e18728 Background: More than 50,000 implantable cardioverter defibrillators (ICDs) are placed in adults aged 65 years and older in the United States each year to manage cardiac arrhythmias. Shared risk profiles between many cancers and heart disease, as well as cardiotoxic effects of anti-cancer treatments suggest individuals with cancer are likely to have ICD devices. Although ICDs have been studied extensively in Europe, little is known about the US population, and even less is known in populations with cancer. ICDs have critical implications for end-of-life care in patients with cancer. Many patients do not include wishes for their ICD in their advanced directives, and end-of-life shocks can cause pain and distress for patients and their families. We evaluated prevalence and end-of-life monitoring of ICDs in a national sample of older adults who had breast, colorectal, or pancreatic cancer. Methods: We analyzed Surveillance, Epidemiology and End Results (SEER) data for first incident breast, colorectal, and pancreas cancer cases linked to Medicare claims (years 2004-2016) in Medicare-beneficiaries aged ³66 years. We evaluated the prevalence of ICDs by cancer stage at diagnosis with descriptive statistics. For patients with late-stage cancer (III or IV) who died in our study period we also evaluated the prevalence of ICD monitoring at the end of life. Because ICD deactivation is not a billable code, we evaluated ICD monitoring as a proxy for having an active device, hence an opportunity to discuss device deactivation. Results: In our sample of 286,015 individuals with cancer, 6.8% of patients had evidence of an ICD in their Medicare claims (6.1% breast, 8.2% colorectal, 5.3% pancreas). Of these, 49.7% had evidence of an ICD before their cancer diagnosis. The prevalence of ICDs was lower among the 60,046 individuals diagnosed with late-stage cancer( p< .001) (5.0% breast, 6.1% colorectal, 4.5% pancreas). Among individuals with late-stage cancer and evidence of an ICD, 27.0% of those who died had ICD monitoring in the last six months of life (21.1% breast, 27.9% colorectal, 29.3% pancreas, p= .001), and 6.2% had monitoring in the last month of life (5.1% breast, 6.5% colorectal, 6.3% pancreas, p= .46). Conclusions: To improve the quality of care at the end-of-life for people with advanced cancer, awareness, communication, and shared decision making about ICD device management options are important considerations. These data suggest there may be opportunities to discuss prognosis and implications of advanced-stage cancer at device monitoring visits or during cardio-oncology visits.