Sensing and Detection With Cardiac Implantable Electronic Devices


In cardiac implantable electronic devices (CIEDs), sensing is the process for identifying the electrical signals (electrograms, EGMs) that correspond to individual atrial or ventricular depolarization. Detection is the process by which device algorithms classify a sequence of sensed signals to determine the cardiac rhythm. Thus basic sensing precedes detection.

Characteristics of Cardiac Electrograms

Electrograms (EGMs) are the graphical display of temporal changes in electric potentials recorded between two points in space. At each point, potential is determined by the superposition of potentials generated by cardiac and noncardiac sources, weighted by the magnitude of the source potentials and the distance from the source to the point. These superimposed signals determine EGM amplitude, duration, and frequency content. Table 4-1 summarizes qualitative determinants of EGM characteristics.

TABLE 4-1
Qualitative Determinants of EGM Characteristics
Effect on EGM
Electrode-Related
Interelectrode distance Wider separation: ↑ EGM duration
Electrode surface area Larger area: ↑ EGM duration
Intracardiac vs. subcutaneous Intracardiac: shorter duration; amplitude greater by ~10×
Postpacing polarization Afterpotentials
Near-Field Vs. Far-Field Near-field: ↓ duration, ↑ amplitude
Biological
Signal source Amplitude: atrial < ventricular
Acute vs. chronic Acute current of injury
Posture and respiratory effects EGMs from extracardiac electrodes rotate relative to the heart.
Cardiac rhythm Fibrillation EGMs show beat-beat variations of amplitude and frequency content
Hyperkalemia ↑ EGM duration, ↑ T-wave amplitude
Other Metabolic Effects
Myocardial disease ↑ EGM duration
Postshock electroporation ↓ EGM amplitude
EGM, Electrogram.

Unipolar Versus Bipolar Electrograms

Recording an EGM requires a pair of electrodes ( Fig. 4-1 ). The terms unipolar and bipolar commonly refer to the number of electrodes in the recording electrode pair that are in direct contact with myocardium. A unipolar EGM is measured between an electrode in contact with myocardium and a remote “indifferent” electrode. The location of the remote electrode has little effect on the cardiac component of the EGM, but it may record noncardiac potentials, such as pectoral myopotentials. The intrinsic deflection of a unipolar EGM is defined as the rapid downward deflection from the largest positive peak. It times with the wavefront of cardiac depolarization passing through the myocardium adjacent to the cardiac electrode. In contrast, the peak of the maximum deflection of a unipolar surface electrocardiogram (ECG) lead is referred to as the “intrinsicoid” deflection because it not in direct contact with the myocardium.

Figure 4-1, Concept Drawing of Spatial and Temporal Relationships for Unipolar and Bipolar Endocardial Electrogram.

A bipolar EGM is the superimposed difference between unipolar EGMs measured at two different electrodes. Usually, closely spaced bipolar EGMs recorded from electrodes in direct contact with normal ventricular myocardium are dominated by signals that originate in local myocardium. Noncardiac signals that originate remotely from closely spaced bipolar electrodes appear nearly identical on both electrodes and are thus eliminated from the EGM by differencing. For closely spaced bipolar EGMs, the intrinsic deflection is the largest and steepest positive or negative deflection. Its peak approximates the time at which the wavefront passes the midpoint between the electrodes.

Electrogram Descriptors

The most commonly used EGM descriptors relate to amplitude and frequency content ( Fig. 4-2 ). Automated CIED measurements of EGM amplitude and operator measurements may differ because of differences in methods and signal filtering. Slew rate, a term borrowed from electrical engineering, is the most commonly used descriptor of frequency content. It is defined as the maximum of the first time derivative of the EGM signal (maximum rate of change) in units of volts per second. The relationship between slew rate and frequency is defined for repetitive signals that have a single frequency and uniform amplitude. For a sinusoidal signal, the maximum slew rate is 2π fA, where f is the frequency and A the amplitude. EGMs are neither sinusoidal nor stationary signals, so a true representation of the signal in the frequency domain must show the evolution of frequency content over time. However, in clinical application EGM amplitude and maximum slew rate provide workable surrogates for signal frequency that can be used to create selective filters for separating the signal of interest, the cardiac activation, from unwanted signals or noise. Table 4-2 summarizes amplitude and frequency characteristics of selected CIED EGMs in baseline rhythm and fibrillation. Overlapping frequency content of signals requires tradeoffs in filter design. EGM temporal duration (“width”) is another surrogate for signal frequency and determines the minimum blanking period required to prevent double-counting of cardiac depolarizations.

Figure 4-2, Electrogram Descriptors.

TABLE 4-2
Approximate Range of Amplitude and Frequency Content of Various CIED EGMs *
Amplitude (mV) Frequency Content (Hz)
Intracardiac Bipolar EGMs
RV bipolar-sinus 2-30 3-75
RV bipolar-VF 2-10 3-15
RA-sinus 0.5-5 3-100
RA-AF 0.1-3 3-10
Far-Field EGMs
RV Coil-Can 2-15 3-50
Subcutaneous closely spaced (38 mm, ILR) sinus 0.1-2 4-60
Subcutaneous widely spaced (10-25 cm, ICD) sinus 0.6-3 3-50
Subcutaneous widely spaced (10-25 cm, ICD) VF 0.2-2 3-10
AF, Atrial fibrillation; EGM, electrogram; ICD, implantable cardioverter-defibrillator; ILR, implantable loop recorder; RV, right ventricular; VF, ventricular fibrillation.

* Ranges are approximate.

Types of Cardiac Implantable Electronic Device Electrograms

Figure 4-3 illustrates EGMs recorded from transvenous CIED leads. In pacemakers unipolar sensing EGMs are recorded between a small tip electrode in the heart and a large remote (indifferent) electrode, typically the metal housing of the pulse generator (the “Can”). Unipolar EGMs include both near-field and far-field components. The location of the remote electrode has little effect on the cardiac component of the EGM, but it may record noncardiac potentials such as pectoral myopotentials.

Figure 4-3, Ventricular Electrograms (EGMs) Recorded From Different Electrode Configurations in the Same Patient.

Bipolar sensing EGMs are recorded between two intracardiac electrodes. Usually, increasing interelectrode separation increases the amplitude of the induced cardiac potential, but also increases the amplitude of interfering signals ( Fig. E4-1 ). In conventional transvenous pacemakers and some ICDs, the bipolar sensing EGM records a local (near-field) ventricular signal between the tip electrode and the adjacent ring electrode. In implantable cardioverter-defibrillators (ICDs), this EGM is referred to as a dedicated bipolar EGM because the ring electrode is dedicated to pace-sense functions, not defibrillation. EGMs recorded between the tip and the right ventricular (RV) coil are referred to as integrated-bipolar EGMs because the RV coil integrates both pace-sense and defibrillation functions. Epicardial sensing systems may be either unipolar or bipolar.

Figure E4-1, Effects of Electrode Configurations on Atrial Endocardial Electrogram.

In ICDs, the shock EGM records a more global (far-field) signal between widely separated, high-voltage electrodes, usually RV coil and Can. Dual-coil defibrillation leads permit recording additional far-field signals, including the shock EGM between RV and superior vena cava (SVC) coils, and the “Leadless ECG” between the Can and SVC coil ( Fig. E4-2 ). The latter often displays a diagnostic quality P wave in a single-chamber ICD. In some ICDs with atrial leads, the leadless ECG may be recorded between the atrial tip and Can. Most cardiac resynchronization therapy (CRT) devices provide a left ventricular (LV) EGM.

Figure E4-2, Electrograms Recorded From Shock Electrodes on Dual Coil Leads.

Near-Field Versus Far-Field Electrograms

In physics the terms near-field and far-field distinguish the properties of electromagnetic fields measured close to or remote from their source, and the potentials induced on electrodes differ between the near field and far field. By analogy, near-field EGMs are recorded from closely spaced bipoles in contact with the source myocardium, whereas far-field EGMs are recorded remotely from their source. Far-field EGMs include ICD shock EGMs, subcutaneous EGMs, and EGMs from remote sources recorded by electrodes in contact with myocardium. The most common example of the last type of far-field EGM is the far-field R wave recorded from atrial bipolar electrodes. In general, near-field EGMs record local myocardial signals, whereas far-field EGMs have a wider “field of view,” with some recording global cardiac signals. Near-field EGMs have higher frequency than far-field EGMs, both because they have greater amplitude as a result of proximity to the source, and because far-field signals superpose multiple out-of-phase signals.

Thus far-field CIED EGMs recorded between two subcutaneous electrodes have a larger field of view in comparison with near-field intracardiac near-field EGMs, but lower amplitude and frequency content ( Fig. 4-4 ). They are thus more similar to surface ECG signals than intracardiac EGMs. ICD shock EGMs are often referred to as far-field EGMs. They are recorded between two large electrodes, one intracardiac and one extracardiac. These shock EGMs have a wide field of view and amplitude closer to near-field EGMs than subcutaneous EGMs (see Fig. 4-4 ).

Figure 4-4, Near-Field Versus Far-Field Electrograms.

Determinants of Electrogram Characteristics

Table 4-1 summarizes the most important electrode-related and biological determinants of EGM characteristics. EGM duration is increased by greater interelectrode distance (see Fig. E4-1 ) or greater surface area because of an increased number of sources being superimposed. Local intracardiac bipolar EGMs have shorter durations and approximately 10 times greater amplitude than more global subcutaneous EGMs as a result of both the distance from the source and superimposing out of phase sources.

Acute EGMs from active-fixation leads display a current of injury that resolves over a period of hours ( Fig. E4-3 ). Modern steroid-eluting electrodes reduce inflammation and edema at the electrode-myocardial interface sufficiently that chronic EGM amplitudes usually are reduced by less than 10% compared with acute amplitudes. Postural changes have clinically significant effects on some far-field EGMs, including ICD shock EGMs used to analyze morphology, and subcutaneous ICD EGMs, but not on intracardiac EGMs during exercise.

Figure E4-3, Acute Current of Injury at Implantation.

Drugs, electrolytes, or fibrosis at the electrode-myocardial interface may slow conduction, reducing EGM amplitude and slew rate, causing undersensing; they may also prolong EGM sufficiently that the EGM is double-counted. Clinically, the most important electrolyte abnormality is hyperkalemia, which prolongs EGM duration and may increase T-wave amplitudes.

The most profound rhythm-related effects occur during atrial or ventricular fibrillation ( Fig. 4-5 ), reducing both mean EGM amplitude and slew rate, while increasing EGM variability. Other effects are more variable. Premature ventricular complexes (PVCs) alter the activation wavefront and may alter recorded EGM, especially on closely spaced bipoles. There are differences in amplitude and frequency content of bipolar atrial EGMs between sinus and retrograde conduction, but to date these differences have not permitted reliable, device-based discrimination.

Figure 4-5, Effect of Filtering on Implantable Cardioverter-Defibrillator Electrograms.

Sensing Basics

Basic sensing determines the timing of depolarizations using filtering, amplitude thresholds, and blanking periods. Accurate sensing results in one sensed event for each cardiac depolarization. Failure to sense a depolarization (undersensing) occurs if the depolarization signal has insufficient amplitude or frequency content to be recognized as a sensed event. Oversensing occurs when signals that do not reflect local myocardial depolarization are sensed. Markers telemetered from the CIED to the programmer indicate the timing of sensing EGMs that reach the amplitude threshold for a sensed event. The marker channel plays a critical role in identifying both oversensing and undersensing.

Sensing Circuitry

Figure 4-6 shows the primary functional operations of sensing systems. The raw signal passes from the leads to the connector, through hermetic feedthroughs with high-frequency filters and high-voltage protection circuitry, before reaching the sensing amplifier.

Figure 4-6, Functional Block Diagram of Implantable Cardioverter-Defibrillator Sense Amplifier.

Amplification and Sampling

Modern CIEDs use digital circuitry because they use less power than their analog counterparts. Sampling is the process of converting an analog EGM signal into a digital sequence of sampled voltages. It must be performed at ≥2 times the source's signal frequency to reproduce the source's information content completely (Nyquist theorem). The fidelity of the sampled signal depends on sampling rate, sampling precision (voltage increment of the least significant bit), and dynamic range of the amplifier. The filtered signal is then amplified. For the true sampling precision to equal the number of bits stored, amplifier noise must be less than the sampling precision. Otherwise, the least significant bits of the sampled signal will be altered by amplifier noise, and the true sampling precision will be less than implied by the number of bits stored. Higher sampling rates (>400 Hz), precision (<10 µV) and dynamic range (>12 bits) may require more power, which is at a premium in implantable devices. Efficient circuit design can mitigate higher current requirements, but compromises in performance are still required to maintain battery longevity. To date, no CIED has successfully used EGM characteristics to discriminate far-field R waves or antegrade versus retrograde atrial EGMs, in part because sufficient sampling rate, precision, and amplifier dynamic range are not practical.

ICD sampling circuitry reproduces cardiac signals faithfully because the low pass filter is at a high enough frequency to include almost all the signal content (39-80 Hz, see Table 4-2 ) and sample above the Nyquist rate (128-400 Hz per channel). However, 128-Hz sampling superimposes a periodic 8-Hz distortion (beat frequency) on 60-Hz signals such as electromagnetic interference (EMI) because the signal is sampled only slightly faster than the Nyquist rate, and provides only slightly more than two samples per cycle ( Fig. E4-4 ). This 8-Hz “signature” uniquely identifies 60-Hz EMI in devices with 128-Hz sampling.

Figure E4-4, Distorted Appearance of a Telemetered Electromagnetic Interference Signal Due to Sampling.

Filtering and Rectification

Once the signal is in the digital domain, it is band-pass filtered to remove baseline wander and select certain ranges of frequencies depending on the application (see Table 4-2 ). For sensing, the key is to filter in such a way as to reduce unwanted signals such as T waves, while retaining as much of the desired signal, the R wave. In ventricular fibrillation (VF), R waves have amplitude and frequency characteristics that overlap with those of T waves ( Fig. 4-7 ). It is essential not to filter out too much of the VF signal, even though the result is occasional T-wave oversensing. When band-pass filtering alone cannot isolate the signal of interest, additional processing such as morphology matching or time-frequency pattern recognition can be applied successfully as described below.

Figure 4-7, Signal Amplitude Versus Frequency.

After the filtered EGM signal has been amplified, it is rectified so that all signals are represented by their absolute magnitudes. This removes polarity information but allows sensing to be accomplished with a single-sided comparison. Sensing occurs when the amplitude of the rectified EGM exceeds the threshold voltage (see Fig. 4-6 ).

Electrogram Amplitude: Effect of Measuring Methods and Signal Processing

It is important to understand how differences in measurement processes can affect measured amplitude to distinguish these from biological changes caused by progression of disease. Figure 4-8 illustrates the primary variables: base-peak versus peak-to-peak, signal filtering, and measurement point. Historically, operators measured peak-peak amplitude “manually” at the intrinsic deflection of unfiltered signals. Usually automated CIED measurements are performed differently. Typically CIEDs measure the base-peak filtered and rectified signal. They measure the maximum amplitude in a short time window (50-120 msec) after the sensed event. Thus if there is a long period of low amplitude signal at the onset the EGM, changing the sensing threshold can alter the measured amplitude. Traditional operator-measured EGM amplitude usually exceeds device-measured EGM amplitude, but automatic measurements usually provide a better indicator of device sensing performance.

Figure 4-8, Effect of Measuring Method on Reported Automated R-Wave Amplitude.

Blanking and Refractory Periods

Blanking Periods

Blanking periods ( Fig. 4-9 ) are designed to ensure that a single event is sensed only once rather than multiple times, by inhibiting sensing after sensed depolarizations, pacing pulses, or shocks. In early CIEDs, blanking was a hardware feature that turned off (“blanked”) the sense amplifier. In modern CIEDs, most blanking is performed using software. Blanking must be longer after paced events than sensed events because there is an activation delay from the paced event to onset of global cardiac depolarization; in contrast, cardiac depolarization begins before sensed events. Some CIEDs have partial blanking periods on the atrial channel, an interval following the blanking period in which the sensing threshold is increased. See subsequent discussion of atrial sensing and blanking.

Figure 4-9, Blanking and Refractory Periods.

Multichamber CIEDs have cross-chamber blanking periods. The ventricular blanking period after atrial-paced events is designed to prevent oversensing that causes crosstalk inhibition of ventricular bradycardia pacing (see Chapter 36 ) and may cause inappropriate detection of ventricular tachycardia/ventricular fibrillation (VT/VF). The postventricular atrial blanking period (PVAB) after ventricular paced or sensed events is designed to prevent cyclical oversensing of far-field R waves that results in inappropriate mode switching and incorrect classification of the atrial rhythm as atrial tachycardia or atrial fibrillation (AF). Unlike same-chamber blanking periods that correspond to physiologic refractory periods, cross-chamber blanking periods have no physiologic correlate. As described below, they may cause functional undersensing of cardiac EGMs in the blanked chamber.

Refractory Periods

Some manufacturers use the term “refractory period” instead of blanking period. Others reserve “refractory period” for an interval following the blanking period in which the sense amplifier is enabled but bradycardia pacemaker functions do not respond to sensed events as they do during typical alert periods (see Fig. 4-9 ).

Refractory-sensed events are not used for basic pacemaker timing cycles (see Chapter 36 ), but they are used for atrial tachyarrhythmia detection and some advanced pacemaker functions. Marker channels denote events sensed during refractory periods. Software blanking may blur the traditional distinction between blanking and refractory periods because software-blanked atrial events are also used for atrial tachyarrhythmia sensing. Software blanking may blur the traditional distinction between blanking and refractory periods as software-blanked and refractory atrial events are both used for atrial tachyarrhythmia detection, but not for pacemaker timing. Marker channels apply different designations to distinguish events during sensing alert periods, refractory periods, and atrial software blanking periods.

Considerations for Different Cardiac Implantable Electronic Devices

Early pacemakers had long same-chamber blanking periods to prevent oversensing from inhibiting pacing. However, CIEDs require short blanking periods to sense all activations during tachyarrhythmias. ICDs have short ventricular blanking periods to sense R-R intervals reliably in VF and modern multichamber pacemakers, and ICDs have short atrial blanking periods to sense short A-A intervals in AF. Some ICDs have longer partial blanking periods in the atrium to prevent far-field ventricular EGMs from interrupting normal pacemaker function. Special considerations that apply to cardiac resynchronization therapy-defibrillators/pacemakers (CRT-D/Ps) are covered in Chapters 18 and 36 .

Sensing Thresholds

Sensing in ICDs and pacemakers share many features, but there are two major differences. First, ICDs need reliable sensing and detection during VF, but pacemakers do not. Second, pacemakers may use unipolar or bipolar sensing, whereas ICDs always use bipolar sensing. These considerations combined with differences in filtering and design philosophies dictate differences in nominal sensitivities among CIEDs ( Table 4-3 ).

TABLE 4-3
Nominal Dynamic Ventricular Sensing Thresholds After Sensed R Waves
Interval From Sensed Event to Decay Onset (ms) Method of Dynamic Sensing Adjustment Initial Threshold (mV) Programmable Minimum Threshold (mV)
Biotronik 110 Discrete steps 50% of peak R wave * 0.8
Boston Scientific 135 Discrete steps 75% of weighted average of peak R waves 0.6
Medtronic 120 Exponential decay 10 × minimum sensitivity up to 3 mV 0.3
St. Jude Medical 125 Linear decay 50% of peak R wave 0.5
Sorin 125 Discrete steps 3 mV for peak R waves ≥ 6 mV; 25-49% of peak R wave <6 mV 0.4

* Programmable to 75% using Enhanced T-wave Suppression.

Value using Low Frequency Attenuation filter, which is nominally ON; without this filter nominal is 0.3 mV.

The coefficient varies linearly from 25% to 49% for R waves between 1.6 and 6.0 mV; it is 25% for R waves <1.6 mV.

Fixed Sensing Thresholds

Early pacemakers used “fixed” sensing thresholds programmable to a constant value. Typically, fixed ventricular sensing thresholds are set higher (less sensitive) than atrial thresholds because of the differences in R-wave versus P-wave amplitude; unipolar sensing thresholds are set higher than bipolar sensing thresholds to reduce oversensing of far-field cardiac and extracardiac signals. Some pacemakers automatically adjust fixed sensing thresholds periodically to adapt to changes in EGM amplitude over time.

Dynamic Sensing Thresholds

High sensitivity is required to ensure reliable sensing of low-amplitude ventricular EGMs in VF and atrial EGMs in VF. Because continuous high sensitivity results in undesirable oversensing, CIEDs that sense AF or VF use feedback mechanisms based on P- or R-wave amplitude that adjust the sensing threshold dynamically ( Fig. 4-10 ). This is referred to by various terms, including automatic adjustment of sensitivity, autoadjusting sensitivity, and automatic sensitivity control. Some early ICDs used automatic step adjustments of sensing-amplifier gain (automatic gain control) as a primary means for avoiding T-wave oversensing and ensuring detection of low-amplitude VF EGMs. This resulted in sensing errors when EGM amplitude changed abruptly and has been abandoned.

Figure 4-10, Dynamic Adjustment of Sensitivity.

Presently, all ICDs and many pacemakers adjust sensitivity “dynamically” in relation to the amplitude of each sensed intrinsic depolarization or pacing pulse. At the end of the blanking period after each sensed event, the sensing threshold is set to a high value. It then decreases with time until a minimum value is reached. In comparison with a fixed sensing threshold, automatic adjustment of sensitivity increases the likelihood of sensing low-amplitude and varying EGMs during AF or VF, while reducing the likelihood of far-field R-wave oversensing in the atrium or T-wave oversensing in the ventricle. After either sensing or pacing, ICDs are most vulnerable to oversensing of low-amplitude signals late in the diastole of bradycardia, when the amplifier sensitivity is maximal for an extended period.

This section considers dynamic sensing in the ventricle. The same considerations apply to dynamic sensing in the atrium, except that in the atrium minimum values of sensitivity are lower and the rate at which the sensing threshold decreases over time is faster.

Postsensing automatic adjustment of ventricular sensitivity.

Details differ among manufacturers (see Table 4-3 ) ( Fig. E4-5 , Fig. E4-6 ), but they all have these common features: (1) ICDs blank the sense amplifier postsense (nominal values 110-135 msec) to prevent R-wave double-counting. (2) The sensing threshold is highest at the end of blanking period (minimum sensitivity). In general, the highest value of the sensing threshold is greater if the R-wave amplitude is greater; some manufacturers limit this value to reduce the likelihood of undersensing VF with highly variable amplitude ( Fig. E4-5 ). No manufacturer displays this value on the programmer. (3) The sensing threshold decreases to a minimum value that determines the smallest EGM amplitude that can be sensed, corresponding to the programmed value of sensitivity (nominal values 0.3-0.8 mV). Depending on the manufacturer, this threshold decrease may occur as a series of step functions, a linear decay, or an exponential decay. (4) The minimum value of the sensing threshold is programmable; in contemporary ICDs, it remains fixed, independent of R-wave amplitude.

Figure E4-5, Dynamic Sensitivity as Implemented by Different Manufacturers.

Figure E4-6, Dynamic Sensitivity.

Programmable and nonprogrammable differences among manufacturers affect the trade-off between preventing T-wave oversensing and ensuring sensing of VF. A combination of a high value for high-pass filter and high starting point for dynamic sensitivity can reduce T-wave oversensing, but rare cases of serious VF undersensing have been reported caused by failure of sensitivity to adjust when rapid changes in EGM amplitude occur. Figure 4-11 shows a clinical example. It is prudent to test sensing of induced VF after programming that increases the risk of VF undersensing if a sensing safety margin was not determined at implant.

Figure 4-11, Dynamic Sensing During Ventricular Fibrillation.

Postpacing automatic adjustment of sensitivity.

Automatic adjustment of sensitivity operates differently after pacing and sensing in ICDs. All manufacturers apply longer postpacing blanking periods than the postsensing blanking periods. The highest value of the sensing threshold at the end of the postpacing blanking period varies among manufacturers. It may be set in relation to an average of recent sensed R waves or to a low value to prevent pacing during VF from causing undersensing (see Fig. E4-5 ). The programmable minimum value of sensitivity is the same as after sensed events.

Postshock Sensing

Postshock sensing is critical for redetection of VF after unsuccessful shocks and for accurate detection of episode termination. Postshock polarization may take seconds to minutes to decay off the electrodes and thus interfere with postshock sensing. Furthermore, the strong electric fields of shocks cause electroporation, the creation of microscopic holes in the cardiac cell membranes. This has been proposed as an additional mechanism for postshock distortion of EGMs recorded from high-voltage electrodes. Because EGMs of dedicated bipolar sensing electrodes are minimally affected by shocks, they became standard for early epicardial ICDs. For transvenous ICDs, postshock sensing recovers more rapidly with dedicated bipolar sensing configurations than with integrated bipolar sensing. This is a minor issue for present integrated bipolar leads with a pacing tip electrode to distal coil spacing of about 12 mm or greater. However, it has reemerged as a potential problem for subcutaneous ICDs that sense using shock electrodes (see below). Transvenous ICDs use the shock EGM to discriminate VT from supraventricular tachycardia (SVT) by ventricular EGM morphology. These morphology discrimination algorithms can be distorted for several minutes postshock.

Analyzing and “Sensing” Other Implantable Cardioverter-Defibrillator Ventricular Electrograms

CIEDs sense other EGMs to perform enhanced sensing functions (ICDs), discriminate VT from SVT (ICDs) and regulate left ventricular (LV) pacing (CRT-D/Ps). Transvenous ICDs do not use the global shock EGM for rate counting because dedicated or integrated bipolar EGMs are less susceptible to extracardiac signals. However, as described subsequently, transvenous ICDs utilize the shock EGM for two purposes: SVT-VT discrimination and withholding inappropriate shocks for oversensing. In contrast to their continuous monitoring of the sensing channel, ICDs monitor the shock channel only as needed to reduce current drain from the shock-channel amplifier.

Ventricular Oversensing

By inhibiting pacing, ventricular oversensing presents as pauses during paced rhythms in pacemaker-dependent patients or loss of biventricular pacing in CRT patients. Oversensing is a greater problem for ICDs than pacemakers because ICDs operate with shorter blanking periods and higher sensitivity to sense VF. In patients with ICDs, oversensing usually presents as alerts or inappropriate detection of VT/VF, often resulting in inappropriate therapy. In pacemaker-dependent patients with ICDs, oversensing may present as syncope resulting from the inhibition of pacing followed by an inappropriate shock. We focus on oversensing in ICDs, but most considerations also apply to pacemakers. Our emphasis is on EGM patterns, clinical context, prevention, and mitigation. See Chapter 38 and recent review for troubleshooting.

Recognition and Classification

Recognition

Oversensing is diagnosed if more than one ventricular sensing marker occurs within one true cardiac cycle. Oversensing of noncardiac signals is diagnosed by the superimposition of these signals on true ventricular EGMs. The diagnosis is most certain when independent, true ventricular EGMs are identified. However, in patients without an intrinsic ventricular rhythm, oversensing inhibits bradycardia pacing, so the absence of true ventricular EGMs or pacing markers complicates the diagnosis. In these patients, oversensing may be confirmed if the oversensed signal is not recorded on the shock channel, even if analysis of the sensing channel alone does not permit definitive differentiation of VF from oversensing.

Classification: Cyclical Versus Noncyclical Oversensing

Oversensing can be classified by EGM morphology, temporal pattern (cyclical versus noncyclical), source type (physiologic versus nonphysiologic), and source location (intracardiac versus extracardiac) ( Fig. 4-12 ). Specific sources generate oversensed signals with characteristic morphological features that differ from true cardiac EGMs in frequency content and amplitude.

Figure 4-12, Classification and common causes of ventricular oversensing in implantable cardioverter-defibrillators. Oversensing atrial electrograms (EGMs) in atrial fibrillation (AF) (e.g., ventricular lead dislodged to atrium) also causes noncyclical oversensing. EMI, Electromagnetic interference.

It is useful to classify the temporal pattern of oversensed signals as varying consistently with the ventricular cycle (cyclical) or independent of the ventricular cycle (noncyclical). A cyclical pattern indicates an intracardiac source of oversensing, but a noncyclical pattern may occur with extracardiac or intracardiac sources. Oversensing of intracardiac physiologic signals (R-wave double-counting or P/T-wave oversensing) usually produces a characteristic pattern of one oversensed event for each true ventricular cycle. Cyclical nonphysiologic, intracardiac signals may present as multiple oversensed events in each ventricular cycle.

In contrast, extracardiac signals (physiologic or nonphysiologic) always produce noncyclical oversensing, with superposed extraneous signals that are dissociated from true ventricular EGMs, analogous to the relationship between ECG artifacts and the cardiac rhythm.

Oversensing Physiologic Intracardiac Signals

Consistent oversensing usually produces characteristic EGM patterns, but atypical patterns also occur, including those caused by intermittent oversensing.

P-Wave Oversensing

Recording an atrial EGM facilitates recognition of P-wave oversensing. During a stable, 1 : 1 rhythm, the device-detected “R-R” pattern consists of alternating sensed P-R and R-P intervals ( Fig. 4-13 ). If first-degree AV block is present, these intervals may be approximately equal. In AF, ventricular oversensing of atrial EGMs may result in multiple oversensed physiologic signals for each true cardiac cycle, producing an interval plot with short, irregular intervals, characteristic of nonphysiologic oversensing.

Figure 4-13, Spectrum of Ventricular P-Wave Oversensing.

Oversensing of spontaneous P waves is rare in adults because the ventricular sensing bipole is usually far from the atrium. Thus early after implant, P-wave oversensing usually indicates RV lead dislodgement to a position closer to the atrium/tricuspid valve. However, P-wave oversensing may occur in children with small right ventricles and in adults with integrated-bipolar sensing. It must be distinguished from atypical, end-diastolic cyclical oversensing in fractures of the cable to the ring electrode (see below).

P-wave oversensing may result in inappropriate detection of VT if the atrial rhythm is a tachycardia and in inappropriate detection of VF if the atrial rhythm is AF (see Fig. 4-13 ). The first approach to significant P-wave oversensing is to reduce the programmed ventricular sensitivity. If this is unsuccessful or sensing of VF is not reliable at the reduced sensitivity, the ventricular lead should be revised. As a temporary measure, P-wave oversensing in sinus rhythm may be mitigated by forced atrial pacing.

Ventricular oversensing of atrial paced events (crosstalk) is addressed in Chapter 36 .

R-Wave Double-Counting

R-wave double-counting occurs if the duration of the sensed EGM exceeds the ventricular blanking period. Consistent double counting results in an alternation of ventricular cycle lengths ( Fig. 4-14 ). The expected sensed event early in the R wave is followed by a second event as soon as the ventricular blanking period ends. The interval between these two events is always short enough to be in the VF zone. The duration of the device-detected interval that begins with the second event approximates the difference between the spontaneous ventricular cycle length and the blanking period of the ICD. This produces a characteristic “railroad track” pattern on a plot of stored ventricular intervals. Of course, any cause of alternating cycle lengths also produces this pattern ( Fig. 4-15 ).

Figure 4-14, R-Wave Double-Counting.

Figure 4-15, Railroad Track Patterns on Plot of Stored Ventricular Intervals.

R-wave double-counting often results from local ventricular conduction delay. It is most common when ICDs with short ventricular blanking periods ≤120 msec (Biotronik and Medtronic) are connected to integrated-bipolar leads. For integrated-bipolar sensing we suggest increasing programmable blanking periods to >120 msec. R-wave double-counting may be precipitated by reversible conduction block caused by hyperkalemia or sodium channel blocking antiarrhythmic drugs. Conduction delays during PVCs or VT/VF may facilitate R-wave double-counting (see Fig. 4-14 ). Most such double counting does not require intervention because it is infrequent (PVCs), occurs only intermittently during VT, or occurs during VF. Rarely, slow or self-terminating tachycardias may receive unnecessary therapy. The primary troubleshooting intervention is to increase the ventricular blanking period. In CRT-Ds, loss of RV capture may present as R-wave double-counting (see Fig. 4-14 ). The ICD counts both the paced ventricular event and the wavefront conducted from LV capture to the RV sensing bipole, if the conduction delay exceeds the ventricular blanking period.

T-Wave Oversensing in Conducted Rhythms

The root cause of T-wave oversensing is the requirement that ICDs reliably sense VF, which is characterized by R-R intervals shorter than the typical Q-T interval in addition to some EGMs with low amplitudes and slew rates. In contrast to the alternating device-measured “R-R” intervals typical of double-counted R waves, the hallmark of T-wave oversensing is alternating sensing EGM morphologies caused by alternation of EGM frequency content ( Fig. 4-16 ). Higher-frequency R waves alternate with the lower-frequency T waves. This may be easier to appreciate on wide-band EGMs than narrow-band EGMs (see Fig. 4-5 ). Device-detected “R-R” interval alternation (R-T versus T-R) is subtle if the Q-T interval is short or the sinus rate is fast; occasionally, it is absent. A simultaneous shock EGM confirms that alternate low-frequency EGMs represent T waves. Bigeminal PVCs during sinus tachycardia or bidirectional catecholaminergic VT may produce a similar pattern on the filtered sensing EGM, but each R wave has a corresponding T wave on the shock EGM. Alternation of EGM amplitude without alternation in morphology is rare and suggests true tachycardia.

Figure 4-16, T-Wave Oversensing.

Oversensing of spontaneous T waves may be dichotomized by the amplitude of the corresponding R waves (see Fig. 4-16 ) because the therapeutic approaches often differ : If the R wave is large and T wave just above the sensing threshold, some T-wave oversensing can be resolved by reducing programmed sensitivity; with small R waves, resolution usually requires one of the alternative approaches discussed below. T-wave oversensing with large R waves is caused by an absolute increase in T-wave amplitude. Clinical correlates include pediatric patients, hypertrophic cardiomyopathy, long QT syndrome, short QT syndrome, hyperkalemia, and rarely, other drug and metabolic abnormalities. In contrast, in the setting of small R waves, oversensed T waves usually have normal amplitude, such as in Brugada syndrome, or advanced myocardial disease. The root cause relates to a fundamental limitation of automatically adjusted sensitivity, which links the initial value of sensitivity to the amplitude of the preceding R wave (see Fig. E4-5A ). Exercise-induced T-wave oversensing may be caused by increase in absolute T-wave amplitude, decrease in R-wave amplitude, or both.

A limitation of dichotomizing T-wave oversensing by R-wave amplitude is that large variations in R-wave amplitude are common in dedicated-bipolar leads. These variations are often abrupt and transient, rather than gradual and progressive. Presently, most cases have no identifiable clinical correlate and thus cannot be predicted, although one case report attributed temporal variation to an inflammatory process (sarcoidosis).

T-Wave Oversensing: Prevention and Remediation

Because T-wave oversensing usually is unpredictable, features that mitigate it should be enabled proactively at implant and operate continuously, providing they do not cause undersensing in VF. Some mitigations may increase undersensing in VF. These should be implemented reactively, after T-wave oversensing has occurred.

Programming sensing parameters.

Often the filtered and rectified amplitude of oversensed T waves barely exceeds the sensing threshold. For ICDs with highly sensitive nominal minimum sensitivity (e.g., ≤0.4 mV), the simplest approach is reprogramming to a less sensitive setting. As noted previously, differences in starting amplitude and time course for increasing sensitivity after a sensed event probably influence the likelihood of T-wave oversensing and the inverse likelihood of VF undersensing. St. Jude Medical offers programmable “Threshold Start” and “Decay Delay.”

Device-based options.

Other approaches ( Fig. 4-17 ) described in a recent review may reduce T-wave oversensing. Biotronik offers programmable rectification options. A higher, nominal high-pass filter (e.g., 20 Hz versus 10 to 15 Hz, St. Jude Medical, and Boston Scientific) reduces T-wave oversensing. Theoretically, a higher high-pass filter might increase undersensing in VF; we are not aware of published data on time to detect VF with different high-pass filters.

Figure 4-17, Features to Minimize T-Wave Oversensing.

If the sensing bipole is programmable, changing from dedicated to integrated-bipolar may prevent T-wave oversensing. SVT morphology discriminators or algorithms that withhold shocks for oversensing related to lead failure do not prevent T-wave oversensing, but they may withhold inappropriate therapies. Algorithms that identify T waves based on frequency content or morphology are discussed under “Enhanced Sensing” below.

A new ICD or pace-sense lead should be inserted if R-wave amplitude is low and T-wave oversensing cannot be prevented without programming that compromises sensing of VF.

Postpacing T-Wave Oversensing

T-wave oversensing is a lesser problem after paced than sensed beats because ICD blanking periods are longer after pacing and because postpacing T-wave oversensing resets the pacing timing cycle, preventing inappropriate detection of VT/VF in pacemaker-dependent patients. In these patients, it is not a clinical problem unless it causes bradycardia pacing or antitachycardia pacing (ATP) at the wrong rate. If AV conduction is intact (e.g., CRT-D patients), consistent postpacing T-wave oversensing causes a repeating sequence of three intervals in which the first two are in the VT or VF zones: pacing stimulus to oversensed T, oversensed T to spontaneous R, and spontaneous R to pacing stimulus. Inappropriate therapy is unusual, but it may occur with a probabilistic counting detection algorithm if only two thirds of intervals are required to be in the VT zone (e.g., 8/12 with fast first and last intervals ) or a few spontaneous premature (PVC) beats. Solutions include adjusting sensitivity and increasing the postpacing blanking period. In CRT-D/Ps, programming V-V pacing delay or LV-only pacing may also resolve T-wave oversensing that inhibits biventricular pacing.

Oversensing Physiologic Extracardiac Signals: Myopotentials

ICDs may oversense myopotentials of diaphragmatic, pectoral, or (rarely) intercostal origin. Skeletal myopotentials have dominant frequencies in the range of 75 Hz, but they have substantial frequency content as high as 100 to 200 Hz and as low as 20 Hz. ICD low-pass filters in the range of 40 to 80 Hz attenuate high-frequency components, but sufficient high-frequency signal passes these filters to give myopotential EGMs a distinctive appearance.

Diaphragmatic Myopotentials

These low-amplitude, high-frequency signals are more prominent on the sensing EGM than the shock EGM because the sensing bipole is closer to the source. Their amplitude varies with respiration, but not the cardiac cycle ( Fig. 4-18 ). Oversensing is most common with integrated-bipolar sensing at the RV apex. It usually begins when sensitivity is maximal, after long diastolic intervals or ventricular paced events, and often ends with a sensed R wave, which reduces sensitivity abruptly. Thus it commonly occurs in pacemaker-dependent patients in whom inhibition of pacing maintains high ventricular sensitivity, resulting in persistent oversensing and inappropriate detection of VF. With chronically implanted leads, oversensing may first occur after the dominant rhythm changes from ventricular sensed to ventricular paced, such as “upgrade” to CRT or AV junction ablation. Oversensing may be reproduced by monitoring real-time EGMs during deep breathing or straining in different positions, after programming VF detection OFF.

Figure 4-18, Oversensing of Diaphragmatic Myopotentials by an Integrated-Bipolar Lead at the Right Ventricular Apex.

To minimize the risk of oversensing proactively in pacemaker-dependent patients, it may be prudent to program a sensitivity of 0.45 to 0.60 mV if this sensitivity is not nominal, especially if integrated-bipolar sensing is used. The nonprogrammable Boston Scientific “noise rejection” algorithm operates continuously and may reduce oversensing. Occasionally, correction requires inserting a new sensing or defibrillation lead away from the diaphragm.

Pectoral Myopotentials

These high-frequency, variable amplitude signals are prominent on EGMs that include the ICD Can, including shock EGMs and leadless ECG. They may be reproduced by pectoral muscle exercise. Because ICDs do not use these signals as primary sensing channels, pectoral myopotentials do not cause oversensing if the lead is intact. However, they may cause misclassification of exercise-induced sinus tachycardia as VT because algorithms that discriminate VT from SVT based on ventricular EGM morphology usually use the RV coil-Can vector. Pectoral myopotentials might also interfere with algorithms that evaluate lead integrity by comparing near-field and far-field signals (see below).

The diagnosis may be confirmed by monitoring the real-time sensing EGM during pectoral muscle exercise. Oversensing of pectoral myopotentials on sensing EGMs typically indicates an in-pocket insulation breach ( Fig. 4-19 and Case Study 4-1 ). Rarely, pectoral myopotentials may be oversensed in pacemaker-dependent patients with intact integrated-bipolar leads. Additionally, reversal of high-voltage connections on DF-1 dual-coil, integrated-bipolar leads may present as myopotential oversensing caused by change of the sensing vector to a far-field signal that includes the SVC coil and Can as electrodes ( Fig. E4-7 ).

Figure 4-19, Pectoral Myopotentials.

Figure E4-7, Sensing Issues in Integrated-Bipolar Leads (Endotak, Boston Scientific).

Case Study 4-1
Implantable Cardioverter-Defibrillator Shock during Exercise

Clinical Summary

Medical History

In 2006, a 48-year-old man had a sustained wide complex tachycardia at a rate of 185 bpm (cycle length 324 msec) while working as an electrician repairing railroad cars for Amtrak. He was hypotensive and required cardioversion. Ten years previously, he had been diagnosed with supraventricular tachycardia (SVT) in his country of origin (Armenia). In 2006, evaluation for structural heart disease identified arrhythmogenic right ventricular cardiomyopathy (ARVC). He underwent electrophysiological study and ablation of monomorphic induced VT with morphology identical to the clinical arrhythmia and a rate of 200 bpm (cycle length 300 msec). He was treated with metoprolol 50 mg twice daily, and a dual-chamber implantable cardioverter-defibrillator (ICD) (Medtronic EnTrust Model D153ATG) was implanted with a dedicated-bipolar lead (Medtronic Sprint Fidelis Model 6949).

In 2009, he received a shock after 30 minutes of vigorous exercise on a rowing machine. In clinic, pacing and sensing thresholds were nominal. Lead impedances were stable on multiple repeat measurements (typical values: atrial 456 ohms, RV Tip-Ring 660 ohms, right ventricular (RV) Coil-Can 54 Ω). A good quality chest x-ray showed no abnormality of the leads or generator. A spot film of the pocket confirmed complete insertion of all lead pins into the header.

Diagnosis and Treatment of Device-Detected Ventricular Tachycardia

Electrogram Analysis and Troubleshooting

Tables E4-1 and E4-2 show programmed detection zones and SVT-VT discrimination criteria. This older ICD lacked a morphology discrimination algorithm. It discriminated VT from SVT by the pattern and cycle lengths of dual-chamber intervals (PR Logic, see Fig. E4-27 ).

TABLE E4-1
Programmed Detection Zones
Detection Interval NID
VF 300 msec 30/40
FVT via VF 220 msec (30/40)
VT 350 msec 24
FVT, Fast ventricular tachycardia; NID, number of intervals to detect ventricular fibrillation; VF, ventricular fibrillation; VT, ventricular tachycardia.

TABLE E4-2
Supraventricular Tachycardia-Ventricular Tachycardia Discrimination Criteria
PR Logic Discriminators SVT Limit 280 msec
Sinus tachycardia ON
Atrial fibrillation ON Stability OFF
Other 1 : 1 SVT ON Onset OFF
SVT, Supraventricular tachycardia.

Figure E4-37 shows the interval plot with events identified by numeric markers. The plot begins with a 1 : 1 tachycardia slightly slower than the VT detection interval of 350 msec (1). The tachycardia accelerates gradually across the detection boundary (2). Simultaneously, atrial intervals become rapid and irregular (3). The ICD identifies a regular ventricular rhythm in the VT zone with dissociated, rapid, irregular atrial rhythm. Based on this analysis of intervals, the ICD classifies the rhythm (Detection) as a dual tachycardia (VT + atrial fibrillation). It delivers three sequences of successively faster burst antitachycardia pacing (ATP), (4)-(6). When tachycardia persists, the ICD charges the shock capacitors.

Figure E4-37, Interval Plot.

Figure E4-38 displays dual-chamber electrograms (EGMs) recorded from the right atrium (RA) and far-field ventricular EGM (RV Coil-Can). The upper panel shows the real-time, sinus-rhythm EGMs recorded in clinic, and the lower panel shows the stored EGM during redetection of VT after the third trial of ATP (TD marker, VT RX 2 CV). By visual inspection, ventricular EGM morphology is identical in the two panels. The combination of EGM morphology match and gradual acceleration (see Fig. E4-37 ) is diagnostic of sinus tachycardia. Although far-field R waves (FFRW) can be identified on the atrial channel, sensing of atrial and ventricular EGMs is accurate in each panel. Nevertheless, intermittent FFRW oversensing is a potential reason for misclassification of sinus tachycardia as dual-tachycardia.

Figure E4-38, Real-Time Electrogram (EGM) Versus Stored EGM During Device-Detected Ventricular Tachycardia.

Figure E4-39 shows the stored EGM at initial detection of VT. Upper and lower panels are continuous. The ventricular markers change from TS to TD as sinus tachycardia accelerates gradually across the sinus-VT rate boundary (1). The marker channels show erratic atrial oversensing. Both EGMs show high-frequency signals that do not represent cardiac EGMs. The differential diagnosis includes electromagnetic interference (EMI) from a poorly grounded rowing machine, pectoral myopotentials, and lead-related oversensing.

Figure E4-39, Stored Electrogram at Initial Detection of Ventricular Tachycardia.

The RV Coil-Can EGM includes a large electrode in close proximity to the pectoral muscle. It commonly records pectoral myopotentials and is not used for rate sensing. The consistent ventricular markers indicate that the dedicated RV tip-ring bipole senses the ventricular EGMs accurately. However, an undamaged closely spaced, intracardiac bipole should not record pectoral myopotentials. Further, myopotentials have approximately constant amplitude on the ventricular channel before ATP (VT Rx 1 Burst in lower panel), but they are absent in sections of the atrial channel (2) and their amplitude varies markedly in others (3). External EMI rarely causes the combination of continuous, constant-amplitude extracardiac signals on the shock channel with intermittent, variable-amplitude signals on the atrial channel. However, this combination is characteristic of an in-pocket insulation breach of the atrial lead.

Figure E4-40 shows real-time EGMs recorded in clinic. The upper panel shows surface ECG lead II and the right atrial (RA) EGM during pocket manipulation. The lower panel shows the right atrial and RV sensing EGMs during isometric arm exercises. Both panels show intermittent, variable amplitude, high-frequency extracardiac signals consistent with pectoral myopotentials (1). Additionally, the upper panel shows a progressive decrease in amplitude of the RA EGM from baseline at left, through an intermediate amplitude in the center (2), to a low amplitude at the right. This reduction in amplitude may reflect the reduced potential difference between the two conductors as each is partially grounded to the potential of extracellular fluid.

Figure E4-40, Real-Time Electrograms During Pocket Manipulation and Isometric Arm Exercise.

Returning to Figure E4-37 , atrial oversensing stops and the ICD senses a 1 : 1 tachycardia reliably as the shock capacitors charge after the third ATP sequence (6). Further, the sinus tachycardia rate slows below the sinus-VT detection boundary (7). Nevertheless, the ICD delivers a shock (35 J). This behavior highlights two limitations of this detection algorithm. First, SVT-VT discriminators do not apply in redetection. Second, in this ICD, shock confirmation requires a cycle length within 60 msec of the VT interval to compensate for potential undersensing in ventricular fibrillation. Newer Medtronic ICDs correct the second limitation and will not synchronize on an interval that exceeds the VT interval. The episode ends after eight consecutive postshock intervals are slower than the VT interval (Term.). Additionally, it is likely that the EGM morphology discriminator would prevent initial inappropriate detection in similar cases.

Clinical Management

At surgery, the atrial lead had a visible, pinhole-sized, outer insulation breach from a coil-can abrasion. It was extracted and replaced. The recalled Fidelis lead was also extracted and replaced.

In some cases of atrial oversensing, it may be preferable to retain the atrial lead (e.g., intermittent FFRW oversensing) or to delay lead revision or replacement. In such cases, single-chamber discriminators may be substituted for dual-chamber discriminators. In present Medtronic dual-chamber ICDs, ventricular EGM morphology is the most useful single-chamber discriminator. In this older model, Onset (sudden onset) would have classified this patient's sinus tachycardia correctly. In other patients, the single-chamber Stability criterion may be used in conjunction with consecutive interval counting to reject rapidly conducted atrial fibrillation, but only in the VT zone.

Summary of Key Points

Oversensing on the atrial channel caused dual-chamber pattern and rate analysis to misclassify sinus tachycardia as a regular VT with coexisting atrial fibrillation. This type of misclassification is unlikely in ICDs with morphology discriminators. Although oversensing stopped before the last redetection, VT was redetected because the ICD does not apply SVT-VT discriminators in redetection. The “conservative” confirmation algorithm delivered the shock despite slowing of sinus tachycardia from the VT zone to the sinus zone. Analysis of the stored atrial and RV Coil-Can EGMs indicated that the atrial rhythm was misclassified because the atrial channel oversensed pectoral myopotentials. In contrast to the RV Coil-Can EGM, atrial EGMs recorded from normally functioning, closely spaced bipoles do not record pectoral myopotentials. Intermittent and variable pectoral myopotentials on a dedicated-bipolar channel are typical of in-pocket insulation failure. This diagnosis was confirmed by real-time maneuvers, despite a normal chest x-ray and normal atrial lead impedance. Lead impedance is measured only a few times per day, and it may not be measured at the moment an intermittent insulation breach opens.

Reference

  • 1.

    Swerdlow CD, Ellenbogen KA: Implantable cardioverter-defibrillator leads: design, diagnostics, and management. Circulation 128:2062–2071, 2061–2069, 2013.

You're Reading a Preview

Become a Clinical Tree membership for Full access and enjoy Unlimited articles

Become membership

If you are a member. Log in here