Date of Award

January 2019

Document Type


Degree Name

Medical Doctor (MD)



First Advisor

Nihar R. Desai


The US Department of Justice (DOJ) investigated implantable cardioverter defibrillators (ICDs) that did not meet the National Coverage Determination (NCD) criteria, resulting in settlements with over 500 hospitals for over $280 million dollars. This study sought to determine whether that investigation was associated with changes in the proportions of devices not meeting the NCD, the volume of ICDs implanted for primary and secondary prevention of sudden cardiac death, and the in-hospital outcomes for those procedures. In this retrospective cohort study of 300,151 ICDs implanted for primary prevention between 2007 and 2015 from 1809 hospitals including 452 that settled with the DOJ, there were significant declines in the proportion of primary prevention ICDs not meeting NCD criteria. In January 2007, the proportion of ICDs not meeting NCD criteria was 25.8% (95% CI, 24.7% - 26.8%) at hospitals that reached settlements with the DOJ and 22.8% (95% CI, 22.1% - 23.5%) at hospitals that did not settle (p < 0.001). Over the study period, there was a 62.7% (95% CI, 59.2% - 66.1%) relative decline and 16.1% (95% CI, 14.8% - 17.5%) absolute decline in the proportion of ICDs not meeting NCD criteria at hospitals that settled compared with 53.2% (95% CI, 50.4% - 56.0%) relative and 12.1% (95% CI, 11.2% - 13.0%) absolute decline at hospitals that did not settle (p < 0.001 for both, pint < 0.001). There were similar declines observed among non-Medicare patients. The decline in the proportion of devices not meeting the NCD was driven by reductions in patients with a disqualifying recent heart failure diagnosis (15.5% to 6.8% for settled hospitals; 13.5% to 7.3% for non-settled hospitals) or recent myocardial infarction (8.4% to 1.3% for settled; 7.4% to 1.5% for non-settled). Across all hospitals, adverse events were significantly higher during the study for ICDs not meeting NCD criteria (adjusted odds ratio of 1.26 for settled, p < 0.001; 1.18 for non-settled, p = 0.001), and the disparity in mortality was even greater (adjusted odds ratio of 1.52 for settled, p = 0.003; 1.54 for non-settled, p < 0.001). The elevated adverse event rates among ICDs that are still implanted outside the NCD criteria highlight the importance of ongoing surveillance and quality improvement efforts.


This thesis is restricted to Yale network users only. This thesis is permanently embargoed from public release.