Assess evidence on respiratory support between prophylactic NIV sessions after extubation, particularly high-flow nasal oxygen versus standard oxygen, and consider the limits of the available comparison.
Approach:
Background: The article distinguishes prophylactic NIV, used immediately after extubation in high-risk patients, from rescue NIV for established respiratory failure, which may delay needed reintubation.
Evidence reviewed: Thille and colleagues pooled individual patient data from two multicentre trials. Neither trial randomized patients to high-flow nasal oxygen versus standard oxygen during NIV breaks.
Analysis: The pooled comparison was observational. The authors report crude and G-computation-adjusted estimates, with adjustment including trial of origin.
Key Findings:
Crude extubation failure by day 7 was 13.7% with high-flow nasal oxygen and 18.5% with standard oxygen (difference −4.7 percentage points; 95% CI −9.4 to −0.3; p = 0.036).
After adjustment, the difference was −3.9 percentage points (95% CI −8.7 to 0.9), and was no longer statistically significant.
Nearly all patients receiving standard oxygen came from one trial, creating substantial confounding between trial of origin and oxygen strategy.
The group-classification rule placed some patients who failed during continuous NIV, before receiving standard oxygen, in the standard-oxygen group.
Interpretation:
The adjusted estimate does not establish equivalence: residual confounding and inadequate power remain possible. The article notes that NIV addresses respiratory mechanics through PEEP and pressure support, and argues that the oxygen modality during NIV breaks may contribute little if NIV reverses the relevant mechanisms during treatment sessions.
Limitations:
The oxygen strategies were not randomized in either trial.
The pooled trials addressed different questions and enrolled heterogeneous patients.
Trial of origin and oxygen strategy were nearly perfectly confounded.
The comparison may be affected by immortal time bias and by classification of early failures into the standard-oxygen group.
Adjustment cannot eliminate unmeasured confounding, and the pooled sample was not powered for this comparison.
Conclusion:
The article states that routine high-flow oxygen during NIV breaks cannot be justified without evidence of benefit and supports a randomized trial, including assessment of which physiological subgroups may benefit.
