Abstract 14712: PCI is Safe and Increasing Among Cancer Patients: Nationally Representative Propensity Score Adjusted and Machine Learning Augmented Study of 100 Million Hospitalizations

Abstract

Introduction: The greater use of antithrombotic and antiplatelet agents in the growing population of patients with cardiovascular disease and cancer contrasts with the limited evidence-base on appropriate revascularization management for these patients. Methods: This is the first nationally representative multi-year cardio-oncology mortality study of thrombocytopenia (platelets <150,000) and coronary intervention including percutaneous coronary intervention (PCI), and the first such study using machine learning augmentation and propensity score (PS). Utilizing the 2016-2018 National Inpatient Sample (NIS), the PS for receiving PCI was generated, balance confirmed, and regression adjustment completed (also including age, sex, race, income, urban, region, metastasis, NIS-calculated mortality risk by disease-related group), and multivariable regression model performance was optimized using backward propagation neural networks. Results: Of the 101,521,656 hospitalized adult patients, 6,456,777 (6.36%) had active cancer of whom 6.14% had thrombocytopenia. Among thrombocytopenia patients, patients with versus without active cancer were significantly (p<0.001) less likely to receive left heart catheterization (1.88 vs. 5.41%) and PCI (0.48 vs. 1.35%). Among cancer patients, patients with versus without thrombocytopenia were more likely to have acute myocardial infarction (4.06 vs. 2.35%) and to receive intravascular ultrasound (62.65 vs. 56.75%) but had comparable rates of fractional flow reserve and optical coherence tomography. PCI nearly doubled from 2016 to 2018 among patients with thrombocytopenia (10.33 vs. 6.16%). In multivariable regression in PCI, neither active cancer (OR 0.89, 95%CI 0.45-1.75; p=0.738) nor any of the 32 tested primary malignancies significantly increased mortality. Conclusions: This nationally representative longitudinal study suggests inpatient PCI is safe and is increasing among patients with active cancer regardless of their primary malignancy.