April 19th, 2011
Non-Evidence-Based ICD Implants: The Debate Continues
A controversial study published earlier this year in JAMA that found that nearly a quarter of all ICD implantations did not meet evidence-based criteria is the subject of further debate this week in the letters section of JAMA.
In the first letter, Jeanne Poole and George Crossley take issue with the designation by the authors of ICD implantations within 3 months of a new HF diagnosis as being non-evidence based. They point out that current guidelines “do not specify a 3-month wait from HF diagnosis before ICD implantation for any etiology of HF.” In response, Sana Al-Khatib and colleagues argue that “recommendations for primary prevention ICDs apply only to patients whose left ventricular ejection fraction remains low despite optimal medical therapy” and that “achieving optimal medical therapy in patients with newly diagnosed HF is an iterative process” that has required more than 3 months to demonstrate full benefits in clinical trials.
In the second letter, Kathleen Blake and Charles Swerdlow write that the NCDR data “is not designed to determine the medical necessity of ICD implants that do not meet evidence-based guidelines.” They describe several scenarios in which non-evidence-based implants may be medically justified. Al-Khatib et al agree that some non-evidence-based implants can be “clinically appropriate” but, they write, “the important question is how many.” They point out that non-evidence-based implantations in hospitals in their study ranged from 0% to 60%, and write that “the proportion of such cases should generally not be high, and if it is, additional investigations are needed to identify reasons.”
In the third letter, Steven Zweibel, Christopher Clyne, and Eric Crespo question the accuracy of the NCDR registry, since “the registry relies on busy clinicians… who may not be completely familiar with the subtleties of clinical care and decision making for patients with cardiovascular disease.” They cite numerous deficiencies in the data collection for the registry at their own institution. In their response, Al-Khatib et al say that the NCDR has numerous mechanisms to ensure the quality of its data, and that a professional audit found a 90% agreement between hospital records and registry data. CMS reimbursement for ICDs requires participation in the registry, “and penalties may be levied if inaccurate data are knowingly submitted.”