Author + information
- Received January 22, 2016
- Revision received May 16, 2016
- Accepted June 2, 2016
- Published online January 16, 2017.
- David F. Katz, MDa,b,∗ (, )
- Pamela Peterson, MDa,b,c,
- Ryan T. Borne, MDa,b,
- Jarrod Betz, MDd,
- Sana M. Al-Khatib, MD, MHSe,
- Paul D. Varosy, MDa,b,f,
- Yongfei Wang, MSg,
- Jonathan C. Hsu, MD, MASh,
- Kurt S. Hoffmayer, MDi,
- Ryan T. Kipp, MDi,
- Carolina Malta Hansen, MDe,
- Mintu P. Turakhia, MD, MASj,k and
- Frederick A. Masoudi, MD, MSPHa,b
- aDivision of Cardiology, University of Colorado, Denver, Colorado
- bColorado Cardiovascular Outcomes Research Group, Denver, Colorado
- cDenver Health Medical Center, Denver, Colorado
- dDepartment of Medicine, University of Colorado, Denver, Colorado
- eDivision of Cardiology, Duke University, Durham, North Carolina
- fEastern Colorado VA Healthcare System, Denver, Colorado
- gYale University and Center of Outcomes Research and Evaluation, New Haven, Connecticut
- hDivision of Cardiology, University of California San Diego, San Diego, California
- iDivision of Cardiology, University of Wisconsin, Madison, Wisconsin
- jVA Palo Alto Health Care System, Palo Alto, California
- kDivision of Cardiology, Stanford University, Palo Alto, California
- ↵∗Reprint requests and correspondence:
Dr. David F. Katz, Division of Cardiology, University of Colorado, 12631 East 17th Avenue, Aurora, Colorado 80045.
Objectives This study sought to define the characteristics and risks of death of patients receiving a physician-designated secondary prevention implantable cardioverter-defibrillator (ICD) in contemporary clinical practice.
Background Data on utilization and outcomes of ICDs for the secondary prevention of sudden cardiac death (SCD) are limited.
Methods Patients enrolled in the National Cardiovascular Data Registry’s (NCDR) ICD Registry from 2006 to 2009 with a physician-designated secondary prevention indication for ICD implantation were identified and linked to the Social Security Death Master File. Those patients with a history either of tachycardic arrest or sustained ventricular tachycardia (SCD/VT) or of syncope without SCD/VT were included. Kaplan-Meier survival analysis was used to assess mortality. Cox proportional hazards survival modeling was used to assess the risk of death in these groups, adjusting for patient characteristics.
Results In the study cohort of 46,685 patients (mean age 66 ± 14 years, 73.5% male, 85% white), 78% had SCD/VT and 22% had syncope. Overall mortality was 10.4% at 1 year and 16.4% at 2 years. Compared with patients having SCD/VT, the adjusted hazard of death at 1 year was lower in the patients having syncope (hazard ratio: 0.89; 95% confidence interval: 0.83 to 0.96) but was not significantly different by 2 years (hazard ratio: 0.96; 95% confidence interval: 0.90 to 1.01).
Conclusions Nearly 9 of 10 patients receiving a secondary prevention ICD in clinical practice are alive 1 year after implantation. The risk of death varies by indication and is highest among patients who survive SCD or sustained VT in the first year after device implantation.
Clinical practice guidelines recommend consideration of an implantable cardioverter-defibrillator (ICD) for either primary or secondary prevention of sudden cardiac death (SCD) in selected patients, provided they have an estimated life expectancy of 1 year with a reasonable quality of life (1,2). Although the efficacy of ICDs for primary prevention of SCD has been established in several randomized controlled trials (RCTs) (3,4) and observational research has demonstrated outcomes in clinical practice that are similar to those observed in the primary prevention RCTs (5), research on the use of ICDs for secondary prevention of SCD is more limited.
Recommendations regarding the use of ICDs for secondary prevention of SCD rely on information from a small number of RCTs that were performed decades ago, with mixed results (6–8). Those trials were restricted to patients with a history of SCD or documented spontaneous sustained monomorphic ventricular tachycardia (VT), and only 1 study included a small subgroup of patients with structural heart disease and syncope (8). Current practice guidelines recommend the use of ICDs for secondary prevention in such patients with syncope and therefore include a broader population than was enrolled in most trials. Furthermore, since publication of these trials, medical therapy, ICD technology, and procedural care have evolved significantly. Despite the evolution of indications for secondary prevention ICDs and improvements in care, a contemporary understanding of the characteristics and outcomes of patients selected for this therapy is lacking.
More than 1 in 5 of the nearly 140,000 ICDs included annually in the National Cardiovascular Data Registry’s (NCDR) ICD Registry are inserted for secondary prevention indications (9). However, this population of patients and their outcomes after implantation have not been well defined. Using data from the registry and from the Social Security Death Master File, this study sought to define the characteristics and risks of death of patients receiving a physician-designated secondary prevention ICD in contemporary clinical practice.
Patients included in this study were enrolled in the NCDR ICD Registry. The registry, formed in partnership between the Heart Rhythm Society and the American College of Cardiology, includes data on patients receiving an implantable device across hospital and payer types. Although not mandated for secondary prevention indications, 91% (1,320 of 1,465) of participating sites enter data on secondary prevention ICDs. Clinical, demographic, and procedural information is collected using standardized data elements and definitions. Data are submitted by participating hospitals using certified software. Data quality is examined using a formal Data Quality Reporting and audit process (10). Mortality data were obtained by linking NCDR registry files with the Social Security Death Master File as previously described (11). The Social Security Death Master File captures 93% to 96% of deaths of individuals ages 65 years or older and has slightly lower capture rates in younger cohorts (12). Analyses of the NCDR ICD Registry were performed under an institutional review board approval by Yale University with a waiver of informed consent because of the design of the study.
All patients enrolled in the NCDR ICD Registry from 2006 to 2009 were identified. Patients were included in the study if they were designated as receiving an ICD for secondary prevention of SCD after documented tachycardic arrest, sustained VT, or syncope. Patients with a previous ICD, those with a recorded history of cardiac arrest because of bradyarrhythmia in the absence of documented arrest for tachyarrhythmia, and those undergoing ICD implantation for primary prevention of SCD were excluded. Survival data were determined through the Social Security Death Master File, and patients with invalid information, such as Social Security number, name, or gender, were excluded.
Indication for ICD implantation
The designation of secondary prevention, as determined by the implanting physician, is captured in the NCDR ICD Registry. Based on contemporary guideline indications for the use of ICDs, patients were categorized into 2 groups: 1) those with tachycardic arrest or a history of sustained monomorphic VT (SCD/VT group); and 2) those with syncope but without documented SCD/VT (syncope group) (1).
In addition to a variable indicating the patient category (SCD/VT vs. syncope), covariates considered included patient, clinician, and hospital characteristics. Patient-level characteristics included demographics (age, sex, race, and insurance payer), comorbidities and risk factors, including 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, and renal failure (hemodialysis); diagnostic information, which included left ventricular ejection fraction (LVEF), whether an electrophysiological study was performed, and, if so, whether a sustaining ventricular arrhythmia was induced, serum creatinine, serum blood urea nitrogen, and serum sodium levels, and systolic blood pressure; and type of ICD implanted (single chamber vs. dual chamber vs. biventricular). Discharge medications considered included angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers, beta-blockers, and antiarrhythmic medications (amiodarone vs. other). Implanting clinician training was considered, as were hospital characteristics including hospital type (private/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).
To avoid case-wise deletions in the multivariable models, specific approaches were used on the basis of the extent of missing values. For most variables missing values were rare (<1% missing values for all variables except for LVEF), and missing values were imputed as the most common values for categorical values and the median for the continuous variables. LVEF, which had 2% of values missing, was treated as a categorical variable with a separate missing category (≤35%, >35%, and missing).
The primary outcome was time to death at 1 year from the procedure. Time to death was also assessed at 3 months and at 2 years.
Patient, physician, and hospital characteristics were compared among the different guideline-based indication groups using the chi-square test for categorical variables and the F test in analysis of variances for continuous variables.
Survival function curves were used to evaluate unadjusted survival among the overall cohort and for the hierarchical groups described earlier. Survival was compared among the hierarchical groups using a Wald chi-square test in Cox proportional hazards survival model analysis.
Factors significantly associated with the time to death outcome were identified using Cox proportional hazards survival analyses in the overall cohort with backward stepwise selection. All of the aforementioned predictor variables were considered. The hazards of death in the syncope group were evaluated adjusting for patient characteristics using Cox proportional hazards survival models sequentially using the SCD/VT subgroup as the referent.
Of the 106,316 patients enrolled in the NCDR ICD registry from 2006 to 2009 with a physician-designated secondary prevention indication, 44,723 were excluded because they were not undergoing first-time implantations. An additional 1,646 patients had a documented bradycardic arrest, and 2,248 patients lacked valid data to link with the Social Security Death Master File. An additional 11,014 patients lacked a guideline-defined secondary prevention indication (SCD/VT or syncope), resulting in a study cohort of 46,685 patients (43.9% of all secondary prevention ICD recipients in the registry) (Figure 1A). The mean patient age was 65.9 ± 14 years (interquartile range), and the population was predominantly white (85.0%) and male (73.5%). Most (63.9%) had ischemic heart disease, and 36.3% had LVEF >35%. The indication for secondary prevention ICD implantation was SCD/VT in 78% and syncope in the remaining 22%.
The characteristics of the cohort and the subgroups are presented in Table 1. The group with SCD/VT was younger (age 65.4 ± 12.8 years), and a large proportion had preserved LVEF (41.9% with LVEF >35%). Cardiac arrest was documented in 65.2% of this group, with the remainder having documented sustained VT. Of note, 41.7% of subjects in the SCD/VT group also had a documented history of syncope but remained in the SCD/VT group according to the hierarchical grouping as previously described. Among patients in the syncope group, prevalent characteristics included nonischemic cardiomyopathy (29.4%), ischemic heart disease (59.0%), and severely depressed left ventricular systolic function (LVEF ≤35% in 62.2%). Inducible sustained ventricular arrhythmias were also common (21.5% of the entire syncope group; 57.6% of patients in the syncope group had undergone an electrophysiological study).
Approximately 78.8% of implantations were performed by board-certified (72.9%) or fellowship-trained noncertified (5.9%) electrophysiologists. Regionally, the fewest procedures were recorded in the Mountain West region (5.6%) and the most in the Southern Atlantic states (21.2%). Most were performed in urban hospitals (63.9%) and at community hospitals (84.5%). Implantations at university hospitals accounted for 14.3%, with only a small fraction performed at government hospitals (1.2%).
ICD type (single chamber vs. dual chamber vs. biventricular) was not associated with mortality at 3 months or at 1 year. At 2 years, recipients of a dual-chamber ICD had lower hazard of mortality (hazard ratio [HR]: 0.89; 95% confidence interval [CI]: 0.84 to 0.94) than did those who received a single-chamber device.
Mortality after ICD implantation in the total study population was 4.1% at 3 months, 10.6% at 1 year, and 16.4% at 2 years (Figure 1B). The risk of death differed between the 2 subgroups (Figure 1C). At 3 months and at 1 year, mortality was greater in the SCD/VT group than in the syncope group (4.4% vs. 3.3%, p < 0.001; and 10.9% vs. 9.2%, p < 0.001), but differences in mortality in these groups had decreased by 2 years after device implantation (SCD/VT 16.5% vs. 16% syncope, p = 0.257). In multivariable models adjusting for demographic and clinical characteristics, patients in the syncope subgroup had significantly lower hazard of death at 3 months (HR: 0.85; 95% CI: 0.75 to 0.96) and at 1 year (HR: 0.89; 95% CI: 0.83 to 0.96) compared with those in the SCD/VT group. At 2 years, there was no significant difference in hazard of death between the SCD/VT and syncope groups (HR: 0.96; 95% CI: 0.90 to 1.01).
In the overall cohort, black race, increased age, and comorbidities including history of heart failure, cerebrovascular disease, lung disease, diabetes, renal failure, atrial arrhythmias, ischemic heart disease, and worse New York Heart Association functional class were associated with higher risk of death. Of note, nonischemic cardiomyopathy was associated with a lower risk of death, as was higher ejection fraction (defined here as >35%) and use of an angiotensin-converting enzyme inhibitor, angiotensin receptor blocker, beta-blocker, or statin. Recent myocardial infarction (<40 days) was associated with a higher hazard of death at 3 months and at 1 year, but this association was no longer present by 2 years. The results of the multivariable models assessing factors associated with time to death at 3 months, 1 year, and 2 years are presented in Table 2.
Among patients enrolled in the NCDR ICD Registry who received an ICD for a physician-designated secondary prevention indication, a substantial majority of patients had a documented guideline-defined secondary prevention indication, and most of those patients (78%) had documented SCD or sustained VT. Mortality approached 10% at 1 year, which, despite advances in medical therapy, is similar to mortality rates described in the secondary prevention RCTs (6–8). Finally, differences in mortality between the subgroups and the persistence of these differences after accounting for differences in patient characteristics demonstrate that the indication for secondary prevention ICD implantation has prognostic value for at least the first year after implantation.
We found that the indications for patients receiving a secondary prevention ICD in clinical practice frequently include syncope without a history of SCD/VT. The RCTs predominantly included patients with prior tachycardic arrest or sustained VT; only 1 trial included a small subgroup of patients with syncope and structural heart disease. The ACC/AHA/HRS 2008 Guidelines for Device-Based Therapy of Cardiac Rhythm Abnormalities state that ICD placement for secondary prevention is a Class I indication (i.e., the procedure should be performed) for survivors of SCD or spontaneous sustained VT and for those with syncope and inducible sustained VT. ICD placement is a Class IIa recommendation (i.e., the procedure is reasonable to perform) for patients with structural heart disease and syncope thought to be arrhythmic in origin, without specification as to whether this should be considered a primary or secondary prevention indication (1). In the NCDR registry, among patients undergoing physician-designated secondary prevention ICD implantation from 2006 to 2009, 11,014 (19%) of enrollees did not have a clearly documented secondary prevention indication, although many of them may have had a primary prevention indication. Of the 46,685 patients with a documented secondary prevention indication, 38,643 (82.8%) met a class I indication for ICD implantation for secondary prevention. The remaining 8,042 (17.2%) had syncope without inducible arrhythmia and therefore met a Class IIa indication for ICD implantation.
The nearly 90% 1-year survival rate reported in this study is commensurate with outcomes reported in a pooled analysis of secondary prevention RCTs (13) but is higher than those reported in the limited observational studies of secondary prevention (14,15). Dutch investigators reported 1-year mortality of 4% to 6% after implantation of a secondary prevention ICD in a slightly younger patient population with a higher burden of ischemic cardiomyopathy and lower prevalence of atrial arrhythmias, with more than half of secondary prevention ICD recipients having appropriate ICD therapy by 5 years of follow-up (14,15). In Ontario, no difference in mortality was observed between men and women, although very elderly secondary prevention patients (>80 years of age) had a significantly higher risk of death (>10%) within 1 year (16,17).
Survival rates in our study population varied by clinical phenotype. At 1 year, patients in the SCD/VT subgroup fared slightly worse than those without this indication. By 2 years the outcome of patients with syncope and structural heart disease was statistically similar to those in the SCD/VT group.
Defining a context within which to understand the mortality rates described in this study is challenging. A recent Danish study of patients who received an ICD for primary or secondary prevention suggested slightly better long-term survival for those treated with an ICD, with similar survival between primary and secondary prevention populations (18). An NCDR study on survival after implantation of a primary prevention ICD reported all-cause mortality at 6 months approached 7%, similar to our findings in a secondary prevention population, but long-term survival was not reported in that study. However, that study examined only Medicare patients, which yielded an older and less heterogeneous population than the one we examined (19). Comparison of populations of identified SCD patients is more challenging. In a cohort of patients receiving early defibrillation in Olmstead County, Minnesota, 39.5% survived to hospital discharge. Forty-seven percent of the patients studied had a reversible cause of SCD, whereas 44.3% went on to receive an ICD. Overall mortality in the cohort receiving an ICD was not reported at 1 or 2 years but appears to be similar to the rates reported in this study, being 21% at 5 years (20).
The RCTs that established the use of ICDs for secondary prevention were published in 1997 and 2000 and may provide the most appropriate context in which to understand the outcomes reported in this study (6–8). A meta-analysis demonstrated that the population of patients enrolled in these trials, the majority of whom had ischemic cardiomyopathy, had low rates of beta-blocker and angiotensin-converting enzyme inhibitor use (15). By comparison, the more heterogeneous population of this study had higher rates of beta-blocker and angiotensin-converting enzyme inhibitor or angiotensin receptor blocker use. Improved outcomes might be anticipated with the improvements in medical care, ICD technology, programming to prevent unnecessary or inappropriate ICD shocks, and ease of implantation (9% of patients were randomized during the era of epicardial ICD placement) in the years separating our analysis from these trials; however, the mortality rates for recipients of secondary prevention ICDs in contemporary clinical practice are similar to those reported in the RCTs.
The results of this study should be considered in the context of specific limitations. Because registry data collection during the time period in question did not include the identification of predisposing conditions to SCD, such as long QT syndrome, hypertrophic cardiomyopathy, or arrhythmogenic right ventricular cardiomyopathy, survival patterns in these subgroups cannot be ascertained. Second, the relationships between the clinical subgroups and mortality may have been subjected to confounding because of unmeasured characteristics or imprecision in the data elements collected in the registry, and use of the Social Security Death Master File precludes examination of cause of death. Third, although inclusion of patients in the NCDR is required nationally as a condition for reimbursement to Medicare beneficiaries undergoing primary prevention ICD implantation, there is no such requirement for secondary prevention ICDs. However, 91% of participating sites include secondary prevention ICDs in their data collection. Fourth, our study does not include a comparison cohort of patients not receiving an ICD. The identification of an equivalent cohort, even if clinically detailed data were available, likely would be challenging. Finally, we were not able to ascertain either mode of death or device therapies, including the incidence of appropriate ICD shocks, which would provide a broader perspective on the efficacy of ICDs for prevention of sudden death and treatment of arrhythmias among their recipients.
This study is the first to describe the characteristics of a U.S. national cohort of patients receiving a secondary prevention ICD in contemporary practice and their outcomes. The clinical indications for use of secondary prevention ICDs in contemporary practice frequently include patients with syncope without prior SCD/VT. One-year survival after implantation of an ICD for secondary prevention approaches 90%. Although mortality is higher for ICD recipients with a history of SCD/VT than it is for ICD recipients with a history of syncope alone at 3 months and at 1 year, mortality rates in these groups were similar by 2 years of follow-up.
COMPETENCY IN MEDICAL KNOWLEDGE: ICDs are commonly inserted in clinical cardiovascular practice. The outcomes of patients undergoing ICD implantation for primary prevention of SCD have been studied extensively, but those for patients receiving a device for a secondary prevention indication are less well characterized. In this U.S. national cohort of 46,685 patients from the NCDR ICD Registry receiving an ICD for a secondary prevention indication, nearly 90% were are alive at 1 year. Despite the greater heterogeneity in indications for implantation, the risk of death in this cohort is similar to that described in the randomized controlled trials that established the use of ICDs for secondary prevention of SCD conducted more than 2 decades ago.
TRANSLATIONAL OUTLOOK 1: The risk of death after implantation of an ICD for secondary prevention of SCD is similar in contemporary practice to that of the randomized controlled trials that established the use of ICDs for this indication; however, further investigation is required to determine the causes of death in this population.
TRANSLATIONAL OUTLOOK 2: Additional research is needed to determine which populations receive the greatest (and the least) benefit from secondary prevention ICD implantation.
Dr. Katz has received product donations from Medtronic for work in Zimbabwe. Dr. Hsu has received honoraria from St. Jude Medical, Medtronic, and Biotronik; and has served on advisory boards for Janssen Pharmaceuticals and Bristol-Myers Squibb. Dr. Hansen receives unrestricted grants from TrygFonden and Helsefonden; and a research grant from The Laerdal Foundation. Dr. Turakhia receives grant support from Medtronic, iRhythm, Gilead Sciences, the American Heart Association, the Department of Veterans Affairs, SentreHeart, and the Dr. Jeffrey Thomas Stroke Shield Foundation; and consultant fees from Medtronic and St. Jude Medical. Dr. Masoudi has a contract with the American College of Cardiology as the Chief Science Officer of the National Cardiovascular Data Registries.
All authors attest they are in compliance with human studies committees and animal welfare regulations of the authors' institutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information, visit the JACC: Clinical Electrophysiology author instructions page.
- Abbreviations and Acronyms
- electrophysiological study
- implantable cardioverter-defibrillator
- left ventricular ejection fraction
- myocardial infarction
- randomized controlled trial
- sudden cardiac death
- ventricular tachycardia
- Received January 22, 2016.
- Revision received May 16, 2016.
- Accepted June 2, 2016.
- 2017 American College of Cardiology Foundation
- Epstein A.E.,
- DiMarco J.P.,
- Ellenbogen K.A.,
- et al.
- Tracy C.M.,
- Epstein A.E.,
- Darbar D.,
- et al.
- Connolly S.J.,
- Gent M.,
- Roberts R.S.,
- et al.
- Kuck K.H.,
- Cappato R.,
- Siebels J.,
- Ruppel R.
- Messenger J.C.,
- Ho K.K.L.,
- Young C.H.,
- et al.
- Hess P.L.,
- Hellkamp A.S.,
- Peterson A.S.,
- et al.
- Hill M.E.,
- Rosenwaike I.
- Connolly S.J.,
- Hallstrom A.P.,
- Cappato R.,
- et al.
- Borleffs C.J.W.,
- van Erven L.,
- Schotman M.,
- et al.
- Yung D.,
- Birnie D.,
- Dorian P.,
- et al.
- Borne R.T.,
- Peterson P.N.,
- Greenlee R.,
- et al.