Purpose: The respiratory rate-oxygenation (ROX) index is used to predict high-flow nasal cannula (HFNC) success in acute respiratory failure, including in Coronavirus disease 2019 (COVID-19) patients. However, no study has described its performance to predict failure of alternating sessions of noninvasive ventilation (NIV) and HFNC in severe COVID-19 patients.
Material and methods: We conducted a monocentric retrospective cohort study. COVID-19 patients admitted in the intensive care unit (ICU) for acute respiratory failure were treated by alternating sessions of HFNC and NIV. The primary endpoint was the ability for ROX index at 2 hours (h) of NIV initiation to predict HFNC/NIV failure defined by orotracheal intubation (OTI) within 7 days after noninvasive support initiation.
Results: One hundred and five patients were included in analysis, of which 47% (n = 49) required OTI by day seven. ROX index values were significantly lower in intubated group at all time points but 24 h. In multivariate analysis, a ROX index at 2 h < 4.88 was associated with a higher risk of HFNC/NIV failure (Hazard Ratio 1.90 [95% Confidence Interval 1.03-3.51], p = 0.039). The area under the receiver operating characteristic curve for ROX index at 2 h was 0.702 [0.608-0.790]. Optimal cut-off value was 5.22. Sensitivity and specificity for predicting intubation with this threshold were 71.4% and 63.3%, respectively.
Conclusions: In our study, the ROX index had a good predictive power for alternating sessions of HFNC and NIV failure in patients with acute respiratory failure due to SARS-CoV-2.
Keywords: ROX index; coronavirus disease 2019; high flow nasal cannula; mechanical ventilation; noninvasive ventilation; severe acute respiratory syndrome coronavirus type 2.