Predicting Failure of Non-Invasive Ventilation With RAM Cannula in Bronchiolitis.

INTRODUCTION: In infants hospitalized for bronchiolitis on non-invasive ventilation (NIV) via the RAM cannula nasal interface, variables predicting subsequent intubation, or NIV non-response, are understudied. We sought to identify predictors of NIV non-response.

METHODS: We performed a retrospective cohort study in infants admitted for respiratory failure from bronchiolitis placed on NIV in a quaternary children's hospital. We excluded children with concurrent sepsis, critical congenital heart disease, or with preexisting tracheostomy. The primary outcome was NIV non-response defined as intubation after a trial of NIV. Secondary outcomes were vital sign values before and after NIV initiation, duration of NIV and intubation, and mortality. Primary analyses included Chi-square, Wilcoxon rank-sum, student's t test, paired analyses, and adjusted and unadjusted logistic regression assessing heart rate (HR) and respiratory rate (RR) before and after NIV initiation.

RESULTS: Of 138 infants studied, 34% were non-responders. There were no differences in baseline characteristics of responders and non-responders. HR decreased after NIV initiation in responders (156 [143-156] to149 [141-158], p < 0.01) compared to non-responders (158 [149-166] to 158 [145-171], p = 0.73). RR decreased in responders (50 [43-58] vs 47 [41-54]) and non-responders (52 [48-58] vs 51 [40-55], both p < 0.01). Concurrent bacterial pneumonia (OR 6.06, 95% CI: 2.54-14.51) and persistently elevated HR (OR: 1.04, 95% CI: 1.01-1.07) were associated with NIV non-response.

CONCLUSION: In children with acute bronchiolitis who fail to respond to NIV and require subsequent intubation, we noted associations with persistently elevated HR after NIV initiation and concurrent bacterial pneumonia.