Troubleshooting Principles

Malfunction of ICDs is uncommon. The most common reasons for inappropriate therapy or the absence of expected therapy relate to inappropriate programming, lead-related complications, imperfect diagnostic specificity, insufficient understanding of the technical specifications of device function, and drug-device interactions. Device-device interactions and true component malfunctions now rarely occur.¹

Troubleshooting of suspected ICD malfunction follows logically from analysis of the index clinical event, whether from a patient report or from a device’s diagnostics. By evaluation of the clinical history, physical examination, radiographic techniques, device memory, and online telemetry, the cause of malfunction can usually be determined. The analysis must take into account the specifics of the ICD as well as the sensing and treatment algorithms used to arrive at an accurate diagnosis. Therapy may be delivered or withheld because of true malfunction of a component or because the device is responding appropriately according to its programming and interaction with the environment (e.g., responses to electromagnetic interference).

Analysis of ICD problems begins with identification of the problem, followed by analysis of its possible cause. Presentation falls into four general areas: multiple shocks, failure to convert ventricular tachycardia (VT) or ventricular fibrillation (VF), failure to detect VT or VF, and problems with pacing.²

Initial Evaluation and Tools

Clinical History

As in all other areas of medicine, the keystone to diagnosis is the clinical history. Historical clues suggesting ICD malfunction, although by themselves nondiagnostic, provide the basis for a presumptive diagnosis and suggest further avenues of inquiry. For example, repeated shocks in an asymptomatic patient suggest false signal detection or an atrial arrhythmia. Certain body positions associated with ICD shocks suggest the possibility of lead fracture or lead instability. ICD model, manufacturer, date of implantation, lead type and location, antiarrhythmic drug history, presence or absence of symptoms of heart failure or angina preceding therapy, and other clues are important, because any or all of these factors may have relevance and can be related to ICD therapy. The presence of an implanted pacemaker with an ICD raises the specter of device-device interaction. Therapy during exercise suggests sinus tachycardia or atrial fibrillation with a rapid ventricular response. Palpitations may signal either supraventricular tachycardia (SVT) or VT.

Physical Examination

The physical examination may be helpful in pinpointing the exact cause of ICD malfunction. For example, for a patient who reports that certain body positions or movements elicit ICD shocks, reproducing the precise maneuver or the exact circumstances known to elicit the shocks, while simultaneously telemetering the device, may prove the diagnosis of lead failure. Occasionally, patients are reluctant to allow the examiner to do this because the prospect of an ICD shock is psychologically threatening.³ In such circumstances, the ICD may be placed in a monitor mode, either by programming or using a magnet to suspend tachyarrhythmia detection, while assuring the patient that shock delivery will not occur. The assessment of recordings from the ICD in these situations is discussed later.

Because a common point of lead fracture occurs where transvenous ICD leads pass between the clavicle and first rib,^4,5 manipulation of the device pocket or the lead entry point in the pectoral area may elicit electrical noise artifact, indicating lead conductor fracture.^6,7 Similar artifacts may occur with a loose connection of the set screw to the lead terminal pin in the ICD pulse generator header. Again, make-break potentials can be demonstrated by noninvasive telemetry, as discussed later.

Other diagnostic clues may be provided by the physical examination, such as detection of an irregular pulse suggestive of atrial fibrillation, which may be readily confirmed by an electrocardiogram. Congestive heart failure is often associated with exacerbation of ventricular arrhythmias.^8–10 Therefore, eliciting symptoms and signs of left ventricular decompensation suggests a possible cause for worsening arrhythmias.

Electrocardiographic Recordings

Objective evidence of the event precipitating ICD therapy, or of an arrhythmia with absence of the expected ICD response, can be extremely helpful and is usually diagnostic (Fig. 30-1). In the days before the availability of stored electrograms (EGMs), such evidence was rarely available unless the patient was hospitalized and monitored at the time of the event. Fortunately, this is now a problem of the past. Nevertheless, even when available, EGMs can sometimes be inconclusive or confusing. In addition, if multiple ICD detections have occurred for any reason, including lead failure, SVT, or a VT or VF storm, EGM documentation of the initial event leading to detection may be overwritten because of limited device memory.

Figure 30-1 Electrogram (EGM) shows sinus rhythm with premature ventricular complexes (PVCs) in a pattern of bigeminy, followed by a rapid, irregular ventricular arrhythmia, then a high-voltage (HV) shock delivered by a Model V193 defibrillator (St. Jude Medical, Sunnydale, Calif.). The three recordings are from the atrial channel (top), marker channel (middle), and ventricular pace/sense channel (bottom). Note that the EGM of the ventricular arrhythmia has a different morphology from that of the preceding sinus rhythm. The first and third beats are PVCs. The second and forth beats are sinus in origin. The fifth beat is a PVC that initiates ventricular fibrillation, typified by a rapid, polymorphic EGM.

Device Telemetry

The advent of device telemetry has revolutionized and greatly simplified analysis of arrhythmias and events suspected of representing ICD malfunction. Early devices (e.g., Ventak, Cardiac Pacemakers [CPI/Guidant], St. Paul, Minn.) were able to emit sounds or beeps synchronous with ventricular electrogram (VEGM) detection, enabling identification of oversensing or undersensing. When recorded with a phonocardiogram, a so-called “beep-o-gram” was produced. Although helpful, these crude attempts at telemetry were quickly supplanted by advances that enabled detailed information, such as R-R intervals and VEGMs, to be telemetered¹¹ (Fig. 30-2). The complexity and sophistication of these diagnostic tools are increasing with the new generation of ICDs. The addition of the atrial channel has greatly simplified analysis of arrhythmic events (Fig. 30-3).

Figure 30-2 The first line of this tracing is an EGM recorded from a single-chamber St. Jude Medical Model V193 ICD. This device monitors the heart rate and has the capability of analyzing the EGM morphology. The bottom lines annotate the ICD sensing function. The checks and 100 numbers indicate the degree of matching of the EGM morphology to a sinus rhythm template that can be used to distinguish atrial from ventricular arrhythmias. The 1047 and 1063 numbers indicate cycle length. S indicates sinus rhythm; R, sensed R waves; and the numbers at the bottom count the seconds of the recording.

Figure 30-3 As in Figure 30-2, this tracing EGM recorded at a routine clinic follow-up visit shows the ventricular EGM (fourth tracing) as well as an atrial EGM (third tracing) from a dual-chamber Model V242 ICD (St. Jude Medical). The first and second tracings are, respectively, a surface electrocardiogram (ECG) and a marker channel. In the marker channel, the checks indicate a template match with sinus rhythm (100% registered below the horizontal marker line). The fourth beat, marked x, is a premature ventricular depolarization with a different morphology than sinus; it is logged as a 0% match. The next two rows of numbers are the P-P and R-R intervals. S indicates sinus rhythm, as defined by the ICD, as opposed to T or F if ventricular tachycardia or fibrillation were present, respectively. R indicates sensed ventricular activity. The data on atrial and ventricular EGM timing and ventricular EGM morphology allow this device to use discrimination algorithms to differentiate supraventricular from ventricular arrhythmias. It can also function as an atrial, ventricular, or dual-chamber pacemaker.

Other diagnostic information is now routinely available from all ICDs, including battery voltage, pacing and sensing lead impedances, charge times, capacitor re-formation times, high-voltage lead impedance, and frequency and timing of ventricular events. The finding of a depleted battery, lead impedance out of range (either too high or too low), and R-R intervals can lead to correct diagnoses. The latter is extremely important, because nonphysiologic intervals (short) raise the suspicion of make-brake electrical noise artifact. One device (Medtronic, Minneapolis) records information regarding both impedance changes (Fig. 30-4) and nonphysiologic R-R intervals.¹²

Figure 30-4 The plot shows decreasing ring-coil impedances logged in a GEM ICD (Medtronic, Minneapolis). The first four impedances readings of less than 15 ohms are circled. The arrow indicates that the low-impedance trigger would have provided an alarm 15 weeks before this patient received an inappropriate shock.

(From Gunderson BD, Patel AS, et al: An algorithm to predict implantable cardioverter-defibrillator lead failure. J Am Coll Cardiol 44:1898-1902, 2004.)

Newer ICDs from all major manufacturers are capable of remote device telemetry from the patient’s home.^13,14 This capability offers two advantages over exclusively office-based interrogation. The first advantage is the ability to interrogate the ICD more quickly and easily after a shock. The second advantage is that most remote monitoring frequently assesses lead impedance changes and other parameters that may herald lead failure. The Lumos-T Reduces Routine Office Device Follow-Up Study (TRUST) showed that remote monitoring with automatic daily surveillance provides early detection of device events.^15–17 Once a potential lead failure is identified by the device, the physician can be notified quickly, often quickly enough to intervene before lead failure produces a clinical event (i.e., shock for lead fracture). The CONNECT Trial demonstrated that remote follow-up of ICD significantly reduced the time from onset of clinical events (new-onset atrial or ventricular arrhythmias, lead or device integrity problems) to clinical decisions compared with usual outpatient clinic follow-up.¹⁸ Unfortunately, some device failures occur with adverse events happening so rapidly, such as inappropriate shocks, that early notification is impossible.¹⁹ Notably, even if mechanisms such as audible tones are available for patient notification, with increasing age, more patients are unable to hear the alerts.²⁰

Radiographic Evidence

Radiographic evidence is helpful if lead malposition, dislodgment, or fracture is suspected.⁷ It is recommended, whenever possible, that chest radiographs taken immediately after ICD implantation be compared with radiographs obtained at the time of a problem. Such comparisons may reveal lead malposition and displacement when none is suspected. Radiographs can also be helpful in demonstrating conductor fracture or pin connectors improperly positioned in the header. Examining the radiographic “signature” of the device can help identify the device type when the patient is unaware of the ICD model or type. Knowledge of unique failure modes such as “migration” of set screws (travel of the right ventricular fixation screw into the channel lumen during shipment) in particular families of ICDs is of course desirable.²¹

Figures 30-5 and 30-6 show examples of lead dislodgment. Because the distal end of the lead is close to the tricuspid anulus (Fig. 30-5, A), both atrial and ventricular signals are recorded (B). This leads to “double-counting” such that, for a given heart rate, the ventricular channel registers twice the actual heart rate, satisfying the high rate detection requirement and leading to administration of a shock. Figure 30-6 shows a similar example, but the electrical artifacts are more disorganized and irregular, presumably because the lead is more free floating. Figure 30-7 shows an example of disruption of a transvenous ICD lead as it passes between the clavicle and first rib. Care must be taken to learn the appearance of the welds of the springs, because they can be mistaken for fracture if their usual appearance of being offset from the central axis of the lead is not appreciated²² (Fig. 30-8). An unusual cause of lead failure, and sometimes failure of telemetry, occurs in twiddler’s syndrome; in these cases, a lead is twisted and fractured or the ICD is inverted^23,24 (Fig. 30-9). Lead perforation can occur subacutely or rarely chronically, leading to several problems, including undersensing of ventricular arrhythmias, failure to convert ventricular arrhythmias, and failure of pacing as the lead tip exits the myocardium (Fig. 30-10).

Figure 30-5 Lead dislodgment.

A, Right ventricular pacing-sensing-defibrillation lead displaced to the tricuspid anulus in biventricular pacing-ICD system (St. Jude Medical Model V340). B, Recordings from the ICD. Top to bottom, Atrial EGM, the marker channels, and the ventricular EGM. In the marker channel area, F indicates fibrillation detection, trigger denotes episode detection and asterisks (*) indicate device charging. During the charging and reconfirmation period, R indicates an interval fast enough to reconfirm, and S indicates an interval below the VT/VF rate. The numbers are atrial and ventricular cycle lengths. Below the numbers, P stands for P wave and R for R wave. Because the distal end of the lead is close to the tricuspid annulus, both atrial and ventricular signals are recorded in the ventricular channel, such that the ventricular rate counter registers twice the actual heart rate. The R-R labels reflect double-counting from the EGM in the bottom tracing recorded from the right ventricular distal tip. This double-counting satisfied high rate detection, and a shock was delivered (HV).

Figure 30-6 Lead dislodgment.

A, Interval plot; B, stored EGM; the right ventricular defibrillation lead was dislodged to the tricuspid valve. Random electrical signals generated by lead movement in the ventricle provided very rapid and irregular artifacts that were interpreted by the Model 7230 ICD (Medtronic) as ventricular fibrillation, and a shock was delivered (CD, charge delivered) in B.

Figure 30-7 Clavicle–first rib lead discontinuity.

A, Posteroanterior chest radiograph of an ICD system from a patient who presented with multiple shocks. The lower lead has a discontinuity where it travels between the clavicle and the first rib. B, Close-up view of a different patient with an identical presentation. The fracture again is at the junction between the clavicle and first rib and is easily identified where the superior, high-voltage lead is bent with a discontinuity at the inferior margin.

(Courtesy of Dr. Paul A. Levine.)

Figure 30-8 Lead fracture.

A, Posteroanterior chest radiograph that was interpreted to show a lead fracture (Endotak, Cardiac Pacemakers, St. Paul, Minn.). The area is the distal portion of the defibrillation coil with apparent lateral displacement of the electrode (arrowhead). B, Coned-down view of the lead, documenting a continuous metallic connection at the level of the transition from a central electrode to the outer coil. This is the typical appearance of a Guidant Endotak lead, and no fracture is present.

(From Kratz JM, Schabel S, Leman RM: Pseudo fracture of a defibrillating lead. Pacing Clin Electrophysiol 18:2225-2226, 1985.)

Figure 30-9 Two examples of “twiddling.”

A, Dramatic example of multiple twists of leads in an ICD implanted in the abdomen. B, Subtler example of twiddler’s syndrome; the ICD has been rotated 180 degrees. This patient’s presentation was inability to interrogate the ICD.

Figure 30-10 Lead perforation.

Chest radiograph obtained 6 weeks after dual-chamber ICD implantation shows subacute lead perforation. The device was implanted uneventfully for primary prevention and to manage sinus node dysfunction. At routine 6-week follow-up, the patient was asymptomatic, the R-wave amplitude had deteriorated, and there was failure to capture. Lead revision was performed without complication.

 Multiple Shocks

Multiple shocks may be the result of incessant VT/VF, repetitive VT/VF, oversensing of T waves, electromagnetic interference, electrical noise artifacts from a fractured conductor or myopotentials (in the case of insulation failure), or SVT, including sinus tachycardia, atrial fibrillation, atrial flutter, or another mechanism (Table 30-1). Sometimes the ICD itself can be proarrhythmic (Box 30-1).

TABLE 30-1 Multiple Shocks

AF, Atrial fibrillation; VF, ventricular fibrillation; VT, ventricular tachycardia.

The occurrence of multiple shocks leads to myriad problems. Shocks are poorly tolerated by patients, families, and referring physicians and can create long-lasting fear and distrust of the ICD requiring professional help. Furthermore, shocks lead to significant increases in cost of therapy by repeated hospitalizations and testing sessions; shocks may jeopardize the physician-patient relationship. ICD shocks, both appropriate and inappropriate, have also been associated with increased mortality.^25–27 The occurrence of repetitive shocks for whatever reason represents a medical emergency.

The management of multiple ICD shocks must address multiple factors: First, the patient must be comforted, and to do so, the underlying rhythm must be determined. Is it sinus, supraventricular, or ventricular? If it is not ventricular, the ICD can be disabled by a magnet or programmer. If it is ventricular, is there a precipitating cause, such as ischemia, heart failure, or lead malfunction? If so, can the shocks be stopped with treatment of the cause, or are antiarrhythmic drugs, β-blockers, or other interventions (e.g., reprogramming, ablation) needed? Usually, the occurrence of multiple shocks requires hospital admission, because a monitored environment is not only helpful diagnostically but also for patient safety and support.

Perhaps the most difficult causes of multiple shocks to treat are incessant VT and VF. On the one hand, delivered therapy is appropriate; on the other hand, it is not only a horrible experience for patients but also reflects a substrate problem. ICDs do not treat heart disease, only its consequences. Initial management centers on the administration of drug therapy,^8,9,28,29 and the ICD usually should be disabled by programming or magnet application. Although antiarrhythmic drugs may be useful, particularly intravenous amiodarone, the judicious use of β-blockers has been found to be extremely beneficial.^9,29 If VT or VF episodes are repetitive or incessant despite drug therapy, it is often advisable to intubate and sedate the patient, disable the ICD, and use an external defibrillator to manage the arrhythmia events. Intra-aortic balloon counterpulsation has been used on occasion with success. With the increased efficacy of radiofrequency ablation, incessant arrhythmias may be so treated.³⁰

Oversensing of P waves, T waves, and even atrial flutter may lead to repetitive shocks.^31–33 Atrial oversensing in the ventricle is uncommon; it is usually seen with integrated bipolar leads when the proximal spring is close to or straddling the tricuspid valve^31,32 (Fig. 30-11). P-wave sensing may also be seen when a lead is dislodged to the right atrioventricular junction, as in Figure 30-5. More often, T waves are sensed because of their large size relative to the R wave, but sometimes simply because the T wave is huge, as in a patient with hypertrophic cardiomyopathy (Fig. 30-12). Rarely, T-wave oversensing occurs during rate-related bundle branch block,³⁴ hyperglycemia,³⁵ and the Brugada syndrome.³⁶

Figure 30-11 Recordings showing counting of both the atrial (P) and ventricular (R) signals from an integrated bipolar ventricular lead (Model RV-1101, Ventritex, Sunnyvale, Calif.) in a radiographically ideal apical position. Because the patient and her heart were small, the distal spring, used both for defibrillation and as the anode for sensing, came to straddle the tricuspid valve, allowing atrial signals to be recorded.

Figure 30-12 Oversensing in patient with hypertrophic cardiomyopathy.

The system comprises a Model V145AC ICD (Ventritex), a bipolar atrial lead, and a Model 6936 defibrillation lead (Medtronic). A, Tracings from top to bottom show the atrial EGM, the marker channels, and the ventricular EGM. There is both double-counting of the R and T waves (registered as VR in third beat in first tracing) and double-counting of multiple components of the R wave itself (registered as RR in both first and second continuous tracings). In addition, there is triple-counting of both R-wave components and the T wave (registered as RR-R) that leads to fibrillation detection (labeled F) and an ICD shock (labeled HV) in sinus rhythm. The R wave was greater than 25 mV, and the T wave greater than 4 mV. Oversensing was corrected by programming; the threshold start (signal amplitude at which sensing sensitivity begins) and the decay delay (time after sensed beat when sensitivity begins to increase) were extended, and the sensitivity was decreased. To ensure proper function, ventricular fibrillation was induced after programming to show that, with more restrictions on sensing, VF could still be sensed appropriately by the ICD. B, The 12-lead ECG from this patient shows marked left ventricular hypertrophy and voltage.

First reported in the mid-1990s, myopotentials can also be sensed and interpreted as a ventricular arrhythmia, leading to inappropriate shock therapy. These may arise from the pectoral muscles^37,38 or from the diaphragm.^39,40 The latter has occurred with integrated bipolar leads that presumably lie near the respiratory muscles (Fig. 30-13). An unusual case of inappropriate shock therapy caused by pectoral muscle myopotential oversensing involving an integrated bipolar lead occurred with high-voltage pin reversal in the device header⁴¹ (Fig. 30-14).

Figure 30-13 Diaphragmatic sensing in patient with high-grade AV block.

The three recordings show the atrial EGM (top), bipolar ventricular EGM from the sensing lead (middle), and the shock EGM (bottom). The marker channel is shown at the bottom. During deep respiration, rapid, high-frequency activity is recorded on the sensing channel and is judged by the ICD as ventricular fibrillation (VF) and triggers ICD charging. This ventricular oversensing also inhibits bradycardia pacing (Guidant ICD, Boston Scientific).

Figure 30-14 Myopotential sensing.

Caused by reversal of the high-voltage pins in the ICD header of a Guidant Model H175 biventricular ICD. The three tracings are electrograms from the right atrial lead, the right ventricular pacing/sensing system, and the high-voltage lead system. Annotation at the bottom indicates transient inhibition of (biventricular) pacing, and the detection of ventricular fibrillation (VF). Reversal of the high voltage pins in the header alters the sensing circuit in such a way as to make the ICD housing a component of the sensing circuit (the other being the ICD lead tip) permitting pectoral myopotentials to be recorded as rapid electrical activity whenever the left arm is raised.

(From Allison JS, Epstein AE, Plumb VJ: Electrical noise artifact: from without or within? J Cardiovasc Electrophysiol 18:332-333, 2007.)

Oversensing can often be rectified by altering the sensitivity or refractory period of the ICD. However, if T waves are large and R waves are small, changing the sensitivity will not suffice. If the sensitivity is decreased, VF may not be identified (Fig. 30-15). Similarly, lockouts imposed by the device to ensure arrhythmia detection limits programming of device refractory periods. One manufacturer provides greater programmability to overcome these problems. Because ICDs, in contrast to pacemakers, vary their sensitivity to identify ventricular activity, altering the amplitude and timing at which sensitivity is adjusted can address this problem.

Figure 30-15 Ventricular fibrillation (VF).

Made during ICD testing, this recording shows VF that was not detected by a Model V232 ICD (St. Jude Medical) because the electrogram (EGM) amplitude was below the sensitivity for detection. The tracings show the surface electrocardiogram (ECG), the marker channel, the atrial EGM, and the ventricular EGM. Notice that in the marker channel, only two intervals labeled F are shown that signified VF detection; for some appropriately detected short intervals, hash marks are present without F annotation since the printer function cannnot accommodate the typed letters. S indicates an EGM interval longer than the tachycardia detection interval. Also note that the zeroes under the marker channel line indicate the absence of a template match with sinus rhythm, but VF is still not diagnosed, because the cycle length is not short enough as assessed by the ICD sensing circuit. The gain of the ventricular EGM recording was 6.0 mV/cm, indicating that many VF signals were less than 0.5 mV.

For the patient with hypertrophic cardiomyopathy shown in Figure 30-12, the R wave was greater than 25 mV and the T wave greater than 4 mV. If the sensitivity were decreased significantly, VF might not be reliably sensed. Furthermore, because of prolonged repolarization and sensing of the terminal portion of the T wave, prolonging the ventricular refractory period would not suffice to correct this problem. In this case, avoidance of T-wave oversensing was achieved by altering the threshold start, the decay delay, and the sensitivity. Alternatively, on rare occasions an ICD lead may need to be repositioned or a separate sensing lead placed in a different area of the right ventricle.

In anticipation of T-wave oversensing in the future, one approach that may be useful is to test VF detection at low sensitivity at implantation. If oversensing occurs, sensitivity may be decreased without the need for retesting of defibrillation sensing. On the other hand, whenever sensitivity is decreased from a level that has not been tested and shown to provide adequate detection of VF, it is imperative to retest VF detection with an electrophysiologic study to demonstrate adequate sensing (see Fig. 30-15).

The sensing of external electrical activity may lead to false detection. Because of the increased use of ICDs and relatively low lead failure rates, the most common cause (after lead failure, discussed later) is probably electrocautery^42,43 (Fig. 30-16). Other causes include electronic surveillance systems,^42,44 razors,⁴⁵ radiofrequency handheld remote controls,⁴⁶ ungrounded electrical tools,⁴² fluid in a connector port,⁴⁷ transcutaneous electrical nerve stimulation (TENS; Fig. 30-17),⁴⁸ personal digital assistants (PDAs),⁴⁹ and even slot machines⁵⁰ and washing machines.⁵¹

Figure 30-16 This stored electrogram (EGM) was recorded from a Guidant Model 1831 ICD (Boston Scientific, Natick, Mass.) when bipolar cautery was delivered to the patient’s ear lobe to remove a skin cancer. Note the electrical noise artifact in the atrial and ventricular sensing-pacing channels (top two EGMs on each tracing, respectively) but only minimally on the EGM from the high-voltage circuit (bottom electrograms), in which sensing does not occur. The atrial and ventricular intrinsic activity can be seen on all channels.

Figure 30-17 Electrogram shows electrical noise artifact from a Guidant ICD (Boston Scientific) recorded as the patient applied his spouse’s transcutaneous electrical nerve stimulation (TENS) to himself in an attempt to treat muscle aches.

A common cause of multiple shocks is lead failure.^6,28,52–60 When make-break potentials are recorded by the ICD, the rapid electrical activity is interpreted as VF. This is one of the easiest and most characteristic problems encountered (Fig. 30-18). Note the nonphysiologic (extremely rapid) electrical activity that is the signature of this phenomenon. Lead failure may become apparent after shocks.⁵⁵

Figure 30-18 Electrogram from one of the early devices capable of storing EGMs, a Model V100 ICD (Ventritex/St. Jude Medical), shows nonphysiologic, rapid make-break potentials characteristic of conductor fracture.

ICD leads have a higher failure rate than pacemaker leads because they are an order of magnitude more complex. In addition, some models have especially high failure rates, including both epicardial leads⁵³ and endocardial leads.^54–56 One reported case had a 37% failure rate (confidence interval, 24%-54%) at 68.6 months; the study showed that failure may not become apparent until late follow-up, but also that a “signature” of failure is the occurrence of electrical noise artifact after shock delivery.⁵⁵ Failure is not predicted by R-wave changes or changes in pacing characteristics. However, combining the occurrence of an abnormal impedance with a counter showing R-R intervals shorter than 140 msec can identify ICD lead failure with high sensitivity and specificity.¹²

The most recent lead to manifest a higher-than-anticipated failure rate is the Medtronic 6949.⁵⁶ Conductor fracture causing make-break potentials interpreted as VF is the most frequent mode of failure for this lead, although high-voltage conductor failure has also been reported. Both inappropriate shocks and inhibition of pacing can occur. As expected, changes in lead impedance and nonphysiologic R-R intervals precede lead failure. For patients with the Medtronic 6949 lead, recommended changes to device programming improved early detection of lead failure and minimized inappropriate shocks. Early detection changes included narrowing the acceptable range for normal lead impedance and tracking the number of sensing integrity counter (SIC) events.⁵⁷ Lead impedances falling outside the established limits or too many SIC events triggered both an audible patient alarm and physician notification by the patient’s remote monitor. Because the make-break signals usually generated by a lead fracture tend to be nonsustained in character, increasing the number of R-R intervals to trigger device detection of a sustained ventricular arrhythmia serves to minimize inappropriate ICD shocks. The safety of such an extension in arrhythmia detection time is discussed later. However, as previously discussed, despite these programming changes, lead failure can occur quickly, and inappropriate ICD shocks might still be delivered.¹⁸

Supraventricular arrhythmias conducting to the ventricles above the ICD detection rate are common causes of multiple shocks.⁶¹ In this instance, device therapy does not reflect malfunction, but rather an appropriate response to tachycardia. The incidence of inappropriate shocks is higher in patients receiving an ICD for primary prevention than in those receiving secondary prevention.^62,63 In the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT), 829 patients with class II or III congestive heart failure received an ICD for primary prevention. Of those, 259 (31%) received an ICD shock, and 32% of the shocks were inappropriate. Therefore, 10% of the population receiving a prophylactic ICD received inappropriate shocks.⁶⁴

Atrial fibrillation (AF) is the most common cause of multiple shocks in patients undergoing ICD implantation for primary prevention.⁶³ AF can be recognized by its EGM characteristics; the VEGM most often has the same morphology as in sinus rhythm (or AF, if it is the baseline) and is irregularly irregular (Fig. 30-19). In contrast, polymorphic VT, although irregularly irregular, has varying EGM morphologies (Fig. 30-20).

Figure 30-19 Electrogram shows atrial fibrillation above the ventricular tachycardia detection rate leading to an ICD shock (Model V199 ICD, Ventritex/St. Jude Medical). The EGM morphology is constant (top tracing), and the notations in the marker channel below signify a 100% match to an EGM template of a signal produced by conduction from the atria to the ventricles (i.e., a “sinus beat”). Therapy was delivered because this device has a timer which, when expired, overrides the supraventricular tachycardia discriminator and releases shock therapy as a safety measure so that therapy, if withheld in error, will not be withheld indefinitely.

Figure 30-20 This stored electrogram (EGM) is from a 12-year-old boy with the familial long QT syndrome who had an ICD implanted after resuscitation from a cardiac arrest. After he collapsed in school, ICD interrogation showed sinus rhythm (note long QT even on EGM) followed by torsades de pointes ventricular tachycardia (VT), characterized by a polymorphic and constantly changing morphology, and degeneration to ventricular fibrillation (VF). The VF was terminated by a shock, which is labeled on the third tracing. In contrast to the tracing shown in Figure 30-1, there was no monomorphic VT preceding VF.

(From Moss AJ, Daubert JP: Images in clinical medicine: Internal ventricular defibrillation. N Engl J Med 342:398, 2000.)

Detection of atrial arrhythmias may be managed by both drug therapy and device programming. Beta-adrenergic blockers are a mainstay of therapy to decrease atrioventricular (AV) conduction and are antiarrhythmic themselves. Antiarrhythmic drugs may be prescribed to decrease the frequency of arrhythmia recurrence. Sotalol⁶⁵ and azimilide⁶⁶ have been efficacious in this regard, and although they avoid the organ toxicities associated with amiodarone, azimilide was never market released because it caused neutropenia. ICD programming to avoid inappropriate therapies for supraventricular arrhythmias depend in part on whether a single- or dual-lead system is present.^67–69 For single-lead systems, detection enhancements to recognize rate stability and sudden onset are useful. For dual-lead systems, these parameters are used in addition to functions that assess AV relationships.

It is controversial whether the addition of an atrial lead improves tachycardia discrimination.^70–72 Multiple reports beginning in the 1990s, when stored EGMs became available, consistently show that the most common cause of inappropriate shocks is SVT.⁶¹ Despite the presence of an atrial lead, the risk of inappropriate shocks from atrial arrhythmias remains about 20% to 30%. In the Antiarrhythmics Versus Implantable Defibrillators (AVID) study, in which single-chamber ICDs were primarily used, the incidence of shocks was 11% for AF and 9% for other SVTs.⁶² In a randomized trial that compared the incidence of inappropriate therapies from single-chamber versus dual-chamber ICDs, 21% of dual-chamber patients received inappropriate therapy despite having detection enhancements enabled. Theuns et al.⁷¹ randomly assigned patients with dual-chamber ICDs to single- or dual-chamber detection. Overall, there was no difference in atrial and ventricular arrhythmia discrimination between the two settings. Figure 30-19 shows an EGM recorded from a patient in whom AF triggered an ICD shock despite a 100% template match with a supraventricular rhythm; despite the match, the timer withholding therapy expired, overriding the discriminator.

Multiple shocks may also be a proarrhythmic response to drugs or to the ICD itself. Repetitive VT and VF can be a consequence of antiarrhythmic drug proarrhythmia. In addition, attention to the history may reveal the ingestion of a cardiac stimulant.^73,74 Proarrhythmia may also be caused by the ICD itself.^75–77 Pinski and Fahy⁷⁶ review types of device proarrhythmia (see Box 30-1). A common cause of VT induction is undersensing of an ectopic beat, followed by a paced beat and subsequent induction of the ventricular arrhythmia (Fig. 30-21). Himmrich et al.⁷⁷ studied patients with a history of tachycardia caused by right ventricular backup pacing and randomized them to no pacing or to VVI backup pacing at 60 bpm. No further pacemaker-induced tachycardias occurred in patients with pacing turned off, but did occur in five of six patients with pacemaker turned on. The induced arrhythmias were both monomorphic and polymorphic VTs.

Figure 30-21 Two examples of pacing-induced ventricular tachycardia (VT).

A, Telemetry recordings from a hospitalized patient. A premature ventricular contraction (second beat) is not sensed by the ICD; the ICD then delivers two pulses for bradycardia backup pacing, which induce VT. B, Sinus rhythm with a P wave blocked in the atrioventricular conduction system, leading to a pause in the ventricular rhythm. The next ventricular depolarization induces VT. The three tracings show EGMs from, respectively, the atrial and ventricular pace/sense channels and the high-voltage channel of a dual-chamber ICD (Guidant, Boston Scientific).

A recent development in programmability to decrease the delivery of shock therapy for VT is the option for antitachycardia pacing (ATP) for extremely rapid VTs. Previously it was believed that such an approach would not only be inefficacious but also unsafe. In a prospective randomized trial comparing ATP with simple shock therapy for fast VTs, however, Wathen et al.^78,79 showed that, for VTs in the range of 188 to 250 beats per minute, ATP was effective in terminating 81% (72% adjusted). Also, empirical ATP was equally safe and associated with improved quality of life compared with shock therapy for fast VTs.

The morbidity of inappropriate shocks, both from supraventricular arrhythmias and from nonsustained ventricular arrhythmias, led to further efforts to define the most appropriate programming for patients receiving ICDs for primary prevention. Wilkoff et al.⁸⁰ showed that empirical programming designed to reduce ICD shocks for SVT and nonsustained VTs as well as empirical ATP performed at least as well as programming tailored to the individual patient by the implanting physician, in terms of reducing unnecessary shocks and avoiding undertreatment of sustained VTs. The authors then extended these findings to a more diverse group of patients undergoing ICD implantation.⁸¹ The effectiveness of longer detection times (30 of 40 beats) was further tested in terms of avoiding unnecessary shocks, as well as the safety of such a detection time in terms of arrhythmic syncope and undertreated ventricular arrhythmias. Compared with a control group of patients with devices programmed to treat much slower arrhythmias, the longer detection times and use of ATP did reduce the number of patients receiving inappropriate therapy, but at a slight increase in the number of patients experiencing arrhythmic syncope.

These findings demonstrate that longer detection times and the use of ATP can, in primary prevention patients, safely reduce the incidence of inappropriate ICD therapy for supraventricular arrhythmias and nonsustained ventricular arrhythmias. As mentioned, these longer detection times can also be used to reduce the incidence of lead fracture associated ICD shocks. Additional trials are under way to further define the programming parameters that will reduce inappropriate ICD shocks.

A strategy of prophylactic catheter ablation was recently shown to reduce ICD therapy by 65% in a group of patients with ischemic cardiomyopathy who received an ICD after presenting with sustained ventricular arrhythmias.⁸² The generalizability of this finding to patients with nonischemic cardiomyopathy and prior sustained ventricular arrhythmias, or patients receiving ICDs for primary prevention, has not been examined.

In some situations, ICDs have been implanted in patients who failed therapy or were deemed poor candidates for medical therapy to treat sustained idiopathic VT. With the increased availability and effectiveness of catheter ablation, many of these patients can now undergo curative ablation procedures, with minimal residual risk of recurrent ventricular arrhythmias. Discontinuing ICD therapy should be discussed with the patient. In these situations the risks and benefits of header and lead extraction must be weighed against continuing therapy.

 Other Presentations

Failure to Convert Ventricular Tachycardia or Ventricular Fibrillation

Problems related to the myocardial substrate, ICD programming, or ICD function are the usual reasons for failure to convert an arrhythmia. Changes in substrate may include ischemia resulting in polymorphic ventricular arrhythmias or VF, worsened heart failure that causes new arrhythmias or conversion to incessant ones, and any combination of events that may increase the pacing, cardioversion, or defibrillation thresholds (Table 30-2).

TABLE 30-2 Nonconversion of Ventricular Tachycardia (VT) or Ventricular Fibrillation (VF)

Cause	Management
Problems with VT therapy	For both VT and VF problems:
Pacing algorithm	Change therapy prescription.
Elevated pacing/cardioversion threshold	Decrease time to therapy. Reverse polarity. Change current path. Change waveform. Change drugs that affect rate or DFT. Treat myocardial substrate.
Incessant VT
Problems with VF therapy
Elevated defibrillation threshold
Fixed voltage/pulse width Incessant VF
Faulty lead system: Dislodgment, fracture, loose connector	Revise lead system
ICD failure	Replace ICD.
Battery depletion	Replace ICD.
ICD-pacemaker interaction	Revise system or reprogram if feasible.
Abandoned electrode shunting energy	Extract extra lead.

DFT, Defibrillation threshold.

Problems with therapy most often include the programming of ineffective pacing or shock therapies. ATP may be either underaggressive or overaggressive, resulting in failure of conversion or arrhythmia acceleration.^75–77 If ventricular arrhythmias persist, they may become self-perpetuating because of ischemia and myocardial stretch exacerbated by hemodynamic compromise. These factors, as well as the arrhythmia duration itself, may also elevate the defibrillation threshold (DFT).⁸³ Other factors that may affect shock efficacy relate not only to the amount of delivered energy but also to the shock waveform. For example, efficacy may be improved by altering the tilt of the shock waveform. Evidence indicates that a 50%/50% tilt is more efficacious than the nominal 65%/65% waveform.⁸⁴ St. Jude Medical devices allow for optimization of the biphasic waveform by varying the duration of each phase of the shock to accommodate the individual patient’s DFT. These observations are not transferable to the waveforms of all manufacturers. For Medtronic devices, a randomized comparison of biphasic waveforms with 65%/65% versus 42%/42% tilt found no difference in the DFT.⁸⁵

A related factor is the position of the ICD lead in the right ventricle. Although adequate defibrillation can occur with many different lead positions, failure to convert VF can sometimes be addressed by moving the lead to a more apical position that then encompasses more of the heart in the shock vector.⁸⁶ Likewise, the efficacy of placement of a subcutaneous lead or array on lowering the DFT can be influenced by the placement of the array and its relation to the heart.⁸⁷ Lead perforation can lead to inadequate defibrillation by altering the shock vector (see Fig. 30-10).

Therapy may also fail if there is component failure of either the ICD (uncommon but reported)¹ or any of its leads. In the past, defective crystal oscillators led to rapid ventricular pacing, VF, and inability to deliver shock therapy.⁸⁸ More recently, electrical overstress (EOS) has led to ICD failure.⁸⁹ EOS is the damage to electrical components caused by high voltages or currents that arc within a pulse generator. In addition, recent occurrences of disabling of ICDs have been caused not only by electromagnetic interference but also by health care professionals unaware that magnet application might permanently inactivate tachycardia detection.⁹⁰

Although interactions between separately implanted pacemakers are less common today, in the past they did interfere with ICD function. Although it was more common for a pacemaker to inhibit detection by asynchronously pacing during VF⁶² (Fig. 30-22), pacing may also reinitiate VT or VF after ICD therapy.

Figure 30-22 Electrogram shows result of intraoperative testing of ICD implanted with unipolar pacemaker. The pacemaker was not fully inhibited during ventricular fibrillation (VF), and the large pacemaker artifacts (arrows in lead V1) were randomly delivered. The ICD failed to recognize VF because it interpreted these large artifacts as a supraventricular rhythm not in the tachycardia detection zone. Sinus rhythm was restored by an external shock (DCCV).

(From Kim SG, Furman S, Waspe LE, et al: Unipolar pacer artifacts induced failure of an automatic implantable cardioverter/defibrillator to detect ventricular fibrillation. Am J Cardiol 57:881-882, 1986.)

Antiarrhythmic drugs are often used in conjunction with ICD therapy (Box 30-2). These agents not only have beneficial effects of decreasing arrhythmia occurrence, but also may affect ICD detection and therapy (Figs. 30-23 and 30-24). Sodium channel blockers and amiodarone may increase the DFT, rendering therapy ineffective.^91–93

Box 30-2
Effects of Antiarrhythmic Drugs on ICDs

Beneficial

Reduce inappropriate shocks

Suppress sustained VT/SVT

Suppress nonsustained VT/SVT

Enhance antitachycardia pacing efficacy

Reduce symptoms by slowing VT rate

Lower DFT (e.g., dofetilide, sotalol)

Adverse

Increase DFT

Increase pacing threshold

Worsen LV function

Proarrhythmia

Increase frequency of VT and/or VF

Organize atrial fibrillation to atrial flutter with a rapid ventricular response

Sensing interference

Increase tachycardia cycle length such that VT not detected

Increase VT cycle length variability so that therapy is inappropriately withheld

Create bradycardia or conduction disturbances, leading to:

Increased pacing and earlier battery depletion

Pacemaker syndrome

Worsened CHF by right ventricular pacing-provoked LV dyssynchrony

Cost

Constitutional side effects

CHF, Congestive heart failure; DFT, defibrillation threshold; LV, left ventricular; VF, ventricular fibrillation; VT/SVT, ventricular tachycardia/supraventricular tachycardia.

Figure 30-23 This 12-lead electrocardiogram (ECG) was recorded in the early 1980s, when ICD rate criteria were not programmable. The slow ventricular tachycardia with a rate 113 beats per minute (bpm) was well below the ICD detection rate of 156 bpm. Despite symptoms, the patient drove for 4 hours to seek help. Even today, if rate criteria are not satisfied, arrhythmia detection will not occur.

Figure 30-24 Amiodarone increasing defibrillation threshold (DFT).

In the absence of amiodarone, DFT was 14 J (A). After amiodarone was started for atrial fibrillation, follow-up testing with shocks of 19.4 J (B) and 33.6 J (C) were ineffective. An external shock was required for restoration of sinus rhythm. The three tracings in all plates show, top to bottom, atrial EGM, ventricular EGM, and marker channel.

(From Rajawat YS, Patel VV, Gerstenfeld EP, et al: Advantages and pitfalls of combining device-based and pharmacologic therapies for the treatment of ventricular arrhythmias: Observations from a tertiary referral center. Pacing Clin Electrophysiol 27:1670-1681, 2004.)

Because remote monitoring has not been available until recently for ICD follow-up^94,95 and warning features (e.g., beeping by the ICD on achievement of the elective replacement indicator),⁹⁶ battery depletion may also prevent delivery of effective therapy.

 Other Issues

As in all fields of medicine, technologic advances and progress in defibrillation and pacing bring new and unanticipated problems. The use of MRI has expanded, and although its use generally is contraindicated in patients with an ICD, studies now report on its safety. A completely protected ICD has yet to be developed. Although some clinicians suggest MRI scanning may be safe with newer devices,^{109,114–116} it is probably premature to make this a general conclusion.^117,118 Not only have interactions been reported,¹²⁰ but series purporting device safety have included only small numbers of patients with relatively few devices. Indeed, specific device programming protocols designed to minimize device malfunction are used at centers evaluating the safety and efficacy of MRI scanning on ICD patients.^{109,114–116,119} To extrapolate these findings to the general population would require much greater numbers. Furthermore, even proponents of MRI safety with ICDs acknowledge that “minimal” changes in pacing threshold and battery function may occur.¹⁰⁹ Although a small change in pacing threshold or battery function may be acceptable most of the time, it can be disastrous if safety margins are small, as when batteries are near their end of life or in patients with borderline pacing thresholds. Safety concerns such as those listed must be carefully weighed against the usefulness of information obtained from the MRI scan. Interactions with new scanners having field strengths greater than 1.5 tesla remain unknown.

Rarely, ICD programmers can malfunction. They are slow, and programming is sometimes not intuitive. In the operating room, these issues are frustrating, and malfunction sometimes can be handled only by powering down and rebooting the system or obtaining a new programmer.

Current ICDs are well shielded from external shocks. Nevertheless, cardioversion or defibrillation may injure the device. Therefore, whenever external shocks are given, paddles must be placed as far from the ICD as possible. In an emergency situation, patient resuscitation obviously takes precedence over saving the device. In addition, other medical equipment may injure an ICD, specifically shock wave lithotripsy.¹²¹ The beam should always be aimed away from the ICD if possible.

Importantly, psychological issues must be considered when taking care of patients with ICDs, and psychological problems certainly fall into the realm of troubleshooting.^122,123 With patients who have received shocks, imagined shocks, and “phantom shocks,” the shock may cause concern for them and their families. Explanation is usually satisfactory, but sometimes fear of shocks can be debilitating, similar to a posttraumatic stress disorder. Indeed, patients receiving ICD shocks can have a worse prognosis than ICD patients who have not received an ICD shock.^25–27 Besides counseling, drug therapy may be warranted to decrease anxiety and depression. Further investigation is needed to determine ways to reduce the morbidity associated with ICD shocks. The combination of a benzodiazepine and serotonin reuptake inhibitor is especially useful.

 Conclusion

Implantable cardioverter-defibrillator troubleshooting requires systematic analysis of evidence based on history, physical examination, and analysis of retrieved ICD information. Problems may be approached from the perspective of multiple shocks or other clinical presentation (failure to detect, failure to convert, problems with pacing) or as those of the system components (ICD battery, programming, integrity, and lead integrity and function). With the remarkable diagnostic capabilities of current devices, the reason for an ICD system problem can usually be identified and, in the absence of component failure, remedied by reprogramming.