Airway Occlusions to Measure Inspiratory Effort, Respiratory Drive, and Lung Mechanics During Noninvasive Ventilation

RATIONALE: In intubated patients, occlusion maneuvers allow noninvasive assessment of inspiratory effort, respiratory drive, and lung mechanics. OBJECTIVES: To assess the feasibility of occlusion maneuvers during noninvasive ventilation (NIV). METHODS: In this multicenter study, 60 hypoxemic patients underwent 2 randomized 1-hour NIV sessions with oronasal and full-face masks after extubation. End-expiratory and end-inspiratory occlusions measured expiratory occlusion pressure (Pocc), 100-ms airway pressure drop (P0.1), and plateau pressure. Esophageal manometry, calibrated before extubation, provided reference values for inspiratory effort, assessed as esophageal pressure swing (ΔPes) and dynamic transpulmonary driving pressure (ΔPL,dyn = pressure support - ΔPes). Interface-specific conversion factors (K) translating Pocc into predicted ΔPes (K × Pocc) and predicted ΔPL,dyn (pressure support - predicted ΔPes) were derived through 100-interaction cross-validation (20-patient derivation set, 40-patient validation set). MEASUREMENTS AND MAIN RESULTS: Pocc was measurable in all patients. Mean K was 0.71 with the oronasal mask and 0.80 with the full-face mask. Predicted ΔPes agreed with observed ΔPes (oronasal: bias -0.41 cm H2O, 95% limits of agreement -2.3 to 1.5; full-face: 0.09, -2.9 to 3.1), and predicted ΔPL,dyn agreed with observed ΔPL,dyn (oronasal bias: 0.03, -2.9 to 2.9; full-face: -0.04, -4.3 to 4.2). Predicted ΔPes identified observed ΔPes ≤ -10 cm H2O, with areas under the receiver operating characteristic curve of 0.98 (oronasal) and 0.97 (full-face). Ventilator-derived P0.1 did not precisely quantify respiratory drive, but values >2.7 cm H2O with the oronasal mask and >3 cm H2O with the full-face mask identified high drive with specificity >90%. Plateau pressure was unstable in 78% (oronasal) and 90% (full-face) of patients. More negative predicted ΔPes, higher predicted ΔPL,dyn, and lower predicted lung compliance (expiratory tidal volume/predicted ΔPL,dyn) were associated with subsequent reintubation. CONCLUSIONS: During NIV, Pocc-derived parameters provide noninvasive estimates of inspiratory effort, lung stress, and mechanics, whereas ventilator P0.1 and plateau pressure are less reliable.