Abstract 14425: Improved Carcinoid Heart Disease Mortality Amid Persistent Race, Income, and Rural Disparities: Propensity Score and Machine Learning Guided Analysis of 55,910 Carcinoid Cases From 2016-2018

Abstract

Introduction: It is unclear what the rates of carcinoid heart disease (CHD) diagnosis, disparities, and inpatient outcomes are. Methods: We performed the first nationally representative longitudinal analysis of the above using Machine Learning-augmented Propensity Score adjusted multivariable regression (ML-PSr) and the 2016-2018 National Inpatient Sample (NIS), the United States’ largest all-payer inpatient dataset. The propensity score was modified for the likelihood of developing malignant carcinoid tumor (MCT). Results: From 2016-2018 among 101,521,656 hospitalizations, 55,910 (0.05%) had a malignant carcinoid tumor (MCT), of whom 4,030 (7.20%) had CHD. The rate of CHD diagnosis remained largely stable from 2016 (18,900 [0.06%]) to 2017 (19,025 [0.05%]) to 2018 (17,985 [0.05%]). Patients with versus without CHD had significantly higher unadjusted mortality in 2016 (4.05% versus 2.19%), 2017 (3.75% versus 1.95%), and 2018 (3.75% versus 1.95%). Of patients with CHD, 0.15% had acute carcinoid syndrome. Compared to the racial distribution throughout the population, the diagnosis of CHD was under-represented in Hispanics (12.83% versus 7.95%), Asians (3.13% versus 2.84%), lowest income quartile (29.47% versus 28.07%), second lowest income quartile (26.90% versus 25.41%), metros of 250,000-999,999 residents (20.79% versus 19.73%) and metros of 50,000-249,999 (9.30% versus 8.56%) (all p<0.001). In ML-PS multivariable regression fully adjusting for age, sex, race, metro density, region, metastasis, and NIS calculated mortality risk by DRG, CHD initially significantly increased mortality in 2016 (OR 1.18, 95%CI 1.10-1.28; p<0001) but decreased it in 2017 (OR 0.76, 95%CI 0.70-0.82; p<0.001) and 2018 (OR 0.78, 95%CI 0.72-0.84; p<0.001). In ML-PSr stratified by income in 2018, there was largely a step wise decrease in mortality with CHD from the lowest income quartile to highest income quartile. Conclusions: Our analysis demonstrates that racial minorities, lower income, and more rural metros have lower rates of CHD diagnosis, its associated mortality appears to be the lowest with higher income, and its mortality over time across all patients in general appears to be decreasing potentially with improved inpatient care.