Impact of defibrillation testing on predicted ICD shock efficacy: implications for clinical practice.

BACKGROUND: Lack of consensus regarding defibrillation testing methods for implantable cardioverter-defibrillators relates to risks of repeated fibrillation episodes. OBJECTIVE: To provide recommendations for testing protocols, repeating testing of patients with high defibrillation threshold (DFT), and interpreting testing after implantable cardioverter-defibrillator system revision. METHODS: We constructed a computer model of defibrillation probability-of-success curves using data from 564 patients. Then, we compared 13 safety margin (SM) or DFT protocols in 50,000 simulated patients to identify those with the best balance of sensitivity and predictive value for detecting patients at high risk for failed defibrillation. Conditional retesting of patients with high DFT was simulated, both without and with revision that lowered defibrillation energy by one-third. RESULTS: SM protocols were more efficient than DFT protocols; 2/2 successes at 20 J or 1/1 at 16 J performed best. Patients who failed testing had a mean probability of defibrillation of 94% at 35 J, but great uncertainty regarding that probability (range 67.0%-100%). When they repeated testing, 62% passed, with 48% owing to regression to the mean. If system revision was performed before retesting, 84% passed; the fraction of patients at high risk reduced (4.7% to 2.7%, with 43% relative reduction); but 3.5% underwent unnecessary revisions. Testing and revision of patients with high DFT benefitted 2.5% of the patients. CONCLUSIONS: SM protocols are superior to DFT protocols for implant testing. For patients who fail testing, there is substantial uncertainty in defibrillation efficacy. After a system revision that does not alter defibrillation efficacy, 62% of these patients pass retesting.

Impact of defibrillation test protocol and test repetition on the probability of meeting implant criteria.

BACKGROUND: Defibrillation testing is a common procedure at defibrillator implant, with the purpose to ensure that each patient receives a device-lead system with a sufficient shock efficacy. The objective of this paper was to study the influence of defibrillation test protocols on the probability of passing implant testing. METHODS: Defibrillation shock efficacy as a function of shock energy was modeled by a dose-response relationship estimated from the clinical data of the PainFREE Rx II study on 564 patients. A Monte Carlo method was used to simulate the outcomes of 12 commonly used defibrillation efficacy test protocols: four safety margin tests and eight protocols estimating the defibrillation threshold (DFT). RESULTS: The probabilities of failing 20-J and 25-J implant criteria for the different protocols ranged from 0.9% to 6.3% for 20 J and 0.3% to 3.4% for 25 J. Large variations in consecutively measured DFT values in the same patients were observed. Best results in the identification of "high risk" patients were obtained with the 2/2 safety margin protocol with an implant criterion of 20 J. The study also showed that the probability of patients inappropriately passing the implant criterion increased when the defibrillation test was repeated after initial failure. CONCLUSION: The defibrillation test protocol greatly influences the probability of meeting implant criterion. Therefore, these test protocols should be standardized. The model developed in this study, could be used to further understand their impact and to derive recommendations.

Estimating the parameter distributions of defibrillation shock efficacy curves in a large population.

Defibrillation efficacy testing is a common procedure at defibrillator implantation, with the objective to ensure that each patient receives a system with adequate shock efficacy. In this study, defibrillation shock efficacy as a function of shock energy was modeled by a dose response relationship. The frequency distributions of the two parameters characterizing this relationship were estimated from a large clinical study on 634 patients. The estimated parameters were then compared to published data on defibrillation studies performing dose response measurements. After having identified the optimal parameter distributions for the whole population, a Monte Carlo method was used to compare the outcomes of two defibrillation efficacy testing protocols: a safety margin test requiring 2 successes out of 2 trials at 20 J, and the binary search protocol used in the clinical study. Results showed good correspondence with clinical observations. The probabilistic nature of defibrillation success and differences in acceptance and rejection rates of the two protocols were highlighted. The model developed in this study could be used to further understand the differences between various defibrillation testing protocols used in clinical practice.

Timing of depolarization and contraction in the paced canine left ventricle: model and experiment.

INTRODUCTION: For efficient pump function, contraction of the heart should be as synchronous as possible. Ventricular pacing induces asynchrony of depolarization and contraction. The degree of asynchrony depends on the position of the pacing electrode. The aim of this study was to extend an existing numerical model of electromechanics in the left ventricle (LV) to the application of ventricular pacing. With the model, the relation between pacing site and patterns of depolarization and contraction was investigated. METHODS AND RESULTS: The LV was approximated by a thick-walled ellipsoid with a realistic myofiber orientation. Propagation of the depolarization wave was described by the eikonal-diffusion equation, in which five parameters play a role: myocardial and subendocardial velocity of wave propagation along the myofiber cm and ce; myocardial and subendocardial anisotropy am and ae; and parameter k, describing the influence of wave curvature on wave velocity. Parameters cm, ae, and k were taken from literature. Parameters am and ce were estimated by fitting the model to experimental data, obtained by pacing the canine left ventricular free wall (LVFW). The best fit was found with cm = 0.75 m/s, ce = 1.3 m/s, am = 2.5, ae = 1.5, and k = 2.1 x 10(-4) m2/s. With these parameter settings, for right ventricular apex (RVA) pacing, the depolarization times were realistically simulated as also shown by the wavefronts and the time needed to activate the LVFW. The moment of depolarization was used to initiate myofiber contraction in a model of LV mechanics. For both pacing situations, mid-wall circumferential strains and onset of myofiber shortening were obtained. CONCLUSION: With a relatively simple model setup, simulated depolarization timing patterns agreed with measurements for pacing at the LVFW and RVA in an LV. Myocardial cross-fiber wave velocity is estimated to be 0.40 times the velocity along the myofiber direction (0.75 m/s). Subendocardial wave velocity is about 1.7 times faster than in the rest of the myocardium, but about 3 times slower than as found in Purkinje fibers. Furthermore, model and experiment agreed in the following respects. (1) Ventricular pacing decreased both systolic pressure and ejection fraction relative to natural sinus rhythm. (2) In early depolarized regions, early shortening was observed in the isovolumic contraction phase; in late depolarized regions, myofibers were stretched in this phase. Maps showing timing of onset of shortening were similar to previously measured maps in which wave velocity of contraction appeared similar to that of depolarization.