Acute to Chronic Changes and Fixation

The amplitude and slew rate of intracardiac EGMs typically decline during the first several days to weeks after lead implantation and then increase to chronic values that are slightly lower than those measured at implantation.⁶ The initial decrease in EGM amplitude is caused by the inflammatory response and edema at the electrode-tissue interface. This gradually resolves and is followed by development of a small, inexcitable fibrotic zone surrounding the electrode tip (Fig. 3-10). This inflammation and fibrotic tissue effectively increases the distance between the surface of the electrode and excitable myocardium that generates the EGM signal. Although chronic EGM amplitudes usually are reduced by less than 10% compared with acute amplitudes, chronic slew rates are reduced by 30% to 40%.⁷

Figure 3-10 Gross microscopic slide specimen shows myocardium remaining after removal of endocardial active-fixation screw lead.

The large dotted line shows where the lead body was located, and the colored staining shows a thin, fibrotic sheath around the lead body. The approximate location of the helical screw-tip electrode is shown by the solid coiled line. The oval shape (dotted line) shows the size of the fibrotic capsule that formed around the helical extended-tip electrode. Most of the tissue outside the dotted lines stained red, indicating that it was active myocardium capable of conducting depolarizations. The tip region of this electrode is similar to that of the tip electrode in Figure 3-1, so propagation of depolarization wavefronts must travel around the tip electrode, in tissue largely out of the field of view on the right side of this figure.

The acute reduction in EGM amplitude is often greater with active-fixation leads than with passive fixation leads. Atrial undersensing can occur during the acute phase despite adequate EGM amplitudes at implantation. To account for these time-related changes in EGM amplitude, the filtered EGM recorded at lead implantation should be at least twice the sensitivity threshold that will be programmed in the pulse generator. Greater sensing safety margins are preferred for active-fixation leads.

The method of lead tip stabilization, active screw-in or passive tines, has had no significant effect on sensing characteristics in most studies.^8,9 Steroid-eluting electrodes reduce chronic pacing thresholds substantially, but also have no significant effects on sensing.^10–13

Metabolic, Ischemic, Aging, and Drug Effects

The effects of metabolic abnormalities and drugs on pacing thresholds are well described. Much less information is available concerning their effects on EGMs and sensing. Factors that reduce EGM amplitude, slow conduction velocity, or diminish slew rate may produce either oversensing or undersensing. By prolonging the intracardiac EGM duration beyond blanking periods, ischemia or antiarrhythmic drugs can produce double-counting of the QRS complex.¹⁴ Similarly, drugs that prolong the PR or QT interval beyond the refractory period may result in oversensing.^15,16

Undersensing may result from reduction in EGM amplitude or slew rate after myocardial infarction at the electrode-tissue interface, from drug and electrolyte effects,^15,16 or from progression of conduction system disease. Acute ischemia causes ST-segment changes that can be detected on VEGMs. Monitoring of EGM ST-segment shifts has been proposed as a method for monitoring ischemia for pacemakers and ICDs.¹⁷ The likelihood of recording abnormal AEGMs (defined as ≥100 msec in duration or having ≥8 fragmented deflections) correlates with age of the patient (r = 0.34; P < .0005).¹⁸

Exercise, Respiratory, and Postural Effects

The effect of exercise on the AEGM amplitude and slew rate is variable. Some studies have reported statistically significant decreases in amplitude that average 10% to 20% but may reach 40% in some patients.^19,20 Other studies did not find significant changes between rest and exercise.^21,22 Decreases in AEGM amplitude were not caused by atrial rate alone or by beta blockade.²³ VDD/R lead studies with “floating” atrial electrodes showed particularly large decreases with exercise.^24,25 Decreases in AEGM amplitude with lead maturation support the programming of a large safety margin for sensing at implantation to offset effects of lead maturation.

P-wave amplitude increases significantly during full inspiration, during full expiration, and with erect posture.²² Respiratory variation averaged 9.7% for unipolar AEGMs and 11.5% for bipolar AEGMs.^25,26 The effect of respiration on VEGMs was less, especially with the unipolar configuration.²⁶

Ventricular Electrograms during Premature Ventricular Complexes, Ventricular Tachycardia, and Ventricular Fibrillation

Premature ventricular complexes (PVCs) may have lower-amplitude R waves than sinus-rhythm R waves, as shown in Figure 3-11, but the reverse may also be true. For monomorphic ventricular tachycardia (VT), mean amplitude decreased only slightly from values in sinus rhythm—14% for epicardial EGMs and 5% for endocardial EGMs.²⁷ In contrast, EGM amplitudes during ventricular fibrillation (VF) decreased by 25% for epicardial and 41% for endocardial EGMs. More importantly, EGMs in VF often have low, highly variable, and rapidly changing amplitudes and slew rates. Figure 3-12 shows endocardial spontaneous VF EGMs from different patients, illustrating variability in intrinsic deflections, amplitudes, slew rates, and morphologies. In a study of induced VF reproducibility, 50% of the variability was caused by interpatient differences and the other 50% occurred among repeated episodes in the same patient.²⁸ In another study, the VEGM amplitude in VF was 1 mV or less in at least one VF episode in 29% of patients.²⁷ If VF lasts for minutes, the amplitude and slew rate of the EGMs decrease.

Figure 3-11 Surface electrocardiogram (ECG) lead II, bipolar right ventricular electrogram, and event markers with downward pulses that show when sensing occurred.

The QRS amplitude on the ECG is about 1 mV, which is typical. The peak-to-peak amplitude of the sinus R waves is about 10 to 12 mV, which is also typical. The slew rate is the maximum slope (dV/dt) of the EGM intrinsic deflection; it is difficult to measure with a paper speed this slow. The two premature ventricular complexes (PVCs) (fourth and sixth) have different amplitudes and shapes on both the ECG and the EGM. The sinus beat in the center, between the two PVCs, has its main intrinsic deflection during the last part of the ECG QRS complex. The left edge of the sense marker indicates the instant that sensing by the ICD occurred. Therefore, EGM morphology and timing of the sense marker may not correspond to the start of the QRS on the ECG as electrocardiographers expect. Each R wave was sensed only once because sensing is blanked by the ICD for 120 msec after each ventricular sense (VS).

Figure 3-12 Variability in electrograms during spontaneous ventricular fibrillation (VF).

Bipolar EGMs recorded during VF detection and charging by Medtronic Gem ICDs are shown for nine patients. The 1-mV calibration markers are the same except for the bottom tracing, where the calibration pulse is about three times larger, meaning that the ventricular EGM for this patient was three times smaller than for the other eight patients. Also note that rapid intrinsic deflections are visible on all tracings, although some substantial beat-to-beat variations in amplitude occurred. The abbreviated sense markers are shown at the bottom edge of each stored EGM strip. The sense amplifier blanking period was 120 msec, and the programmed sensitivity in each case was 0.3 mV, except in the sixth tracing from the top, which had 0.6 mV sensitivity. Note that there was slight VF undersensing on the bottom tracing when large EGMs were followed by smaller EGMs.

Atrial Electrograms during Rhythms Other Than Sinus

Atrial activation from ectopic sites or atrial arrhythmias can alter the amplitude, frequency content, slew rate, and morphology of the AEGM. Retrograde atrial activation during ventricular pacing reduces AEGM amplitude and slew rate by up to 50%.²⁹ These EGM changes are more pronounced in the high right atrium than in the right atrial appendage or low right atrium.³⁰ The frequency content of the AEGM is not significantly altered by retrograde atrial activation.³¹ Analysis of EGM turning-point morphology or the first-differential coefficient of slew rate has been used to discriminate sinus EGMs from those recorded during retrograde and ectopic atrial activation in small groups of patients.³²

Atrial EGMs during atrial fibrillation (AF) are characterized by extreme temporal and spatial variability. EGMs tend to be most organized in the trabeculated right atrial appendage and more disorganized in the smooth right atrium or coronary sinus.^33–35 Thus electrode spacing and positioning of atrial leads influence EGM characteristics during AF^36–38 and may cause inconsistent diagnosis of AF based on rate criteria.

The amplitude of chronic, unipolar pacemaker EGMs was 40% less in AF than in sinus rhythm.³⁹ A comparison of acute AEGM amplitudes recorded with temporary pacing catheters showed that the mean sinus-rhythm EGM amplitude decreased only slightly in atrial flutter but decreased by about 50% in AF.³⁵ Antiarrhythmic drugs may also interfere with sensing during AF by reducing atrial rate, median frequency, and EGM amplitude.⁴⁰

Automatic Adjustment of Sensitivity

Adjustment of Sensitivity in Normal Rhythm

All ICDs automatically adjust sensitivity in relation to the amplitude of each sensed R wave (Fig. 3-19). At the end of the blanking period after each sensed ventricular event, the sensing threshold is set to a high value. It then decreases with time until a minimum value is reached. Compared with a fixed sensing threshold, automatic adjustment of sensitivity increases the likelihood of sensing low-amplitude and varying EGMs, while minimizing the likelihood of T-wave oversensing.

Figure 3-19 Automatic adjustment of sensitivity.

ICDs adjust sensitivity in relation to the amplitude of each sensed R wave. The goal of this feature is to permit sensing of low-amplitude and varying-amplitude EGMs while minimizing T-wave oversensing. The figure shows two sinus beats followed by the onset of ventricular fibrillation (VF). The upper panel shows the unfiltered ventricular EGM. The lower panel shows the corresponding filtered and rectified EGM. After each sensed ventricular event, the sensing threshold is set to a predetermined fraction of the R-wave amplitude. For large R waves, the initial sensing threshold may have a maximum value. The threshold then decreases with time until it reaches a minimum value equal to the programmed sensitivity, which is nominally about 0.3 mV. For sinus beats, the threshold is larger than the T waves, preventing oversensing. When VF begins, the smaller R waves keep the threshold at a lower value, which allows sensing of R waves that are even smaller than the T waves in sinus rhythm.

The methods of the different manufacturers for automatic adjustment of sensitivity perform similarly after small R waves but differently after large R waves. Figure 3-20 shows that after large R waves, the Boston Scientific ICDs increase the sensing floor. In the Cognis/Teligen family, the sensing floor is set to one-eighth the amplitude of the measured R wave if that value is greater than the programmed sensitivity. This prevents T-wave oversensing in the setting of large R waves and reduces oversensing of low-amplitude noncardiac signals (e.g., diaphragmatic myopotentials, EMI). However, it may increase the risk of undersensing during rare episodes of VF with highly variable EGM amplitude.⁴⁸

Figure 3-20 Comparison of automatically adjusted sensitivity after sensed ventricular events for three manufacturers of ICDs after large (10 mV) R wave.

The programmed sensing threshold is approximately 0.3 mV. After sensed ventricular events, Medtronic ICDs reset the sensing threshold to 8 to 10 times the programmed sensitivity, up to a maximum of 75% of the sensed R wave. The value of Auto-Adjusting Sensitivity then decays exponentially from the end of the (sense) blanking period, with a time constant of 450 msec, until it reaches the programmed (maximum) sensitivity. At the nominal sensitivity of 0.3 mV, there is little difference between the sensitivity curves of Medtronic ICDs after large and small spontaneous R waves. If the R wave is large, the entire Auto-Adjusting Sensitivity curve can be altered substantially by changing the programmed value of maximum sensitivity (see Fig. 3-19). At nominal settings, the St. Jude Threshold Start begins at 62.5% of the measured R wave for values between 3 and 6 mV. If the R-wave amplitude is greater than 6 mV or less than 3 mV, the Threshold Start is set to 62.5% of these values (3.75 mV and 1.875 mV, respectively). The sensing threshold remains constant for a Decay Delay period of 60 msec and then decays linearly with a slope of 3 mV/sec. Both the Threshold Start percent and the Decay Delay are programmable, over the range of 50% to 75% and 0 to 220 msec, respectively (see Fig. 3-21). Boston Scientific Cognis-Telegin ICDs set the starting threshold to 75% of sensed R waves with a maximum limit of 3/2 · Peak Running Average. Sensitivity then decays using digital steps, each seven-eighths the amplitude of the previous step. For sensed events, the duration of the first step is 65 msec, and the duration of subsequent steps is 35 msec. This results in a sensitivity of one-half the peak R wave in about 170 msec. (See text for further details.) After a paced ventricular event, all ICDs also adjust sensitivity dynamically, starting at the end of the (pace) blanking period, but the threshold starts at a more sensitive setting.

(Modified from Swerdlow C, Friedman P: Advanced ICD troubleshooting. Part I. Pacing Clin Electrophysiol 28:1322-1346, 2005.)

Postpacing Automatic Adjustment of Sensitivity

After ventricular pacing, all ICDs set ventricular sensitivity to a highly sensitive value to prevent pacing during VF. The sensitivity threshold then decays to the programmed sensitivity level (Fig. 3-21). Thus ICDs are especially vulnerable to oversensing of low-amplitude signals late in diastole during pacing, when the amplifier sensitivity or gain is maximal. Clinically, the most important manifestation is the oversensing of diaphragmatic myopotentials.³

Figure 3-21 Comparison of automatic adjustment of sensitivity after paced and sensed ventricular events.

The filtered and rectified ventricular EGM, the corresponding sensing threshold, and ventricular (V) event markers are shown. Horizontal bars denote postpacing blanking periods. The blanking periods after paced events are longer than those after sensed events (250-350 vs. 120-140 msec), and the initial values of sensitivity are less. In this Medtronic ICD example, the initial sensing threshold after sensed events is 8 times the minimum programmed sensitivity of 0.3 mV, whereas the initial threshold after paced events is 4.5 times that value. The goal of a lower initial postpacing threshold is to prevent pacing into ventricular fibrillation. It compensates in part for the longer postpacing blanking period.

Ventricular Blanking Periods

Ventricular blanking periods prevent ventricular oversensing of same-chamber signals (R-wave double-counting) and cross-chamber signals (atrial pacing pulses and P waves) in regular rhythms. Because reliable sensing in VF requires minimizing blanking periods, blanking periods in ICDs are short and may occasionally be insufficient to prevent oversensing.

Short Same-Chamber Blanking Periods and R-wave Double-Counting

In adults who are not taking antiarrhythmic drugs, inter-EGM intervals for filtered EGMs in VF vary from about 130 to 300 msec, with a peak near 200 msec (Fig. 3-22). Therefore, fixed or nominal blanking ICD periods after ventricular sensed events range from 120 to 135 msec. R-wave double-counting occurs if the duration of the sensing EGM exceeds the ventricular blanking period.

Figure 3-22 Sensed cycle lengths of human ventricular fibrillation (VF) EGMs by ICD sensing system compared with manual cycle-length measurements of same signals from many patients.

Two histograms are superimposed. The red and orange vertical bars show the manually measured cycle lengths during VF (786 intervals); the blue bars show the intervals sensed by the ICD (772 intervals). Intervals of less than 120 msec, the blanking period, were not permitted. Note that there was some oversensing for intervals shorter than 180 msec, a slight amount of undersensing for intervals between 180 and 280 msec, and a small number of long intervals greater than 280 msec, which represent undersensing during VF. The peak in the histograms occurs at about 220 msec.

“Sensing” Other Ventricular Electrograms

Implantable cardioverter-defibrillator may use information derived from other VEGMs (shock-channel EGM, LV EGM in resynchronization systems) to enhance their functionality. The EGM from the shock channel (far-field EGM) is not used by transvenous ICDs for rate counting because bipolar or integrated bipolar EGMs are less susceptible to extracardiac signals. However, some manufacturers (Boston Scientific, Medtronic) use morphologic characteristics of far-field EGMs for discrimination of supraventricular tachycardia (SVT) from VT.

Automatic analysis of the far-field EGM has also been proposed as a method to identify oversensing in ICDs resulting from lead fracture or sensing-lead connection problem. In one approach, the peak-peak far-field EGM amplitude is measured in a small window centered around each sensed event (on near-field channel) to discriminate rapid oversensing from true VT/VF. Oversensing is identified when sensed events on the near-field channel correspond to isoelectric periods on the far-field channel.^55,56 In true VF, isoelectric periods are rare on the far-field channel (Fig. 3-23).

Figure 3-23 Algorithm to discriminate pace-sense conductor failure of true bipolar ICD leads from ventricular fibrillation using far-field EGM (Medtronic Protecta ICDs, currently investigational).

In pace-sense conductor failures, the far-field EGM has long isoelectric segments, which are not present in true VF. Stored EGM tracings during lead noise oversensing (left) and spontaneous VF (right) are shown; top tracings are the near-field (RVtip-RVring) EGM, middle tracings are the far-field EGM (RV coil-can), and the sensing markers are the lower tracing. Oversensing is identified by analyzing the peak-peak amplitudes of the far-field EGM in a 200-msec window centered at each of 12 sense markers (sensing derived from near-field EGM), as shown by the red arrows and boxes. Oversensing is identified when there are at least two far-field EGM analysis windows with isoelectric (<1 mV) amplitudes and at least one analysis window with peak-peak amplitude at least six times greater than the average of the two smallest amplitudes. Lead noise oversensing would have been properly discriminated from VF in these two examples.

Evaluating Sensing of Ventricular Fibrillation at Implantation

Increasing interest in implanting ICDs without assessing defibrillation efficacy has focused attention on the extent to which adequacy of VF sensing can be determined from EGMs recorded in baseline rhythm. Although the statistical correlation between R-wave amplitude in VF and baseline rhythm is weak,^57,58 two studies reported that sensing of VF is adequate with nominal sensitivities near 0.3 mV if the baseline R wave is sufficiently large (≥5 mV or ≥7 mV).⁵⁹ Rarely, clinically significant undersensing of VF or polymorphic VT may occur despite adequate sinus-rhythm R waves.^48,60 In these cases, undersensing occurs because auto-adjusting sensitivity criteria respond inadequately to variations in R-wave amplitude, rather than consistently low-amplitude R waves. The reproducibility of this phenomenon is unknown, as is its predicted extent at implantation.

Therefore, it is uncertain whether clinically appropriate testing at implantation can detect this infrequent cause of undersensing. During ICD implant with true bipolar sensing and current digital sensing amplifiers, clinically significant undersensing of VF is rare and unrelated to sinus-rhythm R-wave amplitude.⁶¹ Undersensing of spontaneous VT/VF in the VF zone is similarly rare. Reliable sensing of VF cannot be predicted from baseline EGMs if the baseline ventricular rhythm is paced. Sensitivity is programmed to a less sensitive value than nominal (e.g., to avoid T-wave oversensing), or patients have other implanted electronic ICDs, such as pacemakers, cardiac contractility modulation devices, or transcutaneous electrical nerve stimulation (TENS) units, that could cause device-device interactions.

Postshock Sensing

Postshock sensing is critical for redetection of VF after unsuccessful shocks and for accurate detection of episode termination. Electroporation, the process by which strong electric fields create microscopic holes in the cardiac cell membranes, has been proposed as the mechanism for postshock distortion of EGMs recorded from high-voltage electrodes.⁶² Because EGMs of dedicated bipolar sensing electrodes are minimally affected by shocks,⁶³ they became standard for early epicardial ICDs. For transvenous ICDs, postshock sensing recovers more rapidly with true bipolar sensing configurations than with integrated bipolar sensing.^64,65 This is a minor issue for current integrated bipolar leads with a pacing tip electrode–to–distal coil spacing of approximately 12 mm.⁶⁶

Postventricular Atrial Blanking and Rejection of Far-Field R Waves

To prevent oversensing of FFRWs, older dual-chamber ICDs had fixed PVAB periods, similar to those in pacemakers (Fig. 3-24). With a fixed blanking period, the blanked proportion of the cardiac cycle increases with the ventricular rate. Atrial undersensing caused by PVAB causes underestimation of the atrial rate during rapidly conducted atrial flutter or AF, resulting in inappropriate detection of VT⁶⁷ (Fig. 3-24, lower panel). Without PVAB, however, atrial oversensing of FFRWs could cause overestimation of the atrial rate during tachycardias with a 1 : 1 AV relationship.⁶⁸ This may result in either inappropriate rejection of VT as SVT, if FFRWs are counted consistently as atrial EGMs, or inappropriate detection of SVT as VT, if FFRWs are counted inconsistently.⁶⁹

Figure 3-24 Effect of postventricular atrial blanking (PVAB) on atrial sensing.

Upper panel shows the surface electrocardiogram (ECG), atrial electrogram (AEGM), and event markers during atrial sensed (As)–ventricular paced (Vp) rhythm. The first segment of each upper horizontal bar denotes the period of PVAB; the second segment denotes the postventricular atrial refractory period (PVARP); FFRW, far-field R wave on the atrial channel. The lower horizontal bars denote the postpacing ventricular blanking periods. With a short PVAB (left), the FFRWs are oversensed. A longer PVABP (right) prevents oversensing of FFRWs. Lower panel shows the ECG, AEGM, and ventricular electrogram (VEGM) tracings from a patient with atrial flutter with 2 : 1 atrioventricular (AV) conduction. The thin horizontal bars on the ventricular channel denote the periods of PVAB, which resulted in atrial undersensing of alternate atrial flutter EGMs (in boxes). The resultant incorrect calculation of atrial rate resulted in inappropriate ventricular therapy for atrial flutter (not shown), because the ICD did not classify the ventricular rate as less than the atrial rate.

Medtronic ICDs also reject FFRWs algorithmically by identifying a specific pattern of atrial and ventricular events that fulfill specific criteria (Fig. 3-25). Intermittent sensing of FFRWs or frequent premature atrial events may disrupt this pattern, resulting in misclassification of a tachycardia. Therefore, it is preferable to reject FFRWs after sensed ventricular events by decreasing atrial sensitivity, if this can be done without undersensing of AF. Atrial sensitivity can be reduced to 0.45 mV with a low risk of undersensing AF. Less sensitive values should be programmed only if the likelihood of rapidly conducted AF is low. FFRW oversensing that occurs only after paced ventricular events (when auto-adjusting atrial sensitivity is maximal) does not cause inappropriate detection of SVT as VT, but it may cause inappropriate mode switching and can contribute to inappropriate detection of AF or atrial flutter.

Figure 3-25 Algorithmic rejection of far-field R waves (FFRW) by pattern analysis.

ICDs with minimum cross-chamber blanking (Medtronic) reject FFRWs by the timing pattern of atrial and ventricular intervals. Atrial events are classified as FFRWs if all the following criteria are met: (1) there are exactly two atrial events for each V-V interval; (2) timing of one P wave is consistent with a FFRW (R-P interval <160 msec); (3) there is a stable interval between the FFRW and the ventricular electrogram (VEGM); (4) there is a short-long pattern of P-P intervals (to distinguish FFRW oversensing from atrial flutter); and (5) the pattern occurs frequently (4 of 12 intervals). A, Atrial electrogram (AEGM), VEGM, and dual-chamber event markers are shown. All five criteria for FFRWs are fulfilled. Horizontal double-ended arrows below event markers denote alternation of long (L) and short (S) atrial intervals. AR, Atrial refractory event; P, P wave; TS, ventricular event in ventricular tachycardia zone. B, Interval plot (left) and stored EGM (right) from episode of sinus tachycardia with consistent FFRW oversensing that is classified correctly. On the interval plot, open squares denote A-A interval and closed circles denote V-V interval. Horizontal lines denote ventricular tachycardia (VT) and ventricular fibrillation (VF) detection intervals of 400 and 320 msec, respectively. Alternating A-A intervals whose sum equals that of the VV intervals produce a characteristic “railroad track” appearance (arrow). The algorithm rejects FFRWs despite that FFRW oversensing does not occur for one V-V interval between seconds 9 and 10. C, Interval plot (left) and stored EGM (right) from episode of sinus tachycardia with consistent FFRW oversensing that was detected inappropriately as VT. Intermittent oversensing of FFRWs occurs as sinus tachycardia accelerates gradually across the VT detection interval of 480 msec (arrow), resulting in inappropriate therapy (“Burst” antitachycardia pacing marker on interval plot, VT marker on EGM event markers).

Medtronic ICDs (starting with Entrust) and Boston Scientific ICDs (starting with Vitality) may use brief atrial blanking or a period of reduced, automatically-adjusting sensitivity (or both) to reject FFRWs without preventing detection of AF (Fig. 3-26). St. Jude ICDs and Medtronic ICDs starting with Entrust provide programmable atrial blanking after sensed ventricular events to individualize the trade-off between oversensing of FFRWs and undersensing of AEGMs in AF. St. Jude ICDs also provide programmable atrial sensing Threshold Start and Decay Delay, corresponding to the same features in the ventricular channel.

Figure 3-26 Atrial sensing features intended to prevent far-field R-wave (FFRW) oversensing while still permitting detection of atrial fibrillation.

These features, which combine limited atrial blanking with automatic adjustment of atrial sensitivity, are less strict than conventional blanking periods. A, Medtronic ICDs, starting with Entrust, have an optional “Partial+” postventricular atrial blanking period (PVAB) setting that increases the sensing threshold for a programmable duration (40-100 msec) after a ventricular event (2). Entrust operation after ventricular sensed events nominally has no absolute blanking or sensitivity changes. With Partial + PVAB enabled, the sensing threshold is raised to 75% of the prior sensed atrial electrogram (AEGM) amplitude to reduce the likelihood of sensing FFRWs. “Absolute” PVAB, another selectable option, blanks all atrial sensed events within the programmable PVAB interval. Because significant atrial undersensing can occur, this method is recommended only for addressing complications not addressed by the Partial + PVAB method. Partial + or Absolute PVAB does not affect atrial sensing operation after sensed events (1) or after atrial paced events (3). AP, Atrial paced event; AS, Atrial sensed event; mS, microseconds; VS, ventricular sensed event. B, Boston Scientific ICDs, starting with Vitality, introduce a brief (15 msec) blanking period after each sensed ventricular event, followed by a decrease in atrial sensitivity to three-eighths the mean P-wave amplitude. This is sufficient to prevent oversensing of FFRWs, provided that the amplitude of the AEGM is at least eight-thirds that of the FFRW.

T-wave Oversensing

T-wave oversensing is the most common clinically-significant cause of intracardiac oversensing. Oversensing of spontaneous T waves may cause inappropriate detection of either VT or VF, depending on the sensed R wave–to–T wave interval and the programmed VF detection interval. T-wave oversensing is typically identified by alternating EGM morphologies.⁷⁰ Device-measured intervals during T-wave oversensing may also alternate, but the degree of alternation often is small and difficult to recognize, especially during sinus tachycardia or if the QT interval is prolonged.

T-wave oversensing may be divided into three classes: postpacing, large R wave (>3 mV) in spontaneous rhythm, and small R wave (<3 mV) in spontaneous rhythm (Fig. 3-29). Postpacing T-wave oversensing can inhibit bradycardia pacing^49,72 or cause antitachycardia pacing to be delivered at the wrong cycle length.⁴⁹ It does not typically cause inappropriate detection of VT but may increment VT or VF counters, making detection of nonsustained VT more likely. It is corrected by increasing the postpacing ventricular blanking period.

Figure 3-29 Classification of T-wave oversensing.

Left panel, Postpacing oversensing results in inhibition of pacing. Inhibition of antitachycardia pacing may also occur. Middle and right panels, T-wave oversensing in spontaneous rhythm can result in inappropriate detection of ventricular tachycardia or ventricular fibrillation. Management often depends on whether the R waves are small (middle) or large (right panel).

Oversensing of spontaneous T waves often occurs in the setting of low-amplitude R waves because the sensing threshold decays from a fraction of the preceding low-amplitude R wave.⁷³ To compound the problem, patients with low-amplitude R waves may require lower minimum sensing thresholds to ensure reliable sensing of VF. T-wave oversensing in these patients is a warning that detection of VF may be unreliable. The ventricular lead should be revised or a separate pace/sense lead added if the safety margin for sensing VF is insufficient.

Specific programming features may be used to reduce T-wave oversensing, provided that detection of VF is reliable, as follows:

Figure 3-30 Specific programmable features to correct T-wave oversensing.

Upper panel, Stored EGM from St. Jude ICD showing inappropriate shock for sinus tachycardia with T-wave oversensing. The lower strip was recorded from the programmer after reprogramming of the Decay Delay from 0 to 220 msec. Lower panel, Diagram of programmable features to avoid and correct T-wave oversensing in St. Jude ICDs. Automatic Sensitivity Control begins to adjust sensing at the end of the 125-msec ventricular blanking period. Both the initial sensitivity (Threshold Start as percentage of R-wave amplitude) and the time delay before onset of linear decrease in sensing threshold (increase in sensitivity) are programmable parameters.

(From Swerdlow CD, Friedman PA: Advanced ICD troubleshooting. Part I. Pacing Clin Electrophysiol 28:1322-1346, 2005.)

Figure 3-31 True bipolar (RVtip to RVring) signal, integrated bipolar (RVtip to RVcoil) signal, and ventricular marker channel during true bipolar sensing shows intermittent T-wave oversensing (arrow) caused by low-amplitude true bipolar signal, despite raising sensing threshold from nominal 0.30 mV to 0.45mV. Programming sensing to the integrated bipolar signal, if available, would prevent T-wave oversensing.

Figure 3-32 Algorithm to reject T-wave oversensing.

Upper tracing, Vtip-Vring EGM with large T waves and T-wave oversensing. Middle tracing, Signal after standard sense amplifier filtering and rectification (purple line), automatic adjusting sensing threshold (red), and peak amplitude at each sensed event (blue; the peak amplitude for each sensed event is held until the next sensed event). Bottom tracing, First derivative of the filtered and rectified EGM (purple line) its peak amplitude at each sensed event (blue), and adaptive threshold used to separate possible R waves and T waves (red horizontal line). The algorithm recognizes oversensed T waves with lower frequency content than the preceding sensed R waves by automatically comparing signal amplitudes in the filtered/rectified sensing EGM and its first derivative (which further attenuates low-frequency signals) to identify patterns of alternating signal frequency content. Using an analysis window of six sensed events, the algorithm first automatically identifies possible R waves and T waves based on comparing peak amplitudes and patterns of the first derivative of the filtered/rectified sensing EGM. For each analysis window, the adaptive threshold is set to a fraction of the three largest first differential EGM peak amplitudes. The algorithm assumes that possible R waves are above threshold and possible T waves are below threshold. An alternating RT pattern with 3Rs and 3Ts or 4Rs and 2Ts in the six sensed events analysis window is required. RT pairs are confirmed when the R-wave/T-wave (R/T) ratio in the filtered/rectified sensing EGM is less than the R/T ratio measured in the first differential filtered/rectified EGM. This approach does not increase the risk of undersensing because the operation of the standard ICD sensing channel is not modified. It does not prevent oversensing of T waves, but rather allows the ICD to withhold therapy when T-wave oversensing pattern is recognized. Testing on more than 1000 VT/VF episodes showed no loss in sensitivity to true VT/VF. Vs, Ventricular sensed event.

R-wave Double-Counting

R-wave double-counting occurs if the duration of the sensed EGM exceeds the short ventricular blanking period in ICDs; currently, this is a rare clinical problem. Consistent R-wave double-counting may result from local ventricular delays in the baseline state in native rhythm or those resulting from use-dependent effects of sodium channel–blocking antiarrhythmic drugs or of hyperkalemia. Double-counting may also occur with loss of RV capture in cardiac resynchronization ICDs: The ICD counts both the paced ventricular event and the conducted wavefront arriving at the RV bipole as a result of LV capture, if the interventricular conduction delay exceeded the ventricular blanking period. R-wave double-counting was a common problem in early ICDs that used Y-adapted or extended bipolar sensing between RV and LV electrodes.⁷⁵ The composite ventricular EGM included deflections from the right ventricle and the left ventricle, both of which could be counted as separate R waves. Occasional R-wave double-counting may occur during PVCs and is more common with integrated bipolar than true bipolar leads.⁷⁶

R-wave double-counting results in alternation of ventricular cycle lengths that produces a characteristic “railroad track” pattern on a plot of stored ventricular intervals. Because the second component of the R wave is sensed as soon as the blanking period terminates, the double-counted RV-RV or RV-LV interval measures within 20 msec of the ventricular blanking period and is always classified in the VF zone (Fig. 3-33). Inappropriate detection of VF may occur despite a true ventricular rate below the VT detection interval.

Figure 3-33 Intermittent R-wave double-counting as a result of loss of RV capture in a cardiac resynchronization ICD patient with left bundle branch block.

Left main panel shows surface ECG, atrial and ventricular marker channels, and true bipolar right ventricular (RV) EGM. The third, fifth, and sixth R waves are double-counted, as shown in insert on right. Simultaneous biventricular pacing activates the left ventricle (LV). After the LV-RV conduction delay, the wavefront arrives at the RV sensing bipole, corresponding to a sensed interval in the ventricular fibrillation (VF) zone (VS) and incrementing the VF counter. This condition alone will not result in VF being detected even if all beats are oversensed, but transient nonsustained ventricular tachycardia (VT) in the VF zone could result in inappropriate detection of VF, since the baseline VF counter may be as high as 50% of its maximum value.

Most consistent R-wave double-counting now results from transient or reversible events such as hyperkalemia, drug effects, or lead failure in cardiac resynchronization ICDs; and most intermittent R-wave double-counting is not clinically significant. Some ICDs permit programming to increase the ventricular blanking period from the nominal value (120-125 msec) to higher values, up to 170 msec. Occasionally, reducing ventricular sensitivity can avoid R-wave double-counting; but ventricular sensitivity should not be reduced unless reliable sensing of VF is confirmed at the reduced level of sensitivity. Rarely, lead revision is required.

P-wave Oversensing

P-wave oversensing may occur if the distal coil of an integrated bipolar lead is close to the tricuspid valve and if the sensed P-R interval exceeds the cross-chamber ventricular blanking period⁴ (Fig. 3-34). It is rare in adults with ventricular sensing electrodes near the RV apex, but it may occur in children or in adults if the RV electrode dislodges or is positioned in the proximal septum or inflow portion of the right ventricle. If P-wave oversensing occurs during a 1 : 1 rhythm, the sensed “R-R” pattern is similar to that of R-wave double-counting, provided that the sensed P-R or R-P interval is less than the VF detection interval. However, oversensing of P waves as R waves can cause inappropriate detection of VF during AF or atrial flutter, independent of the ventricular rate.

Figure 3-34 P-wave oversensing.

The surface electrocardiographic (ECG) lead, integrated bipolar right ventricular electrogram (RV Tip-coil), and ventricular event markers are shown. P-wave oversensing caused by the proximity of the RV coil to the right atrium results in ICD sensing of alternating sinus (ventricular sensed, VS) and ventricular fibrillation (VF) intervals, the latter corresponding to the P-R interval. The sum of the VS and VF intervals equals the true R-R interval.

Consistent oversensing of spontaneous P waves often requires lead revision. One amelioration strategy is to force atrial pacing using DDDR or Dynamic Overdrive modes. This shortens the ventricular cycle length (preventing ventricular sensitivity from reaching its minimum value) and introduces cross-chamber ventricular blanking after each atrial event (reducing the likelihood of oversensing P waves).

Far-Field R-wave Oversensing

The FFRW oversensing on the atrial channel, the inverse of P-wave oversensing on the ventricular channel, shows a pattern of alternating atrial cycle lengths with one sense marker timed close to the ventricular EGM. If rate and duration criteria for VT are fulfilled, FFRW oversensing may confound dual-chamber SVT-VT discrimination. FFRW oversensing does not cause inappropriate detection of VT if the ventricular rate is in the sinus zone, but it may cause inappropriate mode switching.

External Electromagnetic Interference

With oversensing of external electromagnetic interference, signal amplitude is greater on the high-voltage EGM recorded from widely spaced electrodes than on the sensing EGM recorded from closely spaced electrodes. Although the interference signal may be continuous, oversensing is often intermittent because of auto-adjusting sensitivity. Clinical data may suggest a specific identifiable cause (Fig. 3-35). Some manufacturers (St. Jude, Sorin, Boston Scientific) have specific ICD algorithms to prevent inappropriate detection of VF in the presence of EMI (Fig. 3-36).

Figure 3-35 External electromagnetic interference caused by electric drill.

Atrial EGM, integrated bipolar (RV tip-coil) EGM, and high-voltage (Shock) EGM are shown, together with a dual-chamber channel showing event markers. Electromagnetic interference (EMI) is continuous. Oversensing occurs first on the atrial channel, which has a higher sensitivity setting (and gain) than the ventricular sensing channel. Although the amplitude of external EMI is typically less on true bipolar sensing electrodes than on shocking electrodes, it may be similar on integrated bipolar and shocking electrodes.

Figure 3-36 Noise rejection algorithm prevents oversensing of external electromagnetic interference caused by electroconvulsive therapy, despite noise amplitude exceeding programmed sensitivity.

Surface ECG, integrated bipolar sensing EGM, shock EGM, and ventricular marker channel are shown. The Dynamic Noise Algorithm in Boston Scientific Cognis-Telegin ICDs measures the energy in the upper half of the sensing bandwidth (20-85 Hz) and sets a noise-energy–based sensing level (“noise floor”), which is usually lower than the programmed maximum sensitivity. A higher-than-expected fraction of energy in the upper half of the bandwidth suggests the presence of noise, and the ICD responds by increasing the noise floor above the minimum programmed level of sensitivity while noise is present.

(Courtesy Dr. Andrei Catanchin.)

Lead/Connector Problems

Oversensing caused by lead insulation failure or conductor fracture, lead-lead mechanical interactions (“chatter”),⁷⁶ or connector (header, adapter, set-screw) problems is intermittent and may occur only during a small fraction of the cardiac cycle. In ICDs, it is limited to the sensing EGM unless the problem relates to the RV coil in an integrated bipolar lead or to both sensing and high-voltage conductors/connectors. The rate of oversensed signals is always in the VF zone. Intervals below 200 msec are typical. Signals almost always display substantial variability in amplitude or frequency. High-frequency components are common, and high-amplitude signals often saturate the amplifier (Figs. 3-37 to 3-39).

Figure 3-37 Oversensing caused by lead insulation failure.

Top panel shows the atrial (Atip-Aring) EGM, the ventricular true bipolar (Vtip-Vring) EGM, and event markers. Bottom panel shows the interval plot. In the ventricular channel, there are intermittent, erratic signals that saturate the sensing amplifier. The interval plot highlights the wide range of sensed intervals, including many “short” intervals close to the ventricular blanking period of 120 msec. Arrows denote five inappropriate shocks.

Figure 3-38 Ventricular oversensing leads to inhibition of pacing in a pacemaker-dependent patient 2 weeks after replacement of an ICD pulse generator.

Top left, In-hospital electrocardiogram (ECG) monitoring strip with 4.46-second pause in paced ventricular rhythm. The atrial rhythm is atrial fibrillation. Bottom left, Marked variability in right ventricular (RV) pacing impedance with constant impedance in left ventricular (LV) lead. The two right panels show telemetry Holter recordings, including one ECG channel, two ventricular EGMs, and event markers. Top right, Oversensing on tip-ring EGM (arrow, box) but not on tip-RV coil EGM. Oversensing is intermittent and saturates the amplifier. Bottom right, Oversensing on both tip-ring and ring-can EGMs, localizing the oversensing problem to connections related to the ring EGM (box). AS, Atrial sensed event; TS, ventricular interval in ventricular tachycardia (VT) zone; VP, ventricular paced event.

Figure 3-39 Intraoperative fluoroscopic images corresponding to the data presented in Figure 3-38.

The ICD has been removed from the pocket to enhance image quality. Before revision, the lead connector pin was not advanced completely into the header (arrow). The proximal connection between the ring electrode and the header was intermittent, accounting for the high impedance and oversensing illustrated in Figure 3-38. After revision, the ventricular electrode is advanced completely into the header.

In ICDs, extremely short R-R intervals near the ventricular blanking period (120-150 msec) do not represent successive cardiac depolarizations, except occasionally during VF. Medtronic ICDs count these very short intervals as a measure of nonphysiologic oversensing to provide early warning of lead fracture; they store this count as the sensing integrity counter (SIC). A high or rapidly accumulating Sensing Integrity Count (total >300 or 10/day for 3 consecutive days) is a sensitive indicator of pace-sense lead fracture,^3,21,22 but in isolation is nonspecific. The most common cause of isolated, extremely short “R-R” intervals is a combination of an oversensed physiologic event and appropriately sensed R wave.

In P-wave oversensing, the nonphysiologic short interval is the P-R interval. In R-wave double-counting, it is the interval between initial and terminal deflections of a true ventricular EGM. In T-wave oversensing, the interval is bounded by an oversensed T-wave and a PVC.⁷⁶ However, these nonphysiologic combinations of physiologic signals rarely result in repetitive oversensing, which is common in lead or connector problems. Repetitive, transient oversensing may be identified by stored “nonsustained tachycardias.” The combination of isolated, extremely short RR intervals and repetitive rapid oversensing (defined as at least two nonsustained tachycardias with duration ≥5 intervals and mean cycle length ≤220 msec) has a positive predictive value for lead or connector problems of about 80%,^76,78 even with a normal pace-sense impedance. Unusual cases of true VF⁷⁶ and other unusual conditions, such as device-device interactions or header seal-plug problems shortly after implant,⁷⁹ can produce both isolated, very short intervals and rapid, repetitive oversensing.

Role of lead impedance

Modern pacemakers and ICDs perform daily, automated measurements of impedance in lead conductors. A large, abrupt change in impedance associated with rapid oversensing is, for practical purposes, diagnostic of a lead or header-connector problem. However, pace-sense conductor fractures typically present as rapid, repetitive oversensing without abnormal impedance.⁷⁸ A fracture may generate sufficient oversensing to cause inappropriate shocks without an impedance increase or with an impedance increase in the interval between daily impedance measurements.^78,80–82 Rarely, pacing precipitates lead-related oversensing.⁸¹ Lead fractures or header-connector problems also present frequently as abrupt increases in impedance without oversensing.

The Medtronic Lead Integrity Alert (LIA) is triggered by either abnormally high impedance relative to the patient’s baseline impedance or rapid oversensing unlikely to represent a physiologic event^55,78 (Figs. 3-40 and 3-41). Once an alert is triggered, an audible tone sounds immediately and every 4 hours thereafter. LIA alerts also reprogram the number of intervals to detect VF (NID) from the value programmed at the physician’s discretion (nominally 18/24) to 30/40; EGMs are stored for any interval shorter than 200 msec. The purpose of the high NID is to reduce inappropriate shocks caused by transient, fracture-induced oversensing.⁷⁸ The stored EGMs provide a diagnostic clue to determine the cause of rapid oversensing based on EGM characteristics.⁷⁶ In newer ICDs with wireless telemetry, alerts in both groups may also initiate Internet notifications.

Figure 3-40 Stored electrogram shows intermittent oversensing caused by lead fracture.

Pace-sense (V_tip–V_ring), high-voltage (RV_coil-can), and marker channels are shown; VS, TS, and FS, intervals in sinus, ventricular tachycardia (VT), and ventricular fibrillation (VF) zones, respectively; FD, detection of VF. Upper and lower panels are continuous. Pace-sense channel shows intermittent, high-frequency, oversensing of nonphysiologic noise characteristic of lead fracture or header-connector problem. At the programmed number of intervals to detect VF (NID) 18/24, inappropriate detection of VF occurs near left of the bottom panel, resulting in an inappropriate shock (not shown). Coincident with detection of VF, oversensing decreases greatly. VF would not have been detected at higher values of NID. FS, Ventricular interval in VF zone; VS, ventricular sensed event; VP, ventricular paced event.

(From Swerdlow CD, Gunderson BD, Ousdigian KT, et al: Downloadable algorithm to reduce inappropriate shocks caused by fractures of implantable cardioverter-defibrillator leads. Circulation 118:2122-2129, 2008.)

Figure 3-41 Individual patient examples of source data used in retrospective simulation of Lead Integrity Alert (LIA, Medtronic) designed to provide warning of lead failure triggered by either high pace-sense impedance or transient rapid oversensing.

The goal of this feature is to provide patients and physicians with sufficient warning that patients will receive medical attention prior to delivery of inappropriate shocks. A, True-positive impedance trigger: abrupt variations in impedance exceeding 1000 Ω, triggering a LIA alert 12 days before an inappropriate shock. B, False-positive impedance trigger shows gradual increase in impedance. New version of algorithm requires abrupt rise in impedance. C, True-positive oversensing alert triggered 6 days before an inappropriate shock, despite constant impedance. D, LIA was not triggered prior to first inappropriate shock. However, because both impedance and oversensing triggers occurred 1 hour and 45 minutes after that shock, the automatic NID increase would have prevented the subsequent shocks 5 hours later.

(From Swerdlow CD, Gunderson BD, Ousdigian KT, et al: Downloadable algorithm to reduce inappropriate shocks caused by fractures of implantable cardioverter-defibrillator leads. Circulation 118:2122-2129, 2008.)

Myopotential Oversensing

Myopotential oversensing may persist for variable fractions of the cardiac cycle. Diaphragmatic myopotentials are most prominent on the sensing EGM. Oversensing usually occurs after long diastolic intervals or after ventricular paced events when amplifier sensitivity or gain is maximal. It often ends with a sensed R wave, which abruptly reduces sensitivity. In pacemaker-dependent patients, diaphragmatic oversensing causes inhibition of pacing, resulting in persistent oversensing and inappropriate detection of VF (Fig. 3-42). Clinically, this may manifest as syncope from inhibition of pacing followed by an inappropriate shock. This is an exception to the clinical rule that antecedent syncope usually indicates an appropriate shock. A short time constant for automatic adjustment of sensitivity increases the probability of this type of oversensing. It is most common in male patients who have integrated bipolar leads in the RV apex.^3,4 Oversensing of diaphragmatic myopotentials may be corrected by reducing ventricular sensitivity, provided that VF sensing and detection are reliable at the reduced level of sensitivity. In pacemaker-dependent patients, oversensing may also be reduced by pacing at a faster rate. Occasionally, correction requires insertion of a new rate-sensing lead away from the diaphragm.

Figure 3-42 Oversensing of diaphragmatic myopotentials.

Atrial EGM, integrated right ventricular (RV Tip-Coil) bipolar sensing EGM, high-voltage (Shock) EGM, and marker channels from a Boston Scientific Prizm II ICD in a patient with complete heart block are shown. Low-amplitude, high-frequency signals are recorded on the integrated bipolar channel, inhibiting pacing and resulting in inappropriate detection of ventricular fibrillation (VF) at right (labeled Epsd, for episode start/end). Oversensing of diaphragmatic myopotentials usually occurs after long diastolic intervals or after ventricular paced events when amplifier sensitivity or gain is maximal. Repositioning of the lead eliminated the problem. AS, Atrial sensed event; AP, atrial paced event; PVC, premature ventricular complex; PVP→, extension of postventricular atrial refractory period (PVARP); Sr, sensed refractory event; VP, ventricular paced event.

Pectoral myopotentials are more prominent on a far-field EGM that includes the ICD can rather than the near-field EGM. Because ICDs do not use this EGM for rate counting, oversensing of pectoral myopotentials does not cause inappropriate detection, but they may reduce the effectiveness of SVT-VT discrimination by EGM morphology algorithms during exercise-induced sinus tachycardia. Pectoral myopotentials are a major cause of oversensing in unipolar pacemakers and may result in either inhibition of ventricular pacing (if sensed by ventricular channel) or rapid ventricular pacing (if sensed and tracked by atrial channel). Myopotential oversensing may be suspected if symptoms occur during arm motion and can often be demonstrated in the clinic by having the patient forcefully press the hands together while the EGM is monitored.

Ventricular Safety Pacing and Noise Reversion

Ventricular safety pacing prevents inappropriate inhibition of ventricular pacing after atrial paced events. This feature augments the protection provided by cross-chamber blanking after atrial pacing stimuli. If a ventricular event is sensed after the delivery of an atrial pacing stimulus in a “nonphysiologic” period (10-100 msec) during the AV delay, a ventricular pacing stimulus is delivered at the end of the nonphysiologic interval (110-120 msec). This prevents pacemaker crosstalk from causing inhibition of ventricular pacing output, but it may also result in (noncaptured) ventricular pacing immediately after true ventricular events that occur during the AV delay. Ventricular safety pacing is especially likely to occur during atrial undersensing, during AF, or during a junctional rhythm.

Pacemakers also contain algorithms to protect against prolonged inhibition of ventricular pacing caused by oversensing. Noise-sampling windows (30-60 msec in duration) are used during the blanking periods to identify spurious signals and cause the ICD to change to an asynchronous “reversion” pacing mode, to ensure ventricular backup pacing.⁸³ Noise-reversion operation is particularly important for unipolar sensing, which is more likely than bipolar sensing to exhibit oversensing of extracardiac signals⁸⁴ (Fig. 3-43).

Figure 3-43 A, Extension of refractory period resulting from detection of noise during the noise-sampling (NS) portion of the ventricular refractory period; VP, ventricular paced event. The first portion of the ventricular refractory period is absolute (AB). When suprathreshold signals are seen during the NS portion of the ventricular refractory period, a new NS window is initiated, “retriggering” the ventricular refractory period by a duration equal to the NS window. Continuous extension of the ventricular refractory period invokes noise reversion pacing at the programmed reversion rate. A, Normal VVI pacing at 60 bpm on the left. When electromagnetic interference begins, the ventricular refractory period is continuously retriggered, and noise reversion pacing starts at 90 bpm. B, Combination of ventricular inhibition and noise reversion. The noise reversion rate of this VVI pacemaker is 90 bpm, and the base rate is programmed at 70 bpm. Beats with arrows indicate noise reversion pacing caused by continuous retriggering of the ventricular refractory period, as in A. Some beats are slower than 70 bpm because of intermittent (not continuous) oversensing of noise such that ventricular refractory periods are not continuously retriggered.

(B modified from Sweesy MW, Holland JL, Smith KW: Electromagnetic interference in cardiac rhythm management ICDs. AACN Clin Issues 15:391-403, 2004.)

Automatic Mode Switching to Avoid Ventricular Tracking of Atrial Tachyarrhythmias

Pacemakers programmed to dual-chamber synchronous pacing modes may use automatic “mode switching” algorithms to initiate a temporary change to a nontracking pacing mode during paroxysmal atrial tachyarrhythmias, to avoid inappropriately high ventricular pacing rates. Methods of mode switching differ among pacemakers and ICDs and can be classified into five different categories^85,86 (Fig. 3-44). Accurate atrial sensing is critical to appropriate mode switching, because all methods depend on measurement of atrial rate, A-V patterns, or both. The atrial sensing configuration (unipolar vs. bipolar), programmed atrial sensitivity, and atrial blanking periods influence the methods used by the various manufacturers and their performance for detection of atrial tachyarrhythmias. Algorithms to recognize repetitive blanking of atrial events during atrial flutter allow mode switching to occur more rapidly when this condition is confirmed (Fig. 3-45). Whatever method is employed, unreliable atrial sensing, whether from low-amplitude P waves or insensitive programming to avoid FFRW oversensing, will degrade performance of these algorithms.^85–88 ICDs with long PVAB periods or low atrial sensitivity may fail to switch modes if a substantial fraction of atrial events are undersensed.

Figure 3-44 Sensitivity and specificity of automatic mode-switching (MS) algorithms for atrial tachyarrhythmia detection.

Relative performances of the five different classes of algorithms are shown; performance of the mean atrial rate (MAR), rate and count (RAC), and “x out of y” algorithms are shown twice to emphasize performance variability with specific parameter selections. Algorithms that use MAR calculation favor specificity over sensitivity but are generally slower-acting than algorithms that favor sensitivity (e.g., beat-to-beat). Atrial events with short and long atrial cycle lengths equally affect atrial rate estimation in “unbiased” MAR algorithms; long atrial cycle lengths have less impact on atrial rate estimation in “biased” MAR calculations, to reduce the impact of atrial undersensing. Cons, Consecutive; non-cons, nonconsecutive.

(From Israel CW: Analysis of mode switching algorithms in dual chamber pacemakers. Pacing Clin Electrophysiol 25:380-393, 2002.)

Figure 3-45 Algorithm to prevent upper rate limit tracking of alternate atrial EGMs during atrial flutter (Blanked Flutter Search), even if EGMs recorded in postventricular atrial blanking period (PVAB), when sensing cannot occur.

Upper panel, Simulated electrocardiogram (ECG) and ladder diagram. Black horizontal bars denote blanking periods, and white bars denote refractory periods for atrium (upper bars) and ventricle (lower bars). Lower panel, Stored ventricular EGM and dual-chamber EGM markers from a clinical event. Upper panel, The intention of this algorithm is to infer that alternate atrial EGMs are “hiding” in the PVAB. To determine whether this is occurring, the algorithm transiently interrupts atrial tracking to assess the atrial rhythm when it is likely that alternate atrial intervals are occurring in the PVAB. The specific criterion is that the measured A-A interval is both less than twice the sum of the sensed AV interval and the PVAB and less than the mode switch interval for 8 beats. When this occurs, the algorithm extends the postventricular atrial refractory period (PVARP extension). The next atrial EGM that would have been tracked is now in the PVARP (AR marker, first arrow) and is not tracked. The next atrial flutter EGM is sensed because it no longer times in the PVAB (AS marker, second arrow). The pacemaker then switches mode rapidly (DDI). Lower panel, In this stored clinical EGM, ventricular intervals are numbered to correspond to those in the simulation in the upper panel. As in the upper panel, the first arrow (AR) corresponds to PVARP extension, and the second arrow (AS) corresponds to a sensed atrial flutter EGM that would have been blanked in tracking mode. Mode switching follows immediately. VP, Ventricular paced event.

Algorithms to Reduce Ventricular Pacing

Unnecessary ventricular pacing contributes to adverse clinical outcomes (e.g., heart failure hospitalizations) and increased risk of AF.^89,90 Algorithms that lengthen AV intervals adaptively to promote intrinsic conduction reduce ventricular pacing.^91–94 Medtronic and Sorin developed new pacing modes (MVP and SafeR) that operate in AAI(R) mode, with mode switching to DDD(R) to prevent excessive ventricular pauses or asystole.^19,95,96 Single beats of AV conduction failure are allowed by both algorithms, with switching to DDD mode occurring only if higher levels of AV conduction block occurs (Medtronic) or if the PR interval is excessive. Once switched to DDD mode, both algorithms test for AV conduction periodically, to avoid unnecessary ventricular pacing for transient AV block. However, these pacing modes may be associated with pacemaker syndrome in patients with marked first-degree AV block, resulting in unnecessary pauses if complete heart block develops.^97–99

Automatic Assessment of Pacemaker Capture (“Capture Management”)

Automatic sensing of the evoked response (local myocardial depolarization) that results from a pacing stimulus allows for verification of myocardial capture. However, sensing of the evoked response is obscured by the decaying polarization afterpotential that follows the pacing stimulus.¹⁰⁰ This polarization artifact can be attenuated by using high-capacitance, low-polarization leads or using multiphasic pacing impulses to neutralize the postpacing polarization. Figure 3-46 shows an example of the evoked response signal combined with polarization artifact after a pacing stimulus.¹⁰⁰ To minimize the afterpotential from ventricular pacing stimuli and improve detection of the evoked response, some manufacturers use a smaller-output capacitor. This approach further enhances the accuracy of evoked response detection.

Figure 3-46 Local myocardial depolarization (evoked response) is obscured by the postpacing polarization artifact.

A, Pace polarization artifact is observed as a positive deflection that generally decreases toward the initial cell potential within about 200 msec after the pacing pulse. B, Evoked response is present immediately after the pacing pulse and shows an initial negative deflection. C, The signal sensed by the pacemaker is the summation of the pace polarization artifact and the evoked response.

(From de Voogt WG, Vonk BF, Albers BA, et al: Understanding capture detection. Europace 6:561-569, 2004.)

Pacemakers use one of two different ventricular capture verification schemes: beat-to-beat approaches that monitor capture almost continuously and periodic approaches that test pacing thresholds intermittently (e.g., once every few hours). Algorithms that perform periodic capture threshold measurements typically set the pacing output at 1.5 to 2.0 times the threshold to avoid transient loss of capture until the next test. Beat-to-beat capture management algorithms adjust pacing outputs to values that are just above pacing threshold but deliver higher-output backup pulses if loss of capture occurs.¹⁰¹

Ventricular capture verification has been incorporated into some ICDs and resynchronization ICDs. Algorithms for RV capture management use essentially the same methods as pacemakers, except that polarization artifact that challenges evoked response detection in bipolar pacing leads is eliminated or reduced by use of the RV coil-to-can EGM available in ICD systems.¹⁰² LV capture management may be achieved by pacing on the LV lead and analyzing the intraventricular conduction time sensed in the right ventricle. This method has been demonstrated to be safe, reliable, and clinically equivalent to operator-performed LV threshold tests.¹⁰³

Atrial capture management is more challenging than ventricular management because the atrial evoked response is much smaller than the ventricular response. Thus, reliable detection of the evoked response is difficult, and atrial capture detection using the evoked response method can be performed only using low-polarization electrodes.¹⁰⁴ An alternative method relies on critical timing of the atrial and ventricular response to atrial pacing stimuli during normal sinus rhythm (Fig. 3-47). In ambulatory patients, this method has been reported to perform comparably with operator-performed atrial pacing threshold measurements without causing atrial proarrhythmia.¹⁰⁵

Figure 3-47 Atrial capture verification method that relies on atrial or ventricular responses to atrial pacing.

A high-level control algorithm determines which of two timing assessment methods should be used, based on the recent history of atrial pacing, atrial sensing, and atrioventricular (AV) conduction. At least two test paces are required to verify capture or loss of capture. Failed, inconclusive, or aborted threshold tests are repeated after a delay. This method cannot measure atrial capture during atrial paced–ventricular paced (VP) rhythms or rhythms with irregular atrial or ventricular cycle lengths. A, Atrial Chamber Reset method is used if recent history indicates primarily atrial sensed events (AS). A premature atrial test pace (APt) is delivered, and the resulting atrial timing is observed. An atrial refractory event (AR) within 10 bpm after the prior intrinsic atrial rate suggests loss of capture (top panel). The lack of an atrial refractory event in this period suggests resetting of the atrial timing and atrial capture (middle and bottom panels). B, The AV conduction check method is used if the recent history indicates that AV conduction is present. The average AV conduction time is measured before delivery of the APt. A backup atrial pace at the operating atrial output is also delivered 70 msec after the APt. The algorithm looks for ventricular sensed events (VS) in one of two capture verification windows (40 msec in duration) that are placed relative to the APt and the backup atrial paced events based on the average AV conduction time. A ventricular event sensed in the first capture verification window suggests capture by the APt (top panel). Lack of a sensed ventricular event in the first capture verification window and presence of a ventricular event in the second window is evidence of noncapture by the APt (bottom panel).

Monitor-only Zones

Programmable monitor-only zones store EGMs of tachycardias based on rate and duration even if they are slower than the slowest zone in which VT therapy is programmed. If therapy is not programmed “on” for slow VT, the slowest rate zone may be programmed as a “monitor only” zone, with detection “on” and therapies “off.” However, interactions between the counters in the VT zone serving as a monitor zone and the next zone may restrict use of SVT-VT discriminators or decrease the number of intervals required for detection in the therapy zone. ICDs with dedicated monitor-only zones may avoid these potential interactions. However, as noted previously, in St. Jude ICDs the combination of counting method and criteria for resetting the VT counter during sinus tachycardia in the monitor-only zone may result in inappropriate therapy.^109,110

Single-Chamber Ventricular Building Blocks

The first five building blocks in Table 3-1 apply to single-chamber ICDs or the ventricular component of dual-chamber ICDs. The R-R interval + duration analysis building block classifies ventricular intervals into zones by programmable cycle length (or rate) thresholds. All single- and dual-chamber ICDs use cycle length or rate in combination with tachyarrhythmia duration as the most basic method of detection. When used alone, R-R interval + duration analysis provides the highest sensitivity for detection of VT/VF, but cannot discriminate VT/VF from SVT. The R-R regularity building block is used to discriminate between regular ventricular intervals caused by monomorphic VT and irregular ventricular intervals caused by rapidly conducted AF. The R-R onset building block is used to discriminate sudden-onset VT from gradual-onset sinus tachycardia. The VEGM morphology building block classifies tachyarrhythmias as SVT if morphologic characteristics of the ventricular EGM are similar to those of beats of known supraventricular origin. Otherwise, it classifies the rhythm as VT.

Dual-Chamber Building Block: Atrial vs. Ventricular Rate

Direct or indirect comparison of atrial and ventricular rates is the cornerstone of most dual-chamber algorithms. In most studies, the ventricular rate exceeds the atrial rate in more than 80% of VTs in the VT zone of dual-chamber ICDs and a higher fraction of VTs in the VF zone.^68,114 Thus, algorithms that compare atrial and ventricular rates as their first step (Boston Scientific Rhythm ID, St. Jude, Biotronik) apply SVT discriminators to fewer than 20% of VTs; this reduces the risk that they will misclassify VT as SVT, provided that the atrial rate is measured correctly. The Medtronic algorithm indirectly compares atrial and ventricular rates using dual-chamber pattern analysis as the first step to achieve a comparable result. The Sorin algorithm uses a conceptually different approach, by initially detecting tachyarrhythmia based on cycle length, then further classifying the arrhythmia as VT or SVT based on the stability of R-R intervals and an analysis of AV association and tachycardia onset.

Dual-Chamber Building Blocks

The next four building blocks in Table 3-1 are based on combinations of atrial and ventricular information. Their primary role is to discriminate VT with 1 : 1 ventricular-atrial (V-A, VA) conduction from SVT. Their secondary roles are to classify rhythms with stable N:1 atrial-ventricular (A-V, AV) relationships (stable ratio of N atrial events for each ventricular event) as SVT and to classify isorhythmic tachycardias (with AV dissociation and similar ventricular rate) as VT. The P-R dissociation building block can detect the presence of VT during SVT. The P-R patterns/relationships building block applies to stable associations of atrial and ventricular events. For tachycardias with 1 : 1 AV relationships, it analyzes the relative timing of P-R and R-P intervals to classify the rhythm as SVT or VT. It also identifies consistent N:1 AV patterns that occur primarily during atrial flutter. The chamber of origin building block discriminates between VT and SVT with 1 : 1 P-R association by identifying whether the tachycardia originates in the atrium or in the ventricle. In ICD patients, SVT usually begins with an intrinsic atrial event in the interval between the last ventricular event in the sinus rate zone and the first ventricular event in the VT zone. Conversely, at the start of spontaneous VT, there is usually no atrial event in this interval. The response of the opposite chamber to atrial or ventricular pacing has long been a critical tool in the electrophysiology laboratory for diagnosing tachycardias with 1 : 1 AV relationship. A novel ICD algorithm based on this premise has been shown to terminate or discriminate 1 : 1 tachycardias by using simultaneous atrial and ventricular ATP.¹¹⁵

Single-Chamber Atrial Building Blocks

The final four detection building blocks are the atrial correlates of the single-chamber ventricular building blocks: P-P interval (atrial cycle length), P-P regularity, P-P onset, and AEGM morphology. Their primary role is to identify the presence (or absence) of atrial tachyarrhythmias and to invoke (or inhibit) the application of single chamber ventricular building blocks such as interval stability and/or EGM morphology. To date, atrial EGM morphology has not been applied in ICDs to discriminate between antegrade and retrograde conduction.

R-R Interval Regularity

Measures of R-R interval regularity usually are referred to as measures of R-R interval “stability” or “stability algorithms.” Technical details and optimal programmed values differ among manufacturers. Specific measures of R-R regularity interact with the method of counting R-R intervals and the duration required for detection. Requiring consecutive R-R intervals or a higher number of intervals to fulfill the rate and regularity criteria improves rejection of AF but increases the risk that detection of VT will be delayed.^{107,108,116–120}

Regularity criteria can reject AF with ventricular rates slower than 170 beats per minute (bpm) in the absence of antiarrhythmic drugs. At faster rates, these criteria cannot discriminate AF from VT reliably, because R-R intervals in AF are more regular;^116–120 and the criteria may prevent detection of VT in the presence of amiodarone or type IC antiarrhythmic drugs.^117,119 These drugs may cause regular VT to become irregular, or they may cause rapid polymorphic VT to slow from the VF zone, where discriminators do not apply, to the VT zone, where regularity discriminators do apply. R-R regularity may also be used to select the first VT therapy: ATP for regular VT and shock for irregular and presumed VF or polymorphic VT.

Sudden Onset

Measures of abruptness of onset of a tachycardia have high specificity for rejecting sinus tachycardia,^{116,118–120} but may prevent detection of VT that originates during SVT or VT that starts abruptly with an initial rate below the VT detection limit. In the latter case, the ICD misclassifies the “onset” of the arrhythmia as the gradual acceleration of the VT rate across the VT rate boundary. This criterion does not prevent SVT with sudden onset from being treated inappropriately.

Ventricular EGM Morphology

Analysis of ventricular EGM morphology,^{114,120–123} alone or in combination with stability, probably provides the best single-chamber SVT-VT discrimination for initial detection of VT. The morphology building block is the key element of single-chamber algorithms and the central single-chamber component of dual-chamber algorithms with this feature. For this reason, a more detailed analysis is provided.

General

Electrogram morphology algorithms are the most complex and effective building blocks in single-chamber detection algorithms. All morphology algorithms share common steps (Fig. 3-55): (1) record a template EGM of baseline rhythm; (2) construct and store a quantitative representation of this template; (3) record EGMs from an unknown tachycardia; (4) time-align the template and tachycardia EGMs; (5) construct a quantitative, normalized representation of each tachycardia EGM; (6) compare the representation of each tachycardia EGM with that of the template to determine its degree of morphologic similarity; (7) classify each tachycardia EGM as a morphology match or nonmatch with the template; and (8) classify the tachycardia rhythm as VT or SVT based on the fraction of EGMs that match the template. Steps 3 through 8 are performed in real time. Morphology algorithms differ according to EGM source or sources, methods of filtering and alignment, and details of quantitative representations. The features of specific algorithms are described in Figure 3-56.

Figure 3-55 General steps in all morphology algorithms.

See text for details.

Figure 3-56 Specific morphology algorithms.

A, St. Jude MD algorithm. The positive and negative deflections in the sensing EGM are normalized and modeled as a series of three polygons (A, B, and C in the template EGM; A’, B’, and C’ in the tachycardia EGM). The normalized areas of these polygons are then computed. Each tachycardia EGM is compared with the template EGM in three steps. First, the difference in area of corresponding polygons is computed. Second, the absolute values of these differences are summed. Third, a match score is constructed to be inversely proportional to the sum of these differences. If a programmable number (= 4) of 8 complexes in a sliding window exceed the programmable threshold (nominally 60%), the rhythm is classified as supraventricular tachycardia (SVT). If not, it is classified as ventricular tachycardia (VT). B, Boston Scientific Vector Timing and Correlation (VTC) algorithm. This algorithm aligns shocking EGMs of the tachycardia and template based on the peak of the rate-sensing EGM. This method takes advantage of spatiotemporal differences between activation sequences in VT and baseline rhythm that cannot be detected from a single EGM. In this example, the peak of the shock EGM has similar timing to the peak of the rate-sensing EGM in sinus rhythm, but much later timing in VT. The amplitude of the shocking EGM (upper panel) is computed at each of 8 points selected on the basis of their timing relative to the peak of the sensing EGM (lower panel) and extracted as an eight-element “feature set,” corresponding to an eight-dimensional vector. Feature sets of each tachycardia EGM are compared with the template EGM by computing the product-moment correlation coefficient between the two feature sets. This value, named the feature correlation coefficient (FCC), is a measure of mean spatiotemporal difference in ventricular activation between tachycardia and baseline rhythm. A threshold value for the feature correlation coefficient was selected to optimize SVT-VT discrimination on a test data set. This value is neither published nor programmable. If at least 3 of 10 complexes in a sliding window exceed the threshold, the rhythm is classified as SVT; if not, it is classified as VT. C, Medtronic Wavelet algorithm. The algorithm expresses the morphology of ventricular EGMs using the wavelet-transform. Wavelets are functions of constant shape and limited time duration.* They form the basis of a mathematical transformation that represents signals efficiently if they are both highly localized in time and preceded and followed by isoelectric intervals. Wavelets are therefore well suited for representing transient biomedical signals such as ventricular EGMs. The algorithm compares the morphology of ventricular EGMs during a tachycardia with a template recorded during baseline rhythm. This comparison is expressed as a percent-match score that describes the degree of morphologic similarity of the baseline and tachycardia EGMs using a programmable source. In general, the EGM recorded between right ventricular and superior vena cava coils is preferred as a default. This signal combines far-field sensitivity to morphology differences between VT and SVT with resistance to pectoral myopotentials. The algorithm begins processing when a tachycardia fulfills the programmed rate criterion for detection of VT and 8 beats remain to fulfill the programmed duration criterion for detection of VT. Percent-match scores are calculated for each of the last 8 beats before detection. Beats with match scores less than a programmable threshold (nominally 70%) are classified as ventricular. Nominally, a tachycardia is classified as VT if 6 or more of the 8 analyzed EGMs are classified as ventricular. Otherwise, it is classified initially as SVT. If the tachycardia is classified as SVT, the algorithm is applied to each successive 8-beat sliding window until the rate criterion for VT is no longer fulfilled.

(*Meyer Y: Wavelets: algorithms and applications. Philadelphia, 1993, Society for Industrial and Applied Mathematics.)

Limitations of Morphology Algorithms

The START study (SAFARI trial) of arrhythmias recorded at ICD implant reported individual differences for the performance of single-chamber discrimination algorithms based primarily on EGM morphology.¹²⁴ Morphology algorithms share common failure modes for inappropriate detection of SVT as VT: (1) inaccurate template, (2) EGM truncation, (3) alignment errors, (4) oversensing of pectoral myopotentials, (5) rate-related aberrancy, (6) SVT soon after shocks, and (7) inappropriate classification of VT as SVT.

Electrogram Truncation

EGM truncation (“clipping”) occurs when the recorded EGM signal amplitude exceeds the range of the EGM amplifier so that the maximum or minimum portion of the EGM is clipped. This removes EGM features for analysis and alters the timing of the tallest peak, which can affect alignment. The amplitude scale in Medtronic and St. Jude ICDs should be adjusted so that the EGM used for morphology analysis is 25% to 75% of the dynamic range (Fig. 3-57).

Figure 3-57 Electrogram truncation as source of error for morphology algorithms.

A, St. Jude MD template EGM is truncated (arrow) with an amplifier range of 9.8 mV. Truncation was corrected by increasing the range to 14.4 mV. Inconsistent truncation may prevent supraventricular tachycardia (SVT) morphology from matching the template. B, Medtronic Wavelet algorithm. EGMs in rapidly conducted atrial fibrillation exceed the maximum amplitude range of 8 mV, resulting in varying degrees of truncation (“clipping”) of the right ventricular (RV) coil-to-can + superior vena cava (SVC) signal compared with the template. Inappropriate detection of ventricular tachycardia (VT) occurs at right. Complexes below show the last eight EGMs before detection at higher resolution. These tachycardia EGMs are clipped (arrow) so that the peaks are cut off. The rhythm is classified as VT because six of the last eight EGMs have less than 70% match. This problem was corrected by expanding the EGM scale ±16 mV AS, Atrial sensed event; VP, ventricular paced event.

Alignment Errors

Alignment errors prevent match between a tachycardia EGM and a morphologically similar stored EGM. Mechanisms depend on the method used for EGM alignment (Fig. 3-58). Accurate alignment in the St. Jude algorithm is sensitive to the value of the sensing threshold at the onset of the ventricular EGM, as determined by Automatic Sensitivity Control. If a template EGM is acquired at the most sensitive setting of Automatic Sensitivity Control (either because of a slow sinus rate or after a ventricular paced beat), a low-amplitude peak at the onset of the ventricular EGM may be used for alignment. During tachycardia, the next R wave may occur before Automatic Sensitivity Control reaches its sensitive value. In this case, the low-amplitude peak at the onset of the ventricular EGM may not be used for alignment. If identical template and tachycardia EGMs are then compared, their representations in the morphology algorithm will not match. Usually, they are assigned morphology match scores of 0% (see Fig. 3-58). In patients who have dual-chamber ICDs and intact AV conduction, the most reliable template is acquired or verified during atrial pacing at a rate close to the VT rate. In single-chamber ICDs, options may include acquiring or verifying the template during exercise testing and altering the minimum sensitivity, Threshold Start, or Threshold Delay. If a template is acquired during atrial pacing or exercise testing, automatic template updating should be disabled to prevent overwriting the template with one acquired in baseline rhythm.

Figure 3-58 Alignment error in morphology algorithm.

Interaction of Automatic Sensitivity Control and morphology analysis in St. Jude MD algorithm. Left panel, Stored EGM of supraventricular tachycardia (SVT) inappropriately detected as ventricular tachycardia (VT). Right panel, Programmer strip of validated template in sinus rhythm. Despite identical ventricular EGMs, morphology match scores are 0% in SVT and 100% in sinus rhythm. Slanted line denotes slope of Automatic Sensitivity Control. In sinus rhythm, Automatic Sensitivity Control reaches minimum value before the next ventricular EGM, so that the small peak at onset of EGM (arrows) is used for alignment. In SVT, Automatic Sensitivity Control does not reach minimum, and the small peak is not used for alignment.

Medtronic ICDs align EGMs based on their tallest (positive or negative) peaks. If an EGM has two peaks of near-equal amplitude, or if such peaks are caused artificially by truncation of large EGMs that exceed the programmed dynamic range, minor variations in their relative amplitudes may result in an alignment error.¹²⁰ An alternative EGM source should be selected.

The Boston Scientific morphology algorithm (Rhythm ID) simultaneously collects data from two different channels: the local bipolar rate-sense EGM and shock EGM.^125,126 It considers the vector of depolarization wavefront (with shock EGM as the reference), the timing of the peak signal recorded on the rate-sense EGM, and alignment of eight characteristic points of the shock EGM. A correlation equation aligns peaks on the rate-sense EGM for each QRS complex of the unknown rhythm with the rate-sense EGM peak of the sinus template, then compares the eight captured points with the template.

Rate-Related Aberrancy

If complete bundle branch aberrancy occurs reproducibly, the template may be recorded during rapid atrial pacing (Fig. 3-59). However, subtle and varying aberrancy can confound templates acquired during baseline rhythm (Fig. 3-60). If a template is acquired at a fast rate, automatic template updating should be deactivated to prevent subsequent auto-acquisition of a slow, baseline template without aberrancy. During rapidly conducted AF, subtle degrees of aberration often distort the terminal portion of the EGM sufficiently that the percent match is less than the nominal threshold. In St. Jude ICDs, reducing the fraction of EGMs required to exceed the match threshold from 5 of 8 beats to 3 or 4 of 8 beats may reduce this problem without compromising detection of monomorphic VT. Reducing the match percent required to exceed the match threshold may also prevent misclassification of aberrantly conducted SVT as VT, but it results in a greater chance of misclassifying monomorphic VT as SVT than does reducing the fraction of EGMs.

Figure 3-59 Left bundle branch aberrancy error in morphology algorithm.

Programmer strips show electrocardiogram (ECG), event markers, right atrial EGM, and right ventricular true bipolar EGM. Top panel, Measurement of atrial pacing threshold. Captured beats are conducted with rate-related left bundle branch aberrancy resulting in failure of morphology to match template collected in sinus rhythm. The last two ventricular EGMs at the slower sinus rate show 100% template match. A new template was acquired during atrial pacing. Middle panel, Near 100% morphology match to new template during conduction with left bundle branch aberrancy. Bottom panel, EGM morphology templates recorded during baseline sinus rhythm (left) and atrial pacing (right). LBBB, Left bundle branch block; RA, right atrium; RV, right ventricle.

Figure 3-60 Error in morphology algorithm caused by subtle rate-related aberrancy.

Top panel and middle panel, Dual-chamber stored EGM from inappropriately treated sinus tachycardia. Right atrial (RA) EGMs, event markers, and true bipolar right ventricular (RV) EGMs are shown. Values immediately below event markers, corresponding to EGMs in ventricular tachycardia (VT) zone (labeled “T”), indicate the percent match between EGM morphology in VT and sinus morphology (seen in lower left panel). Corresponding “X” labels above event markers indicate that the morphology algorithm classifies the beat as VT because the match is less than 60%. Burst antitachycardia pacing (ATP) is delivered at the right of the first panel. Dotted arrow indicates that panels are not continuous. After several trials of ATP (not shown), a shock is delivered toward the end of the second panel (HV). “Trigger” at right of upper panel indicates (inappropriate) detection of VT. “D =” at right of upper panel indicates that the atrial rate is equal to the ventricular rate. S denotes intervals in the sinus zone above the VT detection interval of 360 msec. Time line is in seconds. Bottom panels show real-time programmer strips. Bottom left panel, 100% template match in sinus rhythm (arrows) and 0% match on premature ventricular complexes (PVCs). However, the bottom right panel, recorded during atrial pacing at a cycle length of 400 msec, shows only two EGMs with adequate match (check marks, arrows). Note that the surface electrocardiogram (ECG) does not show identifiable aberrancy despite sufficient changes in EGM to prevent template match.

Inappropriate Classification of Ventricular Tachycardia as Supraventricular Tachycardia

The St. Jude morphology algorithm, which analyzes only the rate-sensing electrode, continuously misclassifies up to 6% of VTs as SVT (Fig. 3-61). However, if it is restricted to tachycardias (with ventricular rate ≤ atrial rate in dual-chamber ICDs), only 2% of VTs are misclassified.¹¹⁴ The Medtronic morphology algorithm, which usually analyzes high-voltage EGMs, misclassifies about 1% of VTs as SVT.¹²⁷ If misclassification occurs, an alternative EGM source may provide adequate discrimination. Limited data indicate that the corresponding error rate for the Boston Scientific algorithm is comparable or lower. When the St. Jude morphology algorithm is used in a single-chamber ICD, the sustained-duration time-out (Maximum Time to Diagnosis, Section V.F) should be programmed unless the algorithm is known to classify all clinical VTs correctly. Opinions differ about whether comparable features are required for Medtronic and Boston Scientific algorithms.

Figure 3-61 Failure to detect ventricular tachycardia (VT) because of inappropriate classification of VT as supraventricular tachycardia (SVT) by morphology algorithm (St. Jude MD).

Upper panel, Programmer strip from symptomatic tachycardia classified as SVT 3 days after implantation. Electrocardiogram (ECG), event markers, and right ventricular (RV) true bipolar EGM are displayed. Morphology match is 100% on most beats. The 12-lead ECG showed a left bundle branch block–type pattern, a broad R wave in V1 (distinctly different from conducted sinus QRS complexes), and atrioventricular (AV) dissociation. Lower panel, Event markers and RV EGM during tachycardia (left), sinus rhythm after manually delivered antitachycardia pacing (center), and induced VT at predischarge electrophysiologic study (EPS) (right). EGM morphology during spontaneous VT and sinus rhythm are similar and distinct from induced VT at predischarge EPS. True bipolar EGMs fail to distinguish VT from SVT in 5% to 10% of VTs.

Operation for Atrial Rate Less Than Ventricular Rate

Dual-chamber detection algorithms revert to single-chamber operation when the atrial rate is slower than the ventricular rate and generally do not apply any single-chamber discriminators.

Operation for Atrial Rate Equal to Ventricular Rate

The vast majority of tachycardias with 1 : 1 AV relationship are SVT, primarily sinus tachycardia. VT with 1 : 1 VA conduction accounts for less than 10% of VTs detected by ICDs. The building blocks used by each manufacturer for distinguishing 1 : 1 AV conduction of SVT from 1 : 1 VA conduction are summarized in Table 3-2. A dual-chamber onset rule that evaluates both P-R and R-R interval onset improves specificity of R-R interval onset-type building blocks for sinus tachycardia with minimal loss of sensitivity for VT.¹²⁸ Sorin ICDs classify the 1 : 1 tachycardias based on “chamber of acceleration”: the first accelerated cycle is classified as “supraventricular” if two ventricular events are classified as “conducted,” if the R-R interval is greater than 75% of the preceding P-P interval. If the two ventricular events defining the first accelerated interval are ‘ ‘conducted,’ ’ the origin of the acceleration is declared “atrial.” Conversely, if the first accelerated beat is nonconducted, it is considered ventricular in origin. Individual examples are shown in Figures 3-63 to 3-66.

Figure 3-63 Appropriate rejection of supraventricular tachycardia (SVT) with 1 : 1 atrioventricular (AV) conduction by AV relationship and morphology algorithm.

Right atrial (RA), right ventricular (RV) rate-sensing, and Shock EGMs are shown in order with dual-chamber event markers. SVT begins with gradual warm-up, accelerating from a cycle length of approximately 540 msec to 380 msec in 3 seconds. The Boston Scientific Rhythm ID algorithm classifies morphology as SVT based on 1 : 1 AV association combined with constant shock morphology and constant timing relationship between peaks of two ventricular EGMs. The label ATR at bottom of event markers indicates that the rhythm is classified as SVT. Upper panel shows episode summary classifying rhythm as SVT.

Figure 3-64 A, Discrimination of 1 : 1 tachycardias by P-R pattern using the original sinus tachycardia criterion in Medtronic PR Logic (GEM III, Marquis, Maximo, InSync Marquis, InSync Maximo) ICDs. PR Logic uses patterns of A-V, V-A, V-V, and A-A intervals from consecutive ventricular events to form couple codes describing supraventricular tachycardias (SVTs). These couple codes are based on the number of atrial (A) events within the ventricular (V) interval and their relative timing. Discrimination of sinus tachycardia (ST) from ventricular tachycardia (VT) with 1 : 1 retrograde conduction is critically dependent on the programmable 1 : 1 VT-ST boundary, which is defined nominally as 50% of the current R-R interval (with optional values of 35%, 66%, 75%, and 85%). P-R intervals less than the defined percentage of the current R-R interval and greater than 70 msec are considered antegrade (left panel), and P-R intervals equal to or greater than that percentage of the R-R interval and also greater than 40 msec are considered retrograde (right panel). Rhythms in the detection zone with 1 : 1 pattern and antegrade P-R intervals are classified as ST, and therapies are withheld. Rhythms with 1 : 1 pattern and retrograde P-R intervals are classified as VT. B, New adaptive ST rule in PR Logic (Medtronic Entrust DR ICD) no longer uses 1 : 1 VT-ST boundary to discriminate ST from VT with 1 : 1 retrograde conduction. Rather, it uses a combination of pattern, sudden R-R onset, and sudden P-R onset. Operation of the new adaptive ST criterion is illustrated by this example of a spontaneous episode of ST that converted into VT with 1 : 1 retrograde conduction. During the ST, the R-R and P-R intervals occurred within their expected ranges. After initiation of VT, the R-R and P-R intervals occurred outside their expected ranges. Evidence of ST was lost by the third beat of VT. The sudden change in R-R intervals was small (460 msec), so the R-R intervals during the VT occurred within twice the R-R interval expected range, and adaptation of the RR interval expected range continued. Eventually, the R-R interval expected range adapts to accept the R-R intervals of the VT. Conversely, the P-R intervals occurred outside of twice the P-R expected range, and adaptation was inhibited. VT is appropriately detected after 16 consecutive R-R intervals in the VT detection zone despite gradual R-R onset of a 1 : 1 rhythm, because the P-R intervals had sudden onset and were no longer within the expected P-R interval range. VF, Ventricular fibrillation.

(From Stadler RW, Gunderson BD, Gillberg JM: An adaptive interval-based algorithm for withholding ICD therapy during sinus tachycardia. Pacing Clin Electrophysiol 26:1189-1201, 2003.)

Figure 3-65 Appropriate detection of ventricular tachycardia (VT) with 1 : 1 ventricular-atrial (VA) conduction by abrupt change in P-R pattern and ventricular rate.

Upper panel, Dual-chamber stored EGMs and dual-chamber EGM markers. Note that upper numerical values indicate P-R interval, not P-P interval. Lower panel, Interval plot. VT starts in the ventricle with a premature ventricular complex, followed by abrupt change in ventricular rate, ventricular EGM morphology, and the PR-RR relationship (Medtronic PR Logic). Fast VT (FVT) detection at right of EGM is followed by burst antitachycardia pacing (VT Rx 1 Burst), designated by ATP on interval plot. This results in abrupt termination of VT. Note that older versions of this algorithm, which discriminated VT from SVT based only on the PR-RP percentage, may have classified this long-RP tachycardia incorrectly as SVT. AEGM, Atrial electrogram; TF, interval in FVT zone; TS, interval in VT zone; VEGM, ventricular (true bipolar) electrogram; VS, ventricular sensed event.

Figure 3-66 Classification of 1 : 1 tachycardias by chamber of origin.

The Sorin Parad algorithm classifies 1 : 1 tachycardias by chamber of onset. Stored ventricular EGM is shown with dual-chamber event markers (top, atrial; bottom, ventricular). Event log and Event summary are shown above each stored EGM. Upper panel, Supraventricular tachycardia (SVT) is diagnosed by atrial onset. Lower panel, Ventricular tachycardia (VT) is diagnosed by ventricular onset.

As in the electrophysiology laboratory, pacing methods may also be used to discriminate 1 : 1 tachycardias (see Active Discrimination).

Operation for Atrial Rate Greater Than Ventricular Rate

The building blocks used by each manufacturer for distinguishing rapidly conducted AF and atrial flutter from VT during atrial arrhythmias are summarized in Table 3-3. Although various combinations of these building blocks successfully discriminate VT from SVT in VT zones, most VT during AF is sufficiently rapid to be detected in the VF zone,¹²⁹ where some discriminators lose specificity or may not be applied at all. For example, R-R regularity cannot be applied in the VF zone because polymorphic VT has irregular R-R intervals, and undersensing of VF exaggerates the irregularity of measured R-R intervals. AV dissociation occurs uniformly during rapidly conducted AF,⁶⁸ and morphology templates acquired during sinus rhythm often misclassify aberrantly conducted beats. Individual examples are shown in Figures 3-67 to 3-69.

Figure 3-67 Ventricular tachycardia (VT) during rapidly conducted atrial flutter discriminated by morphology of ventricular EGM.

VT occurs during atrial flutter with 2 : 1 atrioventricular (AV) conduction. Time line marks indicate that the ventricular rate had been in the VT zone for 24 seconds at the beginning of the panel. VT therapy is withheld because of both the 2 : 1 AV relationship and the morphology match of ventricular EGMs to baseline sinus-rhythm EGMs (not shown). Numbers below event markers indicate 98% to 100% morphology match in atrial flutter, compared with 40% to 47% match in VT; “<” (box) marks indicate that ventricular rate is less than atrial rate. VT is detected by AV dissociation and morphology match percentage less than the threshold value of 60%. Morphology classification of each beat as SVT (check marks) or VT (X) is indicated above event markers.

(From Swerdlow C, Shivkumar K: Implantable cardioverter defibrillators: clinical aspects. In Zipes DP, Jalife J, editors: Cardiac electrophysiology: from cell to bedside, ed 4, Philadelphia, 2004, Saunders.)

Figure 3-68 Inappropriate detection of rapidly conducted atrial fibrillation (AF) despite rejection by morphology.

Right atrial (RA) EGM, event markers, and true bipolar right ventricular (RV) EGM are shown. Values immediately below event markers corresponding to EGMs in ventricular tachycardia (VT) zone (marked T) indicate the percent match between EGM morphology in tachycardia and sinus morphology. Corresponding check marks above event markers indicate that the morphology algorithm classifies the beats as supraventricular tachycardia (SVT) because the match is greater than 60%. Stability and morphology algorithms are combined with “AND” logic in this St. Jude ICD, so that VT is diagnosed if either discriminator classifies the rhythm as VT. The stability algorithm incorrectly classifies the rhythm because the ventricular cycle lengths regularize. The rhythm would have been classified correctly if the morphology discriminator alone had been programmed. F markers indicate ventricular intervals in ventricular fibrillation (“Fib”) zone. S denotes intervals in the “Sinus” zone above the VT detection interval. T denotes intervals in VT zone. DDI, Mode switch to DDI pacing.

(From Swerdlow CD, Friedman PA: Advanced ICD troubleshooting. Part I. Pacing Clin Electrophysiol 28:1322-1346, 2005.)

Figure 3-69 Appropriate rejection of rapidly conducted atrial flutter by pattern analysis.

Upper panel, Interval plot. Lower panel, Dual-chamber stored EGM with event markers. Atrial flutter is identified by the consistent 2 : 1 PR association of the regular atrial and ventricular rhythms. AF annotations indicate that the number of intervals in the ventricular tachycardia (VT) rate zone required for detection (16) has been reached, but that therapy is withheld because atrial flutter is diagnosed. Interval plot displays 2 : 1 atrioventricular (AV) association until termination of atrial flutter after 12 seconds. AR, Atrial intervals in pacing refractory period; TS, ventricular intervals in VT zone.

Specific Algorithms

Figures 3-70 to 3-74 show block diagrams of the principal dual-chamber detection algorithms in clinical use today. Legends identify unique features of each algorithm.

Figure 3-70 Boston Scientific Rhythm ID algorithm.

The central features of this algorithm are comparison of atrial (A) and ventricular (V) rates (first step) and analysis of ventricular EGM morphology (Vector Timing and Correlation, or VTC). If the V rate exceeds the A rate by at least 10 bpm, the rhythm is classified as ventricular tachycardia (VT). If at least 3 of the last 10 beats are classified as supraventricular by morphology analysis, the rhythm is classified as supraventricular tachycardia (SVT). If morphology analysis does not classify the rhythm as SVT, the A rate and the regularity of the V rate are evaluated. The rhythm is classified as SVT if the A rate exceeds 200 bpm and the ventricular rhythm is irregular (measured ventricular interval stability >20 msec). Otherwise, it is classified as a VT.

Figure 3-71 St. Jude A-V Rate Branch + Morphology Discrimination (MD) algorithm.

Similar to the Rhythm ID, the central features of this algorithm are a comparison of the atrial (A) and ventricular (V) rates (first step) and analysis of ventricular EGM morphology (MD). Rhythms are classified into three Rate Branches: ventricular rate less than atrial rate (V < A), ventricular rate equal to atrial rate (V = A), and ventricular rate greater than atrial rate (V > A). If V < A, the algorithm applies morphology discrimination (and/or both interval stability and N:1 AV association). If V = A, it applies morphology discrimination (and/or sudden onset). All rhythms in the V > A rate branch are treated as ventricular tachycardia (VT). ST, Sinus tachycardia; SVT, supraventricular tachycardia; VF, ventricular fibrillation.

Figure 3-72 Biotronik SMART algorithm.

The high-level structure of this algorithm is similar to that of the St. Jude ICD. SMART first compares atrial (A) and ventricular (V) rates. If the V rate is faster, ventricular tachycardia (VT) is diagnosed. If the V rate is slower, ventricular interval stability and A : V ratio analysis results in discrimination of atrial fibrillation (AFib; unstable ventricular rhythm), atrial flutter (AFlut; stable rhythm, N:1 association), and VT (stable rhythm, AV dissociated). If the A rate and V rate are equal, SMART analyzes the interval stability. If the R-R interval is stable and the P-P interval is unstable, VT is diagnosed. If the PP interval is stable, the algorithm sequentially analyzes AV association (“AV Trend”) and sudden onset as discriminators. Sinus T, Sinus tachycardia; SVT, supraventricular tachycardia.

Figure 3-73 Medtronic PR Logic algorithm.

A, Clinical decision process. The rhythm types are represented using P-R marker diagram examples (top), from least complex (left) to most complex (right). A list of possible PR Logic decisions and the building blocks used are also shown for each of the rhythm types. Note that PR Logic implicitly employs a ventricular (V) rate > atrial (A) rate override (first column), because all rhythms with A rate < V rate and V rate in one of the detection zones are classified as ventricular tachycardia/fibrillation (VT/VF). For tachycardias with A rate = V rate (i.e., 1 : 1 tachycardias, center column), the original PR Logic uses P-R patterns and discriminates 1 : 1 rhythms with critical P-R interval timing zones (see Fig. 3-64, A). The new sinus tachycardia (ST) criterion in PR Logic (Entrust and later) uses the P-R pattern along with P-R and R-R sudden-onset criteria (see Fig. 3-64, B). For rhythms with A rate > V rate, PR Logic uses P-R patterns along with R-R regularity, P-R dissociation, and P-P regularity to ensure detection of double tachycardia (VT or VF during atrial fibrillation [AFib]) and to withhold therapy for 2 : 1 atrial flutter, rapid AFib, and ST with far-field R-wave oversensing (FFRW OS). In the newest Medtronic ICDs (Protecta, currently investigational), EGM morphology analysis (Wavelet) is applied for rhythms with A rate ≥ V rate that are identified by SVT by PR Logic pattern analysis alone. In addition, detection of double tachycardia in the Protecta devices requires that abnormal EGM morphology be present. Integration of Wavelet EGM morphology with PR Logic is designed to improve detection specificity compared to PR Logic interval analysis alone. B, PR Logic computational flow diagram. On each ventricular event, PR Logic processes the new P-R, R-P, P-P, and R-R patterns and timing information for the building blocks. If VT/fast VT (FVT) or VF rate detection criteria are satisfied, the ventricular rate override criterion is checked first. If the median R-R interval is less than the supraventricular tachycardia (SVT) limit, detection occurs through the single-chamber detection criteria without considering the PR Logic discrimination algorithm. If the median R-R interval is greater than the SVT limit, and if double tachycardia (VT/FVT/VF + SVT) is not detected, the three PR Logic criteria for identifying SVTs are tested in the order shown. If any one of the PR Logic SVT criteria is satisfied, inappropriate detection is avoided. If an SVT is not positively identified by PR Logic pattern analysis and the A rate ≥ V rate, EGM morphology analysis (Wavelet) is applied and classify the rhythm as SVT if the EGM morphology during tachycardia matches the template. If none of the SVT discrimination criteria is satisfied, VT/FVT/VF is detected when the R-R interval–based criterion is satisfied. If SVT is identified, the entire process repeats itself on each ventricular event until VT/FVT or VF is detected or the rhythm slows out of the ventricular rate detection zones.

Figure 3-74 Sorin Parad+ algorithm.

If most of the detected R-R intervals are in the ventricular tachycardia (VT) zone, ventricular interval stability is analyzed in a first step, using a histogram of R-R intervals. If the rhythm is irregular, atrial fibrillation (AF) is diagnosed and therapy is withheld. If the rhythm is regular, the atrioventricular (AV) association is assessed by comparing peak amplitudes of R-R and P-R interval histograms. If the rhythm is AV dissociated, VT is diagnosed, unless the Long Cycle Search is activated, which inhibits therapy if a long ventricular cycle (VTLC, characteristic of AF) is identified. If N:1 P-R association is identified, the rhythm is classified as supraventricular tachycardia (SVT). In the presence of 1 : 1 P-R association, the Parad+ evaluates the rate of acceleration of the ventricular rate. If acceleration is gradual, sinus tachycardia (ST) is diagnosed. If acceleration is sudden, Parad+ identifies the chamber of origin and withholds therapy if it is the atrium. AFlutter, Atrial flutter; AT, atrial tachycardia.

Single-Chamber vs. Dual-Chamber Discriminators

Nominal programming of dual-chamber algorithms is safe.^130,131 Dual-chamber stored EGMs provide higher diagnostic accuracy for troubleshooting than single-chamber stored EGMs. However, dual-chamber discriminators cannot be implemented without the complications inherent in atrial leads. Also, dual-chamber ICDs introduce unique risks for underdetection of VT caused by cross-chamber ventricular blanking after atrial pacing (see Intradevice Interactions). In addition, with early dual-chamber algorithms, optimal values of programmable parameters were not known; initial approaches to the problem of atrial blanking versus FFRW oversensing had limited success; and atrial sensing problems and specific design flaws degraded performance. Not surprisingly, clinical studies of early dual-chamber algorithms reported no benefit over single-chamber algorithms.^{67,68,132,133} More recent, randomized prospective studies demonstrate moderate superiority in SVT-VT discrimination for dual-chamber over single-chamber algorithms.^134,135 However, no agreement exists as to when the incremental implant complexity, price, risk of atrial lead complications, and reduced longevity of a dual-chamber ICD are justified for SVT-VT discrimination alone. Generally, dual-chamber ICDs should be considered in patients who are likely to have rapidly conducted SVT, those in whom monomorphic VT and sinus tachycardia rates are likely to overlap, and those who will benefit from additional diagnostics for atrial arrhythmias.

Quantitative Considerations

Detection algorithms must maintain almost 100% sensitivity for detection of VT; but detection of hemodynamically-stable asymptomatic VT is not necessarily synonymous with therapy. If patient populations and programmed detection boundaries are equivalent, positive predictive accuracy may be the most useful statistical measure of algorithm performance.⁶⁸ However, it is highly dependent on the ratio of SVT to VT episodes and therefore on the programmed detection rate and patient population (prevalence of SVT and VT). However, it is almost impossible to obtain a clinically meaningful comparison of different detection algorithms based only on their performance on different sets of data.

Inappropriate Detection, Inappropriate Therapy, and Inappropriate Shocks

All failures of SVT-VT discrimination algorithms do not have equivalent clinical outcomes. ATP is delivered immediately after detection so the numbers of detections and therapies are equivalent. In contrast, shock delivery requires capacitor charging, during which therapy may be aborted. However, inappropriate ATP is often unnoticed. It terminates about 40% of SVTs and reduces the ventricular rate below the VT detection interval in another 20%.¹³⁹ In contrast, inappropriate shocks adversely affect quality of life.¹⁴⁰ Active discrimination blurs the distinction between detection and therapy because only tachycardias that persist immediately after diagnostic pacing are classified.