Clinical predictors
Diagnostic categories were also significant predictors of NIV outcomes. Overall, children with upper airway disorders were less likely to discontinue NIV compared to the other diagnostic categories (0.581 (95% CI 0.348 to 0.971, Figure 1). Children with CNS disorders had 3.3 times higher odds of being hospitalized while using NIV. Specific underlying conditions that were predictors of different NIV outcomes included children with bronchopulmonary dysplasia, who had higher odds of improving their breathing status and discontinuing NIV [4.707 (95% CI 1.306 to 16.962)], and children with obesity/metabolic syndrome who had higher odds of declining NIV therapy [2.077 (95% CI 1.110 to 3.888)].
Parameters of medical complexity including use of other medical technology, number of co-morbidities, and previous surgeries were also significant predictors of long-term NIV outcomes. Children requiring any technology in addition to NIV use had 3.3 (95% CI 2.051 to 5.387) times higher odds of NIV failure (switch to IMV or death) and 2.6 (95% CI 1.526 to 4.340) times higher odds of hospitalization for each added technology (Figure 1). They were also less likely to decline NIV, with a 0.417 (95% CI 0.202 to 0.859) reduction in odds ratio for each additional technology. The presence of comorbidities increased the odds for hospital admissions by 1.2 (95% CI 1.022 to 1.440) times but was not a predictor for NIV failure or any other long-term NIV outcomes. Children who had received a previous adenotonsillectomy surgery had 3.7 times less odds of NIV failure (switching to IMV or dying) compared to children who did not. The presence of previous major surgeries was not a significant predictor of long-term outcomes.