Impact of Age and Medical Comorbidity on the Effectiveness of Implantable Cardioverter-Defibrillators for Primary PreventionCLINICAL PERSPECTIVE
Background— Although implantable cardioverter-defibrillators (ICDs) reduce mortality in primary prevention patients with left ventricular systolic dysfunction, recent studies have questioned their overall role in clinical practice, especially in older patients and those with major comorbid conditions.
Methods and Results— In a prospective cohort of 965 patients with ischemic and nonischemic cardiomyopathies (ejection fraction ≤35%) and no prior ventricular arrhythmias, we compared long-term mortality in patients who did (n=494 [51%]) and did not receive ICDs over a mean follow-up period of 34±16 months. Using a landmark analysis, multivariable Cox proportional hazards models that included propensity scores for ICD implantation assessed the relationship between ICD therapy and mortality in the entire cohort and by age and the presence of major comorbid conditions. Data from these analyses were then used as inputs in a Markov model to generate incremental cost-effectiveness ratios for ICD therapy. Patients who received ICDs were similar in age and prevalence of most major comorbid conditions, including symptomatic heart failure. After multivariable adjustment, ICD therapy was associated with a 31% lower risk for all-cause mortality (adjusted hazard ratio, 0.69; 95% CI, 0.50 to 0.96; P=0.03). The relationship between ICD therapy and lower all-cause mortality was consistent after stratification by age (<65, 65 to 74, and ≥75), ischemic etiology, ejection fraction (>25% versus ≤25%), and the presence of major comorbid conditions (probability values for all interactions >0.05). Incremental cost-effectiveness ratios for ICD therapy were similar between patients aged ≥75 years and younger patients but rose slightly in those with multiple comorbid conditions.
Conclusions— Routine use of ICDs in primary prevention patients with left ventricular systolic dysfunction was associated with lower all-cause mortality, even among older patients and those with major comorbid conditions. Although their use needs to be individualized, our findings suggest that these groups should not be routinely excluded from ICD treatment.
Received July 14, 2008; accepted October 28, 2008.
Although implantable cardioverter-defibrillators (ICDs) have been shown to reduce mortality in primary prevention patients with left ventricular dysfunction,1,2 their effectiveness in settings outside of randomized clinical trials is less clear. Because clinical trial participants are carefully selected, have close follow-up with optimized medical regimens, and are more likely to be enrolled in high-volume centers, results of clinical trials may not always apply to real-world clinical settings.3,4 In addition, the application of an efficacious therapy to community settings may include patients for whom there is less benefit, resulting in a dilution of its effectiveness. The need to establish the generalizability and real-world effectiveness of ICD therapy is particularly important given its cost.
Editorial see p 6
Clinical Perspective see p 16
Recent studies suggest that the use of ICD therapy for primary prevention in routine clinical practice may not be associated with similar mortality benefits as found in randomized clinical trials.5 In particular, it has been suggested that the elderly and those with major comorbid illness may not benefit as much from the use of ICDs.5,6 However, these studies used administrative databases that included a mixed population of primary and secondary prevention patients6 or lacked vital clinical information, such as left ventricular ejection fraction (LVEF), to accurately define a primary prevention cohort.5,6 As a result, they were unable to rigorously evaluate whether patients with higher overall mortality risk also derived substantial mortality benefit from ICD therapy. Although competing risks in older, sicker populations may limit the benefit of ICDs, it is also plausible that these patients may have the most to gain given their higher absolute risk for all-cause and arrhythmic mortality. Growing use of ICDs for primary prevention and the costs associated with this therapy demand the need for comparative effectiveness studies in these high-risk patients.7
To examine the benefit of ICDs in patients at elevated risk for mortality due to advanced age or multiple competing comorbidities, we evaluated the association between the use of ICD therapy in primary prevention patients with known left ventricular systolic dysfunction and mortality in a community-based setting with a particular focus on understanding the effects of ICDs among older and higher risk subgroups.
Although implantable cardioverter-defibrillators (ICDs) reduce mortality in primary prevention patients with left ventricular systolic dysfunction, recent studies have questioned their overall role in clinical practice, especially in older patients and those with major comorbid conditions.
In this study, we found that the use of ICDs in a multisite “real-world” patient population was associated with similar reductions in all-cause mortality as was seen in clinical trials.
Although older patients and patients with major comorbid conditions were found to have higher baseline rates of all-cause mortality, we found that the relative and absolute mortality risk reductions associated with ICD therapy were similar, regardless of age (<65, 65 to 74, and ≥75 years) and the presence of most major comorbid conditions.
Use of ICDs was also found to be cost-effective regardless of age, even among patients 75 years or older. Among patients with multiple comorbidities, ICD therapy was associated with less favorable cost-effectiveness ratios among patients with no or ≥2 comorbidities.
Although decisions about the use of ICDs need to be individualized for a particular patient, our findings suggest that specific high-risk groups should not be routinely excluded from consideration for ICD implantation.
Consecutive patients with left ventricular systolic dysfunction (LVEF ≤35%) of ischemic or nonischemic etiology for at least 3 months were prospectively enrolled from 7 outpatient cardiology clinics by the Ohio Heart and Vascular Center and the Lindner Clinical Trials Center between March 2001 and June 2005 and followed through March 2007. Measurement of LVEF was determined via echocardiogram (>90%), nuclear imaging, or ventriculogram within 6 months of cohort enrollment. The ischemic cohort has been previously described in greater detail and included those patients with ≥70% stenosis on cardiac catheterization in at least 1 coronary vessel, documented myocardial infarction of greater than 30 days, or a history of coronary revascularization.8 Patients had to be 18 years or older and have no history of a prior ventricular arrhythmic event (sustained ventricular tachycardia, ventricular fibrillation, cardiac arrest, or syncope from defined arrhythmia). Patients with terminal illnesses (including malignant cancers) were excluded. Because we were also interested in the prognostic role of microvolt T-wave alternans (MTWA) in risk stratification, patients had to be in sinus rhythm at the time of enrollment. All patients gave informed consent to registry enrollment and follow-up, and the study was approved by the Institutional Review Board at Christ Hospital (Cincinnati, Ohio).
Data on demographic and clinical characteristics were collected at study enrollment and included the following: age, gender, LVEF, QRS duration >120 ms, diabetes mellitus, hypertension, symptomatic heart failure (HF), chronic obstructive pulmonary disease, chronic renal failure, peripheral vascular disease (PVD), and history of myocardial infarction, coronary revascularizarion therapy, stroke or transient ischemic attack, atrial fibrillation, or prior syncope from nonarrhythmic causes. In addition, baseline medication use of aspirin, angiotensin-converting enzyme inhibitor or angiotensin II receptor blocker, β-blocker, digoxin, diuretic, class I or III antiarrhythmic agent, statin, and spironolactone were obtained.
Holter monitoring (24 to 48 hours) was performed on all patients and was defined as positive if nonsustained ventricular tachycardia was >100 bpm for ≥3 consecutive beats and <30 seconds was detected. Baseline assessment for MTWA by treadmill exercise at study enrollment has been previously described,8 and results were interpreted according to standard criteria.9 Finally, patients underwent ICD placement during initial enrollment at the discretion of the attending physician, which was primarily based on established Medicare reimbursement criteria for ICD implant for primary prevention during that time period (eg, a QRS >120 ms on ECG or an abnormal Holter result with subsequent inducible ventricular arrhythmia on electrophysiological study).
Study End Points and Follow-up
The primary end point for the study was all-cause mortality. The secondary end point was cause-specific mortality, which was classified as arrhythmic or nonarrhythmic in etiology using a modified Hinkle-Thaler system.10 Arrhythmic deaths included unwitnessed deaths (if stable when last observed within 24 hours before death), witnessed instantaneous deaths, and sequelae of cardiac arrest. Cause-specific mortality was adjudicated by an independent panel of 2 research investigators blinded to a patient’s clinical characteristics and ICD status. If consensus could not be reached, the decision was referred to a third committee member. Clinical follow-up for mortality end points was achieved for all patients by quarterly office visits (97.3%), telephone contact (99.1%), routine review of office charts, and an annual query of the Social Security Death Index (100%).
The index date was the date of cohort enrollment. Median time from cohort enrollment to ICD implantation was 60 days (range, 32 to 101 days). Seventy-seven (15.8%) patients in the non-ICD cohort with at least 4 months of follow-up subsequently underwent ICD placement after the publication of the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT) in January 2005, which led to the expansion of ICD implant criteria by Medicare. Because we did not prospectively document reasons for ICD placement (eg, SCD-HeFT results, development of ventricular arrhythmia), we took an “intention-to-treat” approach, thereby considering these patients as crossovers but maintaining them in the non-ICD cohort for statistical analyses. As a sensitivity analysis, we found that our results were not meaningfully different when crossovers were considered in the ICD cohort.
Collection of data on clinical variables and outcomes was 100% complete. Baseline characteristics between the ICD groups were compared using ANOVA for continuous variables and χ2 tests for categorical variables. Unadjusted survival curves for the ICD groups were constructed using Kaplan-Meier estimates, and differences in event-free survival were assessed with the log-rank test. The assumption of proportionality was assessed with plots of the log-log (survival) versus log (survival time).
To avoid survival bias (as patients in the ICD group had to survive until the time of ICD implant), a landmark analysis was used. Landmark analysis is a form of survival analysis that classifies patients based on some a priori time point.11 Prognosis is then evaluated from this landmark time point. In our analyses, we defined landmark time as the elapsed time from the date of initial study enrollment (index date). We then defined the landmark time as 4 months, based on our study’s adherence to available clinical trial inclusion criteria at the time requiring patients to be at least 3 months from coronary revascularization before ICD implantation.2 In total, there were 15 patients in the non-ICD group and 6 patients in the ICD group who died during the landmark period and were excluded from the analyses.
Because allocation of ICD treatment was not randomized and because traditional multivariable analyses may not adequately adjust for potential differences between treatment groups, propensity score analyses were then used to evaluate the association of ICD therapy and mortality from the landmark time.12–14 A nonparsimonious propensity score was generated with multivariable logistic regression using all baseline variables to predict ICD placement (C-statistic, 0.78). To further control for potential residual bias, propensity score analyses with quintile subclassification was performed.14,15 The major advantage with this propensity score technique is that it may potentially remove more than 90% of unmeasured bias (ie, bias from unmeasured variables in a given study) and allows for further insights into the benefit of ICD therapy based on the likelihood of receiving an ICD.14,15 To accomplish this, the entire cohort was divided into quintiles based on their propensity scores for receiving an ICD. Multivariable Cox models then compared mortality rates between the ICD and non-ICD groups within each quintile, and a pooled estimate of overall effect was generated from the stratified (by propensity score quintile) Cox models. As a sensitivity analysis, we also performed a time-dependent covariate analysis, whereby the preimplant survival time in ICD patients was counted toward non-ICD survival and found no meaningful difference in the results (results not shown but available from authors).
We also conducted prespecified subgroup analyses to assess whether the benefit of ICD therapy differed by age or risk profile. The prespecified subgroups included the following: age (<65, 65 to 74, and ≥75), ischemic versus nonischemic etiology, LVEF (>25% versus ≤25%), the presence of comorbid conditions with known associations for mortality (symptomatic HF, diabetes mellitus, atrial fibrillation, PVD, prior stroke, chronic obstructive pulmonary disease, renal failure, prior syncope from nonarrhythmic causes), and noninvasive screening tests (abnormal Holter or MTWA findings). To better clarify the benefit of ICDs in patients with high-risk comorbid conditions, we also examined whether the benefit of ICD therapy differed depending on the number of coexisting comorbid conditions (0, 1, 2, and ≥3). In each comparison, adjusted Cox models were constructed with nonparsimonious propensity scores as described earlier, and a formal test of interaction examined whether ICD benefit was significantly different between subgroups.
In addition, to illustrate the absolute gains in survival achieved annually with ICD therapy by subgroup, we determined the subgroup-specific rates of all-cause mortality per year for non-ICD patients and applied the hazard ratios (HRs) from the Cox regression models as described earlier. We then examined the cost-effectiveness of ICD therapy for age and comorbidity subgroups by applying the HRs from our Cox regression models and subgroup-specific mortality rates to a validated Markov model from our prior work.16 In this model, we took a societal perspective on health costs and benefits and applied a 3% annual discount rate, consistent with recommendations for the conduct of cost-effectiveness analyses.17 Our model was based on a lifelong time horizon, and we assumed the benefits and costs of ICD therapy remained constant over time. Our model accounted for the frequency and costs of ICD generator replacement (every 6 years), lead malfunction (3% annually), and lead infection (1% annually).16 Finally, we age adjusted the annual mortality rates for the cohort using life tables from the National Center for Health Statistics.18 From this model, we estimated lifetime costs and quality-adjusted life-years (QALYs) for ICD and non-ICD patients and determined the incremental cost-effectiveness ratio (ICER) for ICD therapy as compared with no ICD therapy within each subgroup.
For all analyses, the null hypothesis was evaluated at a 2-sided significance level of 0.05. All statistical analyses were performed with SAS version 9.1 (SAS Institute, Cary, NC), TreeAge Pro 2006, and R version 2.3.1. The authors designed the study, collected and analyzed the data, and drafted and revised the manuscript. The study was funded by research grants from Medtronic, which had no involvement in the design, collection, management, or analysis of the study or in manuscript preparation. Dr Chan had full access to all of the data and takes full responsibility for the integrity of the data and the accuracy of the data analysis. All authors have read and agreed to the manuscript as written.
The study cohort comprised 965 patients (751 [77.8%] ischemic; 214 [22.2%] nonischemic), of whom 494 (51.2%) received ICDs. Patients in this cohort were significantly older than those in primary prevention clinical trials: SCD-HeFT (median age [interquartile range]: 67.3 [57.7 to 74.5] in this cohort versus 60.1 [51.6 to 68.4] in SCD-HeFT; P<0.001)1 and the Second Multicenter Automatic Defibrillator Implantation Trial (mean age of ischemic patients: 67.2±10 in this cohort versus 64.4±10 in Second Multicenter Automatic Defibrillator Implantation Trial; P<0.001).2
Mean cohort follow-up time was 33±16 months for the non-ICD group and 34±16 months for the ICD group. Overall, patients in both groups were similar with respect to their age and rates of major clinical comorbid conditions, prior revascularization, and medication usage. Compared with non-ICD patients, ICD patients had a lower baseline LVEF and were more likely to have a prolonged QRS duration, a prior history of myocardial infarction or symptomatic HF, and to be on digoxin and aldosterone blocker therapy. They were also more likely to have abnormal Holter and MTWA screening exams (Table 1).
There were 217 deaths (115 [24.4%] in the non-ICD group as compared with 102 [20.6%] in the ICD group). Of these, 105 were arrhythmic (59 [12.5%] in the non-ICD group as compared with 46 [9.3%] in the ICD group) and 112 were nonarrhythmic (56 [11.9%] in the non-ICD group as compared with 56 [11.3%] in the ICD group). Unadjusted Cox models identified a nonsignificant trend toward higher survival in the ICD group for the primary end point of all-cause mortality (unadjusted HR, 0.81; 95% CI, 0.61 to 1.07; P=0.08). Survival curves constructed from Kaplan-Meier analysis are shown in Figure 1.
When adjusted for patient demographics and clinical characteristics using propensity score analyses, ICDs were associated with a 31% lower risk of death from all causes (adjusted HR, 0.69, 95% CI; 0.50 to 0.96; P=0.03) (Table 2). Importantly, there was reasonable distribution of ICD and non-ICD patients within each propensity quintile (Table 3), and comparisons of baseline characteristics of ICD and non-ICD patients within each quintile revealed no significant differences (P>0.05 for all) (see Appendix Table A, available online). In secondary analyses, there was a trend in lower arrhythmic mortality with ICD therapy (adjusted HR, 0.65; 95% CI, 0.40 to 1.03; P=0.07). No significant differences between the ICD groups were seen for nonarrhythmic mortality (adjusted HR, 0.85; 95% CI, 0.57 to 1.29; P=0.39).
Prespecified subgroup analyses found that ICD therapy was associated with lower mortality regardless of age and etiology or degree of left ventricular dysfunction (Figure 2). Similarly, ICD therapy was associated with lower mortality regardless of most coexisting major medical conditions examined (symptomatic HF, diabetes mellitus, atrial fibrillation, PVD, prior stroke, and renal failure) and Holter study result (all interaction term probability values >0.05). When considered as a comorbidity score, ICD therapy was associated with similar differences in mortality among patients with 0, 1, 2, and ≥3 medical conditions (P for trend for interaction of 0.18). Post hoc exploratory analyses, however, did find that there was a trend for greater relative risk differences in mortality with ICD use among patients with 0 or 1 comorbid condition than among patients with 2 or more comorbid conditions (HR, 0.48; 95% CI, 0.29 to 0.80; versus HR, 0.83; 95% CI, 0.60 to 1.15; P for interaction of 0.052). Last, screening with MTWA identified patients where ICD therapy was associated with significant and no mortality differences (abnormal test: HR, 0.59; 95% CI, 0.43 to 0.80; P<0.001; versus normal test: HR, 1.12; 95% CI, 0.64 to 1.93; P=0.69 with P for interaction of 0.02).
Table 4 illustrates the absolute annual survival differences and ICERs associated with ICD therapy by subgroup. Although older patients had higher mortality rates, we found that ICD therapy was associated with significant gains in QALYs and was at least as cost-effective in patients 75 years of age or older as in younger patients (ICER of $82 531 per QALY for patients <65 years, $70 881 per QALY for patients 65 to 74 years, and $37 934 per QALY for patients 75 years and older) (Table 5). Similarly, among patients with major comorbid conditions, ICD therapy was associated with comparable or larger absolute mortality risk reductions annually than in patients without many of these major medical conditions (Table 4 and Appendix Table B [available online]). Finally, ICD therapy was found to be the most cost-effective for patients with 1 comorbidity (ICER of $49 677 per QALY) but was the least cost-effective for patients with 3 or more comorbidities (ICER of $93 485 per QALY).
In this community-based primary prevention cohort with left ventricular dysfunction, we found that ICD therapy was associated with a 31% lower risk of death from all causes. These mortality benefits were similar to those seen in primary prevention clinical trials.1,2 Importantly, we found that among older patients and patients with major medical conditions, ICD therapy was associated with comparable absolute and relative mortality risk reductions despite their higher annual mortality rates. Our findings suggest that the mortality benefits seen in primary prevention efficacy trials for ICDs are generalizable to nontrial populations—in particular, the elderly and those with major medical conditions—without a significant dilution of benefit.
Our study extends the work of prior observational studies of ICD therapy.19,20 Although several studies suggest that ICDs are effective in nontrial populations, these prior studies examined either a secondary prevention population,19 used administrative data sets,19 or had shorter follow-up.20 To our knowledge, this study is the first to examine a clinically well-characterized primary prevention cohort with patients of both ischemic and nonischemic etiology and with long-term follow-up. Because of the sample size of our study, we were also able to evaluate whether the apparent benefits of ICD therapy were confined only to younger and healthier patients. An added strength was that the majority of our study cohort (96% [927 of 965 patients]) was enrolled before the publication of SCD-HeFT and expansion of ICD reimbursement coverage in January 2005, as observational studies performed afterward would likely contain greater bias by indication.
Several recent reports have called into question the role of ICD therapy for primary prevention among older patients and those with major comorbidities.5–7 One study found higher mortality rates among ICD recipients who were older or with PVD, chronic pulmonary disease, prior HF, diabetes, cancer, and renal disease than among ICD recipients who were younger or healthier. Importantly, that study did not have a primary prevention control group and, therefore, could not assess whether ICDs were nevertheless effective in this higher-risk population.6 Another study found no discernible benefit among patients 65 years or older who were followed for 1-year using Medicare administrative datasets.5 Both studies, however, did not have data on LVEF to accurately define a primary prevention cohort. Moreover, the former study enrolled primarily secondary prevention patients (77% of cohort), whereas the latter study may have been limited by only 1 year of follow-up.
Despite these differences, our findings are not inconsistent with these prior studies, but they provide further insights into the role of ICDs in older, sicker patients. We likewise found that patients of older age and with major comorbidities had consistently higher all-cause mortality rates. However, our study also evaluated the relative and absolute mortality risk reductions associated with ICD therapy in these higher risk subgroups and found that the mortality benefit associated with ICD therapy remained similar in older patients as well as in those with diabetes mellitus, symptomatic HF, PVD, atrial fibrillation, stroke, and renal failure. For patients with chronic obstructive pulmonary disease and prior syncope from nonarrhythmic causes, real differences in the benefit associated with ICD therapy may exist, but our interaction analysis may not have been significant because of small numbers of study patients with these disorders.
Indeed, prior studies of ICD therapy have found that high-risk primary prevention patients (eg, those with inducible ventricular arrhythmias and abnormal noninvasive studies) may actually derive greater absolute mortality benefits, despite their higher rates of all-cause mortality.21–23 Our study underscores the fact that a population with higher baseline all-cause mortality risk does not preclude it from receiving substantial benefit from ICD therapy and emphasizes the importance of high-quality comparative effectiveness studies to avoid undertreatment of those most likely to benefit (“risk-treatment paradox”). However, the use of an effective therapy in a population with higher baseline mortality risk does raise potential issues of cost-effectiveness. Although high-risk patient subgroups may derive even larger absolute benefits from a given therapy than low-risk patients, their higher mortality may make such therapies less attractive from a cost-effectiveness consideration. Indeed, we found that ICD therapy was associated with less favorable cost-effectiveness ratios among patients with no or 2 or more comorbidities. This is consistent with a recent Second Multicenter Automatic Defibrillator Implantation Trial substudy that found a U-shaped pattern for ICD efficacy.24 In contrast, although patients 75 years or older also had higher annual mortality rates, ICD therapy remained cost-effective because of a numerically larger relative risk reduction in older patients compared with younger patients.
Although we found that the relative benefit associated with ICD therapy differed by comorbidity burden (0 to 1 versus ≥2), these results should be interpreted with caution as they were not prespecified. There were no significant differences found when comorbidity burden was examined by prespecified groupings (0 versus 1 versus 2 versus ≥3). Moreover, the interaction term probability value of 0.052 in the post hoc analysis was borderline and not adjusted for multiple comparisons. However, there are plausible reasons to believe that patients with multiple severe comorbidities may have substantial mortality risk from these conditions that competes with the potential benefit associated with ICD therapy. In addition, we did find differences in the relative benefit associated with ICD therapy among patients screening high- and low-risk with MTWA. These findings are consistent with earlier reports suggesting this test as a potential risk stratification tool to identify those patients most likely to benefit from ICDs.8,23,25
Our study should be interpreted in the context of the following limitations. As is the case with all cohort studies, our findings are subject to residual confounding. Nevertheless, the use of a landmark analysis and propensity scores with good model discrimination and balance between patients’ baseline characteristics are strengths of our study. Moreover, by enrolling patients from routine clinical practice, we were able to examine the comparative effectiveness of ICD therapy outside of clinical trial settings, a main study objective. Second, although we found that ICD therapy was effective in older patients, most patients within this age subgroup were between 75 and 80 years of age. Although our study enrolled patients up to 90 years of age, inferences regarding the benefit of ICDs in patients older than 80 remain limited. Third, our study had limited data on a few variables of interest, including laboratory data (eg, serum sodium). Fourth, because the study cohort was designed to also examine the role of MTWA in risk stratifying primary prevention patients for ICD therapy, our results may not be applicable to the 8% to 15% of patients with atrial fibrillation at study enrollment who were excluded from our cohort.1,2 Finally, although comparisons of effect sizes for ICD benefit within subgroups did not show statistically significant differences, there remains the possibility that real differences may exist among those medical conditions with low prevalence in our cohort.
Use of ICD therapy for primary prevention in routine clinical practice was associated with lower all-cause mortality. Despite their higher baseline mortality risk, ICD therapy in older patients and those with a limited number of major comorbid conditions was found to have comparable mortality risk reduction and cost-effectiveness when compared with its use in younger, healthier patients. Although decisions about their use will still need to be individualized for a particular patient, our findings suggest that specific high-risk groups should not be routinely excluded from consideration for ICD implantation.
Sources of Funding
Initial cohort development was funded in part by Medtronic, but there was no external funding for this project.
Dr Spertus has a research grant from Medtronic. Dr Masoudi serves on the advisory board of Takeda NA, Amgen, and United Health Care. Dr Kereiakes has research support from Boston Scientific and Medtronic. Dr Chow has research grants from Medtronic, St Jude Medical, Biotronik, and Guidant; has received consulting fees and honoraria from St Jude Medical, Medtronic, Biotronik, Transoma, and Cambridge Heart; and is on the speaker’s bureau for Cambridge Heart, Medtronic, and St Jude Medical. There are no pertinent disclosures for Drs Chan and Nallamothu and Ms Bartone. No sponsor had any involvement in the design, collection, management, or analysis of the study or in manuscript preparation.
Bardy GH, Lee KL, Mark DB, Poole JE, Packer DL, Boineau R, Domanski M, Troutman C, Anderson J, Johnson G, McNulty SE, Clapp-Channing N, Davidson-Ray LD, Fraulo ES, Fishbein DP, Luceri RM, Ip JH. Amiodarone or an implantable cardioverter-defibrillator for congestive heart failure. N Engl J Med. 2005; 352: 225–237.
Moss AJ, Zareba W, Hall WJ, Klein H, Wilber DJ, Cannom DS, Daubert JP, Higgins SL, Brown MW, Andrews ML; Multicenter Automatic Defibrillator Implantation Trial II Investigators. Prophylactic implantation of a defibrillator in patients with myocardial infarction. N Engl J Med. 2002; 346: 870–883.
Hinkle LE Jr, Thaler HT. Clinical classification of cardiac deaths. Circulation. 1982; 65: 457–464.
Anderson JR, Cain KC, Gelber RD. Analysis of survival by tumor response. J Clin Oncol. 1983; 1: 710–719.
Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983; 70: 41–55.
D'Agostino RB Jr. Propensity scores in cardiovascular research. Circulation. 2007; 115: 2340–2343.
Gold M, Siegel J, Russell L, Weinstein M. Cost-Effectiveness in Health and Medicine. New York: Oxford University Press; 1996.
National Vital Statistics Reports: United States Life Tables, 2002. Hyattsville, MD: National Center for Health Statistics, Centers for Disease Control and Prevention; 2002.
Moss AJ, Hall WJ, Cannom DS, Daubert JP, Higgins SL, Klein H, Levine JH, Saksena S, Waldo AL, Wilber D, Brown MW, Heo M; The Multicenter Automatic Defibrillator Implantation Trial Investigators. Improved survival with an implanted defibrillator in patients with coronary disease at high risk for ventricular arrhythmia. N Engl J Med. 1996; 335: 1933–1940.
Chow T, Kereiakes DJ, Bartone C, Booth T, Schloss EJ, Waller T, Chung E, Menon S, Nallamothu BK, Chan PS. Microvolt T-wave alternans identifies patients with ischemic cardiomyopathy who benefit from implantable cardioverter-defibrillator therapy. J Am Coll Cardiol. 2007; 49: 50–58.
Bloomfield DM, Steinman RC, Namerow PB, Parides M, Davidenko J, Kaufman ES, Shinn T, Curtis A, Fontaine J, Holmes D, Russo A, Tang C, Bigger JT Jr. Microvolt T-wave alternans distinguishes between patients likely and patients not likely to benefit from implanted cardiac defibrillator therapy: a solution to the Multicenter Automatic Defibrillator Implantation Trial (MADIT) II conundrum. Circulation. 2004; 110: 1885–1889.
Although implantable cardioverter-defibrillators (ICDs) have been shown in clinical trials to reduce mortality in primary prevention patients, their effectiveness in real-world settings is less clear, especially among the elderly and those with major comorbid conditions. In this study, we examined the clinical effectiveness of ICDs in a prospective primary prevention cohort, with a particular focus on understanding the effects of ICDs among older and higher risk subgroups. We found that ICD therapy in the overall cohort was associated with a 31% reduction in all-cause mortality, driven primarily by a reduction in arrhythmic mortality. The mortality benefit seen with ICD therapy was consistent when stratified by age (<65, 65 to 74, and ≥75 years), ischemic etiology, ejection fraction (<25% versus ≤25%), and the presence of major comorbid conditions (P>0.05 for all interactions). Moreover, the overall cost-effectiveness of ICD therapy was similar between patients ≥75 years and younger patients but rose slightly in those with multiple comorbid conditions. Although decisions about their use will still need to be individualized for a particular patient, our findings suggest that specific high-risk groups should not be routinely excluded from consideration for ICD implantation.
The online-only Data Supplement is available at http://circimaging.ahajournals.org/cgi/content/full/CIRCIMAGING.108.807123/DC1.
Guest editor for this article was William S. Weintraub, MD.