Variation among hospitals in selection of higher-cost, “higher-tech,” implantable cardioverter-defibrillators: Data from the National Cardiovascular Data Registry (NCDR) Implantable Cardioverter/Defibrillator (ICD) Registry

Background New implantable cardioverter/defibrillator (ICD) models are regularly introduced, incorporating technological advantages. The purpose of this study was to determine factors associated with use of a newer, higher-tech/higher-cost device, as opposed to a previously released device, among patients undergoing ICD implantation. Methods We analyzed the 78,494 cases receiving new ICD implants submitted by 978 hospitals to the NCDR ICD Registry between January 2005 and June 2007. Devices were categorized as “previously released” 3 months after a new model from the same manufacturer was released. A nonparsimonious model including all demographic, clinical, provider, and hospital characteristics was created using logistic regression to predict use of a previously released device. Results Overall, 36% of implants involved previously released devices. However, no demographic (race, gender, payor status), clinical, or provider characteristics had a meaningful impact on use of previously released devices. The model C-statistic was 0.602, suggesting that measured characteristics had a limited ability to differentiate those receiving a previously released device. However, individual hospitals varied greatly in use of “previously released” devices, from 3% in the lowest decile to 91% in the top decile. Among physicians implanting at >1 hospital, there was minimal correlation between use of previously released devices between hospitals, suggesting hospital policies or culture, rather than physician preference, drives the large interhospital variation seen. Conclusions The use of “previously released” devices is influenced minimally by measured patient or provider characteristics. Rather, the main determinant of whether patients receive the newest, versus a previously released device, appears to be practice patterns at individual hospitals.