OUP user menu

Does change in device detected frequency of non-sustained or diverted episodes serve as a marker for inappropriate shock therapy? Analyses from the INTRINSIC RV and ALTITUDE-REDUCES Trials

Renee M. Sullivan, Milan Seth, Kellie Berg, Kira Q. Stolen, Paul W. Jones, Andrea M. Russo, F. Roosevelt Gilliam, Brian Olshansky
DOI: http://dx.doi.org/10.1093/europace/eut426 668-673 First published online: 31 January 2014


Aims Implantable cardioverter-defibrillators (ICDs) treat ventricular tachycardia or fibrillation but may also deliver unnecessary shocks. We sought to determine if the frequency of ICD-detected non-sustained or diverted (NSD) episodes increases before appropriate or inappropriate ICD shocks.

Methods and results We evaluated NSD episodes in the INTRINSIC RV Trial and their relationship to ICD shocks (appropriate and inappropriate). Time from NSD to shock was analysed. Results were validated by utilizing 1495 adjudicated ICD and cardiac resynchronization therapy-defibrillator shocks following NSD episodes collected through the LATITUDE remote monitoring system as part of the ALTITUDE-REDUCES Study. In INTRINSIC RV, 185 participants received 373 shocks; 148 had at least 1 NSD episode. Non-sustained or diverted frequency increased 24 h before the first shock for those receiving an inappropriate (P < 0.01) but not an appropriate shock (P = 0.17). Patients with NSD episodes within 24 h of a shock were significantly more likely to receive inappropriate therapy [odds ratio (OR) = 3.12, P < 0.01]. At the receiver operator curve determined optimal cutoff, an NSD episode within 14 min before shock strongly predicted inappropriate therapy (sensitivity 48%, specificity 91%; OR = 8.8, and P < 0.001). The 14 min cut-off evaluated on an independent dataset of 1495 shock episodes preceded by an NSD in the ALTITUDE-REDUCES Study confirmed these results (sensitivity = 47%, specificity = 85%, OR = 5.0, and P < 0.001).

Conclusion Device-detected NSD episodes increase before inappropriate but not appropriate shocks. Novel alerts or automated algorithms based on NSD episodes may reduce inappropriate shocks.

  • Implantable cardioverter defibrillators
  • Inappropriate activations

What's new?

  • Inappropriate ICD shocks are common, potentially harmful, and difficult to prevent.

  • This study sought to determine whether non-sustained or diverted episodes (NSDs) of tachycardia detected and stored in the ICD were associated temporally with an increased risk of inappropriate (or appropriate) shock delivery.

  • NSD episodes did not increase before the appropriate ICD shocks, but increased significantly in the 24-h period before inappropriate shocks, particularly within 14 min before an inappropriate shock.

  • While the time window between NSD episodes and inappropriate shocks was short, knowledge of NSD episodes may allow for future device enhancements that can reduce the risk of inappropriate shocks.


Implantable cardioverter-defibrillators (ICDs) can prevent death from ventricular tachycardia (VT) or fibrillation (VF). Appropriate treatment by an ICD requires accurate detection of the episodes of VT or VF that do not terminate spontaneously. Despite efforts to decrease their occurrence, inappropriate ICD shocks remain a vexing problem that is challenging to prevent13 and may be associated with an increased risk of mortality.48 Various algorithms and approaches have been utilized or considered911 to attempt to reduce inappropriate ICD shocks. Methods to programme devices have been touted12 and clinical variables have been considered predictors13 but inappropriate shocks delivered for arrhythmias that are not VT or VF persist. Warning arrhythmias, such as non-sustained supraventricular, ventricular, and sinus tachycardia, may predict longer, more sustained, and associated episodes that could lead to an inappropriate, unnecessary, or necessary shock. Therefore, we hypothesized that non-sustained or diverted tachycardia episodes (NSD) were related temporally to the delivery of ICD shocks. This study sought to determine if such NSD episodes stored in the ICD were associated with an increased risk of inappropriate or appropriate shock.


The Inhibition of Unnecessary RV Pacing with AV Search Hysteresis in ICDs (INTRINSIC RV) study randomized patients who received dual-chamber ICDs (VITALITY AVT; Boston Scientific) for standard indications to VVI-40 or DDDR with AV search hysteresis programming.14 Patients with longstanding atrial fibrillation were excluded. For the present analysis, we sought to assess the relationship between recorded, stored but non-treated arrhythmia episodes and the adjudicated rhythm observed in the first arrhythmic episode treated with an ICD shock.

Non-treated tachycardia episodes were those in which the ICD recognized a tachycardia, based on 8 of 10 beats during a rolling window at rates above the specified rate criteria in any programmed detection zone, with higher rate zones (i.e. VF vs. VT) having precedence in cases where the criteria for multiple zones was satisfied (Figure 1). Once the detection criteria were met, a period began during which each subsequent rolling detection window had to contain at least 6 of 10 beats above the rate threshold for a programmable detection period, nominally set at 2.5 s for the VT and the VT-1 zones, and 1 s for the VF zone, respectively (Figure 1).

Figure 1

Device detection windows to determine VF or VT.

If there were fewer than 6 beats of tachycardia in any rolling 10-beat window during the detection period, the episode was declared non-sustained, therapy was not delivered, and the episode was stored in the device memory. Once the episode criteria were met, capacitor charging began. During, and immediately following, capacitor charge, monitoring for spontaneous conversion of the arrhythmia occurred; if there was spontaneous conversion and the episode ended without shock delivery, it was classified as a diverted episode.

Electrograms (EGMs) representing the first delivered shock therapy for each patient, and when available, the most recent NSD episode before the first shock, were adjudicated (AMR, BO). Appropriate shocks were ICD shocks delivered for VT or VF that did not terminate before the shock. Inappropriate shocks were shocks delivered for reasons other than VT or VF.

The only protocol requirement for the tachyarrhythmia programming was that at least one shock-only zone be set to 185 b.p.m. Programming of the other tachycardia zones, detection duration, and therapy in each zone was determined by the Principal Investigator at each site. Implantable cardioverter-defibrillator events were recorded on device data discs at each visit.

A validation dataset consisted of 2345 first-shock episodes adjudicated as part of the ALTITUDE-REDUCES Study reported previously.15 Of these, 1495 episodes had preceding NSD data. These data, described previously,16 were obtained from the ALTITUDE project,1517 which includes data from the Boston Scientific Corporation ICD and cardiac resynchronization therapy-defibrillator devices that is regularly transmitted into the LATITUDE system. The EGMs were adjudicated by a panel of nine cardiac electrophysiologists. Non-sustained or diverted episodes were not adjudicated.


The Wilcoxon rank sum test compared NSD frequency between patients with appropriate vs. inappropriate shocks. A signed rank test compared the frequency of NSD episodes within 24 h of first shock to all prior timepoints. Logistic regression analysis and Fisher's exact tests were used to predict time from NSD to inappropriate shock. Receiver operator curves (ROCs) were utilized to evaluate the change in NSD frequency prior to a shock, and the time from the latest NSD to shock occurrence as univariate predictors of inappropriate therapy. Area under the ROC curve (AUC) was estimated by using the trapezoidal method, with 95% confidence intervals constructed with the Delong method. An optimal cut-off point for predicting inappropriate shocks was estimated from the ROC curve by determining the point on the curve geometrically closest to the upper left corner of the graph.1821


In INTRINSIC RV, of the total population of 1530 patients followed over 11.5 ± 3.1 months, 1345 patients received no shock, 93 (6%) patients received an inappropriate shock as the first shock, and 92 (6%) patients received an appropriate shock as the first shock. There were no differences between the patients with appropriate vs. inappropriate first shocks with respect to age, gender, hypertension, coronary artery disease, the presence of VT or VF (secondary prevention device indication), or the presence of congestive heart failure (Table 1). Persons with diabetes had a higher incidence of appropriate first shock similar to a prior report.22 There were no differences in β-blocker, diuretic, or spironolactone utilization between the groups; antiarrhythmic drug use was significantly greater in participants receiving an appropriate first shock (Table 1).

View this table:
Table 1

Baseline characteristics of the patients receiving inappropriate shock vs. appropriate shock vs. no shock

VariableInappropriate first shock n = 93 (6.1%)Appropriate first shock n = 92 (6%)No shock n = 1345 (87.9%)Total n = 1530P value
Age63.2 ± 12.464.0 ± 11.865.6 ± 11.865.3 ± 11.90.67
Male, N (%)74 (79.6%)72 (78.3%)1091 (81.1%)1237 (80.8%)0.83
Hypertension43 (46.2%)47 (51.1%)687 (51.1%)777 (50.8%)0.51
Diabetes14 (15.1%)31 (33.7%)375 (27.9%)420 (27.5%)<0.01
Coronary disease57 (61.3%)51 (55.4%)919 (68.3%)1027 (67.1%)0.42
VT or VF43 (46.2%)53 (57.6%)660 (49.1%)756 (49.4%)0.12
NYHA class
 Unknown2 (2.2%)4 (4.3%)25 (1.9%)31 (2.0%)0.84
 Class I18 (19.4%)22 (23.9%)274 (20.4%)314 (20.5%)
 Class II50 (53.8%)45 (48.9%)738 (54.9%)833 (54.4%)
 Class III22 (23.7%)20 (21.7%)292 (21.7%)334 (21.8%)
 Class IV1 (1.1%)1 (1.1%)16 (1.2%)18 (1.2%)
Baseline therapy
 β-Blocker68 (73.1%)65 (70.7%)1035 (77.0%)1168 (76.3%)0.71
 ACE inhibitor57 (61.3%)68 (73.9%)849 (63.1%)974 (63.7%)0.07
 Antiarrhythmic18 (19.4%)30 (32.6%)186 (13.8%)234 (15.3%)0.04
 Diuretic57 (61.3%)49 (53.3%)704 (52.3%)810 (52.9%)0.27
 Spironolactone11 (11.8%)12 (13.0%)176 (13.1%)199 (13.0%)0.80

Of the enrolled participants, 807 had a one-zone device, 447 had a two-zone device, and 109 had a three-zone device (Table 2). A total of 373 ICD shocks were delivered in 185 patients for which EGMs were stored and available (related to the storage capacity of the devices and the frequency of saving to disc). Of these, 148 participants experienced at least one NSD. Data regarding NSD prediction were not analysed based on the number of zones or by zones.

View this table:
Table 2

Tachycardia rate and zone programming (in b.p.m.)

Zone 1Zone 2Zone 3
Patients, N (%)807 (59.2)447 (32.8)109 (8)
Rate threshold VT-1 (b.p.m.)NANA153 ± 14.0
VT (b.p.m.)NA176 ± 12.7186 ± 11.7
VF (b.p.m.)183 ± 7.0207 ± 15.8212 ± 14.7

Non-sustained or diverted episodes were categorized as occurring >24 h before the first shock or within 24 h of the shock. The frequency of NSD episodes did not increase in the 24 h time period before appropriate shocks (Table 3, P = 0.17), but was significantly greater within 24 h of an inappropriate shock (Table 3, P < 0.01). Non-sustained or diverted episodes preceded 54.8% of the inappropriate shocks and 28% of the appropriate shocks (P = 0.001) (Table 3), within the 24 h window.

View this table:
Table 3

Relationship of NSD episodes to shocks

 Appropriate shock (n = 75)Inappropriate shock (n = 73)P value
NSD frequency median, (Q1, Q3)NSD frequency median, (Q1, Q3)
NSD > 24 h before shock0.004, (0, 0.033)0.016, (0, 0.041)0.12
NSD ≤ 24 h before shock0, (0, 1)1, (0, 4)<0.01
Change in NSD frequency 24 h vs. others0, (−0.021, 1) (P = 0.17)0.99, (−0.012, 3.18) (P< 0.01)<0.01
Patients with NSD within 24 h of shock (%)21 (28.0)40 (54.8)<0.01

Receiver operator curves, represented in Figure 2, were analysed: (i) the increase in frequency of the NSD in the 24 h preceding a shock episode (ROC1) and (ii) the time from the most recent NSD to a shock (ROC2) as a predictor of inappropriate shock therapy. Both approaches demonstrated diagnostic value, with the AUC values for both significantly >0.5 [ROC1: AUC = 0.63 (0.54–0.71), ROC2: AUC = 0.69 (0.64–0.74)]. The optimal cutoff identified for the NSD frequency change (ROC1) was at an increase of 0.1 NSD/day. At this cutoff, a sensitivity of 55% and a specificity of 73%, respectively, were achieved.

Figure 2

Receiver operator curves (ROCs) showing the relationship of change in NSD frequency to inappropriate shock and time from previous NSD to inappropriate shock.

A similar performance was achieved at the point on the ROC2 curve corresponding to a 24 h time cutoff between last NSD and shock (sensitivity = 55%, specificity = 72%). The optimal cutoff for time from the previous NSD to shock (ROC2) was at 14 min (sensitivity = 48%, specificity = 91%). The last NSD episode occurred within 14 min of the shock in 35 (48%) of the 73 patients with inappropriate episodes, while only 7 (9%) of the 75 patients with VT or VF had an NSD within the 14 min timeframe. A prior NSD within 14 min of a shock was associated with a marked increase in the odds of the shock being inappropriate (odds ratio = 8.8, P < 0.001). The positive-predictive value was 83.3% and the negative-predictive value was 64.2%, respectively (Table 4). For 91 patients, an EGM was available for adjudication of at least one NSD preceding the first shock. The adjudications for these last available NSD EGMs before the first shock are summarized in Table 5 (none of the 91 adjudicated episodes were noise or artifact).

View this table:
Table 4

The time from all the last NSD episodes (dichotomized at 14 min) to appropriate, inappropriate, and total shocks is displayed

Time from last NSD episode to shockInappropriate shockAppropriate shockTotal
≤14 min35742
>14 min3868106
View this table:
Table 5

Summary of the 91 adjudicated NSD episodes before first shock (dichotomized at 14 min)

Non-sustained/diverted episodes
Shock adjudicationAppropriateInappropriateTotal
Shocks >14 min from NSDAppropriate24832
Shocks ≤14 min from NSDAppropriate314

The time from prior NSD episodes and, specifically, the performance of the 14 min optimal cutoff determined from the analysis dataset was evaluated as a predictor of inappropriate first-shock therapy in the validation dataset, ALTITUDE-REDUCES (ROC Figure 3). Performance of the 14 min cut-off (sensitivity = 46%, specificity = 86%) in the validation dataset was consistent with the data from INTRINSIC RV. The area under the ROC curve for the time from the prior NSD in the validation set was 0.70 (0.67–0.72), similar to that observed in INTRINSIC RV.

Figure 3

Receiver operator curve depicting 14 min cutoff point from NSD to inappropriate shock delivery in the ALTITUDE-REDUCES Study.


In this analysis of data from INTRINSIC RV, NSD episodes detected by the ICD served as a marker for the subsequent delivery of inappropriate shock therapy. This finding represents the first attempt, to date, to identify a specific variable that can predict which patients are about to receive an inappropriate ICD shock.

Inappropriate (‘unnecessary’) therapy remains one of the most critical issues, and deterrents, when considering ICD implantation and programming. Not only do such shocks cause pain and adversely affect the quality of life, but they can be pro-arrhythmic, drain ICD battery, and be potentially life-threatening. Recent data from the ALTITUDE Study showed that patients with ICDs who received their first shock appropriately for ventricular rhythms or inappropriately for atrial fibrillation had an increased risk of death (compared with no shock). Inappropriate shocks for sinus tachycardia, noise, artifact, or oversensing were not associated with survival.23 Unnecessary shocks may also be delivered for episodes of otherwise self-terminating VT.

The algorithms assessing tachycardias have not yet proven themselves capable of reliably discriminating supraventricular tachycardia from VT and thus inappropriate ICD shocks may occur. Non-sustained or diverted episodes predict inappropriate shock delivery. Such episodes may be useful for developing future algorithms that could auto-adjust and lengthen the detection window temporarily to reduce the frequency of inappropriate and thus unnecessary ICD shocks. Through remote monitoring, NSD detection may provide a red flag such that early surveillance may allow healthcare providers a window to adjust antitachycardia or other programming before an inappropriate shock ensues. Other methodologies have been considered in attempts to reduce inappropriate ICD shocks. The Primary Prevention Parameters Evaluation (PREPARE) study utilized a long detection window and specific high-rate parameters to reduce episodes of inappropriate ICD shocks.24 This, plus utilization of antitachycardia pacing algorithms, which delay the delivery of a shock, allowing more time for potential spontaneous conversion that may obviate the need for a shock, can reduce appropriate and inappropriate shocks and may reduce total shock burden.25

Devices differ in how they classify appropriate and inappropriate shocks. In the Rhythm ID Going Head to Head Trial (RIGHT),26 inappropriate detections were seen with both Medtronic and Boston Scientific devices. More inappropriate detections were seen with Boston Scientific devices, but that effect was highly dependent on the detection rate of the arrhythmia with more differences seen with slower rhythms. To date, no device is 100% accurate in its ability to discriminate supraventricular tachycardias from VT. Discrimination algorithms, including the analyses of atrial and ventricular dissociation, are not 100% effective to determine when a patient should or should not receive an ICD shock.2729 It is not completely clear if a dual-chamber device that superior to a single-chamber device that determine the need for an ICD shock.

The Multicenter Automatic Defibrillator Implantation Trial—Reduce Inappropriate Therapy (MADIT-RIT),30 showed that programming tachycardia detection criteria with high-rate cut-off or prolonged duration can increase the time to first inappropriate ICD activation substantially for patients who receive dual-chamber (or cardiac resynchronization) ICDs for primary prevention purposes. Nevertheless, over the long-term, inappropriate ICD activations remain a problem for patients who receive primary or secondary prevention ICDs.


This is a post hoc analysis derived from a study not specifically designed to assess endpoints reported within this manuscript. However, this is a large population of patients, representing a variety of programming characteristics (with varying tachycardia rates and zones), allowing for real-life assessment of outcomes with regard to appropriate and inappropriate shocks. The goal of the present study was not to assess specific mortality or heart failure outcomes but to look at the methods to detect presence or absence of an inappropriate or appropriate shock based upon NSD episodes as a predictor. Most patients did not receive ICD activations, in particular, shocks. Therefore, despite the large size of this study, the dataset of the participants receiving shocks is rather small. It is possible that the study was underpowered to utilize NSD to predict the appropriate shocks; however, we found similar results when evaluating the LATITUDE dataset. We did not assess medication adherence which may also affect the outcomes. We do note that more participants receiving appropriate therapy were prescribed antiarrhythmic drugs; this may reflect a higher likelihood that these patients had important VT or VF episodes or may have potentially decreased the likelihood of developing NSD episodes.


Inappropriate ICD shocks are common, potentially harmful, and difficult to prevent. Non-sustained or diverted episodes did not increase before appropriate ICD shocks but increased significantly in the 24 h period before inappropriate shocks, particularly within 14 min prior to an inappropriate shock. While the time-window between NSD episode and inappropriate shock was short, knowledge of NSD episodes may allow for future device enhancements that can help reduce the risk of inappropriate shocks.

Conflicts of interest: R.M.S. has no conflicts to declare; M.S. has received salary (>10 000) from Boston Scientific; K.B. has received salary (>10 000) from Boston Scientific; K.Q.S. has received salary (>10 000) from Boston Scientific; P.W.J. has received salary (>10 000) from Boston Scientific; A.M.R. has received honoraria from Medtronic, St Jude, Boston Scientific, Biotronik; she was on the advisory board of Cameron Health and conducted research for Cameron Health; she also conducted research for Medtronic (each < 10 000). F.R.G. was a consultant to Boston Scientific and Cameron Health (each < 10 000); B.O. was a consultant to Medtronic and also conducted research for Medtronic, consultant to Boston Scientific and also conducted research for Boston Scientific—DSMB, consultant to Amarin—DSMB, consultant to BioControl and also conducted research for BioControl, consultant to Sanofi-Aventis—DSMB, Boehringer Ingleheim, and Executive Health Resources (each < 10 000).


View Abstract