Accelerometer Systolic Time Intervals as Fast-Response Sensors of Upright Posture in the Young
Background Sensors of posture may improve rate-adaptive pacing in syndromes where syncope occurs in the upright posture, particularly in the young. No sensor of posture has been described to date. Previous studies suggest that two sensors currently under investigation (preejection period [PEP] and left ventricular ejection time [LVET] systolic time intervals [STIs] and accelerometers) may be affected by posture. A PEP-sensing pacemaker is available commercially in which heart rate (HR) decreases with an increase in PEP (∂[HR]/∂[PEP]<0). In patients with upright syncope, it is not known how such algorithms respond to posture. Also, it is not known whether STIs correlate with posture independent of autonomic tone.
Methods and Results We studied accelerometer-derived STIs in head-upright tilt-testing with β-blockade and catecholamine stimulation in patients with syncope or presyncope using an ultra-low-frequency accelerometer placed on the chest. Thirty-two patients age 6 to 22 years with unexplained recurrent syncope or presyncope underwent tilt-testing involving two to four tilts (60°) at baseline, during esmolol infusion (500 μg/kg load, 50 to 140 μg/kg per minute), after esmolol withdrawal, and during isoproterenol infusion if not contraindicated. PEP, LVET, and other indexes were quantified, and their relations to posture and to autonomic state were determined. With tilt, PEP increased from 98.9±2.2 to 109.1±2.8 msec (P<.0001), and LVET decreased (supine-to-upright) from 295.5±4.5 to 247.2±4.7 msec (P<.0001). PEP/LVET changed from 0.337±0.01 to 0.45±0.02 (P<.0001). Similar postural changes were observed during tilt with β-blockade and esmolol withdrawal, and during isoproterenol infusion. STI changes occurred immediately on postural change and were stable. Postural change of PEP was greater than the β-adrenergic effect by 6:1. Postural change of STIs was independent of vagal tone.
Conclusions First, accelerometer-derived STIs detect postural changes. Because these changes are independent of autonomic tone and are rapid and stable, they may be useful as fast-response sensors of upright posture in rate-adaptive pacemakers. Second, with postural change, HR increases when PEP increases. However, PEP-sensing pacemakers presently under investigation assume the opposite (inverse) mathematical relationship. Therefore, current PEP-sensing pacemakers use an incorrect algorithm for physiological postural responses in syncope patients. These data predict a paradoxical tachycardic response to the supine posture in patients implanted with these devices.
Syncope caused by severe bradycardia or asystole may be treated with a permanent pacemaker in adults as well as in children, whether related to atrioventricular block, sick sinus syndrome,1 2 3 or cardioinhibitory vagal syndromes.4 5 6 7 8 9 10 11 12 13 14 15 Syncope usually occurs in the upright posture.
Vagally mediated syncopal syndromes (hypersensitive carotid sinus syndrome being the prototype)11 12 13 14 15 may be most frequent in the pediatric population, eg, cardioinhibitory (vagal) syncope and convulsive syncope.4 5 6 7 8 9 10 11 16 17 Although rarely requiring permanent pacemakers,4 5 6 8 9 10 11 18 19 these syndromes focus the clinician’s attention on upright posture as a precondition for syncope because they are characterized by abnormal heart rate response to upright posture in the absence of structural heart disease or detectable rhythm or conduction tissue disease. The several pacing approaches used to treat these syndromes have numerous drawbacks because they fail to sense upright posture to provide a more rapid rate when the patient stands.
Another group of patients likely to benefit from sensors of posture are those with atrial chronotropic incompetence,20 21 22 23 24 in whom the limited metabolic needs of light activity are poorly or tardily sensed by present rate-adaptive pacemakers. A sensor detecting upright posture could alleviate this problem and improve exercise tolerance without resorting to nonphysiological use of other sensors.
This raises the question of whether pacemaker research should assign priority to a sensor of upright posture for use in rate-adaptive pacing. Would machine perception of posture improve pacemaker therapy?
Sensors described for rate-adaptive pacing include fast-response sensors that detect onset of exertion (eg, vibration [“activity”] sensors, stimulus-T [“QT”] sensors, O2 saturation sensors inter alia) and a smaller number of slow-response sensors (correlating with degree of sustained exertion, eg, minute ventilation/respiratory rate sensors, temperature sensors). None of these detect posture as such. However, the recent introduction of two new fast-response sensors, PEP25 sensors and accelerometers (both single axis and multiple axis) raises the possibility that a sensor may be used to derive an output that correlates with upright posture.26 27 28 A PEP-sensing pacemaker is available commercially in which heart rate decreases with PEP increase25 29 30 (ie, heart rate–PEP relationship: ∂[HR]/∂[PEP]<0).29 However, PEP responses to postural stress and variation of autonomic state have never been reported for syncopal individuals.
In the present investigation, we studied PEP and LVET in a large group of younger patients with clinical syncope and presyncope with and without heart disease by using the conditioned signal of an ultra-low-frequency accelerometer fastened to the chest to derive STI.31 32 We have determined STI in relation to posture in the resting state and with β-blockade, β-adrenergic stimulation, and vagal stimulation as part of autonomic testing employing head-upright tilt-testing.19
Patient Selection and Evaluation
The patient population consisted of 32 consecutive individuals with unexplained recurrent upright syncope or presyncope with historical features not consistent with simple faint who underwent head-upright tilt-testing and consented to accelerometer study. Blood pressure measurements were taken in the lying, sitting, and standing positions. The patients had 12-lead ECG and two-dimensional echocardiography; 24- to 48-hour Holter monitoring and exercise tests were performed if indicated. Clinical evaluation and diagnostic tests did not establish the cause of the symptoms. For all except 2, tilt-testing was recommended as a diagnostic test before consideration of catheterization, electrophysiological testing, or treatment. In 2 competitive athletes who had syncope during exertion, electrophysiological testing was performed (and found to be negative) before tilt-testing.
Head-upright tilt-testing with esmolol and esmolol withdrawal was performed as described previously.19 Testing was started at midday 30 to 60 minutes after intravenous cannula insertion and noninvasive ECG and blood pressure monitoring (Quinton Q-5000, Colin Pulsemate BX-5 or equivalent unit) using a motorized tilt-table with foot-plate support and a piezoelectric accelerometer (PCB-Piezotronics 336A) affixed to the sternum. An external pacemaker (Zoll) was attached to permit immediate pacing in patients with a history of convulsions.
Baseline tilt4 19 33 was maintained for 49 minutes unless syncope or intolerable symptoms developed, at which time the patient was returned to the supine position. Esmolol tilt19 34 was performed after supine rest for 20 to 30 minutes. Esmolol (Brevibloc) was administered intravenously with a 500 μg/kg loading dose followed by a 50 μg/kg per minute infusion. Maintenance infusion rate was increased by up to 20% every 2 to 4 minutes to 140 to 150 μg/kg per minute provided the heart rate exceeded 50 bpm. Sixty-degree head-upright tilt was then initiated after 3 to 10 minutes when the heart rate was stable. This position was maintained either for 15 minutes (if baseline tilt was negative) or for the full time of baseline tilt (59 minutes) plus 5 to 10 minutes (if baseline tilt had been positive) or until syncope. For esmolol-withdrawal tilt,19 at the end of the esmolol tilt, esmolol infusion was acutely discontinued while maintaining the tilt. This position was maintained for an additional 29 minutes, or until syncope or intolerable symptoms occurred. Isoproterenol tilt was performed at least 45 minutes after discontinuation of esmolol if the three previous tilt-tests were negative and there were no contraindications to isoproterenol. Isoproterenol was given at a dose of 0.01 μg/kg per minute, titrated upward every 1 to 2 minutes to a heart rate ≥120% baseline before 60° head-up tilt was initiated. The infusion rate was reduced if the heart rate exceeded 150 bpm. The tilt was continued for 15 minutes or until syncope or intolerable symptoms occurred.
A positive response to a tilt-test was defined as one that duplicated the patient’s symptoms in association with hypotension or bradycardia and required recumbency for recovery. If cardiopulmonary resuscitation or transcutaneous pacing was required for resolution of symptoms, the protocol was aborted, and the patient was considered “positive” for that tilt-test.
Accelerometer-Based Measurement of STI
An externally applied ultra-low-frequency piezoelectric accelerometer (PCB Piezotronics Flexcel 336A; frequency range, 1 to 2000 Hz; resolution, 0.0005 g peak; sensitivity, 102 mV · m−1 · s−2) was used in conjunction with a dedicated personal computer–based customized modification of a commercially available signal-averaging system (Seismed Instruments Corporation) to derive the PEP, LVET and their ratio, PEP/LVET.
The accelerometer, which weighs 4.5 g, was fastened securely either by tape or snap electrode to the sternum or anterior precordium within 3 cm of the xiphisternum. This method of attachment is similar to that described for noninvasive testing of piezoelectric motion sensors35 36 and piezoresistive accelerometers37 of commercially available rate-adaptive pulse generators. The simultaneous accelerometer signal and a surface ECG lead were recorded and stored in the computer-based signal-averaging system.
The Q wave and well-defined points on the signal-averaged accelerometer waveform were used for derivation of STIs. PEP is measured from onset of the Q wave to the accelerometer correlate of the opening of the aortic valve (the AO point). LVET is measured from the AO point to the closure of the aortic valve (the AC point).31 32 The PEP/LVET ratio is determined directly from the measurement values not corrected for heart rate. Recordings (20 to 60 seconds long) of accelerometer signal waveforms are cross-correlated beat-to-beat, and families of beats are identified. Tracings were excluded if they did not have at least one family with ≥25% of the beats. With this system, STIs may be obtained both continuously in real time (with 30-second delay) and from later analysis of files containing the digitized accelerometer signal. When obtained from later analysis of files, time resolution is limited to 4 msec.
Before tilt, a 1-minute interval of accelerometer data was obtained after 10 to 15 minutes of rest in the supine position. This was repeated within 1 minute of assumption of 60° head-upright tilt in the baseline state. One-minute recordings of the accelerometer signals were obtained every 3 to 5 minutes through the procedure. Initial tilt and periodic recordings during tilt were repeated during esmolol tilt, esmolol-withdrawal tilt, and isoproterenol tilt-testing if performed.
Systolic and diastolic blood pressures, heart rate, and respiratory rate were recorded for the immediate pretilt, immediate posttilt, immediate presymptomatic period, and symptomatic period when the test was terminated, as well as every 30 to 60 seconds during the tilt before symptoms and termination.
From the signal-averaged accelerometer tracings during tilt, the PEP, LVET, and PEP/LVET were derived. STI records were further reviewed to determine time after tilt onset at which a significant change occurred (time when trend was noted as well as time when STI stabilized at new level) and stability of trend over serial records. The partial derivative ∂[HR]/∂[PEP] (in ms−1 · min−1) was evaluated for each tilt-test both at a time immediately after tilt and for the time of maximal postural stress or β-adrenergic stress, as reflected in maximal induced heart rate.
Relative magnitudes of STI change caused by alterations in β-adrenergic state versus those caused by change in posture were discriminated by computation of the ΔPEP (change in PEP) specific to each of these interventions: ΔPEPβ-blockade, ΔPEPposture, esmolol HUT, ΔPEPposture, ΔPEPβ-blocker withdrawal, and ΔPEPposture, esmolol-withdrawal HUT (see text for further details).
Student’s paired t test or the Wilcoxon test was used for paired comparisons, with significance assigned to P<.05. One-factor ANOVA was used to compare hemodynamic variables among the groups and to compare ΔPEPsupine-to-tilt, initial ∂[HR]/∂[PEP], and maximal ∂[HR]/∂[PEP] inter alia.
The study population included 32 patients with unexplained recurrent syncope or presyncope despite standard workup (Table 1⇓). The patients ranged in age from 6 to 22 years at time of presentation, with a median of 14 years. (All except 3 were 10 to 18 years old.) Twenty-eight patients had syncope, and 4 had presyncope only. Thirteen of the patients with syncope had a history of seizures associated with syncope. No patient had isolated loss of consciousness triggered by fright or surprise (simple faint). Two-dimensional echocardiography was performed in 31, and 24- to 48-hour Holter monitoring in 26. Those with symptoms during exercise underwent treadmill testing; treadmill results were nondiagnostic in all. Twelve had heart disease believed not to be hemodynamically significant. Six of the children with heart disease and an additional 15 without structural heart disease had arrhythmias or conduction system disturbances believed not to be causing syncope (1 had a pacemaker).
Clinical Observations on Tilt-Testing
Syncope or intolerable vagal symptoms were provoked in 21 of the 32 patients (positive test). In 11 patients, all tilt-tests were negative.
In 20 patients, the baseline, esmolol, or esmolol withdrawal test was positive. Of the patients with at least one positive test, 8 had positive baseline tilt, 9 had positive baseline and positive esmolol or esmolol-withdrawal tilt, 3 had negative baseline but positive esmolol- withdrawal tilt, and 1 had positive isoproterenol tilt. No patient had negative baseline but positive esmolol tilt. In the patients with any positive test, syncope was preceded by hypotension before marked decrease in heart rate in 16; in 5, the heart rate decreased or there was asystole that preceded hypotension.
Results in 10 patients suggested that syncope might respond to β-blockade and in 3 patients that β-blockade would have no beneficial effect. Two patients with seizure presentation were treated with pacemakers. Six patients had tilt-positive syncope where the data were inadequate to predict response to β-blocker therapy. The 11 patients with all negative tests underwent further evaluation.
Three patients underwent repeated tilt investigation to reconfirm diagnosis. Two patients had cardioinhibitory syncope with asystole and convulsions (Fig 1⇓). Both were treated with permanent pacemakers (DDD) and, on repeat tilt, no longer had syncope or convulsions. In follow-up, both have remained seizure free. The third patient who underwent repeat tilt was the one patient who had positive isoproterenol but negative baseline, esmolol, and esmolol-withdrawal tilts; the patient had syncope during two trials of oral β-blockers, and repeat tilt was negative during isoproterenol tilt, while not on β-blockers. The original positive isoproterenol tilt may have represented a false-positive. (Follow-up has been reported for 17 patients.19 38 )
Heart Rate and Blood Pressure Responses in Early Tilt
Heart rate and blood pressure responses at onset of tilt are depicted in Table 2⇓. Heart rate increased significantly, but insignificant initial changes in systolic and diastolic blood pressures were recorded in tilt-tests in all β-adrenergic states. Maximal heart rate changes (maximal heart rate increase on tilt) were compatible with the β-adrenergic stimulatory character of esmolol withdrawal and isoproterenol tilt-testing as reported previously.19
Only early tilt responses are documented in Table 2⇑, as these are reflective of postural circulatory stress independent of tilt-test outcome.
Accelerometer-Derived STI Postural Change in Various β-Adrenergic States
Accelerometer-derived STI measures were recorded for each of the four tilt-tests and are reported in Table 3⇓. On baseline tilt, PEP immediately increased with assuming the upright posture in 30 of 32 patients (P<.0001, Wilcoxon). In the other 2, the increase was delayed 3 and 10 minutes, respectively. LVET diminished immediately on assuming the upright posture in 32 of 32 patients (P<.0001, Wilcoxon). With baseline tilt, the accelerometer-derived PEP increased from 98.9±2.2 to 109.1±2.8 msec (P<.0001), LVET decreased (supine-to-upright) from 295.5±4.5 to 247.2±4.7 msec (P<.0001), and PEP/LVET changed from 0.337±0.01 to 0.45±0.02 (P<.0001). Examples of PEP and LVET changes with posture in tilt-testing are depicted in Figs 2 through 4⇓⇓⇓, and statistical results are presented in Table 3⇓.
Similar postural changes were observed with the esmolol tilt, reflecting β-blockade. PEP increased immediately in 25 of 30 and did not change in 3 of 30 (P<.0001, Wilcoxon). A missing supine or tilt state accelerometer tracing (one each) prevented data for 2 patients from being evaluated. The 3 patients in whom PEP did not increase immediately did experience subsequent increase. LVET diminished immediately with tilt in 28 of 30 (P<.0001, Wilcoxon).
Similar results were observed with the two forms of catecholamine stimulation studied at the time of maximal β-adrenergic effect. With the esmolol withdrawal tilt-test, PEP increase was observed in 22 of 25 (P<.0001, Wilcoxon), with no change in 2 of 25 and noncollected data in 1 of 25. LVET decrease was observed in 24 of 25 (P<.0001, Wilcoxon) with noncollected data in 1 of 25. It may be noteworthy that supine heart rate in this small group of isoproterenol-tilted patients was 117±16 bpm (range, 96 to 134 bpm) and ranged on tilt from 119 to 169 bpm, with 1 patient experiencing junctional tachycardia.
In all tilt-tests, the PEP increased significantly. By ANOVA, there was no difference in the increment in PEP compared among the different physiological states represented by the different tilt-tests. Examples of PEP and LVET changes with postural stress in esmolol and esmolol withdrawal tilt-tests are depicted in Fig 3⇑.
Thus, PEP increased with upright postural change in all tilt-tests, LVET decreased, and PEP/LVET increased. Changes were invariant with respect to β-adrenergic state.
Rapidity and Stability of Accelerometer-Derived STI Changes
The STI changes described were detected in the initial accelerometer recording (started at 0 seconds of tilt) in 30 of 32 for PEP and in 32 of 32 for LVET. In 2 of 32, the accelerometer-derived PEP increase was delayed by 3 and 10 minutes, respectively. Thus, in 30 of 32, tilt-induced STI change had occurred for both PEP and LVET by the time of first posttilt determination and could be documented to have occurred as early as 20 seconds after assuming 60° tilt posture.
After the posture-related PEP increase had occurred, the PEP remained elevated in 26 of 32 patients, within the ±2 msec resolution of the system (evaluating six serial measurements). In 6 of 32, there was spontaneous late variation of PEP not associated with postural stress with cumulative PEP decrease similar in magnitude to postural increase.
After the initial decrease in LVET, the LVET remained depressed in 32 of 32 patients. Thus, the STI measures are characterized by rapidity and excellent initial stability.
Magnitude of PEP Change Related to β-Adrenergic State Versus Change Related to Posture
The change in PEP caused by change in β-adrenergic state was compared with the change in PEP due to postural stress in two different ways. First, the change in PEP caused by β-blockade in the supine position was quantified by computing the difference between the supine PEP with β-blockade to the supine PEP without β-blockade. This change in supine PEP related only to β-blockade was ΔPEPβ-blockade=1.79±1.6 msec. This number is to be compared with the change in PEP due to postural stress in the same tilt-test, ΔPEPposture, esmolol HUT=12.1±1.9 msec. Clearly, the change due to postural stress far exceeds the magnitude of change due to change in β-adrenergic state. The ratio of the change in PEP due to postural stress versus the change in PEP due to variation in β-adrenergic state is approximately 7:1.
Alternatively, the ΔPEPβ-blockade could be compared with the mean ΔPEPposture for esmolol tilt and baseline tilt (11.2±1.1 msec). In this case, the ratio of the change in PEP due to postural stress versus the change in PEP due to variation in β-adrenergic state is approximately 6:1, with confidence interval of (6.1±2.2) to 1.
We also compared the change in PEP due to β-adrenergic stimulation, ΔPEPβ-blocker withdrawal=3.0±1.3 msec, with the corresponding PEP change due to posture, ΔPEPposture, esmolol-withdrawal HUT=15.5±2.3 msec. Again, the ratio of variation of PEP related to posture, to the variation in PEP due to change in β-adrenergic state, was 5:1 or, with confidence interval, (5.2±2.6) to 1. PEP changes not associated with a change in posture were not significant. The changes of PEP due to change in β-adrenergic state without postural influence are displayed in Fig 5⇓.
The magnitude of any PEP change related to posture is clearly significantly greater than the changes related to β-adrenergic state by a factor of from 5:1 to 7:1. Catecholamine change of PEP fails to attain statistical significance in any measure.
STI and Heart Rate Changes in Relation to Tilt-Test Outcome
Postural change of STIs was independent of vagal tone reflected in clinical symptoms (impending vagal syncope versus none, P=NS). Heart rate and maximal heart rate increase related to postural change early in tilt-testing were similarly invariant with respect to tilt-test outcome. The STI changes appeared to reflect the postural state of the patient and little else.
The correlation of STI with posture is clear through all the data. Although upright posture may be a necessary precondition for syncope, it was not sufficient to provoke syncope in all patients in all tilt-tests with the elaborate and internally redundant tilt protocol used in the present study. Therefore, a spurious correlation of these STI changes with syncope was avoided.
Estimation of ∂[HR]/∂[PEP]
Directly measured, ∂[HR]/∂[PEP]=4.06±1.04 ms/min under conditions of maximal postural stress on baseline tilt (before hemodynamic decompensation) and 0.64±0.40 ms/min immediately on tilt (n=32). At time of maximal postural stress in esmolol tilt, ∂[HR]/∂[PEP]=1.52±0.63 ms/min (n=25), reflecting the reduced heart rate increase. At the time of maximal β-adrenergic stimulation under esmolol withdrawal, ∂[HR]/∂[PEP]=3.50±0.82 ms/min (n=23), and at the time of maximal heart rate during isoproterenol tilt, ∂[HR]/∂[PEP]=1.03±0.44 ms/min (n=6), omitting 1 patient who had a complication of isoproterenol (junctional tachycardia). The ∂[HR]/∂[PEP] measures are depicted in Table 4⇓ and Fig 6⇓. For all tilt-tests, ∂[HR]/∂[PEP] is positive. The initial as well as the maximal ∂[HR]/∂[PEP] were compared among the different tilt-tests, with no significant differences by ANOVA.
It is noteworthy that mean ∂[HR]/∂[PEP]>0 (ie, the sign is positive) in response to postural stress in all physiological states tested (P<.0005 for positivity, Fig 6⇑). Thus, the positivity of ∂[HR]/∂[PEP] appears to be a physiological constant, independent of autonomic tone.
The most important finding of the present study is that accelerometer-derived systolic time intervals PEP and LVET are sensitive sensors of upright posture with fast response in this highly selected group of younger patients with upright syncope and presyncope. These data show that STI changes occur rapidly on assumption of the upright posture and are invariant with respect to β-adrenergic state (β-blockade, β-adrenergic stimulation) and to vagal tone. The esmolol/esmolol-withdrawal tilt protocol was used because it is a rigorous tilt-testing protocol that exposes the patient to multiple autonomic stresses, including two states of β-adrenergic stimulation and one state of β-blockade, and appears to provoke an extreme hypervagotonia that can lead to syncope of several types (cardioinhibitory, vasodepressor, and mixed) as well as to convulsions in patients with convulsions during syncope16 17 19 (Fig 1⇑).
The findings support the conclusion that accelerometer-derived STI may be an effective fast-response sensor of upright posture independent of autonomic tone, with utility in rate-adaptive pacemaker therapy.
PEP and the Physiological Response to Tilt
The finding that accelerometer-derived PEP increases on assuming the upright posture during tilt and that LVET diminishes is different from recent reports, particularly with regard to the direction of change of PEP. PEP has been reported to decrease with tilt,29 30 whereas the data show that in our patients, the predominant response of PEP was to increase (Fig 6⇑). These results corroborate a similar finding in healthy humans.28 39 In the present study, the determinations are made in a presymptomatic phase in patients who may go on to lose consciousness from vagal or other mechanisms. The measurements represent determinations made during a compensated phase of postural stress when intrinsic postural reflexes (eg, heart rate elevation caused by arterial baroreceptor withdrawal) are still functioning adequately to avoid loss of consciousness, although these compensatory reflexes may be just about to fail. Examples of this are provided by the two individuals who were diagnosed as having a seizure disorder in whom measurements were made just before the sudden occurrence of protracted asystole. At the time of the measurements, these patients were asymptomatic with postural stress; shortly thereafter, during asystole and on recovery from asystole, each experienced convulsions similar to his or her clinical syndrome. Recovery from asystole was assisted in one case by cardiopulmonary resuscitation (Fig 1⇑), where a 33-second period of asystole was observed, not heralded by hypotension. Thus, PEP measurements were taken in the upright posture in the asymptomatic state at a time when it may have been desirable for pacing to be instituted. (That pacing before syncope may be helpful in avoiding clinical syncope and convulsions is evidenced by asymptomatic responses to tilt-test after pacemaker implantation in these patients and symptom-free follow-up).
Such physiological measurements have not previously been reported.
Relation of Heart Rate to PEP
We have demonstrated that the heart rate and PEP increase together with postural change, ie, ∂[HR]/∂[PEP]>0. This relation has been shown to hold independent of autonomic state. In these patients subjected repeatedly to postural stress under varying autonomic conditions, the relationship ∂[HR]/∂[PEP]>0 was shown to hold for 30 of the 32 on baseline tilt (Fig 6⇑), and remarkably, when studied during other tilt-tests, the relation was found for all patients who underwent β-adrenergic stimulation (esmolol withdrawal or isoproterenol, Fig 6⇑). Furthermore, we have shown that the magnitude of change of PEP under postural change was greater than that observed under varying β-adrenergic state and specifically that the variation of PEP from supine-to-tilt was greater (by a factor of between 5:1 and 7:1) than the PEP change between states of β-blockade and β-adrenergic stimulation.
These relationships stand in evident contrast to the published relation assumed in the algorithm of the existing PEP-sensing pacemakers, ie, ∂[HR]/∂[PEP]<0. In a patient with such a pacemaker (∂[HR]/∂[PEP]>0), based on the data presented in this study, it would be predicted that the heart rate would decrease on going from supine to tilt and, conversely, that heart rate would increase on lying down, causing inappropriate pacemaker-mediated tachycardia in the supine position if the pacemaker is functioning correctly. Such tachycardias have been described with other sensors40 but have never previously been attributed to such a mechanism.
Recent reports suggest that this prediction is borne out in patients who have received this type of pacemaker. Ruiter et al29 presented data from patients implanted with such pacemakers, where heart rate changes are inversely related to PEP changes ∂[HR]/∂[PEP]<0. In their published raw data, 7 of 10 patients had an increase of PEP contrary to what was anticipated when standing from supine position (ie, similar to what was observed in the present study but contrary to what was anticipated by those authors and contrary to the assumptions underlying the pacemaker algorithm for determining heart rate). Several patients experienced tachycardia and palpitations in the supine posture. This necessitated disabling the sensor (ie, reprogramming to DDD) in at least two cases. On the basis of our data, we suggest that this behavior is due not to any idiosyncrasy of those patients but rather to the use of a faulty algorithm in the pacemaker. Such a paradoxical response to the supine posture may require disabling of the rate-response circuitry in other patients who may already have received such units. Such observations lead to the consideration that pacemakers of this particular type are probably of no significant benefit over DDD (atrioventricular universal) pacemakers and may be detrimental.
Pacing and Upright Syncope
In the absence of structural heart disease, vagally mediated syndromes (eg, hypersensitive carotid sinus syndrome, neurocardiogenic syncope, and convulsive syncope) represent important causes of syncope and related syndromes. Symptoms in the hypersensitive carotid sinus syndrome (where upright posture is a necessary condition for syncope) may respond to dual-chamber pacing.11 12 13 14 15 41
Syncope in neurocardiogenic syncope (without convulsions) typically does not respond to pacing, although some discrepancies appear to persist in American versus British and continental medical literature4 5 6 7 8 9 10 11 18 19 41 42 regarding (1) the ability of pacing to abolish spontaneous syncope versus syncope that occurs during provocative testing and (2) the ability of pacing to reduce syncope in some cases of mixed syncope (cardioinhibitory and vasodepressor features). In the benchmark article on the failure of vasodepressor vagal hypotension and syncope to respond to pacing,18 it is noteworthy that no patient had convulsive syncope and no patient had a primary cardioinhibitory response to orthostatic stress; in the individuals who had mixed syncope, the cardioinhibitory component appeared later and was mild.
Convulsive syncope (ie, in which a convulsion is related to asystole) may respond to pacing. Although convulsions occur in several vagal syndromes spanning the timeline from infancy to adulthood, the best-studied convulsive syndromes are those occurring in cardioinhibitory syncope (either pure cardioinhibitory syncope or mixed syncope with severe and early cardioinhibitory component). In our19 experience and that of several researchers,9 10 these convulsive syndromes appear to respond to pacing. In our modest experience with approximately 25 patients with syncope and convulsive symptoms, it has been observed that in cardioinhibitory syncope, permanent dual-chamber pacing abolishes convulsions. Because typically these patients present with generalized motor convulsions (the clinical features are those of a new-onset seizure disorder), these syndromes are not likely to be confused with vasodepressor neurocardiogenic syncope; rather, they are mistaken for idiopathic epilepsy of recent onset. The convulsion typically occurs at the time of reperfusion when the heart starts beating again after a period of asystole; thus, the definition of convulsive syncope must be taken to assume a very particular temporal association between asystole and the convulsion. Successful pacemaker therapy of convulsive (asystolic) syncope may not alleviate all syncopal and presyncopal symptoms. Indeed, in some patients, new presyncope may arise concomitant with successful therapy of the seizures, which will have disappeared with the institution of permanent pacing. Pacing is highly effective in abolishing the convulsions (at the expense of replacing those convulsions with presyncope or asymptomatic hypotension). Presumably, vasodepressor responses are still present. That some of these children have been diagnosed only after aborted sudden cardiac death10 is a further argument for pacing, although it is not specifically known whether pacing confers increased longevity in the majority of patients with these rare syndromes.
The objective of developing a posture sensor applicable to permanent rate-adaptive pacing is not designed specifically for the benefit of the rare syncopal patient with cardioinhibitory convulsive syncope, although such an individual may benefit significantly. Rather, it is intended for the benefit of all patients with symptoms in the upright posture that are related to asystole, bradycardia, or relative bradycardia. (Because increase in heart rate on assumption of the upright posture in the normal patient is mediated by parasympathetic withdrawal via the arterial baroreceptor reflex, the development of a posture sensor in rate-adaptive pacing is the equivalent of restoration of an autonomic reflex in some patients and is the equivalent of the modulation of this reflex for others by manipulation of open-loop gain.43 )
Sensors: Fast Response
Fast-response sensors detect the onset of exertion44 and may permit graded response correlating with degree of exertion. These include vibration (“activity”) sensors, stimulus-T interval (QT or ST) sensors, O2 saturation sensors, and less-familiar sensors of paced depolarization integral, dp/dt, and others. None of these sensors will change pacing parameters based on posture alone.
A significant development has been the introduction of accelerometers as fast-response sensors, an area of much current theoretical26 and clinical27 45 46 47 48 49 50 51 interest. Although prototypes based on various technologies exist for single-axis and for multiple-axis acceleration detection, the most widely used are based on piezoelectric accelerometers (eg, Medtronic, Inc and Intermedics, Inc). These differ from commonly used vibration sensors where simple algorithms are used to process a piezoelectric crystal signal reflecting vibration and not acceleration. Significant departures from piezoelectric technology in accelerometer design that have not yet been incorporated in pacemakers include advanced piezoresistive and variable capacitance accelerometer technologies52 (although Biotronik, Inc has incorporated one piezoresistive sensor in a commercial unit). Sensors based on these technologies may be smaller (750 μm) and have superior low-frequency response. Suitable filtering of the digitized signal of such an accelerometer placed in proximity to the heart could permit multiple types of information to be derived from the same accelerometer signal, eg, acceleration in three dimensions, vibration, and STI. Accelerometer-derived STI could be used alone as a posture sensor or in combination with another type of information, an arrangement that might be particularly efficient in maintaining rate-adaptive responses in varying physiological states and in conditions of noise. Other sensor types that may be used as combination sensors to derive STIs include impedance and intracavitary pressure.
Pacemaker sensors based on STIs may be the first effective sensors of upright posture.
Sensors: Slow Response and Combination
A small number of slow-response sensors exist correlating with degree or duration of exertion: minute ventilation (or respiratory rate) types on the one hand, and temperature on the other. Each fails to detect the onset of exertion, but later in exercise, the correlation with O2 consumption is excellent. (Minute ventilation sensors have a component of their signal contaminated by ipsilateral arm movement when the pulse generator is in the prepectoral position,53 54 and these pacemakers may behave as if they had a combination fast-response and slow-response sensor during exercise involving arm motion.) The temperature sensor suffers the additional drawback that a monotonic relation between temperature excursion (increase) and degree of exertion exists only for high levels of exertion; at low levels of exertion, the chief temperature change registered is a depression of temperature caused by return of cool blood from the extremities.55 Therefore, slow-response sensors have been coupled with fast-response sensors in pacemakers currently under investigation. The minute ventilation sensor has been coupled with an “activity” (vibration) sensor56 (although the latter is used primarily to detect onset of exercise, and any fast-response sensor might be substituted [E. Alt, personal communication]), and the temperature sensor has been coupled with an O2 saturation fast-response sensor requiring a cumbersome lead. Hardware space requirements are prohibitive for rate-adaptive pacing in children.
These comments are relevant to the discussion of accelerometer technology because in piezoresistive sensor technology, the bridge resistance varies with temperature. Therefore, the bridge current may be monitored to sense temperature simultaneous with normal operation of the device.57 58 In a constant voltage mode (Fig 7⇓), the voltage across a resistor placed in series with the bridge reflects bridge current and therefore is an effective temperature sensor. This would permit a combination sensor, where the fast-response sensor is an accelerometer-based technology (acceleration, vibration, or STI as proposed in the present article as a sensor of posture) and the slow-response sensor is a piezoresistive accelerometer bridge current-based NTC-class thermistor device operating using algorithms previously validated in other temperature-sensing rate-adaptive pacing systems. Such an approach has never been used in any pacemaker or artificial organ implant and therefore represents a novel approach.
In conclusion, machine perception of posture is achievable at the present time with existing hardware, either with accelerometer-derived STI or with STI determined using other technologies such as impedance or intracavitary pressure sensing. This may permit significant benefits for rate-adaptive pacing, particularly for rate-adaptive pacing of the young.
Selected Abbreviations and Acronyms
|bpm||=||beats per minute|
|HR||=||heart rate (in equation)|
|LVET||=||left ventricular ejection time|
|STI||=||systolic time intervals (including PEP and LVET)|
We gratefully acknowledge the dedication and expert secretarial and technical assistance of Eleanor LiCalsi, Edith Makler, and the staff of the Exercise Laboratory of the University of Arizona.
- Received April 4, 1995.
- Revision received May 1, 1995.
- Accepted May 3, 1995.
- Copyright © 1995 by American Heart Association
Alpert MA, Curtis JJ, Sanfelippo JF, Flaker GC, Walls JT, Mukerji V, Villareal D, Katti SK, Madigan NP, Krol RB. Comparative survival after permanent ventricular and dual chamber pacing for patients with chronic high degree atrioventricular block with and without preexistent congestive heart failure. J Am Coll Cardiol. 1986;7:925-932.
Alpert MA, Curtis JJ, Sanfelippo JF, Flaker GC, Walls JT, Mukerji V, Villareal D, Katti SK, Madigan NP, Morgan RJ. Comparative survival after permanent ventricular and dual-chamber pacing for patients with chronic symptomatic sinus node dysfunction with and without heart failure. Am Heart J. 1987;113:958-965.
Arnold LW, Gamble W, O’Connor B, Saul JP, Walsh E. Pacemaker treatment for toddlers and children with severe neurocardiogenic syncope and cardiac asystole. PACE Pacing Clin Electrophysiol. 1992;15:507. Abstract.
Benditt DG, Remole S, Asso A, Hansen R, Lurie K. Cardiac pacing for carotid sinus syndrome and vasovagal syncope. In: Barold SS, Mugica J, eds. New Perspectives in Cardiac Pacing, 3rd ed. Mt Kisco, NY: Futura; 1993:15-28.
Blanc JJ, Boschat J, Penther P. Hypersensibilité sino-carotidienne: évolution à moyen terme en fonction du traitement et des symptômes. Arch Mal Coeur. 1984;77:330-335.
Peretz DI, Abdulla A. Management of cardioinhibitory hypersensitive carotid sinus syncope with permanent cardiac pacing: a seventeen year prospective study. Can J Cardiol. 1985;2:86-91.
Deschamps D, Richard A, Citron B, Chaperon A, Binon JP, Ponsonaille J. Hypersensibilité sino-carotidienne: évolution à moyen terme et à longue terme des patients traités par stimulation ventriculaire. Arch Mal Coeur. 1990;83:63-67.
Grubb BP, Gerard G, Roush K, Temesy-Armos P, Elliott L, Hahn H, Spann C. Differentiation of convulsive syncope and epilepsy with head-up tilt testing. Ann Intern Med. 1991;115:871-876.
Samoil D, Grubb BP, Kip K, Kosinski DJ. Head-upright tilt testing in children with unexplained syncope. Pediatrics. 1993;92:426-430.
Ovadia M, Thoele D. Esmolol tilt testing with esmolol withdrawal for the evaluation of syncope in the young. Circulation. 1994;89:228-235.
Prior M, Masterson M, Blackburn G, Maloney J, Wilkoff B. Critical identification of patients with sinus node dysfunction for potential sensor-driven pacing. PACE Pacing Clin Electrophysiol. 1988;11:512. Abstract.
Daubert JC, Mabo PH, Druelles PH, Ritter PH, Paillard F. Dysfonction sinusale et anomalies de conduction auriculoventriculaire. Stimucoeur. 1988;16:206-210.
Heinz G, Öhner T, Lackovics A. Sinus node dysfunction after orthotopic heart transplantation: comparison of electrophysiologic and clinical parameters in predicting long-term outcome. PACE Pacing Clin Electrophysiol. 1989;12:673. Abstract.
Lau C-P. Activity-sensing rate-adaptive pacing. In: Alt E, Barold SS, Stangl K, eds. Rate Adaptive Cardiac Pacing. Berlin: Springer-Verlag; 1993:83-98.
Stafford RW, Harris WS, Weissler AM. Left ventricular systolic time intervals as indices of postural circulatory stress in man. Circulation. 1970;41:485-492.
Chirife R. Impedance-derived cardiac signals: preejection interval, stroke volume, peak ejection rate for rate adaptive pacing and antitachyarrhythmia devices. In: Alt E, Barold SS, Stangl K, eds. Rate Adaptive Cardiac Pacing. Berlin: Springer-Verlag; 1993:152-162.
Salerno DM, Zanetti J. Seismocardiography: a new technique for recording cardiac vibrations. Concept, method and initial observations. J Cardiovasc Technol. 1990;9:111-118.
Crow R, Salerno DM, Hannan P, Zanetti J. Seismocardiography for measurement of left ventricular performance at rest and immediately post exercise. J Am Coll Cardiol. 1991;17:352A. Abstract.
Strathmore N, Mond H, Cheong S, Hunt D. Rate responsive pacing: comparison of a minute ventilation sensor with an external activity-sensing pacemaker. Aust NZ J Med. 1989;19:576. Abstract.
Mond HG. Do we really need multiple sensors for optimal rate-adaptive pacing? In: Barold SS, Mugica J, eds. New Perspectives in Cardiac Pacing, 3rd ed. Mt Kisco, NY: Futura; 1993;310, 313.
Ovadia M, Thoele D. Esmolol withdrawal during esmolol tilt testing is a superior alternative to isoproterenol for noninvasive provocative testing of syncope in the young. J Am Coll Cardiol. 1993;21:295A.
Lewis RP, Rittgers SE, Forester WF, Boudoulas H. A critical review of the systolic time intervals. Circulation. 1977;56:146-158.
Sgarbossa EB, Pinski SL, Ching E, Maloney JD. Adverse effects and limitations of rate-adaptive pacing. In: Barold SS, Mugica J, eds. New Perspectives in Cardiac Pacing, 3rd ed. Mt Kisco, NY: Futura; 1993:383-423.
Petersen MEV, Chamberlain-Webber R, Fitzpatrick AP, Ingram A, Williams T, Sutton R. Permanent pacing for cardioinhibitory malignant vasovagal syndrome. Br Heart J. 1994;71:274-281.
Sagawa K. Baroreflex control of systemic arterial pressure and vascular bed. In: Shepherd JT, Abboud F, eds. Handbook of Physiology, Section 2: The Cardiovascular System, Volume III, Peripheral Circulation and Organ Blood Flow. Bethesda, Md: American Physiological Society; 1983:453-496.
Alt E, Matula M, Schrepf R, Schömig A. Accelerometer controlled rate adaptation: effect on performance and symptoms. PACE Pacing Clin Electrophysiol. 1993;16:1194. Abstract.
Delay M, Verboven Y, Mounier C, Jonot F. Analysis and optimisation of an accelerometer-based rate response for the pacemaker patient population. PACE Pacing Clin Electrophysiol. 1993;16:1197. Abstract.
Soussou AI, Helmy MG, Guindy RR, Greco EM. Exercise sensitivity and RR settings optimization of an acceleration driven pacer. PACE Pacing Clin Electrophysiol. 1993;16:1198. Abstract.
Petrucca A, Occhetta E, Rognoni G, Pistis G, Fazzari M, Baduini G, Rossi P. Clinical evaluation of a rate-responsive pacemaker using a gravimetric acceleration sensor. PACE Pacing Clin Electrophysiol. 1993;16:1198. Abstract.
Binner L, Kochs M, Wieshammer S, Lippert R. Accelerometer based single and dual chamber rate adaptive cardiac pacing. PACE Pacing Clin Electrophysiol. 1993;16:1198. Abstract.
Link B. Field-quantified silicon accelerometers: from 1 milli g to 200,000 g. Sensors. 1993;3:28-33.
Alt E, Stangl K, Theres H. Central venous blood temperature. In: Alt E, Barold SS, Stangl K, eds. Rate Adaptive Cardiac Pacing. Berlin: Springer-Verlag; 1993:128-136.
Alt E, Fotuhi P, Combs W, Wilhaus R, Gastmann U. First experience with a rate adaptive pacemaker controlled by body activity and ventilation. Circulation. 1993;88(suppl I):I-1-I-20. Abstract.
Bryzek J. The evolution of smart sensor and transducer design. Sensors. 1993;10:12-22.
Fraden J. AIP Handbook of Modern Sensors: Physics, Designs and Applications. New York, NY: American Institute of Physics; 1993:234-238.