Abstract
BACKGROUND: Non-invasive assessment of respiratory drive and effort in spontaneously breathing ARDS patients is challenging, yet clinically relevant. We explored whether hierarchical clustering applied to electrical impedance tomography (EIT- a radiation-free non-invasive lung imaging technique) identifies ARDS sub-phenotypes with increased drive and effort. RESULTS: Thirty intubated patients with ARDS on assisted mechanical ventilation were monitored by EIT and esophageal pressure during a decremental positive end-expiratory pressure (PEEP) trial. A comprehensive EIT assessment was made (computed variables n = 180) during tidal breathing at different PEEP levels. Agglomerative nesting was applied to scaled data distances. Three clusters of ARDS were identified: inhomogeneous ventilation, unmatched V'/Q, and mismatched V'/Q. The unmatched V'/Q cluster had the highest respiratory drive (p = 0.045) and effort (p = 0.021) at lower PEEP, and experienced longer length of ICU stay (p = 0.019). CONCLUSIONS: Higher PEEP levels reduced drive of the unmatched V'/Q cluster, mitigating the physiological differences. Clustering approaches to EIT data identify physiologically and clinically relevant sub-phenotypes of ARDS.