When appraising the result of a noninferiority trial, both the risk difference and its 95% confidence interval are considered. Although our best estimation of the risk difference was 8.4 percentage points, we acknowledge that the upper limit of the 95% confidence interval was close to the margin of noninferiority (20 percentage points) and that slightly different failure rates in either group could have altered our conclusions. Further research on the use of high-flow nasal cannulae in preterm and term newborn infants is required. Since such cannulae are used with increasing frequency in the nontertiary care setting, randomized trials of high-flow nasal cannulae as primary respiratory support from birth will be important. We conclude that treatment with high-flow nasal cannulae was noninferior to the use of nasal CPAP as respiratory support after extubation in very preterm infants. Since our trial was underpowered to show noninferiority in infants with a gestational age of less than 26 weeks, the use of high-flow nasal cannulae as first-line respiratory support after extubation in this extremely preterm group requires caution
cannulae as primary respiratory support from birth will be important. We conclude that treatment with high-flow nasal cannulae was non inferior to the use of nasal CPAP as respiratory support after extubation in very preterm infants. Since our trial was under powered to show non inferiority in infants with a gestational age of less than 26 weeks, the use of high-flow nasal cannulae as first-line respiratory support after extubation in this extremely preterm group requires caution