Recently, it has been shown that atrial and dual-chamber pacemakers can improve short-term hemodynamics [9], exercise tolerance, and well-being [10, 11] in patients with the sick sinus syndrome when compared with fixed-rate ventricular pacing. It has also been suggested that these more physiologic pacing modes may decrease serious morbidity [12, 13] and improve survival [14]. This hypothesis was supported by several studies without internal controls [15-17], but in none of them were multivariable techniques used to adjust for baseline characteristics of patients treated with different pacing modalities, nor were large groups of patients followed for more than 5 years.

Several authors no longer recommend single-chamber ventricular pacing as permanent therapy for the sick sinus syndrome [12, 18, 19]. However, it is still the most frequently used pacing modality in the United States [1] due to its wider availability, ease of implantation, and lower costs. Because of the recent emphasis on the monitoring of outcomes of medical interventions to determine their cost-effectiveness [20], we examined the potential benefit on long-term survival of physiologic pacing modalities in patients with the sick sinus syndrome.

Methods

Top Methods Results Discussion Author & Article Info References

Study Patients

Between 1 January 1980 and 31 December 1989, we implanted an initial pacemaker in 507 adult patients (age > 18 years) with the sick sinus syndrome. The sick sinus syndrome was defined by the presence of inappropriate, persistent sinus bradycardia (rate < 50 beats/min), sinus pauses greater than 3 seconds, or sinoatrial block. Patients in permanent atrial fibrillation at the time of implant were excluded from the study. Patients with the sick sinus syndrome and complete atrioventricular (AV) block or type II second-degree AV block (unless resulting from AV junctional ablation in a patient who fulfilled criteria for the sick sinus syndrome before the procedure), were also excluded. All patients were symptomatic, required bradycardia-producing drugs for treatment of tachyarrhythmias, or both [21]. Documentation of symptomatic bradycardia (for example, its simultaneous occurrence with syncope), was considered desirable but not imperative for pacemaker implantation.

Baseline variables describing cardiac disease, concomitant diseases, and electrocardiographic and echocardiographic findings were analyzed (Table 1). Concomitant diseases considered were systemic hypertension, diabetes, cerebrovascular disease, and peripheral vascular disease. Diagnostic criteria for diabetes were fasting glycemia greater than 140 mg/dL or chronic treatment with insulin or oral hypoglycemic agents. Cerebrovascular disease was defined by a history of stroke, transient ischemic attacks, carotid endarterectomy, or severe (70% or more) carotid stenosis by angiography. Peripheral vascular disease was defined by presence of intermittent claudication, aortic aneurysm, or a history of a peripheral revascularization procedure. Left ventricular function was assessed in 383 patients (76%) using contrast ventriculography (60%), echocardiography (34%), or radionuclide ventriculography (5%).

Additionally, the initial pacing mode was analyzed. The pacing mode was selected in a nonrandomized fashion, based on the attending physician's appraisal of the patient's needs. Ventricular pacemakers were implanted in 112 patients (22%), atrial pacemakers in 19 patients (4%), and dual-chamber pacemakers in 376 patients (74%). For the purposes of this analysis, atrial and dual-chamber pacemakers (defined as physiologic pacing) were compared with ventricular pacemakers. Due to technologic improvements during the time frame of the study, the year of implantation was the strongest determinant of type of pacemaker implanted, with a progressive increase in the number of dual-chamber pacemakers over time. Otherwise, patients with physiologic and ventricular pacemakers were similar for most of the variables analyzed, except for cerebrovascular and peripheral vascular disease (more prevalent in patients with ventricular pacemakers) and complex ventricular arrhythmia (more prevalent in patients with physiologic pacemakers) (see Table 1). Only 31 (6%) of the devices had sensors to provide rate responsiveness, too few to justify a separate statistical analysis.

The mean AV delay that was programmed at discharge in patients with dual-chamber pacemakers was 198 ± 43 msec. The mean PR interval in this subgroup was 180 ± 52 msec. The relatively long AV delay was programmed to allow for spontaneous ventricular activation whenever possible.

Follow-up

Patients were followed for a mean of 66 ± 38 months. Follow-up began on the date of pacemaker implant and ended on the date of the patient's death or the end of the study (31 December 1990). Survival status, causes of death, and reprogramming of the initial pacing mode were ascertained through review of medical records, through questionnaires completed by the patient's private physician, and through telephone interviews with the patients or their families. Follow-up was complete in 98.5% of the patients. Eight patients were lost to follow-up after a mean of 56 months and had their follow-up censored at time of last contact.

Causes of death were grouped as cardiovascular-related (sudden death, congestive heart failure, refractory ventricular tachyarrhythmia, myocardial infarction, stroke, peripheral embolism), noncardiovascular (any other documented cause), or unknown.

Statistical Analysis

Continuous variables are presented as mean ± 1 SD. Actuarial curves for total and cardiovascular mortality were calculated with the method of Kaplan and Meier. The primary analysis was done using an "intention-to-treat" approach. In a secondary analysis, patients were censored at the time of crossover to the alternative pacing mode (that is, upgrading from ventricular to dual-chamber pacing or permanent reprogramming of a physiologic pacing mode to a ventricular pacing mode).

The expected survival for an age- and sex-matched general population was compiled from data of the National Vital Statistics System. A standardized score ("z" score) was calculated to test for differences in survival between the study group and the general population at each yearly interval (year 1 to year 10). To adjust for the multiple comparisons, a P value 0.005 was considered significantly different (the Bonferroni correction).

Analyzed variables were screened by univariate statistical methods to identify those associated with total and cardiovascular mortality. Kaplan-Meier actuarial curves for survival were constructed for each variable on the basis of natural dichotomies (for example, the presence or absence of coronary artery disease) or stratifications in the data (for example, New York Heart Association functional class) and were compared by means of the log-rank test. Multivariable regression analysis, done with the Cox proportional-hazards model, was applied to all variables that had at least marginal univariate predictive value (P < 0.10). The variables were entered into the regression equation in a forward stepwise manner. Variables with statistically independent predictive value (defined as P < 0.05) were identified, and their hazard ratio (that is, the relative risk attributable to various levels of each variable, all other variables being unchanged) was estimated. The assumption of proportional hazards for the exposure of interest was tested, and it was not violated. Statistical analyses were done using EGRET software (Statistics and Epidemiologic Research Corporation, Seattle, Washington) [22]. Covariates-adjusted curves for total survival according to pacing modality were constructed with the "corrected group prognostic curves" method described by Lee and colleagues [23].

Results

Top Methods Results Discussion Author & Article Info References

Total and Cardiovascular Survival

During the follow-up period, 130 patients (26%) died (46 patients with ventricular pacemakers and 84 patients with physiologic pacemakers). Seventy-seven patients (59%) died from cardiovascular-related causes (26 patients with ventricular pacemakers and 51 patients with physiologic pacemakers), including congestive heart failure (48%), sudden death (23%), miscellaneous (for example, myocardial infarction [21%]), and stroke (8%). Causes of death were noncardiovascular in 41 patients (32%) and unknown in 12 patients (9%). Actuarial survival probability for the entire study group was 0.95 at 1 year (CI, 0.93 to 0.97); 0.80 at 5 years (CI, 0.76 to 0.83); and 0.61 at 10 years (CI, 0.55 to 0.67).

Freedom from cardiovascular-related mortality was 0.96 (CI, 0.94 to 0.98); 0.88 (CI, 0.84 to 0.91); and 0.76 (CI, 0.70 to 0.81), at 1, 5, and 10 years, respectively.

Expected survival for an age- and sex-matched general U.S. population was 0.97 at 1 year, 0.84 at 5 years, and 0.70 at 10 years. Differences were not statistically significant during the first 5 years of follow-up. After the fifth year, survival in the study group was consistently poorer (P = 0.002 at 6 years; P = 0.004 at 10 years) (Figure 1).

View larger version (15K): [in this window] [in a new window]   Figure 1. Actuarial survival in the study group compared with the expected survival for an age- and gender-matched general population. Dots represent 95% confidence intervals for each year. Difference first reaches statistical significance at 72 months (P = 0.002) and continues to be significant thereafter (P = 0.005 at 10 years). The numbers beneath the graph are the numbers of paced patients with the sick sinus syndrome who were at risk at each point.

Prediction of Total and Cardiovascular Mortality Using Univariate Analysis

Qualitative overlap occurred between univariate predictors of total and cardiovascular death (Table 2). Although the ventricular pacing mode was associated with 43% higher mortality than was physiologic pacing (total death: P = 0.053; hazard ratio = 1.43; CI, 0.99 to 2.07; cardiovascular death: P = 0.15; hazard ratio = 1.41; CI, 0.87 to 2.29), this variable was not statistically different for either total or cardiovascular death.

During follow-up, only 9 patients (8%) who received a ventricular pacemaker were upgraded to a dual-chamber mode. At the same time, 59 patients (16%) with dual-chamber pacemakers were reprogrammed to ventricular pacing because of loss of atrial capture or development of chronic atrial fibrillation. Ventricular pacing was not a univariate predictor of total mortality (P = 0.07; hazard ratio = 1.42; CI, 0.96 to 2.11) or cardiovascular mortality (P > 0.2; hazard ratio = 1.26; CI, 0.76 to 2.11) when patients were censored at the time of crossover.

Prediction of Total Mortality Using Multivariable Analysis

Independent predictors of total mortality by multivariable analysis in the entire study group were 1) high functional class [hazard ratio = 1.67/class; that is, for each unit increase in functional class, the adjusted relative risk for mortality increases by 67%]; 2) age [hazard ratio = 1.62/SD of age; that is, for each 12-year increase in age, the adjusted relative risk for mortality increases by 62%]; 3) peripheral vascular disease [hazard ratio = 2.21; that is, patients with peripheral vascular disease have 2.21 times greater risk for mortality than patients without peripheral vascular disease]; 4) bundle branch block [hazard ratio = 2.04]; 5) coronary artery disease [hazard ratio = 1.66]; and 6) valvular heart disease (hazard ratio = 1.71) (Table 3). Ventricular pacing was not an independent predictor of mortality even after forced entry in the Cox regression model (P = 0.17; hazard ratio = 1.29; CI, 0.89 to 1.88 for total mortality (Figure 2); P = 0.23; hazard ratio = 1.35; CI, 0.83 to 2.20 for cardiovascular mortality).

View larger version (23K): [in this window] [in a new window]   Figure 2. Total actuarial survival for patients with physiologic pacing and ventricular pacing after adjustment for all covariates. The numbers beneath the graph are the numbers of patients in each group who were at risk at each point. P = 0.17.

Predictors of Cardiovascular-related Mortality Using Multivariable Analysis

Independent predictors for cardiovascular death by multivariable analysis in the entire population were similar to the independent predictors for total death (see Table 3). In addition, cerebrovascular disease had borderline statistical significance for cardiovascular-related death (P = 0.047; hazard ratio = 1.69; CI, 1.00 to 2.86). The same seven variables remained the only independent predictors whether all 12 deaths of uncertain cause were analyzed as cardiovascular-related or as noncardiovascular deaths.

Value of Echocardiographic Data

When the analysis was restricted to the 245 patients with complete echocardiographic data, independent predictors for total and cardiovascular mortality were New York Heart Association functional class (P = 0.007 and P < 0.001, respectively); age (P < 0.001 and P < 0.001, respectively); coronary artery disease (P = 0.003 and P < 0.001, respectively); and left ventricular function (P = 0.038 and P = 0.002, respectively).

Discussion

Top Methods Results Discussion Author & Article Info References

Current emphasis on cost-effectiveness in the United States demands that more expensive therapies be used only when there is clear evidence of benefit [24, 25]. During the past decade, dual-chamber pacing was introduced as a new therapy and was rapidly accepted in academic medical practice without a rigorous evaluation of its potential advantages on large populations under controlled conditions. Many authors consider now that ventricular pacing should be avoided in patients with the sick sinus syndrome [12, 18, 19]. The current American College of Cardiology/American Heart Association Task Force Guidelines for pacemaker implantation recommend ventricular pacing in this disease as a class I indication only for patients with chronic atrial flutter or fibrillation or giant atria [26]. However, a recent survey of cardiac pacing practices showed that ventricular modes (that is, VVI and VVIR [27]) are still the most frequently used in the United States [1]. These contradictory facts would suggest that suboptimal care is being provided to many patients with the sick sinus syndrome.

Our study and a critical review of the literature indicate that the survival benefit of physiologic pacing over ventricular pacing appears to be inconclusive. Studies suggesting a deleterious effect from ventricular pacing were retrospective [28-33], follow-up was incomplete [31], and multivariable statistical analysis to adjust for known prognostic factors was not used or was only partially done [31, 32]. European studies have reported survival rates in paced patients with the sick sinus syndrome similar to that of the general population [2, 33, 34]; however, long-term survival in our study patients was poorer than in an age- and sex-matched U.S. population. Although the reasons for this discrepancy are unknown, the occurrence of a decreased survival rate in paced patients has clinical implications. As the average life expectancy of a 65-year-old person in the United States is an additional 17 years [35], prolongation of life by a few years remains a reasonable, but still elusive, goal of therapy in patients with the sick sinus syndrome [36].

Although the prominence of survival as an end point for assessing therapeutic outcomes is undisputed [20], quality of life is another variable that deserves scrutiny because it may not always parallel survival [37]. The effects of alternative therapies on quality of life, functional capacity, employment status, and medical costs, together with patients' values and preferences, should therefore be considered [37]. Surprisingly, few randomized studies have addressed the effects of different pacing modes on quality of life in patients with the sick sinus syndrome [10, 38]. Dual-chamber pacing (DDD mode) has been associated with fewer symptoms than VVI pacing [38], but the maximum exercise tolerance remained similar. Similarly, the advantages of dual-chamber rate-responsive pacing (DDDR) compared with VVIR pacing in terms of cardiopulmonary function are trivial [39]. Other authors have reported increased incidence of heart failure [15, 30], chronic atrial fibrillation [15, 17], and thromboembolism [17] in patients with ventricular pacemakers compared with physiologic pacemakers. Because these studies were not randomized, the possibility of selection bias exists.

In contrast with the uncertainty about long-term clinical benefits of dual-chamber pacemakers for patients with the sick sinus syndrome, the high costs associated with these devices are well recognized. In a recent time-engineering study done at our hospital [40], procedural times for dual-chamber implants were longer than for single-chamber ventricular pacemakers (184 ± 54 minutes compared with 157 ± 59 minutes; P = 0.002), which results in higher operating room charges ($416). It has been estimated that dual-chamber pacing costs approximately $9427 at the time of implant, compared with $6924 for single-chamber pacing [41]. The difference in cumulative cost increases as much as $5100 by the 12th year ($16 506 compared with $9392), reflecting the higher cost of follow-up and the shorter functional life, with the need for earlier reoperation. Although eventual reoperation costs are similar for both single- and dual-chamber pacemakers, the added likelihood of generator replacement for dual-chamber systems has been conservatively estimated around 24% (assuming a high mortality rate of 15% at 1 year), compared with only 8% for single-chamber systems [41]. However, operating costs will also increase for ventricular-paced patients who need upgrading to dual-chamber pacing because of the development of the pacemaker syndrome [42]. Estimates on the incidence of this complication vary widely [43, 44], but it may be between 5% and 10% [42], which agrees with our rates. Because the pacemaker syndrome has also been observed with physiologic pacing [45, 46], the impact of this complication on costs should be prospectively evaluated.

Limitations of Our Study

The major limitation is the retrospective, nonrandomized nature of the study, with a relatively small proportion of patients with ventricular pacing. Thus, healthier patients were probably not selected for ventricular pacing.

The cause of death could not be ascertained in 12 patients. However, sensitivity analysis showed that the predictors for cardiovascular death did not change when all 12 deaths of unknown cause were reanalyzed as being cardiovascular related.

Fewer than half of the patients had a complete echocardiographic assessment available. It is possible that echocardiographic variables could have become stronger predictors if they were available for all patients.

Few patients had single-chamber atrial pacemakers (AAI/R) [27], a relatively inexpensive alternative for patients with the sick sinus syndrome, which is commonly used in Europe. Although the long-term stability of AV conduction in selected patients with atrial pacing is well established [47], concerns about the subsequent development of high-degree AV block have been a reason for its limited use in the United States [1]. In our study, a considerable number of patients not only have bundle branch block but also have paroxysmal atrial tachyarrhythmias, with an imperative need for negative-dromotropic drugs. In patients paced in the AAIR mode, this could lead to a lack of physiologic adaptation of the AV interval to the sensor-driven pacing rate and result in a form of the pacemaker syndrome [48]. Consequently, AAI or AAIR may be less than optimal pacing modes in many patients with the sick sinus syndrome.

Our observations pertain only to patients with isolated sick sinus syndrome and not to those with AV block or carotid sinus hypersensitivity. Because patients with the sick sinus syndrome are probably paced less frequently throughout the day, findings from studies that mainly include patients with high-degree AV block may not be extrapolatable to patients with the sick sinus syndrome.

Need for a Randomized Study

Not surprisingly, mortality in patients with the sick sinus syndrome is strongly predicted by preimplant clinical markers of cardiac and vascular disease. Multivariable analytic methods (including the Cox model) are appropriate to find the most powerful predictor variables in a given population [49]; however, they may not be as effective in identifying the less cogent predictors [49]. Thus, our study is not conclusive about the effect of single-chamber ventricular pacing mode on mortality. A 25% to 35% benefit on long-term survival (beyond 5 years) from physiologic pacing cannot be excluded. This is precisely the magnitude of a difference between therapies likely to be detected only by adequately sized randomized trials [50]. The medical and industrial pacemaker community needs to apply the same rigorous, ethical methods [51] used in the last two decades in other areas of cardiovascular medicine, for the critical evaluation of pacing therapies. A multicenter, randomized study is needed to determine if statistical and clinical long-term differences exist in outcome (mortality and morbidity) between ventricular and physiologic pacing in patients with the sick sinus syndrome.