Using real-world data (RWD) from anintegrated platform for rapid analysis of patients withcancer with and without COVID-19 across distinct healthsystems

ABSTRACT

Introduction: Reports suggest worsened outcomes in patients with cancer (pts) and COVID-19 (Cov), varying bygeography and local peak dynamics. We describe characteristics and clinical outcomes of pts with and without Cov. Methods: RWD at 2 Midwestern health systems from the Syapse Learning Health Network were used to identifyadults with active cancer (AC) or past history of cancer (PHC). AC pts were identified by encounters with ICD-10code for malignant neoplasm or receipt of an anticancer agent within 12 months prior to February 15, 2020;PHC pts were identified by encounters with an active cancer code from May 15, 2015 to February 15, 2019 and no receipt ofanticancer therapy within the prior 12 months. Cov was defined by diagnostic codes and laboratory results fromFebruary 15 to May 13, 2020. Comorbidities were assessed prior to February 15, 2020;hospitalizations (hosp), invasive mechanical ventilation (IMV), and all-cause mortality (M) were assessed from February 15 to May 27, 2020. Results: We identified 800 pts with Cov (0.5%) out of a total of 154,585 pts with AC or PHC. Compared to AC pts without Cov (AC WO, 39,402), AC pts with Cov (AC Cov, 388) were more likely to be non-Hispanic Black (NHB,39% vs. 9%), have renal failure (RF, 24% vs. 12%), cardiac arrhythmias (33% vs. 19%), congestive heart failure(CHF, 16% vs. 8%), obesity (19% vs. 14%), pulmonary circulation disorder (PCD, 9% vs. 4%), and a zip code withmedian annual household income (ZMI) <$30k (18% vs. 5%). Comorbidity and income were similarly distributed forPHC pts with Cov (PHC Cov, 412). Compared to PHC pts without Cov (PHC WO, 114,383), coagulopathy (coag)was more common in PHC Cov pts (10% vs. 5%). Hosp for AC Cov pts was higher than for AC WO pts (81% vs.15%). Hosp for PHC Cov pts was also higher than for PHC WO pts (68% vs. 6%). Hosp was highest for NHB pts inboth AC Cov and PHC Cov groups (88% and 72%) and for AC Cov pts in low ZMI (94% in <$30K). Pts <50 yearsold had hosp rates of 79% (AC Cov) and 49% (PHC Cov). IMV rate for AC Cov pts was higher than for PHC Cov pts(21% vs. 14%). Rates of IMV for AC Cov pts were highest in low ZMI (27%) and in pts with coag (36%). M by group was AC Cov 16%;AC WO 1%;PHC Cov 11%;PHC WO 1%. Among AC Cov pts, M was higher for men (19% vs.13%) and pts with PCD (31%), RF (25%), or diabetes (DM, 24%);among PHC Cov pts, M was also higher for men(14% vs. 8%) and pts with coag (30%), valvular disease (27%), or PCD (24%). Increasing age, DM, RF, and PCD were associated with increased risk of M for AC Cov pts in age, race/ethnicity, and comorbidity-adjusted logisticregression;increasing age and coag were associated with M in PHC Cov pts. Conclusion: In this rapid characterization from RWD, pts with Cov have higher rates of pre-existingcardiopulmonary/vascular and renal conditions and increased risk of hospitalization, IMV, and mortality than pts without Cov. Higher Cov risk and worse outcomes in NHB and lower-income pts suggest health care disparities.Whether these outcomes are due to comorbidities or acute sequelae merits further study, as does investigation ofalternative definitions for real-world populations and outcomes.