Study population and setting
This study consisted of 5700 sequential patients (median age: 63 years [range 0 -107 years], 60% male), with lab-confirmed SARS-CoV-2 infection admitted to one of Northwell Health’s hospitals in New York City and surrounding areas between March 1 and April 4, 2020. 2634 patients died or had been discharged by the study endpoint on April 4, 2020, while the remaining 3066 remained hospitalized.
Summary of Main Findings
At hospital presentation, only 31% of patients had measured fever. Most patients had significant underlying medical conditions, including high prevalence of hypertension (57%), diabetes (34%), and obesity (42%). Among patients with available outcome data, 21% died, 14% were admitted to an intensive care unit (ICU), and 12% required mechanical ventilation. Mortality was closely linked to age: there were no deaths in those younger than 20, while 43% of those over 70 years died. Among the 1151 patients requiring mechanical ventilation, 24.5% died, 3.3% were discharged alive, and 72.2% remained hospitalized. Therefore, of mechanically-ventilated patients with observed outcomes (n=320), 88.1% died (n=282). Finally, 2% of all discharged patients were readmitted.
Large sample size, detailed patient information from electronic medical records, follow-up of patients after discharge.
Only 46% of patients had been discharged or died by the end of the study period. Those remaining in the hospital (especially with longer admissions) may be systematically different from those with available outcomes. This would bias the observed risk of mortality in various patient groups. For example, ICU patients (including those on ventilators) whose condition improved may have extended admissions and may still be in the hospital. These patients are likely to have lower mortality risk than patients with observed outcomes, most of whom died. Alternatively, however, patients with extended hospital or ICU stays face increased risks of mortality related to infection or other hospital-associated complications. Insufficient data are presented to evaluate the likely direction and magnitude of the bias. Secondly, the lack of multivariable analysis makes any patterns in potential risk factors, including medication use, difficult to discern. Finally, the short duration of follow up after hospital discharge (median 4.4 days) likely results in underestimates of readmission rates.
This is the largest study to date of hospitalized COVID-19 patients in the United States, and suggests dramatically higher mortality, particularly among patients on ventilators, than has previously been reported.