OUP user menu

Failed anti-tachycardia pacing can be used to differentiate atrial arrhythmias from ventricular tachycardia in implantable cardioverter-defibrillators

Kevin A. Michael, Andres Enriquez, Adrian Baranchuk, Charlotte Haley, Jane Caldwell, Christopher S. Simpson, Hoshiar Abdollah, Damian P. Redfearn
DOI: http://dx.doi.org/10.1093/europace/euu169 First published online: 16 July 2014


Background Atrial fibrillation/tachycardia (AF/AT) may result in inappropriate therapies in implantable cardioverter-defibrillators (ICDs). The post-pacing interval (PPI) and tachycardia cycle length difference (PPI − TCL) has been previously demonstrated to indicate the proximity of the pacing site to a tachycardia origin.

Aims We postulated that the PPI and PPI − TCL would be greater in AT/AF vs. ventricular tachycardia (VT) after episodes of failed anti-tachycardia pacing (ATP).

Methods and results This was a single-centre, retrospective study evaluating consecutive patients implanted with dual (DR)/biventricular (BIV) ICDs. Stored electrograms were used to determine whether the ATP captured the arrhythmia and the arrhythmia did not present with primary or secondary termination. Measurements were done using manual calipers. A total of 155 patients were included. There were 79 BIV and 76 DR devices. In total, 39 episodes were identified in 20 patients over a 23-month follow-up period. A total of 76 sequences of ATP (burst/ramp) were delivered, 28 (37%) of them inappropriate. Fifty-one events (18 AT/AF and 33 VT) were compared. The mean PPI was 693 ± 96 vs. 512 ± 88 ms (P < 0.01) and the mean PPI − TCL was 330 ± 97 vs. 179 ± 103 ms (P < 0.01) for AT/AF and VT, respectively. Cut-offs of 615 ms for the PPI [area under curve (AUC) 0.93; 95% confidence interval (CI): 0.84–1.00; P < 0.01] and 260 ms for PPI − TCL (AUC 0.86; 95% CI: 0.74–0.98; P < 0.01) were identified.

Conclusion The PPI and PPI − TCL after failed ATP differs significantly between AF/AT and VT and are therefore useful indices to discriminate between supraventricular tachycardia and VT in ICDs.

  • ICD
  • Ventricular tachycardia
  • Anti-tachycardia pacing
  • Inappropriate therapies
  • SVT–VT discrimination

What's new?

  • We describe the clinical utility of the post-pacing interval (PPI) and PPI − TCL (PPI minus tachycardia cycle length difference) after episodes of failed anti-tachycardia pacing to discriminate between atrial tachycardia/fibrillation and true ventricular tachycardia in patients with implantable cardioverter-defibrillators.

  • This element should be particularly useful in patients with single lead devices, in which the analysis of atrial electrograms is not available and probably also in patients with permanent atrial fibrillation.


Atrial fibrillation or atrial tachycardia (AF/AT) with fast conduction to the ventricles are frequent causes of inappropriate therapies in patients with implantable cardioverter-defibrillators (ICDs).1,2 The rate of the conducted supraventricular tachycardia (SVT) may exceed the upper detection interval at which discriminators and morphology templates are programmed, leading to inappropriate shocks that are usually preceded by an episode of anti-tachycardia pacing (ATP).3

If the ATP is ineffective in terminating the rhythm, the pause until resumption of the tachycardia, derived from the intracardiac electrograms (EGMs), may be used to suggest the origin. The concept of entrainment has been previously defined as a basic electrophysiological manoeuvre to indicate the proximity of a roving pacing catheter to a macro-reentrant circuit or focal tachycardia.4,5 These fundamental principles may be extrapolated to device-based EGM interpretation.

We postulated that the post-pacing interval (PPI) following a failed episode of ATP for true ventricular tachycardia (VT) is significantly shorter than the PPI for AF/AT and therefore may be used to discriminate the origin of the tachycardia.


This study was a retrospective descriptive analysis of consecutive patients implanted with ICDs at a single centre. Events were adjudicated by two observers (K.A.M., C.H.), who were blinded to the clinical background of the patients. The clinical records of all consecutive patients implanted with dual-chamber (DR) and biventricular (BIV) ICDs, were examined over a 2-year period. Only patients with apical lead placement were included in this study. All patient-related device therapies were flagged in a database and were then examined. These were then classified into appropriate or inappropriate therapies.

Exclusion criteria were:

  1. Patients with single-chamber devices or with the atrial port of the device plugged. This was done so that only episodes with a corresponding atrial EGM would be analysed for the proof of concept.

  2. Evidence of atrioventricular block.

  3. All non-physiological events such as oversensing phenomena were excluded from the analysis.


The PPI was defined as the return tachycardia cycle length (TCL) after ATP (burst/ramp) as measured in the ventricular channel (Figure 1). A further correction factor of the PPI minus the mean TCL (PPI − TCL) was also applied. The mean cycle length of the tachycardia for analysis was determined as the average of five consecutive cycle lengths of the ambient tachycardia rate prior to ATP. This was derived from the marker channel of the device with verification from the corresponding ventricular EGM ensuring that there was no undersensing of the sampled V-sensed events.

Figure 1

The response to ATP illustrating dissociation of the ventricle (V) during an episode of atrial tachycardia (A). The PPI is indicated. Note that there is a ‘pseudo VAAV’ atrial response because of failure to capture the atrium during pacing.6

The following exclusion criteria were applied to the remaining datasets to ensure that the delivered ATP did not significantly perturb the tachycardia to account for the episode as ongoing:

  1. The pre-ATP and post-ATP mean TCL varied >50 ms. This criterion was only applicable if the preceding tachycardia appeared to be regular. The considerations for a stable rhythm were VT, AT, or pseudo-regularization of rapidly conducted AF. This did not pertain to irregular atrial events which were diagnosed as AF.

  2. If ATP terminated the episode.

  3. A ventricular paced event occurred at the lower programmed rate immediately after ATP as this would have truncated the true PPI.

  4. Anti-tachycardia pacing-induced VT events were excluded and this was based on EGM morphology change and/or TCL acceleration.

All events were evaluated in a standardized manner:

  1. The scatter plot and local as well as the far-field EGMs were viewed collectively to diagnose the underlying tachycardia.

  2. Events were evaluated in chronological order (per episode) and then sequentially (per sequence of ATP) in each individual patient.

  3. Each episode was categorized as appropriate or inappropriate according to predetermined criteria for VT or AF/AT (Table 1).

View this table:
Table 1

Criteria used to distinguish VT from AF/AT using ICD scatter plots and intracardiac EGMs

Ventricular tachycardiaAtrial fibrillation/Atrial tachycardia
Onset of tachycardia in ventricleOnset of tachycardia in atrium
V rate > A rateA rate > V rate
A and V dissociationV rate dependent on A events for 1 to 1 tachycardias
Stable V–V relation (delta V ≤40 ms)Variable V–V intervals (delta V >40 ms)
V–A timing constant and <50% of TCL compatible with retrograde conductionVariable V–A timing suggesting AT

Statistical analysis

All continuous variables were represented as mean ± standard deviation while categorical variables were recorded as numbers and/or percentages. Data were analysed and presented using Matlab (Mathworks).

An unpaired t-test was used to assess significance and a P < 0.05 was regarded as significant. Receiver operator curves (ROCs) were used to determine the optimal cut-off values that would discriminate AF/AT from VT using the PPI and PPI − TCL after ATP delivered in the ventricular channel. These data are presented with a cut-off value and area under curve (AUC) with P < 0.05 as significant. Sensitivity and specificity along with positive predictive value (PPV) and negative predictive value (NPV) are reported with 95% confidence intervals (CIs).


A total of 155 patients were included. Baseline characteristics are summarized in Table 2. Mean age was 68.1 ± 12.4 years and 126 were male (81.3%). Mean left ventricular ejection fraction was 28.5 ± 11.6% and aetiology was ischaemic in 71.6%, idiopathic dilated cardiomyopathy in 21.3%, and others (hypertrophic cardiomyopathy, Brugada syndrome, long QT, arrhythmogenic right ventricular cardiomyopathy) in 7.1%. The indication for ICD was primary prevention in 69% and secondary prevention in 31% of patients. Devices implanted were DR in 76 patients (49.0%) and BIV ICDs in 79 (51.0%). All defibrillators were from the same manufacturer; Medtronic (Medtronic Inc.) including the following models Entrust, Virtuoso, Concerto, and Protecta. A 31.6% had prior history of AF or flutter.

View this table:
Table 2

Clinical characteristics (n = 155)

VariableMean ± SD or %
Age68.1 ± 12.4 years
Male sex81.30%
LVEF28.5 ± 11.6%
Ischaemic aetiology71.60%
 Primary prevention69.00%
 Secondary prevention31.00%
Type of device
 Dual chamber49.00%
History of AF31.60%
  • LVEF, left ventricular ejection fraction; AF, atrial fibrillation.

Twenty out of the 155 patients (12.9%) had ATP for detected ventricular arrhythmia over a follow-up period of 24 months. In this group, a total of 39 ATP episodes were identified and 76 sequences of ATP (burst/ramp) were delivered within these episodes. Fifty-one sequences of ATP (18 AT/AF and 33 VT) were available for analysis after strictly applying the exclusion criteria (attrition rate of 33%). In the nine patients that presented with AT/AF episodes, the EGMs were suggestive of AF in five patients and of AT or flutter in the remaining four.

Advanced discriminators were programmed ON in all but one patient who had inappropriate therapies delivered by the device. Onset and stability parameters were either turned off or programmed to monitor only.

All elements of ‘PR logic’ (Medtronic Inc.) were enabled in the ICDs except for ‘other 1 : 1 SVTs’ which was nominally programmed off in these devices. All inappropriate therapies that occurred were detected in the ‘ventricular fibrillation’ (VF) zone of the devices, where only a rudimentary rate-driven detection algorithm was applied.

All devices had a single burst of ATP before/during charge programmed at 88% coupling interval in the fast VT zone (≥188 b.p.m.). Other ATPs within the VT zones were individually programmed to include burst and ramp programming per physician decision in the case of secondary prevention indications. However, in the case of primary prophylactic implantation criteria, ATP and zone programming were modelled on programming in the PAINFREE trials.79 The SVT limit was nominally set to 340 ms (i.e. upper extent at which advanced discriminators were applied).

The mean PPI was 693 ± 96 vs. 512 ± 88 ms (P < 0.01) and mean PPI − TCL was 330 ± 97 vs. 179 ± 102 ms (P < 0.01) for AT/AF and VT, respectively (Figure 2A and B). Receiver operator curves were applied to both PPI and PPI − TCL measurements (Figure 3). A PPI < 615 ms favoured VT as the diagnosis rather than AF/AT with a sensitivity of 77.8% (95% CI: 58.6–97.0%) and a specificity of 87.5% (95% CI: 76.0–99.0%). The PPV for AT/AF detection was 77.8% (95% CI: 58.6–97%) and the NPV was 87.5% (95% CI: 76.0–99.0%).

Figure 2

(A) Distribution of data and the respective mean values using PPI as a discriminator for VT and AF/AT. Boxes and bars represent standard deviation and range. (B) Distribution of data and the respective means using PPI − TCL as a discriminator for VT and AF/AT. Boxes and bars represent standard deviation and range.

Figure 3

The ROC analysis of PPI and PPI − TCL parameters. The respective cut-off values have been ringed.

Similarly, a PPI − TCL <260 ms was more likely to be VT than AF/AT with a sensitivity of 72.2% (95% CI: 51.5–92.9%) and a specificity of 78.1% (95% CI: 63.8–92.4%). The PPV was 65.0% (95% CI: 44.1–85.9%) for AT/AF and the NPV was 83.3% (95% CI: 70.0–96.7%) (Figure 4).

Figure 4

(A) The scatterplot illustrates an episode of tachycardia in a dual-chamber ICD. Anti-tachycardia pacing (burst and the ramp) entrains the ventricle without termination of the arrhythmia. (B) The EGM shows PPI and PPI-TCL calculated as (470 − 400) ms = 70 ms, which is compatible with VT. The V–A relationship (>1) confirms the diagnosis


Anti-tachycardia pacing has been shown to successfully terminate VT in over 90% of cases, making this a very successful and painless therapy in ICDs.10 As a result, programming strategies now utilize ATP as initial therapy even in fast VT episodes if the rhythm is determined to be stable.11

If ATP fails to terminate the tachycardia, there is usually an escalation in therapies to cardioversion. Some manufacturers also resort to committed therapies if there is failure to terminate the episode based on the original detection of the rhythm satisfying criteria for VT or VF. Thus, if there was an incorrect classification of a supraventricular arrhythmia at the outset, multiple inappropriate shocks may be delivered by the device. Inappropriate therapies are more likely to occur in patients with underlying AF because of fast conduction to the ventricle outside of the range of discriminators.12 The QRS morphology detection algorithm has been shown to improve rhythm detection; however, this is also prone to sampling errors which may still result in inappropriate ATP and/or shocks.13

The overall incidence of device therapies in this study (12.9%) is lower than is quoted in other studies, but this may be because of a shorter follow-up period. However, the proportion of inappropriate therapies is comparable (16 vs. 15%, respectively) to those found in the PAINFREE Rx II trial.10 The detrimental effects of shocks both appropriate and inappropriate have been shown to portend a poor prognosis with worsening heart failure symptoms.14 Moreover, programming strategies aimed at reducing unnecessary ICD therapies result in significant reduction in mortality, with no increase in the risk of syncope.15,16 This emphasizes the need to refine ICD algorithms to minimize inappropriate shocks and to enhance ATP as first-line therapy.

This study analyses the PPI after a failed episode of ATP using the response to pacing to discriminate between SVT and VT. The PPI is only considered if the underlying tachycardia is not terminated by the ATP and provided that the ventricles are captured during pacing. In electrophysiological terms, a PPI equal to the TCL suggests that the pacing stimulus is within the macro-reentrant substrate of the tachycardia.17 We did not assume to obtain matched PPI and TCL values given the location of the RV lead being apically placed in all patients analysed; however, deductions can be made based on the relative durations of the PPI. The longer mean PPI of 693 ± 96 ms observed in AF/AT can be explained by retrograde penetration of the atrioventricular node (AVN) and subsequent delay in resumption of antegrade conduction.18,19 We postulated that this delay would be greater than the PPI after entrainment of VT by ATP. Our results support this hypothesis, with significant differences demonstrated in the PPI between inappropriately treated atrial arrhythmias (AF/AT) and true VT that failed to terminate with ATP.

The ROC cut-off values suggested that a PPI < 615 ms and a PPI − TCL <260 ms was more likely to be VT rather than AF/AT (Figure 3). The ROC analysis for these parameters revealed a greater AUC for PPI than for PPI − TCL. This implies that the former is a more robust parameter for rhythm discrimination. This is then advantageous as it suggests that the correction factor is not essential for making a meaningful deduction from this observed interval. There was, however, 5 of 33 (15%) VT episodes with PPI ≥ 615 ms overlapping into the predicted AF/AT range. However, with exception of ventriculoatrial dissociation, no single discriminator has an absolute specificity and sensitivity, and they present additive value when used in combination. Possible explanations for this overlapping would include location of the ventricular substrate at a distant site from the pacing lead such as the basal–lateral left ventricular segments or due to slow anisotropic inter-myocyte conduction in scarred, dilated ventricular myocardium.6 A shorter PPI than the TCL may also be encountered in cases of AF/AT if there is failure to capture the myocardium during ATP. Alternatively, this may represent the influence of catecholamines increasing AVN conduction time.

The coupling interval for the ATP sequences is typically calculated as a percentage of the TCL and there may be a programmed decrement between each burst or between individual ATP pulses referred to as a ‘ramp’. These pulses may be delivered at an interval shorter than the effective refractory period of the ventricular myocardium resulting in failure to capture. Because of this, we prefer evaluating the PPI after ATP that is delivered in a burst format. However, for the purposes of this study the PPI was evaluated after both ramp and burst ATP but after ensuring capture of myocardium as per examination of the device EGMs.

The use of the response to pacing as a discriminator in ICDs has been previously studied by Saba and colleagues.20,21 Their analysis for 1 to 1 (A to V) tachycardias relied on pacing the atria and ventricle simultaneously. They then determined in which chamber the return of tachycardia was recorded first. Our study differs in that discrimination of SVT from VT is reliant upon just a single ventricular lead, making it applicable to single-chamber devices as well multi-channel ICDs. Ridley et al. described and classified atrial pacing responses during ventricular pacing, to distinguish an AT from other mechanisms of supraventricular arrhythmias in DR ICDs.21 The atrial response to ATP is not essential for the methods outlined in this study and it also is not applicable to atrioventricular reentrant tachycardias (i.e. atrioventricular nodal and atrioventricular tachycardia). The proposed observation is however only intended to distinguish organized AF/AT from VT. This may appear to be limiting, but the incidence of AF/AT is arguably more likely to be encountered in the population of patients receiving ICDs as a cause of inappropriate therapies than the reentrant mechanisms of SVTs. Our results may have particular relevance in patients with single-chamber ICDs, in which atrial EGMs are not available for analysis, and in patients with chronic AF to differentiate rapidly conducted AF from dual tachycardia.

Ultimately, the PPI and PPI − TCL parameters have widespread acceptance in interventional electrophysiology and therefore can be easily extrapolated to the ICD EGM interpretation context and can be adjunctively applied to assist the device specialist during troubleshooting.22,23 It may have the potential to be incorporated into ICDs as a method of redetection or active/dynamic discrimination of the underlying rhythm and a complement to the conventional passive algorithm. This would however require a larger study to validate the concept for this application.


The sample size of this study is small and our findings need to be validated in a larger cohort of patients. For the same reason, it was not feasible to perform subgroup analysis by aetiology (ischaemic vs. non-ischaemic), indication of the ICD (primary vs. secondary prevention), or medications. No patients with dual tachycardias (e.g. VT during AF) were included and our findings are not applicable in this subgroup. There was a high attrition rate of data with 33% of measured PPI values not being included in the final analysis. This is because the exclusion criteria were strictly adhered to evaluate this proof of concept. Although only a single manufacturer's devices were used in this study, it afforded a homogeneous patient population and served to reduce observer-related errors. Finally, given the retrospective nature of the study, criteria for ATP programming were not standardized for all patients and future prospective designs will be necessary to assess the impact of the aggressiveness (% of cycle length) or modality (burst vs. ramp) of ATP.


The PPI and PPI − TCL difference are established electrophysiological concepts that indicate the proximity of a pacing site to the source of a tachycardia. This concept was applied to differentiate VT from AT/AF followed by failed episodes of ATP, showing significant differences in the mean values for both these tachycardia sources when identified in DR and BIV ICDs.


The authors wish to thank Mark Brown PhD (Medtronic Inc. MN, USA) for his kind assistance and advice in preparing this manuscript.

Conflict of interest: none declared.


View Abstract