Cardiac Electrophysiology Review 2001;5:41–46©C 2001 Kluwer Academic Publishers. Manufactured in The Netherlands.

Comparative Features, Functions and Longevitiesof Implantable Cardioverter Defibrillators

Mohammad Saeed, Craig Swygmanand N.A. Mark Estes IIINew England Cardiac Arrhythmia Center, New EnglandMedical Center, Boston, MA

Introduction

Implantable cardioverter defibrillator (ICD) tech-nology has rapidly evolved since being introducedclinically in 1980 [1]. Advances in several key ar-eas have enabled ICDs to provide more sophisti-cated rhythm management [2]. Several companiesmanufacture ICDs, and each company has its mod-els and versions. Although the devices have cer-tain similarities, and generalization can be maderegarding the functions and operational charac-teristics, each device has unique features andphysical characteristics. Knowledge of these dif-ferences can help in selecting the appropriate ICDsuited to the individual patient needs. The pur-pose of this review is to compare and highlight themajor differences among the current ICDs man-ufactured by various companies. The differenceswill be discussed in general, as they pertain toa device platform made by a particular manufac-turer. Wherever it is necessary, specific individualexamples will be used.

Tachyarrhythmia Detection

The basic tachyarrhythmia detection in all the cur-rently available ICDs is based on the cardiac cyclelength. Although similar in this respect, each man-ufacturer has certain additional criteria for initialdetection and redetection that are unique to theirdevices.

The CPI Ventak Prism™ (Guidant, St Paul,MN, USA) uses both rate and duration for tachy-arrhythmia detection. Each programmed tachy-arrhythmia zone has an associated detection win-dow that consists of 10 most recent R-R intervals.Each new interval is compared to the programmedrate threshold and classified as either “fast” or“slow”. When 8 of 10 intervals in a zone’s windowhave been classified as “fast”, the detection win-dow is considered satisfied. The detection will re-main satisfied as long as 6 of 10 intervals remainclassified as “fast”. A duration timer (1–60 sec) be-gins when its respective zone detection windowis satisfied. When a zone’s duration timer expires

and if the last detected interval is in that zone,detection is considered met and therapy is deliv-ered. Shock therapy will be delivered, after capac-itor charge, reconfirmation and R wave synchro-nization. The same number of events are requiredto satisfy the redetection window, but the dura-tion timer is typically shorter (ventricular fibrilla-tion [VF] = 1 sec, ventricular tachycardia [VT] =1–15 sec).

The Medtronic GEM II™ (Medtronic Inc.,Minneapolis, MN, USA), detects VF when the cou-pling interval of 75% of a programmable numberof signals (NID = number of intervals to detect,nominal = 18/24) is within the programmable VFzone. The first shock is delivered following the ca-pacitor charge time, reconfirmation and R wavesynchronization. The redetection algorithm is dif-ferent from the detection algorithm in that the re-detect NID is smaller (nominal = 12/16), the sec-ond and subsequent shocks are committed, andshock synchronization is not required. In the caseof R-R intervals falling in the VT zone, the ICDcounts consecutive VT events towards the VT ini-tial or VT redetect NID, as opposed to countingthe ratio as is the case with VF zone. The VT eventcounter resets to zero whenever an interval fails tosatisfy the VT interval criteria, while it remainsat its current value if an interval is shorter andfalls into the VF zone.

The Ventritex Profile™ (St. Jude Medical,Sylmar, CA, USA) follows a different algorithm fortachyarrhythmia detection. A sensed event countstowards the initial detection of VF if: (1) the cou-pling interval is less than the fibrillation detectioninterval (FDI), and (2) the running average of thelast four intervals (internal average) is also lessthan FDI. Once a programmable number of indi-vidual events (nominal = 12) satisfy these crite-ria, VF is detected. Events that do not meet these

Address correspondence to: Mohammad Saeed, M.D., NewEngland Medical Center, 750 Washington Street, Box 197,Boston, MA 02111. E-mail: msaeed@lifespan.org

41

42 Saeed, Swygman and Estes III CEPR 2001; Vol. 5, No. 1

criteria are either counted towards “redetection”of sinus rhythm (if both the interval and the inter-val average rate are >50 ms above the FDI), or areignored (if one criteria and not the other is satis-fied). If a programmable number of events satisfysinus rhythm criteria (nominal = 5, “slow” = 7),the fibrillation detection event counter is reset tozero. A similar interval and interval average cri-teria is used for redetection, except only six eventsneed to be counted for redetection. After success-ful detection or redetection, the capacitors chargeand an R wave synchronous charge is deliveredafter VF is reconfirmed.

The detection enhancements available in mostcurrent ICDs, e.g., onset and stability may be usedduring initial detection to add specificity and limitinappropriate therapy for sinus tachycardia orrapidly conducting atrial fibrillation [3–7]. Theiruse in actual practice has been limited so farbecause of the potential risk of lowering thesensitivity of treating ventricular tachyarrhyth-mia [8]. By using sudden onset criteria underde-tection of VT may range from 0.5–5.0% in sinusrhythm and to 13% in atrial fibrillation [9,10]. Alsolong-short initiation mechanism of VT may resultin underdetection of VT in some cases. In a studyutilizing sudden onset criterion in combinationwith the stability criterion, the sensitivity andspecificity of correctly identifying VT was 90% and93% respectively [11]. In dual-chamber devices,algorithms utilizing atrial rate and relationshipof atrial to ventricular electrograms can add fur-ther discriminatory power for diagnosing VT [12].Other available detection enhancements to dis-criminate supraventricular from ventriculartachyarrhythmias include electrogram width cri-teria on some Medtronic devices and beat to beatmorphology discriminator on some St. JudeMedical and Medtronic devices.

In order to prevent unnecessary shocks for non-sustained ventricular arrhythmias, all current de-vices are designed to reconfirm the tachyarrhyth-mia by measuring the R-R interval during thecharging process and before delivering therapy.Reconfirmation criteria are typically less stringentthan initial detection, which can make them moresusceptible to errors. There have been reports ofdelivery of shock therapy for non-sustainedepisodes because of presence of paced beats or pre-mature beats during reconfirmation period [13]. Ifthe first shock fails to terminate the arrhythmia,devices are designed to redetect and deliver ther-apy up to 6–8 times. In case of Medtronic prod-ucts, all the therapies except for the first shockare committed and do not require reconfirmation,while in case of Guidant and St. Jude Medical allthe therapies are reconfirmed before delivery oftherapy.

Tachyarrhythmia Therapy

The modern defibrillator has come a long way fromthe early devices that were capable of recogniz-ing only VF and delivered committed high-energyshock therapy. Current ICDs are not only more so-phisticated in detecting and discriminating amongvarious tachycardia, but can also deliver tieredtherapy consisting of anti-tachycardia pacing(ATP), low energy cardioversion (LES) and highenergy defibrillation shock. Ventricular rate sta-bilization feature available in some Medtronic de-vices is designed to prevent the occurrence ofshort-long-short sequences of ventricular cyclelengths that have been clinically observed to pre-cede the onset of some spontaneous ventriculartachyarrhythmias [14–15].

The different devices available vary in the num-ber of maximum high energy defibrillation shocksthat can be delivered after first failed shock. TheCPI Ventak Prism™ can deliver up to a total of 8high energy shocks for a single tachyarrhythmiaepisode, out of which first two are programmable,while Medtronic GEM II™ and Ventritex Profile™can deliver up to a total of 6 programmable highenergy shocks. Also there is slight variation in thecharge time and maximum energy output amongthe devices from various manufacturers (seetable 1) [16]. This can translate into clinically im-portant difference for patients with poorly toler-ated arrhythmias or high defibrillation threshold.

Most of the current devices allow the type(monophasic or biphasic) and the polarity of theshock wave to be programmed. In devices manu-factured by Guidant the distal shocking coil actsas the cathode, while the proximal shocking coiland the pulse generator act as the anode, in nom-inal configuration. This configuration can be re-versed to make the distal coil act as the anode andthe combination of proximal coil/pulse generatoract as the cathode. The nominal polarity settingin Medtronic devices includes the pulse generatorand the proximal coil as cathode, and distal coilas anode (AX → B, A = active can, X = proximalcoil, and B = distal coil). This setting is programm-able and can be reversed (B → AX). St. Jude Med-ical devices also use distal coil as anode, while theproximal coil and the pulse generator serve as thecathode in the nominal setting. Polarity can alsobe reversed in St. Jude Medical devices.

Bradycardia Therapy

All present day ICDs are equipped with ventricu-lar antibradycardia pacing. This back up VVI pac-ing is useful for intermittent or post-shockbradyarrhythmias but may be hemodynamically

CEPR 2001; Vol. 5, No. 1 Comparison of ICDs 43

Table 1. Comparative features of implantable cardioverter defibrillator

Manufacturer Max† Detection Stored EGM Vol Wt(Model) DF/CV/ATP Pacing Output (J) Waveform Enhancements EGM Storage (min) (cc) (gm)

CPI Guidant

Ventak P2/P3* +, +, − VVI 39 mono/bi none none none 105 233Ventak PRx* +, +, + VVI 39 mono/bi stb, sud none none 130 220Ventak PRx II* +, +, + VVI 39 mono/bi stb, sud none none 130 220Ventak PRx III* +, +, + VVI 39 mono/bi stb, sud none none 105 182Ventak Mini* +, +, + VVI 33 mono/bi stb, sud V 5 68 125Ventak Mini II* +, +, + VVI 30 mono/bi stb, sud V 5 59 115Ventak Mini III +, +, + VVI 31, 36 (HE) mono/bi stb, sud V 5 48 90Ventak Mini IV +, +, + VVI 31 mono/bi stb, sud V 5 39 78Ventak VR +, +, + VVIR 31 mono/bi stb, sud V 16 51 94Ventak AV II DR +, +, + DDDR 31 mono/bi stb, sud, AV A, V 16 73 136Ventak AV III DR +, +, + DDDR 31 mono/bi stb, sud, AV A, V 16 58 108Ventak Prism VR +, +, + VVIR 31 mono/bi stb, sud V 19 38 98Ventak Prism DR +, +, + DDDR 31 mono/bi stb, sud, AV A, V 19 39 99

Medtronic

PCD* +, +, + VVI 34 mono stb, sud none none 119 197Jewel* +, +, + VVI 34 mono/bi stb, sud V — 80 129Jewel CD* +, +, − VVI 34 mono/bi stb, sud V — 80 129Jewel Plus* +, +, + VVI 34 mono/bi stb, sud V 15 80 131MicroJewel +, +, + VVI 34 mono/bi stb, sud V 15 78 125MicroJewel II +, +, + VVI 30 mono/bi stb, sud V 15 54 97Gem +, +, + VVI 35 mono/bi stb, sud V 15 49 90Gem DR +, +, + DDDR 35 mono/bi stb, sud, AV A, V 22 62 115Gem II DR +, +, + DDDR 30 mono/bi stb, sud, AV A, V 22 39.5 77Gem II VR +, +, + VVVR 30 mono/bi stb, sud V 22 39 77Jewel AF +, +, + DDD 27 mono/bi stb, sud, AV A, V 22 55 93

Ventritex

Cadence* +, +, + VVI 42 bi none V 1.5 132 198Cadet* +, +, + VVI 42 bi none V 8 73 129Contour +, +, + VVI 42 bi stb, sud V 16 57 109Contour II +, +, + VVI 42 bi stb, sud V 16 57 109Contour MD +, +, + VVI 42 bi stb, sud, MD V 16 57 109Angstrom II +, +, + VVI 31 bi stb, sud V 16 44 90Angstrom MD +, +, + VVI 31 bi stb, sud, MD V 16 44 90Profile MD +, +, + VVI 32 bi stb, sud, MD V 16 34 70Photon¶ +, +, + DDDR 33 mono/bi stb, sud, MD A, V 25 46 85

ELA Medical

Defender IV DR +, +, + DDDR 33 bi stb, sud, AV A, V varies 75 140

Biotronik

Phylax XM +, +, + VVI 30 mono/bi stb, sud V 16 69 104Microphylax +, +, + VVI 30 mono/bi stb, sud V 16 54 89Phylax AV +, +, + DDD 30 mono/bi stb, sud, AV A, V — 69 109

†Stored energy, ¶To be released in 2nd half of 2000, *Product no longer available.A = atrial, ATP = antitachycardia pacing, AV = atrio-ventricular discrimination algorithm, CV = cardioversion, DF = defibrillation,EGM = electrogram, J = joules, MD = morphology discrimination, stb = stability, sud = sudden onset and V = ventricular.

detrimental in patients who require frequent pac-ing. Until recently, up to 15% of patients requiredimplantation of a separate dual-chamber or rate-responsive pacemaker. Newer dual-chamber ICDshave a rate-responsive dual-chamber pacemaker

incorporated in them. With the availability ofICDs capable of dual-chamber pacing the need fora separately implanted permanent dual-chamberpacemaker, which is associated with a potential forpacemaker-ICD interactions has been eliminated

44 Saeed, Swygman and Estes III CEPR 2001; Vol. 5, No. 1

[17]. Of the three leading manufacturers in UnitedStates, Guidant and Medtronic already have dual-chamber ICDs approved for clinical use while St.Jude Medical has a dual-chamber device in clin-ical trial which will be available in later part of2000. (See Chapter 3).

Sensing

To sense low-amplitude signals reliably duringventricular fibrillation and avoid sensing of Twaves or extracardiac noise, the sensing circuit inthe ICD automatically adjusts either the gain orthe sensing threshold [18]. Devices manufacturedby Guidant and St. Jude Medical both utilize au-tomatic gain control systems, with one differencebeing that in St. Jude Medical ICDs the gain isnot adjusted on a beat-to-beat basis, but instead isadjusted based on a sampling of the amplitude ofthe previously sensed signals. Medtronic deviceson the other hand use an automatically adjust-ing sensitivity on a beat-to-beat basis. In future,newer St. Jude Medical devices (Photon™) willhave a new sensing algorithm that utilizes auto-matic sensitivity adjustment.

There have been reports of oversensing in de-vices that utilize automatic gain control for sen-sitivity adjustment [19–20]. The risk is probablylower in devices that adjust automatic gain on abeat-to-beat basis, and in devices with more rapiddecrement of automatic gain level [21]. Sensing ofmyopotentials during bradycardia pacing (maxi-mal gain) is common in ICDs systems that uti-lize automatic gain control [22]. Oversensing maylead to clinically significant events, such as sup-pression of bradycardia pacing or triggering of in-appropriate tachycardia therapy in the absenceof ventricular tachyarrhythmias in pacemaker-dependent patients.

Another feature available in recent devices isthe ability to measure high voltage lead systemimpedance using pacing pulses, rather than de-livering low energy synchronized test shocks [23].This is helpful not only at the time of implant butis also useful in following the impedance changesin the high voltage lead system over time whichmay occur clinically, thus eliminating the need todeliver a shock for impedance measurement.

Physical Characteristics

Size and shape of the device is sometimes an im-portant factor when selecting the device for indi-vidual patients. With the advances in ICD tech-nology the size of the devices has been graduallyreduced. In addition to making them smaller, at-tention has also been placed on making them morephysiologic in shape, so as to minimize patient dis-

comfort and enable easier implantation. Of all thesingle-chamber ICDs currently available in clini-cal practice, the Ventritex Profile™ is the smallestin size, while among the dual-chamber ICDs theMedtronic GEM II DR™ is the smallest (see table).With future development further reduction in sizeis expected.

Diagnostic Storage

There has been a rapid and significant evolution inthe stored diagnostic information available fromICDs [24]. With the advent of dual-chamber ICDs,it is now possible to store atrial rates and electro-grams (EGM) with marker channels. This diag-nostic information has greatly enhanced the clin-ician’s ability to determine the rhythm triggeringdevice therapy as well as to identify potential prob-lems with the ICD system. Furthermore, this in-formation may be useful in identifying triggers ofventricular arrhythmias in patients at high riskfor sudden death [25].

In CPI Ventak Prism™, the stored EGM is par-titioned into 3 segments; episode onset, pre-therapy and post-therapy. Episode onset refers tothe 10 sec of EGM prior to the episode declaration.Pre-therapy storage provides up to 10 sec of in-formation ending with therapy delivery (includescharge and reconfirmation). If the pre-therapy timeexceeds 10 sec, the last 10 sec of data is retained.The post-therapy EGM storage starts followingtherapy delivery and stores up to 10 sec of EGM.The EGM source can be ventricular or shock elec-trode, or both. In case of dual-chamber device (Ven-tak Prism DR™) additional option of storing EGMfrom atrial electrode as a third channel is alsoavailable. In addition EGM from high rate atrialepisodes leading to mode switching can also berecorded separately. A total of 19 minutes of elec-trogram can be recorded from a single source. Us-ing multiple sources will decrease the total storagecapacity. When the storage memory is full, the de-vice overwrites the older EGM data segments.

The Medtronic GEM II™ allows EGM storageof up to two channels. The storage is triggered bythe third arrhythmia interval. EGM can be storedduring charging, or before onset of arrhythmia orboth. Prolonged EGM storage of up to 100 R-Rintervals prior to last tachyarrhythmia onset isalso available but decreases device longevity. TheEGM source has 5 different programmable options(10 options in dual-chamber device). With variousconfigurations dedicated bipolar, integrated bipo-lar, far-field or unipolar sensing is possible. Thedevice can store up to 12 minutes of EGM fromtwo channels or 22 minutes of EGM from a singlechannel.

CEPR 2001; Vol. 5, No. 1 Comparison of ICDs 45

The Ventritex Profile™ can store up to 16 min-utes of EGM from a single channel. The EGM stor-age is programmable and can be triggered byeither third beat of tachyarrhythmia, delivery oftherapy or redetection of sinus rhythm after atachyarrhythmia. The single EGM source can beselected as either near-field or far-field ventricu-lar. Stored EGM are annotated with marker chan-nel data and state of automatic gain. Once storagememory is full, the earlier events will be replacedwith new events.

Longevity

It is very difficult to objectively compare thelongevity of different ICDs available because ofdearth of scientific data on this subject, exceptfor the information provided by the device manu-facturers themselves about their products. In ad-dition because of lack of a standardized way ofreporting expected device longevity among themanufacturers, it is hard to draw any firm conclu-sions. Part of the problem is due to the fact thatbattery life can be influenced by several differentfactors, including pacing parameters and pacingrate, frequency of therapy and amount of diag-nostic information storage enabled. The longevityof the pulse generator decreases with an increasein the pacing rate, pacing pulse amplitude, pac-ing pulse width, percentage of bradycardia pacedto sensed events, or charging frequency or with adecrease in pacing impedance. Higher frequencyof capacitor reformation will also adversely affectbattery life. Capacitor reformation is a program-mable feature in all the devices. Guidant deviceshave a nominal setting of 3 months, whileMedtronic and St. Jude Medical have nominal set-ting of 6 and 3 months respectively.

Conclusion

Present day ICDs are sophisticated rhythmmanagement devices with better diagnostic, ther-apeutic and storage capabilities than their prede-cessors in spite of overall reduction in size andprolongation of battery life. Although all ICDs aregenerally similar, specific differences exist amongdevices from various manufacturers. Knowledgeof the various features available, and an under-standing of differences among the current devicescan aid in selecting the appropriate ICD for an in-dividual patient.

References

1. Mirowski M, Reid PR, Mower MM, Watkins L, GottVL, Schauble JF, Langer A, Heilman MS, KolenikSA, Fischell RE, Weisfeldt ML. Termination of ma-

lignant ventricular arrhythmias with an implantedautomatic defibrillator in human beings. N Engl JMed 1980;303:322–324.

2. Morris MM, KenKnight BH, Warren JA, Lang DJ.A preview of implantable cardioverter defibrillatorsystems in the next millennium: An integrative car-diac rhythm management approach. Am J Cardiol1999;83(5B):48D–54D.

3. Swerdlow CD, Chen PS, Kass RM, Allard JR,Peter CT. Discrimination of ventricular tachycar-dia from sinus tachycardia and atrial fibrillationin a tiered-therapy cardioverter-defibrillator. J AmColl Cardiol 1994;23:1342–1355.

4. Brugada J, Mont L, Figueiredo M, Valentino M,Matas M, Navarro-Lopez F. Enhanced detectioncriteria in implantable defibrillators. J CardiovascElectrophysiol 1998;9(3):261–268.

5. Barold HS, Newby KH, Tomassoni G, Kearney M,Brandon J, Natale A. Prospective evaluation of newand old criteria to discriminate between supraven-tricular and ventricular tachycardia in implantabledefibrillators. PACE 1998;21(7):1347–1355.

6. Weber M, Bocker D, Bansch D, Brunn J, CastrucciM, Gradaus R, Breithardt G, Block M. Efficacy andsafety of the initial use of stability and onset cri-teria in implantable cardioverter defibrillators. JCardiovasc Electrophysiol 1999;10(2):145–153.

7. Nanthakumar K, Paquette M, Newman D, DenoDC, Malden L, Gunderson B, Gilkerson J, GreeneM, Heng D, Dorian P. Inappropriate therapy fromatrial fibrillation and sinus tachycardia in auto-mated implantable cardioverter defibrillators. AmHeart J 2000;139(5):797–803.

8. Reiter MJ, Mann DE. Sensing and tachyarrhyth-mia detection problems in patients with im-plantable cardioverter defibrillators. J CardiovascElectrophysiol 1996;7:542–558.

9. Neuzner J, Pitschner HF, Schlepper M. Pro-grammable VT detection enhancements in im-plantable cardioverter defibrillator therapy. PacingClin Electrophysiol 1995;18:539–547.

10. Swerdlow CD, Ahern T, Chen PS, Hwang C, GangE, Mandel W, Kass RM, Peter CT. Underdetectionof ventricular tachycardia by algorithms to enhancespecificity in a tiered-therapy cardioverter defibril-lator. J Am Coll Cardiol 1994;24:416–424.

11. Brugada J. Is inappropriate therapy a resolved is-sue with current implantable cardioverter defibril-lators? Am J Cardiol 1999;83:40D–44D.

12. Nair M, Saoudi N, Kroiss D, Letac B. Automatic ar-rhythmia identification using analysis of the atri-oventricular association: application to a new gen-eration of implantable defibrillators. Circulation1997;95:967–973.

13. Grimm W, Menz V, Hoffmann J, Maisch B. Failure ofthird-generation implantable cardioverter defibril-lators to abort shock therapy for nonsustained ven-tricular tachycardia due to shortcomings of the VFconfirmation algorithm. PACE 1998;21(4, Part 1):722–727.

14. Kay GN, Plumb VJ, Arciniegas JG, Henthorn RW,Waldo AL. Torsades de pointes: The long-short ini-tiating sequence and other clinical features: Obser-vations in 32 patients. J Am Coll Cardiol 1983;2:132–133.

46 Saeed, Swygman and Estes III CEPR 2001; Vol. 5, No. 1

15. Cranefield PF, Aronson RS. Torsades de pointesand other pause-induced ventricular tachycardia:The long-short-long sequence and early after-depolarizations. PACE 1998;11:670–678.

16. Mann DE, Kelly PA, Robertson AD, Otto L, ReiterMJ. Significant differences in charge times amongcurrently available implantable cardioverter defib-rillators. PACE 1999;22(6 Pt 1):903–907.

17. Brode SE, Schwartzman D, Callans DJ, GottliebCD, Marchlinski FE. ICD-antiarrhythmic drug andICD-pacemaker interactions. J Cardiovasc Electro-physiol 1997;8:830–842.

18. Jones GK, Bardy GH. Considerations for ven-tricular fibrillation detection by implantable car-dioverter defibrillators. Am Heart J 1994;27:1107–1110.

19. Peters RW, Cooklin M, Brockman R, Shorofsky SR,Gold MR. Inappropriate shocks from implanted car-dioverter defibrillators caused by sensing of di-aphragmatic myopotentials. J Intervent CardiacElectrophysiol 1998;2(4):367–370.

20. Babuty D, Fauchier L, Cosnay P. Inappropriateshocks delivered by implantable cardiac defibrilla-tors during oversensing of activity of diaphagmaticmuscle. Heart 1999;81(1):94–96.

21. Mann DE, Damle RS, Kelly PA, Landers M, Otto

L, Reiter MJ. Comparison of oversensing duringbradycardia pacing in two types of implantablecardioverter-defibrillator systems. Am Heart J1998;136(4 Part I):658–663.

22. Rosenthal ME, Paskman C. Noise detection dur-ing bradycardia pacing with a hybrid nonthora-cotomy implantable cardioverter defibrillator sys-tem: Incidence and clinical significance. PACE1998;21(7):1380–1386.

23. Mehdirad AA, Stohr EC, Love CJ, Nelson SD,Schaal SF. Implantable defibrillators impedancemeasurement using pacing pulses versus shock de-livery with intact and modified high voltage leadsystem. PACE 1999;22(3):437–441.

24. Sarter BH, Callans DJ, Gottlieb GD, Schwartz-man DS, Marchlinski FE. Implantable defibrillatordiagnostic storage capabilities: Evolution, currentstatus, and future utilization. PACE 1998;21(6):1287–1298.

25. Auricchio A, Hartung W, Geller C, Klein H.Clinical relevance of stored electrograms forimplantable cardioverter-defibrillator (ICD) trou-bleshooting and understanding of mechanismsfor ventricular tachyarrhythmias. Am J Cardiol1996;78(Suppl 5A):33–41.