Author + information
- Received April 13, 1998
- Revision received November 18, 1998
- Accepted December 24, 1998
- Published online April 1, 1999.
- Arshad Jahangir, MD∗,
- Win-Kuang Shen, MD, FACC∗,*,
- Sharon A Neubauer∗,
- David J Ballard, MD, PhD†,
- Stephen C Hammill, MD, FACC∗,
- David O Hodge‡,
- Christine M Lohse‡,
- Bernard J Gersh, MB, DPhil, ChB, FACC∗ and
- David L Hayes, MD, FACC∗
- ↵*Reprint requests and correspondence: Dr. Win-Kuang Shen, Mayo Clinic, 200 First Street SW, Rochester, Minnesota 55905
This study analyzes the relationship between pacing mode and long-term survival in a large group of very elderly patients (≥80 years old).
The relationship between pacing mode and long-term survival is not clear. Because the number of very elderly who are candidates for pacing is increasing, issues related to pacemaker (PM) use in the elderly have important clinical and economic implications.
We retrospectively reviewed 432 patients (mean age, 84.5 ± 3.9 years) who received their initial PM (ventricular in 310 and dual chamber in 122) between 1980 and 1992. Follow-up was complete (3.5 ± 2.6 years). Observed survival was estimated by the Kaplan-Meier method. Age- and gender-matched cohorts from the Minnesota population were used for expected survival. Log-rank test and Cox regression hazard model were used for univariate and multivariate analyses.
Patients with ventricular PMs appeared to have poor overall survival compared with those with dual-chamber PMs. Observed survival after PM implantation in high grade atrioventricular block (AVB) patients was significantly worse than expected survival of the age- and gender-matched population (p < 0.0001), whereas observed survival of patients with sinus node dysfunction was not significantly different from expected survival of the matched population (p = 0.413). By univariate analysis, ventricular pacing in patients with AVB appeared to be associated with poor survival compared with dual-chamber pacing (hazard ratio [HR] 2.08; 95% confidence interval [CI] 1.33 to 3.33). After multivariate analysis, this difference was no longer significant (HR 1.41; 95% CI 0.88 to 2.27). Independent predictors of all-cause mortality were number of comorbid illnesses, New York Heart Association functional class, left ventricular depression and older age at implant. Pacing mode was not an independent predictor of overall survival. Older age at implantation, diabetes mellitus, dementia, history of paroxysmal atrial fibrillation and earlier year of implantation were independent predictors of ventricular pacemaker selection.
After PM implantation, long-term survival among very elderly patients was not affected by pacing mode after correction of baseline differences. Selection bias was present in pacing mode in the very elderly, with ventricular pacing selected for sicker and older patients, perhaps partly explaining the apparent “beneficial impact on survival” observed with dual-chamber pacing.
More than 85% of pacemaker (PM) recipients in the United States are older than 65 years (1). As life expectancy of the population steadily increases, issues related to PM use in the elderly will have a greater socioeconomic impact (2). The selection of the pacing system has important clinical and economic implications—sophisticated devices are more expensive and complex when compared with single-chamber PMs and require more frequent follow-up. Whether the increased cost, complexity and frequent follow-up required by these pacing systems offset the known or suspected clinical benefits in the very elderly is not clear.
The beneficial impact of pacing mode in symptomatic conduction system disease has been well documented (3–8). By maintaining atrioventricular (AV) synchrony, atrial or dual-chamber pacing has been suggested to confer a physiologic advantage (9)over single-chamber ventricular pacing. Physiologic pacing may reduce morbidity by decreasing the incidence of atrial fibrillation (10–13), thromboembolic phenomena (6,13)and congestive heart failure (12,14)and affect the quality of life by improving effort tolerance and general well-being (9). However, the effect of pacing mode on long-term survival, especially in the very elderly, is not clear. The impact of pacing mode in octogenarians and nonagenarians is more difficult to establish because of the high prevalence of coexisting illnesses, the nonspecific nature of symptoms, shorter life expectancy and limited availability of long-term follow-up data. Most of the survival studies in this age group are short-term, retrospective analyses of small numbers of patients with incomplete follow-up. Moreover, multivariate analytic techniques have not been used to adjust for baseline differences between patient groups with different pacing modalities. Hence, the impact of pacing mode on overall survival in the very elderly remains uncertain.
In the present study, the differential impact of pacing mode (ventricular vs. dual chamber) on long-term survival in a large group of octogenarians and nonagenarians was examined. Multivariate analysis was used to adjust for baseline characteristics. Factors predictive of long-term survival and those influencing the selection of the pacing mode were also assessed.
The study population consisted of all patients 80 years and older who received their initial PM at the Mayo Clinic between January 1980 and December 1992. These patients were categorized according to the presence of sinus node dysfunction (SND) or high grade atrioventricular block (AVB) (Table 1). Sinus node dysfunction was defined by the presence of inappropriate persistent sinus bradycardia (<50 beats/min), sinus pauses longer than 3 s or sinoatrial block. High grade AVB was defined by the presence of complete AVB or type II second-degree AVB. Patients in chronic atrial fibrillation were excluded from pacing mode comparison. The pacing mode was selected in a nonrandomized fashion on the basis of the pacing physician’s appraisal of a patient’s need.
Data collection and follow-up
Data were obtained retrospectively from a centralized system of records of PMs implanted and followed up at the Mayo Medical Center. Each patient was followed up after PM implantation up to December 31, 1993, or the time of death before January 1, 1994. Follow-up information was obtained by review of the medical records or by telephone interview of patients and their families. Causes of death were grouped as “cardiovascular-related” (sudden death, refractory ventricular arrhythmia, heart failure, myocardial infarction, peripheral embolism), “cerebrovascular disease” (stroke), “noncardiovascular” (any other documented cause) or “unknown.”
Multivariate and univariate models to predict pacing mode were completed using logistic regression models. Survival after permanent pacemaker implantation was estimated with the Kaplan–Meier method. Expected survival was completed separately for each subgroup of patients with AVB or SND based on life tables for death from all causes from the Minnesota population for people of like age, gender and calender year of birth. Within-group comparisons of observed to expected survival were completed using the one-sample log-rank test. Two-group survival curve comparisons were based on the two-sample log-rank test. Univariate and multivariate associations of baseline variables with survival were assessed using the log-rank test and the Cox regression model (15). Multivariate models are summarized in the form of point estimates and 95% confidence intervals on the multivariate hazard ratios. Baseline variables considered as potential prognostic factors are listed in Table 2.
Patient demographics and pacemaker selection
A total of 584 patients 80 years and older received a permanent PM at the Mayo Clinic between January 1980 and December 1992. All patients with chronic atrial fibrillation or carotid sinus hypersensitivity (n = 152) were excluded from pacing mode comparison, leaving 432 patients for analysis (51% men and 49% women) (Table 1, Fig. 1). ⇓The mean age of the study group was 84.5 ± 3.9 years (range, 80 to 99 years). There were 276 patients (64%) with high grade AVB, 141 (33%) with SND and 15 (3%) with other indications, including syncope of undetermined cause (Table 1). In patients with high grade AVB, the pacing mode was ventricular in 196 (71%) and dual chamber in 80 (29%). In patients with SND, the pacing mode was ventricular in 101 (72%) and dual chamber in 40 (28%). The proportion of dual-chamber PMs implanted increased each year, from 5% in 1980 to 33% in 1985 to 39% in 1992. The baseline characteristics of patients who received single-chamber or dual-chamber pacing devices are summarized in Table 2.
The patients were followed for a mean of 3.5 (±2.6) years. For those who survived, mean follow-up was 4.5 (±2.5) years, with maximum follow-up to 10.9 years. Up-to-date medical records of all 432 patients were reviewed. All surviving patients (n = 194) and physicians or family members of deceased patients (n = 238) were followed by telephone interview.
Overall survival and survival from cardiovascular mortality
Observed overall survival in the very elderly with symptomatic conduction system disease who had ventricular pacing devices was significantly worse than that of patients with dual-chamber pacing devices (p = 0.0003). Overall survival for the entire study group with respect to pacing mode is shown in Figure 2. Observed survival in the ventricular pacing group at 1, 3 and 5 years was 80%, 57% and 45%, respectively, compared with 92%, 76% and 58% for the dual-chamber pacing group.
Observed survival curves after PM implantation for patients with AVB and those with SND and expected survival curves for their age- and gender-matched control populations are shown in Figure 3. In patients with AVB, cumulative survival after permanent PM implantation was significantly worse than expected survival in the general population (p < 0.0001, Fig. 3A). Observed survival at 1, 3 and 5 years was 83%, 61% and 47%, respectively, compared with 90%, 71% and 54% for the control population. In elderly patients with SND, observed survival after PM implantation was comparable to expected survival for the matched population (p = 0.413, Fig. 3B). At 1, 3 and 5 years, the observed survival after PM implantation for patients with SND was 83%, 63% and 52%, respectively, compared with 90%, 72% and 55% for the matched population.
The impact of pacing mode on long-term survival of very elderly patients with high grade AVB is shown in Figure 4. Patients who had single-chamber ventricular pacing had significantly worse survival outcome compared with patients with dual-chamber pacing (p = 0.0001; risk ratio 0.445; 95% confidence interval [CI] 0.293 to 0.674). Observed survival of patients with AVB and dual-chamber pacing at five years was 62%, compared with 41% for those with ventricular pacing. For patients with SND, no significant difference was seen in overall survival between the two pacing modes (risk ratio 0.711; 95% CI 0.418 to 1.210) (Fig. 5).
During follow-up, 238 patients died (48 patients with dual-chamber PMs and 190 with ventricular PMs). The causes of death are summarized in Table 3. One hundred fifteen patients (48.3%) died of cardiovascular-related causes (22 with dual-chamber PMs and 93 with ventricular PMs), 121 (50.8%) of noncardiovascular causes and 2 (0.8%) of unknown cause. There was no difference in the causes of death between the two pacing groups. Survival from cardiovascular-related mortality was 86%, 73% and 67% at 1, 3 and 5 years, respectively, in the ventricular pacing group and 96%, 89% and 74% in the dual-chamber pacing group. The most common cause of cardiac death in this patient population was congestive heart failure (49% of cardiovascular deaths) (Table 3). However, there was no significant difference in terms of mortality due to congestive heart failure between the two pacing groups.
Univariate and independent predictors of overall survival
The univariate predictors of overall and cardiovascular mortality are summarized in Table 4. To minimize the effect of selection bias, the Cox proportional hazard regression model was used for adjustment of possible baseline differences and to identify independent predictors of poor survival. The independent predictors of overall survival and survival from cardiovascular-related death, with risk ratios and 95% confidence intervals, are summarized in Table 5. The mode of pacing (ventricular vs. dual chamber) was not an independent predictor of overall survival. The difference in overall survival in patients with AVB initially seen between the two pacing mode groups disappeared after this adjustment. Independent predictors of poor survival at the time of initial PM implantation were the number of comorbid illnesses, New York Heart Association (NYHA) functional class, left ventricular depression and older age.
Predictors of PM selection
Independent predictors associated with PM selection identified by multivariate analysis are summarized in Table 6. Patients who received a dual-chamber PM were younger than those who received a ventricular PM, with a mean age of 83.3 years compared with 85 years in the ventricular PM group (p < 0.001) (Fig. 1). There was no significant difference in PM selection between the two genders. The dual-chamber system was selected more frequently for patients who had previous cardiac surgery (p = 0.01) or prosthetic valve (p = 0.003). Patients with diabetes mellitus (p = 0.02) or a history of stroke or transient ischemic attack (p = 0.05) or organic brain syndrome (p = 0.001) or those with comorbid illnesses (p = 0.02) more frequently received single-chamber ventricular devices. There was no significant difference between the two groups (Table 2)with regard to the underlying cardiac disease.
The results of the present study indicate that overall survival of very elderly patients with high grade AVB was worse than the expected survival of an age- and gender-matched population. However, observed survival after PM implantation in patients with SND was comparable to the expected survival of the matched population. In elderly patients with SND, no significant difference in overall, cardiovascular or noncardiovascular mortality was observed in patients receiving a ventricular or a dual-chamber PM. In patients with high grade AVB, ventricular pacing was associated with an increased risk of overall and cardiovascular mortality by univariate analysis. However, after adjustment of the baseline differences with multivariate analysis, the mode of pacing was no longer an independent predictor of overall mortality. Age at implantation, left ventricular depression, NYHA functional class and number of comorbid illnesses were independent predictors of poor overall survival.
Long-term survival in patients with sinus node dysfunction
Several studies in patients with SND treated with single-chamber fixed-rate ventricular pacing have demonstrated an apparent increase in complications compared with those treated with atrial or dual-chamber pacing (13,14,16). These complications include PM syndrome, lack of AV synchrony and physiologic heart rate response to exercise, development of atrial fibrillation and systemic embolization (6,10,11). Physiologic pacing (atrial or dual chamber) has been shown to improve hemodynamics and exercise tolerance (9)in the short term and has been suggested to reduce morbidity by decreasing the occurrence of atrial fibrillation (6,11), heart failure (12,14)and thromboembolic phenomena (6,7,16–19). Critical review of the literature of SND indicates that the clinical and survival benefit of physiologic pacing over ventricular pacing is apparently inconclusive (18–21). The conclusions of most of the studies that suggest a deleterious effect from ventricular pacing should be interpreted in light of their limitations; most were nonrandomized, retrospective studies limited by the inability to control for important baseline clinical differences between patients with different pacing modes (6,11,22). In most of the studies, the rationale for selecting different pacing modes was not documented, and differences in cardiovascular and noncardiovascular comorbid conditions were not considered. The increased morbidity and mortality in patients with ventricular pacing may well be related to the underlying cardiovascular disease, coexisting noncardiac condition and age rather than the pacing system selected and, thus, may reflect selection bias. Furthermore, the study follow-up periods for different pacing modes were either incomplete or different and, therefore, difficult to compare. Recent studies in younger patients with SND have used multivariate techniques to control for baseline differences and have demonstrated no significant worsening effect of ventricular pacing on heart failure (20)or on overall or cardiovascular mortality (10,19,21). Our results confirm and extend these findings to octogenarians and nonagenarians, namely, that ventricular pacing does not appear to have a negative impact on overall survival or survival from cardiovascular mortality in patients with SND.
Long-term survival in patients with AV block
The one- and five-year cumulative survival rates of 83% and 47%, respectively, observed in our study of patients with AVB, are worse than the expected survival of 90% and 54%, respectively, of an age- and gender-matched Minnesota population. The observed survival rates in our group with AVB are comparable to those reported by Strauss and Berman (3)in the Canadian population and by Elizabeth and Green (23)in the British population. These survival rates are better than that reported by Breivik and Ohm (4)in the Norwegian population. The explanation for the different outcomes is not known, but it most likely reflects different indications for pacing, patient selection criteria, presence and severity of underlying heart disease and dates of the studies. The causes of mortality in patients with AVB in our study were multifactorial, with a significant number dying of noncardiovascular causes.
In a recent retrospective analysis of a random sample of 36,312 Medicare beneficiaries 65 years and older, dual-chamber pacing was an independent predictor of survival at one and two years after PM implantation (24). In other studies of patients without congestive heart failure, however, the survival rate was not significantly different for patients with ventricular pacing and those with dual-chamber pacing (25,26). With preexisting heart failure, patients with dual-chamber pacing appear to have a better survival outcome compared with those with ventricular pacing (26). In the present study of patients 80 years and older, the relative survival after PM implantation for AVB appears to be poor regardless of the presence of heart disease. Observed survival was significantly worse than expected, both for patients with and for those without associated underlying heart disease. This is consistent with the observation that the causes of mortality in the very elderly are likely multifactorial, as reflected in the significant number of deaths due to noncardiac causes (∼51% of all deaths) and will not be significantly affected by PM (Table 3).
Advancing age in our very elderly population was associated with decreased use of dual-chamber pacing. This is consistent with findings by others (24,27)and reflects the existing practice that emphasizes the use of more expensive and complicated devices and interventions in active, younger patients. Despite demonstrable hemodynamic advantages and suggested clinical benefits of dual-chamber pacing, ventricular pacing is still the most frequently used pacing modality (1), especially in the very elderly. This discrepancy between the suggested potential benefits and the pacing practice suggests that physicians implanting PMs are often less aggressive in the very elderly and choose simpler, less expensive devices, perhaps because of the lack of data specifically addressing the elderly or because of the attitude that more complex, expensive, sophisticated interventions are of “limited use” in this group (28). Factors other than the chronologic age, such as functional ability, activity level, coexisting illnesses and life expectancy, are important in selecting a pacing mode.
In our study, dual-chamber pacing mode in the very elderly population was not an independent predictor of overall survival or survival from cardiovascular mortality. The apparent differences in overall survival seen with pacing mode in patients with AVB reflected a selection bias of choosing ventricular pacing for sicker patients. In this study, we did not analyze the effect of PM mode on the incidence of atrial fibrillation, thromboembolism, congestive heart failure or quality of life.
With the current cost-containment environment, increase in the aging population and the competition for shrinking resources, long-term clinical outcomes provide needed information in guiding therapeutic recommendations. Despite lack of data (at this time) convincingly showing a beneficial effect on survival, chronologic age alone should not be considered a bar to the use of sophisticated PMs when a clear beneficial effect can be established.
Octogenarians and nonagenarians belong to a highly heterogeneous group with respect to the presence of coexisting medical illnesses, severity of cardiac diseases and functional capabilities. The relative survival is inversely related to coexisting underlying heart disease and medical illness. Pacing mode selection in the elderly or in any population must be individualized—with careful prescription according to activity level, underlying rhythm disorders and cardiovascular and noncardiac comorbidities—to provide optimal functional benefit with minimal complications related to inappropriate pacing and cost. In patients with AVB and SND with coexisting heart disease, the assessment of benefits from physiologic pacing is an important area for further investigation. Results from ongoing prospective trials comparing single-chamber with dual-chamber pacing will help to better understand the functional status, quality of life, cost-effectiveness, morbidity and survival benefits after long-term ventricular and physiologic pacing in the overall and rapidly expanding elderly populations. Cost reductions may be achieved by minimizing the use of complex systems in the very elderly who have limited mobility (e.g., musculoskeletal disease or neurologic deficits), a terminal illness or symptoms that are infrequent, intermittent or vague. For others, improvement in cardiac output and quality of life with physiologic pacing may be a crucial factor in allowing continued independence and potential decrease in the number of hospital visits and medical treatment for atrial fibrillation, heart failure (with systolic or diastolic dysfunction), thromboembolic phenomena or other morbidity. It is important to note that no significant quality of life benefits were observed in the elderly with dual-chamber pacing mode from a recent prospective randomized trial (29).
Our observations and conclusions should be interpreted in light of the limitations imposed by the retrospective nature of the study. Pacemaker selection was not randomized, and unknown, undocumented confounding variables that could not be determined by a retrospective review of the medical records may have led clinicians to select one device over the other. The multivariate model was used in our study to minimize the effect of baseline differences. Although all the clinical records were interpreted and translated into a standard data format, most of the information was qualitative. In particular, left ventricular function was not routinely measured in all patients, and the anatomical severity of coronary artery disease was not routinely determined. Moreover, considerable advances were made in PM and lead technology in the 1980 to 1992 period, along with improvement in management of patients with cardiovascular disease, which could have affected clinical outcomes and overall survival.
In contrast to many of the previously reported studies, a strong point of our analysis is the use of multivariate techniques to adjust for baseline characteristics of the patients to minimize the impact of selection bias. In addition to being a complete sample (no patient was lost to follow-up) of all initial PM implantations in a relatively large group of octogenarians and nonagenarians, another strength of our study is the careful evaluation of the impact of coexisting heart disease and medical illnesses on survival after pacemaker implantation.
In our study, after correction for baseline differences, long-term survival among octogenarians and nonagenarians after PM implantation for AVB or SND was not affected by pacing mode. Selection bias was present in the very elderly, with ventricular pacing selected for older, sicker patients, perhaps partly explaining the apparent “beneficial impact on survival” observed with direct chamber pacing in healthier, “younger” elderly patients.
☆ Dr. Jahangir is supported by a CR20 award from the Mayo Foundation for Education and Research.
- atrioventricular block
- confidence interval
- New York Heart Association
- sinus node dysfunction
- Received April 13, 1998.
- Revision received November 18, 1998.
- Accepted December 24, 1998.
- American College of Cardiology
- Hesselson A.B,
- Parsonnet V,
- Bernstein A.D,
- Bonavita G.J
- Cox D.R
- Sgarbossa E.B,
- Pinski S.L,
- Maloney J.D,
- et al.
- Elizabeth J.E,
- Green G.J
- Lamas G.A,
- Pashos C.L,
- Normand S.L,
- McNeil B
- Alpert M.A,
- Curtis J.J,
- Sanfelippo J.F,
- et al.
- de Belder M.A,
- Linker N.J,
- Jones S,
- Camm A.J,
- Ward D.E