The purpose of the study is to describe early predictors and to develop a prediction tool that accurately identifies the need for mechanical ventilation (MV) in pneumonia patients with hypoxemic acute respiratory failure (ARF) treated with high-flow nasal cannula (HFNC).Materials and methods:
This is a 4-year prospective observational 2-center cohort study including patients with severe pneumonia treated with HFNC. High-flow nasal cannula failure was defined as need for MV. ROX index was defined as the ratio of pulse oximetry/fraction of inspired oxygen to respiratory rate.Results:
One hundred fifty-seven patients were included, of whom 44 (28.0%) eventually required MV (HFNC failure). After 12 hours of HFNC treatment, the ROX index demonstrated the best prediction accuracy (area under the receiver operating characteristic curve 0.74 [95% confidence interval, 0.64-0.84]; P < .002). The best cutoff point for the ROX index was estimated to be 4.88. In the Cox proportional hazards model, a ROX index greater than or equal to 4.88 measured after 12 hours of HFNC was significantly associated with a lower risk for MV (hazard ratio, 0.273 [95% confidence interval, 0.121-0.618]; P = .002), even after adjusting for potential confounding.Conclusions:
In patients with ARF and pneumonia, the ROX index can identify patients at low risk for HFNC failure in whom therapy can be continued after 12 hours.