Multistate Modeling of COVID-19 Patients Using a Large Multicentric Prospective Cohort of Critically Ill Patients
Moreno Ursino, Claire Dupuis, Niccolò Buetti, Étienne de Montmollin, Lila Bouadma, Dany Golgran-Toledano, Stéphane Ruckly, Mathilde Neuville, Yves Cohen, Bruno Mourvillier, Bertrand Souweine, Marc Gainnier, Virginie Laurent, Nicolas Terzi, Shidasp Siami, Jean Reignier, Corinne Alberti, Jean‐François Timsit, on behalf of the OUTCOMEREA Study Group
Abstract
The mortality of COVID-19 patients in the intensive care unit (ICU) is influenced by their state at admission. We aimed to model COVID-19 acute respiratory distress syndrome state transitions from ICU admission to day 60 outcome and to evaluate possible prognostic factors. We analyzed a prospective French database that includes critically ill COVID-19 patients. A six-state multistate model was built and 17 transitions were analyzed either using a non-parametric approach or a Cox proportional hazard model. Corticosteroids and IL-antagonists (tocilizumab and anakinra) effects were evaluated using G-computation. We included 382 patients in the analysis: 243 patients were admitted to the ICU with non-invasive ventilation, 116 with invasive mechanical ventilation, and 23 with extracorporeal membrane oxygenation. The predicted 60-day mortality was 25.9% (95% CI: 21.8%-30.0%), 44.7% (95% CI: 48.8%-50.6%), and 59.2% (95% CI: 49.4%-69.0%) for a patient admitted in these three states, respectively. Corticosteroids decreased the risk of being invasively ventilated (hazard ratio (HR) 0.59, 95% CI: 0.39-0.90) and IL-antagonists increased the probability of being successfully extubated (HR 1.8, 95% CI: 1.02-3.17). Antiviral drugs did not impact any transition. In conclusion, we observed that the day-60 outcome in COVID-19 patients is highly dependent on the first ventilation state upon ICU admission. Moreover, we illustrated that corticosteroid and IL-antagonists may influence the intubation duration.