Characteristics of COVID-19 readmissions in an urban community hospital

Rationale: COVID-19 has produced a global pandemic resulting in a widespread increase in hospitalizations for both pneumonia and multiorgan dysfunction with frequent readmissions. As we brace for a second surge of infections and hospitalizations, understanding the natural history of this unique viral infection is crucial. Shared patterns likely exist in the disease course of both new hospitalizations and readmissions. We hypothesize that common characteristics in patients readmitted with COVID-19 pneumonia include patient demographics, laboratory results, chest imaging, comorbidities, and underlying risk factors. Recognition of these common features can help develop targeted interventions for patients identified as high risk for readmission in order to reduce hospitalizations and improve upon allocation of limited resources. Methods: We conducted a singlecenter retrospective analysis of patients with COVID-19 infection confirmed by nasal PCR, age greater than 18 years, admitted to a New York City hospital between March and May 2020 with discharge to home or subacute rehabilitation facility, who were readmitted less than 30 days after discharge. Patients discharged to long term acute care facilities (i.e. Javits Center, USS Comfort), home hospice, or those that left against medical advice were excluded. Data collected included patient demographics (sex, race, age, body mass index, comorbidities), oxygen requirements on discharge and readmission, presence of bilateral infiltrates on imaging, and inflammatory markers (c-reactive protein, d-dimer). Results: 102 patients were readmitted during the study period, 50 of which met inclusion criteria. In our cohort, 56% were Caucasian and 36% were African American, 56% were male, 66% had body mass indices >25, and 62% were over 70 years of age. Comorbidities in our cohort included 64% with hypertension, 40% with hyperlipidemia, 24% with diabetes, 20% with congestive heart failure, and 18% with coronary artery disease. 85% of patients who were readmitted were prescribed supplemental oxygen on discharge and required it on return. Additionally, 82% of patients had bilateral infiltrates on initial admission. We did not see a significant correlation with inflammatory markers and readmission. Readmission diagnoses commonly encountered were hypoxia, acute encephalopathy, and acute renal failure. Conclusion: In our analysis, we found that comorbid conditions, particularly congestive heart failure and diabetes, presence of bilateral infiltrates on imaging, and ongoing oxygen requirements at the time of initial discharge were strong predictors for readmission. Based on this information, we are now developing a targeted readmission reduction program for those identified as high risk for readmission.