- Cappato R.,
- et al.
- Epstein A.E.,
- DiMarco J.P.,
- Ellenbogen K.A.,
- et al.
- Tracy C.M.,
- Epstein A.E.,
- Darbar D.,
- et al.
- Messenger J.C.,
- Ho K.K.,
- Young C.H.,
- et al.,
- NCDR Science and Quality Oversight Committee Data Quality Workgroup
- Healey J.S.,
- Hallstrom A.P.,
- Kuck K.H.,
- et al.
- Yung D.,
- Birnie D.,
- Dorian P.,
- et al.