Author + information
- Received April 24, 1996
- Revision received October 25, 1996
- Accepted November 5, 1996
- Published online March 1, 1997.
- Panagiotis T Panotopoulos, MDA,
- Kathi Axtell, RNA,
- Alfred J Anderson, MSA,
- Jasbir Sra, MD, FACCA,
- Zalmen Blanck, MD, FACCA,
- Sanjay Deshpande, MDA,
- Michael Biehl, MDA,
- Edward T Keelan, MDA,
- Mohammad R Jazayeri, MDA,
- Masood Akhtar, MD, FACCA and
- Anwer Dhala, MD, FACCA,*
- ↵*Dr. Anwer Dhala, 2901 West KK River Parkway, No. 470, Milwaukee, Wisconsin 53215-3660.
Objectives. We sought to assess the effect of advanced age on the outcome of patients with an implantable cardioverter-defibrillator (ICD).
Background. ICDs are effective in preventing sudden cardiac death in susceptible patients, but their beneficial effect on survival is attenuated by the high rate of nonsudden cardiac death in those treated. Although advanced age is an important variable in determining cardiovascular mortality, its impact on the outcome of patients with an ICD has been inadequately studied.
Methods. We performed multivariate analysis of a data base consisting of 769 consecutive patients with an ICD. Seventy-four patients ≥75 years old at ICD implantation (Group 1) were compared with the remaining 695 patients (Group 2).
Results. The two groups were similar in clinical presentation, left ventricular function and gender distribution. The mean follow-up time was 29 and 42 months, respectively, for patients in Group 1 and Group 2. Actuarial survival at 4 years was 57% in Group 1 versus 78% in Group 2 (p = 0.0001). This difference was primarily due to a higher rate of nonsudden cardiac death in Group 1. On multivariate analysis, age ≥75 years, New York Heart Association functional class III, left ventricular ejection fraction <30% and appropriate shocks during follow-up were independently associated with increased mortality (odds ratio 3.56, 1.8, 1.6 and 1.39, respectively).
Conclusions. Among patients with similar functional class and ejection fraction, the mortality risk is increased threefold in those ≥75 years old at the time of ICD implantation. Extrapolation of results from younger patients is likely to overestimate ICD benefit in the elderly.
(J Am Coll Cardiol 1997;29:556–60)
In a society where health care cost is rapidly becoming a major consideration in decision-making, it is important to know precisely the effect of innovative and costly treatment modalities on patient groups with factors thought to be at the extremes of the spectrum in terms of risk stratification. Advanced age is such a factor and its relevance to clinical decision-making is obvious given the increasingly aging population in modern Western societies ([1, 2]). Sudden cardiac death (SCD) remains a significant health care problem with an annual incidence of ∼300,000 in the United States (). Authorities () appear to agree that the implantable cardioverter-defibrillator (ICD) reduces the incidence of SCD in high risk patients. This consensus, along with the ease of ICD implantation with minimal complications, technologic advances in ICDs and physicians’ tendency to overtreat rather than undertreat patients at risk for SCD, has led to an exponential increase in rates of ICD implantation. However, some investigators ([5–9]) have questioned whether reducing the risk of SCD leads to prolongation of life, as many patients at increased risk of SCD also have advanced heart disease and significant comorbid illnesses. Although this issue may be particularly relevant in elderly patients, little has been published () on the effect of age on the benefit derived from ICDs. The present study is an attempt to assess this effect.
Seven hundred sixty-nine consecutive patients who underwent ICD implantation between September 1983 and September 1995 were included. Of these, 74 patients (Group 1) were ≥75 years old at the time of implantation; the remaining 695 patients form Group 2. The end point of follow-up was March 31, 1996 (which assured a minimal follow-up period of 6 months) or death, if it occurred before that date. For patients lost to follow-up, the end point was the last follow-up contact.
The clinical characteristics of the two patient groups are summarized in Table 1. The majority of patients (90% in Group 1) had coronary artery disease; fewer patients had nonischemic cardiomyopathy, valvular heart disease or other underlying conditions. Most patients had an ICD implanted because of aborted SCD, spontaneous ventricular tachycardia (VT) or syncope with inducible VT in the setting of left ventricular (LV) dysfunction. Before ICD implantation, the nature and severity of the underlying heart disease were assessed with studies that included, when indicated, cardiac catheterization, echocardiography and nuclear imaging studies. Implantation techniques evolved from thoracotomy approaches used initially to the current transvenous systems.
The patients were followed up at 2- to 3-month intervals clinically and with device interrogation. To minimize the frequency of device discharge, antiarrhythmic medications were used as needed to control supraventricular or ventricular arrhythmias.
Whenever a device discharge occurred in an unmonitored situation, the patient was interviewed in an attempt to assess the appropriateness of the discharge. The discharge was considered appropriate if there was electrocardiographic documentation of arrhythmia at the time of discharge or if the shock had been preceded by syncope or symptoms consistent with cerebral hypoperfusion that resolved immediately afterward. Shocks that were not preceded by symptoms, that occurred in association with palpitation but not dizziness or that occurred during intense physical activity were not considered appropriate unless a qualifying arrhythmia was documented by an event monitor, stored electrogram or similar means. Finally, shocks attributed to device-induced proarrhythmia (i.e., VT acceleration or degeneration into ventricular fibrillation [VF]) were also not considered appropriate.
Deaths were classified as cardiac, sudden cardiac, nonsudden cardiac or noncardiac on the basis of data derived from postmortem device interrogation, history obtained from family members or witnesses, hospital records and involved physicians. Sudden cardiac death (SCD)was defined as death that occurred unexpectedly, within 1 h of symptom onset, or that was unwitnessed, as during sleep, in a patient who had been witnessed to be well just before death. Cardiac death (CD)was defined as any death due to cardiac causes and included both SCD and nonsudden CD. Noncardiac deathincluded all deaths not classified as CD ().
Data were expressed as mean value ± SD or as a percent. Comparisons were made by using the Student ttest, the chi-square test or the Fisher exact test as appropriate. Survival curves were presented by using the Kaplan-Meier method. The log-rank method was used for statistical analysis of differences between survival curves. Independent predictors of mortality and odds ratios were determined by using the Cox proportional hazards model. The SAS system (SAS Institute) was used for statistical analysis. Differences were considered statistically significant at p < 0.05.
The mean age at ICD implantation was 77.2 years (range 75 to 84) in Group 1 and 60.8 years (range 15 to 74) in Group 2. The mean left ventricular ejection fraction (LVEF) and the percentage of patients with LVEF <30% were almost identical in the two groups. Most patients in either group were in New York Heart Association (NYHA) functional class I or II. There were no patients in NYHA class IV at the time of implantation and there were no significant differences in the presenting arrhythmia in the two groups. The percentage of patients undergoing concomitant coronary bypass surgery or cryoablation was similar in the two groups as was the mode of implantation (Table 1).
The mean follow-up interval was 29.1 ± 23.0 months (range 1 day to 115 months) for Group 1 patients and 42.2 ± 31.7 months (range 0 day to 151 months) for Group 2 patients. Eleven patients were lost to follow-up (1 [1.4%] from Group 1 and 10 [1.4%] from Group 2). These patients were censored at the point they were lost to follow-up. Perioperative mortality was very low: One elderly patient (1.4%) and five younger patients (0.7%) died within 1 month of ICD implantation. Survival data are shown in Table 2. At 2 and 4 years of follow-up, actuarial survival was 80% and 57%, respectively, in Group 1 versus 90% and 78% in Group 2 (p = 0.0001). Actuarial freedom from SCD in the two groups was 100% and 98%, respectively, at 2 years and 100% and 97% at 4 years (p = NS). Actuarial freedom from cardiac death was 86% and 76% at 2 and 4 years, respectively, in Group 1 versus 93% and 87% in Group 2 in the same time period (p < 0.001). Percentage free from noncardiac death at 2 and 4 years was 93% and 75%, respectively, in Group 1 and 97% and 91% in Group 2 (p < 0.01). Kaplan-Meier survival curves are depicted in Fig. 1.
2.2 Device use.
At 4 years of follow-up actuarial freedom from appropriate shock, as defined previously, was 58% in the elderly and 57% in the younger patients (p = NS). Kaplan-Meier curves of appropriate shocks delivered by the device are shown in Fig. 2. During the entire period of follow-up, 27 elderly patients (36.5%) received appropriate shocks, 8 (10.8%) received indeterminate shocks and 39 (52.7%) received no shocks. Among the younger patients, 261 (37.6%), 127 (18.3%) and 307 (44.2%), respectively, received appropriate, indeterminate and no shocks. The mean time from ICD implantation to first appropriate shock was 10.2 ± 14.5 and 13.3 ± 18.1 months (p = NS) and the mean time from first appropriate shock to death or last follow-up was 22.4 ± 19.0 and 37.7 ± 30.2 months, respectively, in Group 1 and Group 2 (p < 0.001) (Table 3).
2.3 Predictors of mortality.
The results of a multivariate analysis to determine which factors, if any, were independently associated with increased mortality are shown in Table 4. Age ≥75 years at implantation was the most powerful predictor of an adverse outcome (odds ratio 3.56, confidence interval 2.28 to 5.54). Other factors independently associated with mortality were ejection fraction <30% at implantation, NYHA class III and appropriate shocks during follow-up. Factors considered in this analysis, but not found to be of independent prognostic significance, included clinical presentation, type of heart disease, gender, history and location of myocardial infarction.
In this large single-center experience the ICD conferred near elimination of SCD risk in the elderly, to the same extent as in younger patients. However, the elderly had a significantly higher total mortality rate. Although the higher death rate in this age group would seem expected and obvious, we found that it was primarily due to an increase in nonsudden cardiac death. In fact, the variable with the highest odds ratio for total as well as cardiac mortality in our ICD patients, as assessed by a multivariate analysis model, was age ≥75 years at implantation.
Some investigators ([6, 8]) have questioned the benefit of the ICD in prolonging life in any patient age group, an issue that may be particularly relevant in the elderly at risk for SCD. Several randomized studies are underway to assess the efficacy of ICD therapy versus antiarrhythmic medications in prolonging life, but it is unlikely that these studies will have sufficient numbers of patients to address the use of ICDs in the elderly. Although our study is retrospective, it is the first large study that has examined the pattern of ICD usage and outcome in patients >75 years old. We arbitrarily chose 75 years as the age cutoff point in an effort to identify a variable that would be useful in clinical decision-making. Currently 13% of the population in the U.S. are >65 years old and 3% are >80 years old (). Among our ICD patients, ∼10% are >75 years old. The average life expectancy for the general population in this age group is 10.7 years (). An appropriate intervention in this age group could therefore have a significant impact on survival.
3.1 Age and cardiovascular mortality.
Mortality due to cardiovascular disorders even in the absence of arrhythmias is known to be adversely influenced by advanced age. For example, the 21-day mortality rate in the control group of the Gruppo Italiano per lo Studio della Streptochinasi nell’Infarto Miocardico (GISSI)-1 trial () rose from 7.7% in patients <65 years old to 33.1% in those >75 years old. Similarly, the 1-year mortality rate after hospital discharge was 6% in patients <55 years old and 33% in patients ≥75 years old in the Worcester Heart Attack Study (). Among patients undergoing coronary artery bypass (CABG) surgery, perioperative mortality was 1.9% in those <65 years old versus 9.5% in patients 75 to 84 years old (). Data from the Coronary Artery Surgery Study (CASS) registry show a 91% 5-year survival rate in patients <65 years old versus 70% in patients >75 years old (). To some extent this increased mortality in the elderly is attributable to poorer baseline values for variables known to affect survival in these settings. However, another important factor is physician reluctance to implement appropriate therapeutic strategies that are associated with more frequent complications in the elderly. In fact, appropriately aggressive interventions such as thrombolytic therapy, for example, may result in a greater mortality reduction in the elderly than in younger, less high risk patients (). Whether an intervention such as the ICD would similarly benefit elderly patients needs to be determined.
3.2 Assessment of ICD benefit in the elderly.
In the absence of randomized prospective data two methods have been used to try to assess ICD benefit in the elderly. The first compares historical control subjects with ICD patients ([17–19]). However, identification of historical control subjects is problematic in studies that extend over a period of time during which significant changes in both anti-ischemic and antiarrhythmic management have occurred. The second method involves the assumption that the time of the first appropriate shock would have been the time of death had the ICD not been implanted ([5, 20–23]). Criticism of this method is based on the fact that some of the appropriate shocks may have been given for nonsustained events by committed devices, for hemodynamically tolerated events or for fast VT or VF precipitated by attempts by the device to pace-terminate slower VT ([6–8]). Because of these methodologic limitations, along with problems often associated with accurately determining the mode of death, it has been suggested () that total mortality should be considered the primary end point in survival studies of patients with an ICD.
The present study highlights the difficulties associated with ICD use in the elderly. Our data suggest that elderly and younger patients at risk for SCD have a similar frequency of arrhythmic events. The almost total absence of SCD death in both groups implies that the ICD confers its stated benefit of effectively treating arrhythmias in these patients. However, this benefit appears to be significantly attenuated in the elderly because of a threefold increase in overall mortality in this group. Our findings, therefore, caution against extrapolation to the elderly of ICD outcome studies performed predominantly in younger patients. A true assessment of the ICD benefit in the elderly can be obtained only by a prospective randomized trial. Because of the higher total mortality among the elderly, the importance of such a trial is obviously greater in this group than in younger patients. In the absence of such a trial, the challenge to the implanting physician will be to identify among the elderly those patients at relatively low risk for nonsudden cardiac death, who are likely to derive the most benefit from the device.
3.3 Study limitations.
The retrospective design of this report is its major limitation. Selection bias may have been involved during decision-making for ICD implantation in the elderly, particularly during the period before nonthoracotomy devices became available. We did not attempt to introduce concomitant pharmacologic management (including antiarrhythmic and anti-ischemic or heart failure treatment) as a variable in our study. Although this is an important omission, over the follow-up period significant changes occurred in what was considered state of the art treatment as a result of the Cardiac Arrhythmia Suppression Trial (CAST), introduction of angiotensin-converting enzyme inhibitors and recognition of the role of beta-blockers. The changes made over time in each individual patient’s regimen made such a comparison impossible.
The ICD can be placed with minimal risk in elderly patients and is as effective in preventing sudden cardiac death in this group as in younger patients. However, among patients with similar functional class and ejection fraction, the risk of nonsudden cardiac death is increased threefold in patients ≥75 years old at ICD implantation. Extrapolation of results from younger patients is likely to overestimate the benefit of ICD implantation in the elderly. Conversely, inclusion of a large number of elderly patients in such studies is likely to result in underestimation of the potential benefits of ICD implantation in younger patients.
- coronary artery bypass graft surgery
- Coronary Artery Surgery Study
- cardiac death
- implantable cardioverter-defibrillator
- left ventricular
- left ventricular ejection fraction
- New York Heart Association
- sudden cardiac death
- ventricular fibrillation
- ventricular tachycardia
- Received April 24, 1996.
- Revision received October 25, 1996.
- Accepted November 5, 1996.
- The American College of Cardiology
- National Center for Health Statistics: Advanced report of final mortality statistics, 1988. Monthly Vital Statistics Report. Vol. 39, No. 7 suppl. Hyattsville, MD, Public Health Service, pp 1–48.
- Gillum RF
- Sweeney MO,
- Ruskin JN
- Myerburg RJ,
- Luceri RM,
- Thurer R,
- et al.
- ↵Kim SG, Fogoros RN, Furman S, Connoly SJ, Kuck KH, Moss AJ. NASPE Policy Statement. Standardized reporting of ICD patient outcome: The report of a North American Society of Pacing and Electrophysiology Policy Conference, February 9–10, 1993. Pacing Clin Electrophysiol 1993;16:1358–62.
- ↵U.S. Bureau of the Census, Spencer G. Projections of the population of the United States by age, sex and race: 1988 to 2080. Current Population Reports. Series P-25, No. 1018. Washington, D.C., Government Printing Office 1989;1–17.
- Gersh BJ,
- Kronmal RA,
- Frye RL,
- et al.
- Powell AC,
- Fuchs T,
- Finkelstein DM,
- et al.
- Fogoros RN,
- Elson JJ,
- Bonnet CA,
- Fiedler SB,
- Burkholder JA
- Böcker D,
- Block M,
- Isbruch F,
- et al.
- Levine JH,
- Mellits ED,
- Baumgardner RA,
- et al.