Double-lumen InSurE met non-inferiority—but its margin did not match the protocol

In 138 analyzed preterm infants, ventilation by 72 hours occurred in 14.5% with NDT-InSurE and 18.8% with LISA. The confidence interval clears a 10-point boundary, but the paper calls 20 points prespecified while the protocol describes 10—and 13 randomized infants were not analyzed.

THE HOT TAKE

The short-term numbers support non-inferiority. The reporting around the margin and the randomized set does not support calling the methods equivalent.

Investigators at six tertiary NICUs in Yunnan Province randomized 151 infants born at 32 weeks or less with respiratory distress syndrome. Eligible infants were on non-invasive respiratory support and needed FiO₂ of at least 0.30 to keep SpO₂ at 90%–95%. The published analysis included 69 infants per group after 13 post-randomization exclusions.

NDT-InSurE used a patented double-lumen tracheal tube designed by the study team: T-piece positive-pressure ventilation continued through the main lumen while surfactant passed through a 0.2 mm side channel, followed by extubation. LISA used a thin catheter while the infant remained on non-invasive support and breathed spontaneously. The primary endpoint was invasive ventilation within 72 hours because non-invasive support failed.

The endpoint occurred in 10/69 NDT-InSurE infants and 13/69 LISA infants. The reported risk difference was −4.3 percentage points for NDT-InSurE minus LISA, with a 95% confidence interval from −12.5 to +4.0 points. The P value of .493 tests superiority; it is not evidence of equivalence. Non-inferiority comes from comparing the confidence interval with a boundary chosen before the result is known.