Predictors of Hospitalization for Patients Presenting to Emergency Department with COVID-19 Infection

Background: Predictors of morbidity and mortality in hospitalized COVID-19 patients have been extensively studied. However, comparative analyses of predictors for hospitalization versus discharge from the emergency department remain limited. Methods: This retrospective study evaluated predictors of hospitalization among adults (≥18 years) presenting to the emergency department with COVID-19 infection between 1 March 2020 and 15 June 2020. Data were obtained from electronic health records across five hospitals within the Johns Hopkins Health System, encompassing 2513 beds. Multivariable logistic regression models were employed to assess the association between sociodemographic characteristics, clinical symptoms, and comorbidities with hospitalization. Results: Of the 2767 patients presenting to the emergency department, 1678 (61%) were hospitalized, while 1089 (39%) were discharged from the emergency department. Hospitalized patients were older (mean age 61.8 years, SD 18), more likely to be African American and White, non-Hispanic, unemployed or on disability, medically insured, had access to primary care, and presented on weekends. Smoking status, alcohol use, and higher comorbidity burden (mean age-adjusted Charlson Comorbidity Index > 3) were also more prevalent with hospitalization. Dyspnea was a prominent clinical feature among hospitalized patients. After adjusting for sociodemographic and clinical risk factors, significant predictors of hospitalization included health insurance (OR 3.44; 95% CI: 1.98-6), having a primary care (OR 1.85; 95% CI: 1.33-2.59), presentation from a non-home locale (OR 4.04; 95% CI: 1.93-8.47), age-adjusted CCI > 3 (OR 1.72; 95% CI: 1.11-2.68), dyspnea (OR 2.22; 95% CI: 1.56-3.17), neutrophil-to-lymphocyte ratio ≥ 3 (OR 2.17; 95% CI: 1.54-3.06), and an abnormal chest radiograph findings (OR 6.17; 95% CI: 4.40-8.66). Interestingly, obesity, defined as a BMI ≥ 30 kg/m² (OR 0.45; 95% CI: 0.32-0.64), and the presence of fever (OR 0.64; 95% CI: 0.43-0.95) were found to be associated with a decreased likelihood of hospitalization. Conclusions: Future studies are warranted to further explore predictors of COVID-19 hospitalization, with particular focus on the implications of weekend presentations and the paradoxical relationship of obesity with COVID-19 health outcomes. These findings could inform the development of triage models to enhance preparedness for future pandemics.