Skip to main content

Factors associated with hospitalization and critical illness among 4,103 patients with COVID-19 disease in New York City

Our take —

In New York City, SARS-CoV-2 infection appears to be associated with more severe outcomes than has been reported from similarly large studies in China. As previously described, older age and pre-existing medical conditions are associated with the need for hospitalization. Hypoxemia at admission and inflammatory markers may have strong prognostic value.

Study design

Case Series

Study population and setting

All 4,103 patients who tested positive for SARS-CoV-2 infection and who were treated within the NYU Langone Health System in New York City, between March 1,2020 and April 2, 2020. Follow-up was through April 7, 2020. Authors used logistic regression model to estimate the risk of two outcome: hospitalization and critical illness conditional on hospitalization.

Summary of Main Findings

Among the 7,719 potentially eligible patients tested for SARS-CoV-2, 49% (4,103) were positive. Among those with a diagnosed infection, 49% were hospitalized. In total, 79% had an outcome determined by the end of follow-up, of whom 59% were discharged home, 41% experienced critical illness, and 15% died or were discharged to hospice. Important predictors of hospitalization were older age, obesity, and comorbid chronic kidney disease or heart failure. Important predictors of critical illness after hospitalization were low oxygen saturation (hypoxemia) at admission and high concentrations of d-dimer, ferritin, or C-reactive protein (all inflammatory markers).

Study Strengths

The study sample is large and includes both admitted and discharged patients. The statistical methods were appropriate and robust.


Over 20% of patients had not yet experienced an outcome at the time of analysis, which may bias results in the critical illness model, particularly if there were differences in outcomes over time. Clinical factors were defined and categorized in the models a priori based on clinical judgement, which may mask important threshold values for clinical prognostication. While the sample includes non-admitted patients, those presenting for care to a hospital or clinic are unlikely to represent the full spectrum of illness in the community. The timing of laboratory testing was not standardized, thus weakening the prognostic value of laboratory parameters, since they may have been obtained after clinical status was deteriorating. Finally, given that smoking was not observed to be associated with increased risk of hospitalization or severe illness, further discussion about the potential for reporting bias and the disaggregation of current and former tobacco use in multivariable models are warranted.

Value added

This is the largest case series to date of hospitalized Covid-19 from the United States. The study provides strong support for previous findings on the heightened risk of hospitalization and severe disease for the elderly and those with underlying medical comorbidities.