Physician Credentials and ICD Implantation: Title and subTitle BreakCertified “Electricians” Best Deal With Electrical Problems

On January 27, 2005, the Centers for Medicare & Medicaid Services published the national coverage decision, which expanded indications for implantable cardioverter-defibrillator (ICD) implantation based largely on the results of the Sudden Cardiac Death in Heart Failure Trial.¹ The Centers for Medicare & Medicaid Services required that patients receiving an ICD for primary prevention of sudden cardiac death be enrolled in a registry.² Under the leadership of the Heart Rhythm Society and the American College of Cardiology Foundation, the National ICD Registry was developed.³

Hospitals have been encouraged not to limit the registry to Medicare beneficiaries but to include all ICD recipients in the registry. In fact, approximately 75% of hospitals have elected to enter all ICD implantations into the registry and 88% of all implants have been included in the registry. Approximately 10 000 ICD implants are added to the registry each month and as of June 2008, 270 373 implants from 1448 hospitals in the United States were entered into the registry.⁴ The registry provides a unique opportunity to examine whether current clinical practice will mirror the results obtained in the clinical trials that have led to the current guidelines for ICD implantation.

Although most ICDs are implanted by board-certified electrophysiologists, a significant proportion are implanted by nonelectrophysiologists and, of those physicians, a considerable number have not fulfilled the requirements for alternate track training proposed by the Heart Rhythm Society.^{4 - 6} The rationale for having nonelectrophysiologists implant ICDs includes the need to increase access to ICDs and to reduce wait times. However, to ensure optimal patient care, it is necessary to maintain training and proficiency standards for physicians who implant ICDs. In the first 2 years of the registry, 3899 physicians implanted 206 604 ICDs with the vast majority (82.8%) of the implants performed by physicians who completed an electrophysiology fellowship (57.5% of all physicians who performed ICD implantation procedures). An additional 16.5% of physicians had met the criteria set out in the Heart Rhythm Society's guidelines and accounted for 9.1% of implants. In addition, 11.3% of physicians who implanted ICDs completed a thoracic or cardiac surgery residency and accounted for 2.3% of the implants.⁴ Moreover, 5.8% of the ICD implantations were performed by physicians who were not certified in the above categories (14.6% of all physicians who performed ICD implantation procedures).

In this issue of JAMA, Curtis and colleagues⁷ used data from the registry to retrospectively evaluate the rates of procedural complications and implantation of ICDs delivering cardiac resynchronization therapy (CRT-D) by physician board certification. Physicians who performed ICD implantation procedures were categorized into 4 mutually exclusive groups (electrophysiologists with board certification, nonelectrophysiologists with cardiovascular board certification, board-certified thoracic surgeons, and other specialists who either had never been certified or allowed their board certification to lapse). Overall and major complication rates were lowest for the electrophysiologist group and highest for the thoracic surgery group but there were major differences in patient characteristics among the 4 groups. Nevertheless, in a multivariable analysis, the adjusted risk of complications was significantly higher among the nonelectrophysiologist cardiologists and thoracic surgeons than the electrophysiologists. For single-chamber ICDs, the total complication rates did not differ among physician certification categories. In a propensity analysis comparing electrophysiologists with nonelectrophysiologist cardiologists and electrophysiologists with thoracic surgeons, the absolute differences in complication rates were comparable with the overall analysis but did not reach statistical significance, perhaps due to smaller sample sizes. Because a prior study suggested an increase in number of complications in the lowest-volume implanting physicians, the authors addressed the volume of ICD implants performed by each physician by eliminating all physicians who performed fewer than 10 implants.⁸

The second outcome analyzed in this study was the difference among the 4 groups in implanting CRT-D in patients who met criteria for CRT-D according to the 2008 guidelines of the American College of Cardiology, the American Heart Association, and the Heart Rhythm Society.⁹ This outcome is not a true measure of efficacy but rather a surrogate for appropriateness. The authors found that nonelectrophysiologists were less likely to implant CRT-D than electrophysiologists in patients who met criteria for CRT-D. While it is possible that the differences in complication rates may be directly related to the training and skill of the implanting physicians, the decision to implant an ICD should not be the sole decision of the implanting physician, but rather a joint decision among the patient's cardiologist, heart failure expert, or both, and the physician performing the ICD implantation. Not all patients eligible for CRT-D benefit from it; the best way to measure dyssynchrony is a matter of debate¹⁰ and factors not apparent in the database such as functional status and patient preference may influence the decision. It is also possible that the low percentages for thoracic surgeons may be biased by the exclusion from this analysis of the 2689 implants with epicardial leads.

Based on the thorough and insightful analysis by Curtis et al, a compelling argument can be made based on the outcome measure of procedural complications, whenever possible, a board-certified electrophysiologist should be implanting ICDs. Curtis et al found that access to electrophysiologists is not a major factor in the implantation of ICDs by nonelectrophysiologists because two-thirds of the implants by nonelectrophysiologists were performed in hospitals that had electrophysiologists on staff and the distance to a hospital with a board-certified electrophysiologist was only a factor in a small percentage of cases. The present analysis cannot determine whether nonelectrophysiologists who have met the guidelines suggested by the Heart Rhythm Society and the American College of Cardiology have outcomes significantly different from electrophysiologists.

The findings raise 2 concerns about the practice of ICD implantation. First, 83.1% of the patients who received an ICD for primary prevention met the inclusion criteria of the key primary prevention trials. It is noteworthy that 17% of all patients receiving ICDs for primary prophylaxis did not appear to meet appropriate criteria for ICD implantation. Further, because the majority of ICDs were implanted for primary prophylaxis, it is disappointing that only 25% of the implants were single-lead devices when all the data for primary prophylaxis used single-lead devices and, unless justification is provided, the Centers for Medicare & Medicaid Services requires a single-lead device for this indication.²

Additionally, the registry can be used to assess the economic impact of ICD implantations. Given the higher rate of complications with dual-chamber devices, which are even higher in the nonelectrophysiologist groups, it would have been interesting to analyze the economic effects of using dual-chamber devices. Were dual-chamber devices used appropriately by all groups? A potential advantage of having certified electrophysiologists performing ICD implantation might be a lower reliance on having industry sales representatives be present during the procedure and follow-up, which may also affect the price of these devices.

Finally, the registry data are currently limited to data collected during the implantation admission. To obtain maximum value from the database, long-term follow-up is needed. This can be done either by expanding the database or combining it with data from other sources such as the Centers for Medicare & Medicaid Services or the National Death Index. A pilot study, the Longitudinal Registry Study, is planning to follow up 3500 patients from the registry for 3 years for events and 5 years for survival.⁴ In the future, this registry could be used to explore the utility of providing individual physicians who perform ICD implantation procedures with data about their performance relative to national or regional peers. Until those data are available, the findings of Curtis et al suggest when patients need an ICD implanted, their physicians should select a certified “electrician,” but should also have input into device selection, taking into consideration both the guidelines and the distinct clinical profile of each individual patient.