Author + information
- Bruce D. Lindsay, MD∗ ()
- ↵∗Reprint Requests and Correspondence:
Dr. Bruce D. Lindsay, Cleveland Clinic Foundation, Cardiovascular Medicine, 9500 Euclid Avenue/J2-2, Cleveland, Ohio 44195.
From the time of the initial clinical report in 1980 (1), implantable cardioverter-defibrillators (ICDs) have undergone testing to confirm that the devices can detect and terminate ventricular fibrillation (VF). The initial devices were implanted in the operating room and tested before discharge in the electrophysiology laboratory. As the technology evolved, defibrillation threshold testing was performed in the operating room or the electrophysiology laboratory at the time of implantation. The concern was that an ICD might interact with a coexisting pacemaker, that the ICD might not sense ventricular tachycardia that fell below the rate detection criterion, that it might not detect low-amplitude signals during VF, or that the energy required for defibrillation might exceed the maximal output of the ICD (2). With the development of improved lead technology, biphasic shocks, programmable waveforms, higher delivered energy, programmable sensitivity, and integration of pacemakers into ICDs, the probability of failure has been reduced.
In recent years, the need to perform defibrillation threshold testing has been questioned because failure rates are low, the results of testing may not predict clinical outcomes accurately, and there is some additional risk (3–5). Moreover, the concept of testing the defibrillation threshold is only an estimate of success, because the defibrillation threshold is a probabilistic term as opposed to a fixed amount of energy. A common practice is to test 1 or more shocks at an energy at least 10 J lower than the maximal output of the ICD system. Advocates of testing termination of VF by an ICD argue that even if the failure rate is low, the reported incidence of failure to convert with a <10-J safety margin is in the range of 4% to 11% (3,4,6). One may argue that physicians who implant ICDs have an obligation to determine whether the selected lead configuration can terminate VF, because there are techniques to improve thresholds. An extensive study in Canada demonstrated that the risk for serious complications or mortality associated with testing conversion of VF is limited to 0.042% (7). Those who advocate defibrillation threshold or defibrillation safety margin (DSM) believe that the benefits outweigh the risk.
If we knew which patients were most likely to have low energy requirements for defibrillation testing and which were at risk for failed conversion, we could focus testing on those at greatest risk. Hsu et al. (8) analyzed the National Cardiovascular Data Registry ICD Registry to identify factors predictive of an inadequate DSM, which was defined as lowest energy tested <10 J from the maximal device output, and they evaluated the association of an inadequate DSM with adverse events. Among the 12,397 patients in whom DSMs were tested, they were inadequate in 9.4%. Of the 337,547 patients considered for analysis, the investigators excluded patients who underwent generator replacement without new leads, those in whom defibrillator threshold testing was not performed (17.7%), and a small number in whom the results were not known. The analysis was focused on the remaining 132,477 patients (39% of the registry patients considered for analysis) from 1,457 facilities in whom new leads were inserted and testing was performed. Within this group, they found that inadequate DSMs were observed in 9.4%. Although the percent of patients with inadequate DSMs is higher than in some reports, it is within the range of values from other published studies.
There are several factors that cannot be determined from the ICD Registry. The rationale for testing some patients but not others is not available, and the impact of drugs on defibrillation thresholds cannot be assessed, because it is not a required database element. There could be a bias in the selection of patients for defibrillation testing, yet we are left with useful data on a large number of patients. Moreover, even within the limitations of the registry data, it is evident that an inadequate DSM remains problematic.
In this issue of the Journal, Hsu et al. (8) found that the attributes associated with an inadequate DSM are complex and include age, sex, race, New York Heart Association functional class, absence of ischemic heart disease, current renal dialysis, indication for secondary prevention, and even whether the ICD was a single-chamber device or a cardiac resynchronization therapy defibrillator (8). Using a scoring based on 16 variables, they found that the risk for an inadequate DSM increased from 4.9% in the patients at lowest risk to 24.5% in those with high risk scores. The ICD Registry does not include structural data derived from echocardiography or any estimates of left ventricular mass. Although a scoring system such as this is imperfect, it would allow some risk estimation that could guide physicians on the need for testing, appropriate precautions, and the possibility that a subcutaneous coil might be required. The investigators also determined that higher risk scores were associated with slight increases in procedural complications, length of stay, and in-hospital mortality. On the basis of these findings, some low-volume centers might chose to refer patients with higher risk scores to more experienced centers for implantation procedures.
In response to the Centers for Medicare and Medicaid Services mandate for an ICD registry, the American College of Cardiology and the Heart Rhythm Society accepted the responsibility to track implantation data by developing the National Cardiovascular Data Registry ICD Registry. The expectation was that clinical insights could be gained from analysis of a large number of patients and that it might serve to measure quality outcomes. The sheer volume of data acquired by the ICD Registry dwarfs the numbers generated by single-center or even multicenter studies. The limitations are that the database elements do not meet the needs for all inquiries, and only about 10% of the data are audited for accuracy. Nonetheless, audits have determined that the quality of the data is quite good, and within the limits of any registry, it lends itself to useful clinical studies.
The observations reported by Hsu et al. (8) undoubtedly will provoke some debate, but it is a legitimate line of inquiry that serves as an example of how the ICD Registry can be used to develop strategies that improve patient outcomes.
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
Dr. Lindsay has served on the Medtronic Scientific Advisory Board and has contributed as a speaker to educational programs, honoraria related to these activities are <$10,000; and he also serves on the National Cardiovascular Data Registry Management Board.
- American College of Cardiology Foundation
- Winkle R.A.
- Healey J.S.,
- Gula L.J.,
- Birnie D.H.,
- et al.
- Strickberger S.A.,
- Klein G.J.
- Hsu J.C.,
- Marcus G.M.,
- Al-Khatib S.M.,
- et al.