Radiological phenotyping in COVID-19 Acute Respiratory Distress Syndrome: a secondary analysis of a retrospective cohort study

ObjectiveTo assess whether focal and non-focal COVID-19 ARDS exhibit different respiratory mechanics and arterial blood gas (ABG) trajectories during the first extended prone positioning (PP) session. DesignPost-hoc analysis of a previously published retrospective monocentric cohort study. SettingA university-affiliated intensive care unit in Paris (France) between March 2020 and April 2021. PatientsSeventy-four adult patients with moderate-to-severe COVID-19 ARDS who underwent extended prone positioning (PP) and had chest imaging (CT or X-ray) performed within five days before the first PP session. InterventionsNone MeasurementsChanges in compliance, driving pressure, PaO2/FiO2 ratio, and ventilatory ratio between pre-PP and end-of-PP; chest X-rays and CT scans diagnostic agreement between reviewers assessed by crude agreement and Cohens kappa coefficient. Main resultsDiffuse ARDS predominated, identified in 91% by the intensivist and 86% by the radiologist. Crude diagnostic agreement was 89% (95% CI 79.8-95.2), while interrater reliability was only fair ({kappa} = 0.44, 95% CI 0.08-0.81). Discrepancies mainly involved focal classifications on chest X-rays, whereas agreement was perfect when CT scans were available. The small number of focal cases precluded comparative analysis of PP trajectories. ConclusionsPP indication or duration should probably not be based on ARDS phenotypes and when differentiating focal from diffuse COVID ARDS is necessary, chest CT scans should be preferred to ensure accurate and reproducible phenotyping.