Connolly SJ, Sheldon R, Thorpe KE, Roberts RS, Ellenbogen KA, Wilkoff BL, Morillo C, Gent M, for the VPS II Investigators . Pacemaker Therapy for Prevention of Syncope in Patients With Recurrent Severe Vasovagal SyncopeSecond Vasovagal Pacemaker Study (VPS II): A Randomized Trial. JAMA. 2003;289(17):2224-2229. doi:10.1001/jama.289.17.2224
Author Affiliations: Departments of Medicine (Drs Connolly and Morillo) and Clinical Epidemiology and Biostatistics (Messrs Thorpe and Roberts, and Dr Gent), McMaster University, Hamilton, Ontario; Department of Cardiovascular Medicine, University of Calgary, Calgary, Alberta (Dr Sheldon); Department of Medicine, Medical College of Virginia, Richmond (Dr Ellenbogen); Department of Cardiovascular Medicine, Cleveland Clinic, Cleveland, Ohio (Dr Wilkoff); and Department of Cardiology and Cardiovascular Sciences, Instituto del Corazon, Floridablanca, Colombia (Dr Morillo).
Context Three previous small randomized trials have reported that pacemaker
therapy is beneficial for patients with severe recurrent vasovagal syncope.
However, because these trials were not double blind, they may have been biased
in their assessment of outcomes and had a placebo effect of surgery.
Objective To determine if pacing therapy reduces the risk of syncope in patients
with vasovagal syncope.
Design, Setting, and Patients A randomized double-blind trial of pacemaker therapy in outpatients
referred to syncope specialists at 15 centers from September 1998 to April
2002. In the year prior to randomization, patients had had a median of 4 episodes
of syncope. Patients were followed up for up to 6 months.
Intervention After implantation of a dual chamber pacemaker, 100 patients were randomly
assigned to receive dual-chamber pacing (DDD) with rate drop response or to
have only sensing without pacing (ODO).
Main Outcome Measure Time to first recurrence of syncope.
Results No patients were lost to follow-up. Of the 52 patients randomized to
ODO, 22 (42%) had recurrent syncope within 6 months compared with 16 (33%)
of 48 patients in the DDD group. The cumulative risk of syncope at 6 months
was 40% (95% confidence interval [CI], 25%-52%) for the ODO group and 31%
(95% CI, 17%-43%) for the DDD group. The relative risk reduction in time to
syncope with DDD pacing was 30% (95% CI, –33% to 63%; 1-sided P = .14). Lead dislodgement or repositioning occurred in
7 patients. One patient had vein thrombosis, another had pericardial tamponade
leading to removal of the pacemaker system, and a third had infection involving
the pacemaker generator.
Conclusions In this double-blind randomized trial, pacing therapy did not reduce
the risk of recurrent syncope in patients with vasovagal syncope. Because
of the weak evidence of efficacy of pacemaker therapy and the risk of complications,
pacemaker therapy should not be recommended as first-line therapy for patients
with recurrent vasovagal syncope.
Vasovagal syncope, also known as neurally mediated syncope, is a common
problem for which no highly effective pharmacological treatments are available.
Because vasovagal syncope episodes are often associated with bradycardia,
pacemakers have been proposed as a potential treatment. After some uncontrolled
follow-up studies reported a benefit,1,2 3
small randomized controlled trials of pacemaker therapy were performed.3- 5 All 3 trials reported
a reduction in syncope recurrence with pacing. However, treatment in all 3
studies was not blinded, such that patients and their physicians knew whether
the patient had received a pacemaker. It is possible that the reported benefit
of pacemaker therapy was due in part to a psychological or emotional effect
related to receiving a device by means of an invasive procedure. To exclude
this possibility and to provide a stronger level of evidence that prevention
of bradycardia by means of pacemaker therapy reduces the risk of recurrent
syncope in patients with vasovagal syncope, we conducted a randomized double-blind
trial of pacing.
Patients were eligible for this study if they were older than 19 years
and if they had a typical history of recurrent vasovagal syncope with at least
6 episodes of syncope ever, or at least 3 episodes in the 2 years prior to
enrollment. In addition, patients had to have a positive head-up tilt table
test result with a heart rate × blood pressure product of less than
6000/min × mm Hg. Each center used its own tilt study protocol. Although
there was variation in the head-up tilt table test protocols used, considerable
uniformity existed between protocols at the 15 centers in Canada, Australia,
the United States, and Colombia. A passive head-up tilt table test was conducted
at 60° to 80° for between 15 and 30 minutes and then an isoproterenol
infusion was administered at doses varying from 1 to 4 µg/min for 5
to 15 minutes. Nitroglycerine was used by some centers instead of or with
isoproterenol. The protocol was approved by a research ethics board at each
center and each patient provided signed informed consent. Patients were excluded
from the trial if any other cause of syncope was evident. They were also excluded
if they had important valvular, coronary artery, or myocardial disease; an
electrocardiographic abnormality; or any major noncardiovascular disease.
The trial was conducted from September 1998 to April 2002.
After implantation of a dual-chamber pacemaker (Medtronic Kappa, Medtronic
Inc, Minneapolis, Minn), patients were randomized by a central process to
dual-chamber pacing (DDD) or sensing without pacing (ODO) (Figure 1). An unblinded nurse or physician, who had no other patient
contact, did all the programming. The patients' physicians, the patients,
and all other study personnel remained blinded to treatment allocation. Blinded
study personnel and physicians were asked not to perform routine electrocardiograms.
Patients randomized to DDD also received rate drop response pacing, a feature
of the pacemaker that instituted rapid DDD pacing if the device detected a
rapid decrease in heart rate. The protocol specified that the initial rate
drop response parameters should be a drop size of 20 beats, a drop rate of
70/min, and an intervention rate of 100/min for 2 minutes.
The primary study outcome was syncope defined as a transient loss of
consciousness with prompt spontaneous recovery. Patients were requested to
report syncope episodes as soon as possible after the syncopal event occurred.
Evidence of syncope was collected including signs of injury and reports from
witnesses. A blinded committee of investigators adjudicated all reports of
syncope. The study follow-up period was 6 months or up to the time of occurrence
of the first episode of recurrent syncope.
This study was designed to have 80% power to detect a 50% relative reduction
in the risk of recurrent syncope from a rate of 60% in the control group to
30% in the treatment group. To achieve this, a study population of 80 patients
was planned. However, after enrollment of 60 patients, the combined event
rate of the 2 treatment groups was lower than anticipated, so the study target
enrollment was increased to 100 patients.
The primary analysis of the study was planned as a comparison of the
cumulative risk of syncope between the 2 treatment groups using a log-rank
test. All randomized patients had complete data for the primary outcome (recurrence
of syncope) and were analyzed according to the intent-to-treat principle.
Thus, all outcomes were attributed to the randomly assigned treatment groups
regardless of compliance to assigned treatment. The randomization schedule
was stratified by center and used randomly varying block sizes of 2 and 4.
The centers were not aware of the block sizes. The individual responsible
for randomization in the center was not involved in patient recruitment. All
patients received their allocated treatment assignment.
The relative risk reduction (RRR) was calculated as 1 − the hazard
ratio from a Cox model. The Cox model was also used for subgroup analyses.
A 1-sided test was specified for the primary analysis because it was judged
that there was no conceivable interest in or plausible potential for an increase
in syncope to occur with pacing. The 95% confidence intervals (CIs) are 2-sided.
All statistical analyses were performed using SAS (Version 8, SAS Institute
Inc, Cary, NC) and S-Plus (Version 6, Insightful Corp, Seattle, Wash) software.
A total of 137 patients met the inclusion criteria (Figure 1). Two patients had exclusion criteria prohibiting enrollment
(1 patient had 2 exclusions). Of the remaining 135 patients who met all eligibility
criteria, 100 were enrolled in the study and randomized. There were 48 patients
randomized to the DDD group and 52 to the ODO group. Baseline clinical characteristics
of the patients are shown in Table 1.
There were more men in the ODO group than in the DDD group, but otherwise
the 2 treatment groups were well matched. Patients had many prior syncope
episodes and a median of 4 in the year prior to randomization. Presyncope
episodes were also common.
By protocol design, all patients had had a positive head-up tilt table
study result. The mean duration of the tilt test was 30 minutes in both groups.
Syncope occurred during this test in 60% of the patients randomized to the
DDD group and 71% in the ODO group. The mean lowest heart rate recorded was
53/min in both groups and the mean lowest systolic blood pressure was 63 mm
Hg in both groups.
Many patients had previously tried medication to control syncope and
presyncope. The 2 most commonly used drugs were β-blockers and fludrocortisone. Table 1 shows the associated medical conditions
and the consequences of syncope in these patients. Many patients had previously
sustained injuries secondary to syncope episodes, had missed time from work,
or had had driving privileges restricted because of recurrent syncope.
Patients were randomized centrally via the telephone after implantation
of a dual-chamber pacemaker. The median time from implantation to randomization
was 1 day (maximum 4 days) and the median duration of hospitalization for
pacemaker implantation was 1 day. All patients randomized to the ODO group
received ODO programming. However, 46 of 48 patients randomized to the DDD
group received DDD pacing. The other 2 patients in the DDD group received
dual-chamber inhibited pacing. Rate drop response was activated initially
in all DDD patients. The median low rate programmed was 50/min.
During follow-up, no patients randomized to ODO pacing had pacing functions
activated before having an outcome event. No patient was lost to follow-up.
Several patients in the DDD group had programming changes. One patient had
the pacemaker system removed. Two patients had their pacemakers reprogrammed
to ventricular pacing due to atrial lead sensing and pacing problems. Concomitant
pharmacological therapy for vasovagal syncope was used in some patients during
study follow-up. Twelve percent of the patients in the ODO group compared
with 19% in the DDD group received β-blockers; fludrocortisone, 10% vs
2%; and selective serotonin reuptake inhibitors, 12% vs 13%.
A total of 38 patients had syncope during the 6-month follow-up period.
Of the 52 patients randomized to ODO, 22 had recurrent syncope within 6 months
compared with 16 of 48 patients in the DDD group. The median duration of syncope
reported was 2 minutes in the ODO group and 1 minute in the DDD group. Syncope
resulted in injuries with bruising or bleeding in 3 patients in each group;
no other injuries were reported. Syncope was witnessed in 12 patients in each
The Kaplan-Meier plots showing time to first episode of syncope, based
on intent-to-treat analyses, are shown in Figure 2. The cumulative risk of syncope at 6 months was 40% (95%
CI, 25%-52%) for the ODO group and 31% (95% CI, 17%-43%) for the DDD group.
The RRR in time to syncope with DDD pacing was 30% (95% CI, − 33% to
63%; 1-sided P = .14). Another treatment analysis,
which excluded 1 patient who deviated from allocated therapy (this patient
was randomized to the DDD group and had the pacemaker removed during the study),
showed an RRR with DDD pacing of 35% (95% CI, −26% to 66%; 1-sided P = .10).
To explore whether subgroups of patients benefited from a pacemaker,
exploratory analyses were performed (Figure
3). Age, duration of tilt test before syncope, and minimum heart
rate of less than 50/min during the tilt test did not define which patients
would benefit from pacing. Patients who received isoproterenol during the
tilt study were significantly more likely to benefit from pacemaker therapy
than those who did not require isoproterenol during the tilt study. Sex and
history of vehicular collision were also examined as potentially prognostic
baseline characteristics but were not associated with an increased or decreased
risk of syncope.
Information on presyncope was collected from patient diaries. Presyncope
was defined as a feeling of impending loss of consciousness that does not
result in complete syncope. Patients were instructed to record every episode
of presyncope and to grade each episode on a scale of 1 to 5, in which 5 was
the most severe. There were 49 (94%) patients with any presyncope in the ODO
group and 46 (96%) in the DDD group (P>.99). The
median reported episodes of presyncope per 100 days of follow-up were 16 in
the ODO group and 13 in the DDD group. Of those patients who recorded any
presyncope, the median maximum severity of presyncope was 4 for both groups.
Pacemaker complications occurred in several patients (Table 2). One patient had infection requiring reimplantation of
the pacemaker generator and another had pericardial tamponade leading to removal
of the pacemaker system.
Three previous randomized studies have reported that pacemaker therapy
reduces the risk of recurrent syncope in patients with vasovagal syncope.
In the first randomized trial, the Vasovagal Pacemaker Study (VPS I),1 patients were randomized to receive or not receive
a pacemaker. This pilot trial of pacing was terminated early when a large
treatment effect in favor of pacing was observed after enrollment of just
54 patients. Subsequently, 2 other randomized trials of pacing for vasovagal
syncope were also terminated early. One of these trials randomized 42 patients
to receive a pacemaker or not.2 The other trial
randomized 93 patients to receive a pacemaker or to receive a β-blocker
(atenolol).3 All 3 studies observed statistically
significant reductions in the risk of syncope in patients who received pacing.
However, all 3 of these trials were unblinded. Patients and physicians
knew whether pacing therapy was being used or not. Any open-label trial has
the potential for bias in reporting and assessment of outcomes. For trials
in which the outcomes of interest are major morbid events such as stroke or
death, the risk of bias in outcome assessment is minor. However, syncope is
an outcome that has a major subjective component and is difficult to verify
objectively. It is possible in the unblinded studies that some patients, hoping
to have received a pacemaker and disappointed by being randomized not to receive
one, may have been more prone to report syncope. The double-blind trial design
removes this type of potential bias to a considerable extent.
Vasovagal syncope episodes can be aggravated by adverse experiences
such as anxiety and fear. It is possible that the disappointment of being
randomized not to receive a pacemaker in unblinded trials actually increased
the likelihood that patients would have recurrent syncope. On the other hand,
it is well accepted that surgical procedures can have a placebo effect independent
of the actual surgical care received.6- 8 Patients
receiving a pacemaker may have benefited from the psychological effects of
receiving a surgical procedure. Although this effect is an accepted part of
medical care,6 it is important to know whether
pacemaker implantation is beneficial because of its physiological effects
on the heart or because of the psychological effects of surgery, or both.
In the 3 unblinded studies, it was not possible to determine whether patients
benefited from the actual pacing therapy.
To reduce potential for bias and confounding psychological effects,
we performed this double-blind study. We expected that the risk of syncope
in the control group would be reduced to some extent by the receipt of a device,
even if it was not actually pacing, and we increased the study's statistical
power accordingly. The VPS II is, to our knowledge, the first double-blind
randomized trial of pacing, and also is the largest of the randomized pacemaker
trials for vasovagal syncope.
The main finding of this double-blind trial was that a statistically
significant benefit was not found for pacemaker therapy for prevention of
syncope in patients with vasovagal syncope. The main difference between the
results of this trial and the nonblinded VPS I is the observed risk of syncope
in the nonpaced group. Whereas in the VPS I study almost 80% of nonpaced patients
had syncope by 6 months, in this study only 41% of ODO patients had syncope
by 6 months. The 6-month rates of syncope in the patients receiving pacing
therapy in the 2 studies were more similar; 20% in VPS I and 31% in the present
study. Another difference between this trial and all 3 previous trials was
that the previous trials were all terminated prematurely. Early termination
of a trial for unexpected efficacy tends to overestimate the treatment effect.
This study was designed to detect an RRR with pacing of 50%. The observed
RRR was 30% with a wide 95% CI. An RRR of 50% with pacing is unlikely but
still plausible. However, the large RRRs (in the range of 80%), which were
observed in the 3 unblinded randomized trials, are unlikely. These RRRs are
not included in the 95% CI of the RRR observed in this study, the upper limit
of which was 63% RR. This trial was designed to have reasonable power to detect
an RRR of 50%, which we believed to be the minimum effect size that would
justify this invasive treatment. The RRR of 30% observed in this study, if
it were real, might be considered by some to be a reasonable benefit to obtain
by pacing. Based on the results of this study, a benefit of this magnitude
is plausible but not proven.
The rationale for the use of a pacemaker in vasovagal syncope is that
bradycardia often occurs at the time of syncope. Prevention of bradycardia
is the main physiological mechanism by which a pacemaker can prevent attacks
of syncope. However, patients with vasovagal syncope often experience reductions
in blood pressure at the beginning of a syncopal episode and heart rate changes
later.9,10 If profound hypotension
has already occurred, pacing therapy will not help patients even if bradycardia
or asystole has been demonstrated at the time of syncope. It is possible to
capture the marker channel information recorded in the pacemaker at the time
of syncope if the patient activates the pacemaker to do so shortly after the
syncope event. Several patients made such recordings and when these data are
analyzed, it may provide information to help understand how the pacemakers
were functioning at the time of syncope.
Of the subgroup analyses performed, 1 was statistically significant
and it suggested that patients who required isoproterenol during their tilt
test were more likely to respond to pacing therapy than those who did not
receive isoproterenol. It is difficult to find a biologically plausible reason
why this should be so, and it is possible that this finding is due to the
play of chance.
Small differences in study design between this study and VPS I are unlikely
to explain the different results, but should be noted. The VPS I study only
enrolled patients with a minimum heart rate below 60/min during the tilt test
and the present study did not specify any minimum heart rate during the tilt
test. Both studies required a heart rate blood pressure product below 6000/min
× mm Hg. The percentages of patients with a heart rate of less than
40/min during the tilt test were similar in both studies. In VPS I, 12 (22%)
of 54 patients enrolled had a rate of less than 40/min during the tilt test
compared with 19 (19%) of 100 patients in VPS II. Therefore, there were similar
numbers of patients with extreme bradycardia at the time of positive tilt
test results in both studies, and this minor difference in study design is
not a factor in the different results observed. Moreover, our study allowed
investigators to use their own institutional protocol for head-up tilt table
testing. This reflects the fact that there is considerable variation in the
details of the tilt test procedure. This increases the generalizability of
the results of this study.
This study is unique among pacemaker and device trials because of the
use of a double-blind study design, which removed the potential bias that
could occur if patients and physicians knew which treatment a patient had
received. Although the use of placebo-controlled surgery trials can be criticized
as unethical, this was less of a concern in this study because patients receiving
the pacemaker could have it activated to deliver pacing therapy once study
participation had been completed.11
Considering the risk of complications, the rate of recurrence of syncope
in the patients receiving a pacemaker and the weak evidence for any true benefit
of pacing, pacemaker therapy should not be recommended as first-line therapy
for patients with vasovagal syncope.