Statistical analysis

A sample size of 390 participants per group was chosen to have 80% power to demonstrate that the modified HFNC group was superior to the COT group for the primary measure (proportion of patients with an SpO₂ < 90% during bronchoscopy), with the use of a margin of 0.08 based on an observed 28% of patients with an SpO₂ < 90% during bronchoscopy for the COT group in a previous study and an assumed 20% for the modified HFNC group [16].

The results for continuous variables are shown as either means (± standard deviation) or medians (with interquartile ranges). Groups were compared using either Student’s t test or the Mann–Whitney U test, as appropriate. For categorical variables, the percentage of patients in each category was compared using a chi-square test or Fisher’s exact test. The overall time course for vital signs and FiO₂ was compared using two-way analysis of variance for repeated measures.

The independent predictors were assessed for risk factors associated with hypoxemia during bronchoscopy in the modified HFNC group via a univariate analysis. The statistically significant (p ≤ 0.1) variables from the univariate analysis were included in a multivariate analysis. The multivariate analysis was assessed using multiple logistic regression based on backward stepwise selection. We used a receiver operating characteristic (ROC) curve to confirm the cutoff value of patients with hypoxemia in the modified HFNC group.

All p values were two-sided, and values < 0.05 were considered significant. Data were analyzed using statistical software (SPSS 21.0; IBM Corp., Armonk, NY, USA).