Covid-19 Incidence, Mortality, and Disparities for Cancer, Cardio-Oncology, and The General Population: Machine Learning National Analysis of 27 Million Hospitalizations in 2020

Introduction: The paucity of large, longitudinal COVID data for higher risk cancer and cardio-oncology patients limits sufficient preparedness for the next pandemic. Therefore, questions regarding the incidence, outcomes, and disparities for the above patients need to be studied. Methods: BAyesian Machine learning-augmented Propensity Score translational (BAM-PS) statistics with multivariable regression was performed on the largest all payor US inpatient dataset, the National Inpatient Sample. Results: Of the 27,667,386 adult hospitalizations in 2020, 47,820 (0.17%) had COVID, of whom 2,430 (5.08%) had active cancer and 765 (1.60%) had cardio-oncology disease. In 2020, COVID was significantly less likely in patients with versus without active cancer (0.12% versus 0.18%, p<0.001), with the most common malignancies in COVID including leukemia (22.22%), multiple myeloma (14.20%), lung (13.79%), and non-Hodgkin lymphoma (11.93%). The odds of being treated inpatient for COVID was significantly reduced by active cancer (OR 0.66, 95%CI 0.60-0.73, p<0.001). In active cancer, COVID significantly increased mortality (OR 2.02, 95%CI 1.52-2.68, p<0.001), though within COVID, neither active cancer nor cardio-oncology disease significantly increased mortality. In active cancer with COVID, mortality was significantly increased for the East South Central region (OR 0.08, 95%CI 0.01-0.95, p=0.045) compared to the New England region, though there were no other disparities by race, insurance, income, or urban density (with no disparities in cardio-oncology). Conclusions: This large nationally representative machine learning analysis suggests that cancer patients with COVID were less likely to be treated inpatient despite increased mortality. Cardiovascular disease does not significantly contribute to the mortality of these already extremely frail and complex patients.