Table 2.
The outcome effect by the implementation of humidified high-flow nasal canula on transport (n = 3022)
| Outcome | Cohort | Level change between 2010–2014 and 2015–2019a | Difference in level changes between two cohortsb | p |
|---|---|---|---|---|
| PICU length of stay, ratio (95% CI)c | ||||
| Transported | 0.69 (0.54–0.87) | 0.64 (0.49–0.83) | 0.001 | |
| Non-transport | 1.07 (0.89–1.28) | – | ||
| Estimated PICU length of stay, absolute difference, hour (95% CI)d | ||||
| Transported | − 9.1 (− 14.8 to − 3.3) | − 10.6 (− 16.9 to − 4.3) | – | |
| Non-transport | 1.5 (− 2.8 to 5.9) | – | ||
| IV + NIV hours, absolute difference, hour (95% CI)e | ||||
| Transported | − 17.0 (− 37.1 to 3.1) | − 25.1 (− 47.9 to − 2.3) | 0.03 | |
| Non-transport | 8.1 (− 8.0 to 24.2) | – | ||
| IV hours, absolute difference, hour (95% CI)e | ||||
| Transported | − 11.5 (− 21.1 to − 1.9) | − 22.4 (− 33.8 to − 10.9) | < 0.001 | |
| Non-transport | 10.9 (-2.0 to 19.7) | – | ||
–, not applicable
PICU paediatric intensive care unit, CI confidence interval, IV invasive ventilation, NIV non-invasive ventilation
aLevel change between the pre-intervention era (2010–2014) and post-intervention era (2015–2019)
bThis is the main result of the study, presenting the outcome effect by the intervention by subtracting the outcome level changes between pre- and post-intervention era in non-transport children from the outcome level change in transported children
cThe ratio was estimated based on a linear regression model with the log-transformed outcome
dThe absolute difference of PICU length of stay between pre- and post-intervention era was predicted by the fitted linear regression model with the covariate adjustment
eThe absolute difference of the outcome was predicted based on a zero-inflated binomial regression as a substantial proportion of observed outcomes were zero