Purpose: Coronavirus disease 2019 (COVID-19) is creating an unprecedented healthcare crisis. Understanding the determinants of mortality is crucial to optimise intensive care unit (ICU) resource use and to identify targets for improving survival.
Methods: In a multicentre retrospective study, we included 379 COVID-19 patients admitted to four ICUs between 20 February and 24 April 2020 and categorised according to time from disease onset to ICU admission. A Cox proportional-hazards model identified factors associated with 28-day mortality.
Results: Median age was 66 years (53-68) and 292 (77%) were men. The main comorbidities included obesity and overweight (67%), hypertension (49.6%) and diabetes (30.1%). Median time from disease onset (i.e., viral symptoms) to ICU admission was 8 (6-11) days (missing for three); 161 (42.5%) patients were admitted within a week of disease onset, 173 (45.6%) between 8 and 14 days, and 42 (11.1%) > 14 days after disease onset; day 28 mortality was 26.4% (22-31) and decreased as time from disease onset to ICU admission increased, from 37 to 21% and 12%, respectively. Patients admitted within the first week had higher SOFA scores, more often had thrombocytopenia or acute kidney injury, had more limited radiographic involvement, and had significantly higher blood IL-6 levels. Age, COPD, immunocompromised status, time from disease onset, troponin concentration, and acute kidney injury were independently associated with mortality.
Conclusion: The excess mortality in patients admitted within a week of disease onset reflected greater non-respiratory severity. Therapeutic interventions against SARS-CoV-2 might impact different clinical endpoints according to time since disease onset.