B-7 | Impella Has Greater Mortality Benefit For Patients With Versus Without Cancer: Machine Learning Supported Propensity Score Case-control Analysis of 101 Million+ Hospitalizations From 2016-2018

First 2 authors: co-first authors contributing equallyBackgroundThis is the first nationally representative multi-year nation analysis of mortality, cost, and procedural volume of Impella mechanical circulatory support for patients with and without active cancer.MethodsThis case-control study of the above outcomes was conducted using the United States’ largest all-payer hospitalized dataset, the 2016-2018 National Inpatient Sample (NIS). Machine Learning-augmented Propensity Score adjusted multivariable regression (ML-PSr) was utilized, weighted by the NIS complex survey design, and adjusted for known confounders (including NIS-calculated mortality risk by DRG and the likelihood of undergoing Impella versus medical management).ResultsAmong 101,521,656 hospitalizations, 136,974 (0.05%) underwent Impella among whom 60,060 (0.02%) had active cancer. Patients with versus without cancer were significantly less likely to receive Impella in 2016 (0.01 versus 0.04%), 2017 (0.02 versus 0.02 versus 0.04%), and 2018 (0.03 versus 0.06%) concurrent with the linear increase in Impella procedures across all patients in the same time period (all p<0.001) 2018. Across all years, this disparity was similar in cardiogenic shock (1.62 versus 4.19%) and clinical severity-matching (by age and mortality risk: 3.77 versus 11.54%), though rates were comparable for septic shock (all p<0.001). There were no significant procedural disparities among patients with versus without cancer by sex, race, income, or urban density, but there were for regions (with the highest Impella volume in patients with cancer being done in the Northwest [5.96%] and Mountain [4.04%], p=0.026). In multivariable regression across all years, Impella significantly reduced mortality for patients with cancer (OR 0.39, 95%CI 0.19-0.82; p=0.012) along with an increased cost of $7,627.79 (95%CI 7394.51-7,861.07; p<0.001) for patients with versus without cancer.ConclusionsThis national propensity score analysis suggests Impella has a disproportionate mortality benefit for patients with versus without cancer, though the former receives them significantly less than the latter despite an otherwise comparable risk profile and gradually increasing rate of use.DisclosuresS. A. Arain: Teleflex: Speaker Bureau; K. A. Honan Nothing to disclose. D. J. Monlezun Nothing to disclose. A. Badalamenti Nothing to disclose. J. W. Kim Nothing to disclose. V. Liu Nothing to disclose. A. Javaid Nothing to disclose. S. Chauhan Nothing to disclose. C. A. Simbaqueba Clavijo Nothing to disclose. D. V. V. Balanescu Nothing to disclose. N. Palaskas Nothing to disclose. M. Cilingiroglu Nothing to disclose. A. Dhoble Nothing to disclose. K. Charitakis Nothing to disclose. K. D. Boudoulas Nothing to disclose. K. Marmagkiolis Nothing to disclose. C. A. Iliescu Nothing to disclose. First 2 authors: co-first authors contributing equally BackgroundThis is the first nationally representative multi-year nation analysis of mortality, cost, and procedural volume of Impella mechanical circulatory support for patients with and without active cancer. This is the first nationally representative multi-year nation analysis of mortality, cost, and procedural volume of Impella mechanical circulatory support for patients with and without active cancer. MethodsThis case-control study of the above outcomes was conducted using the United States’ largest all-payer hospitalized dataset, the 2016-2018 National Inpatient Sample (NIS). Machine Learning-augmented Propensity Score adjusted multivariable regression (ML-PSr) was utilized, weighted by the NIS complex survey design, and adjusted for known confounders (including NIS-calculated mortality risk by DRG and the likelihood of undergoing Impella versus medical management). This case-control study of the above outcomes was conducted using the United States’ largest all-payer hospitalized dataset, the 2016-2018 National Inpatient Sample (NIS). Machine Learning-augmented Propensity Score adjusted multivariable regression (ML-PSr) was utilized, weighted by the NIS complex survey design, and adjusted for known confounders (including NIS-calculated mortality risk by DRG and the likelihood of undergoing Impella versus medical management). ResultsAmong 101,521,656 hospitalizations, 136,974 (0.05%) underwent Impella among whom 60,060 (0.02%) had active cancer. Patients with versus without cancer were significantly less likely to receive Impella in 2016 (0.01 versus 0.04%), 2017 (0.02 versus 0.02 versus 0.04%), and 2018 (0.03 versus 0.06%) concurrent with the linear increase in Impella procedures across all patients in the same time period (all p<0.001) 2018. Across all years, this disparity was similar in cardiogenic shock (1.62 versus 4.19%) and clinical severity-matching (by age and mortality risk: 3.77 versus 11.54%), though rates were comparable for septic shock (all p<0.001). There were no significant procedural disparities among patients with versus without cancer by sex, race, income, or urban density, but there were for regions (with the highest Impella volume in patients with cancer being done in the Northwest [5.96%] and Mountain [4.04%], p=0.026). In multivariable regression across all years, Impella significantly reduced mortality for patients with cancer (OR 0.39, 95%CI 0.19-0.82; p=0.012) along with an increased cost of $7,627.79 (95%CI 7394.51-7,861.07; p<0.001) for patients with versus without cancer. Among 101,521,656 hospitalizations, 136,974 (0.05%) underwent Impella among whom 60,060 (0.02%) had active cancer. Patients with versus without cancer were significantly less likely to receive Impella in 2016 (0.01 versus 0.04%), 2017 (0.02 versus 0.02 versus 0.04%), and 2018 (0.03 versus 0.06%) concurrent with the linear increase in Impella procedures across all patients in the same time period (all p<0.001) 2018. Across all years, this disparity was similar in cardiogenic shock (1.62 versus 4.19%) and clinical severity-matching (by age and mortality risk: 3.77 versus 11.54%), though rates were comparable for septic shock (all p<0.001). There were no significant procedural disparities among patients with versus without cancer by sex, race, income, or urban density, but there were for regions (with the highest Impella volume in patients with cancer being done in the Northwest [5.96%] and Mountain [4.04%], p=0.026). In multivariable regression across all years, Impella significantly reduced mortality for patients with cancer (OR 0.39, 95%CI 0.19-0.82; p=0.012) along with an increased cost of $7,627.79 (95%CI 7394.51-7,861.07; p<0.001) for patients with versus without cancer. ConclusionsThis national propensity score analysis suggests Impella has a disproportionate mortality benefit for patients with versus without cancer, though the former receives them significantly less than the latter despite an otherwise comparable risk profile and gradually increasing rate of use. This national propensity score analysis suggests Impella has a disproportionate mortality benefit for patients with versus without cancer, though the former receives them significantly less than the latter despite an otherwise comparable risk profile and gradually increasing rate of use.