To test the hypothesis that in neonates on mechanical ventilation, heart rate characteristics index (HRCi) can be combined with a clinical model for predicting extubation outcomes in neonates.

HRCi and clinical data for all intended intubation-extubation events (episodes) were retrospectively analyzed between June 2014 and January 2015. Each episode started 6 hours pre-extubation or at the time of primary intubation if ventilation duration was shorter than 6 hours (baseline). The episodes ended at 72 hours postextubation for successful extubations or at reintubation for failed extubations. Mean of 6 hourly epoch HRCi-scores (baseline) or fold-changes (postextubation) were analyzed. Results are expressed as medians (IQR) for continuous data and proportions for categorical data. Multivariable logistic regression mixed model was used for statistical analysis.

Sixty-six infants contributed to 96 episodes (18 failed extubations, 78 successful extubations) in the study. Failed extubations had significantly longer duration of ventilation (65.3 hours, 19.94-158.2 vs 38.4, 16.5-71.3) and more culture positive sepsis (33.3% vs 3.8%) than successful extubations. Baseline HRCi scores (1.68, 1.29-2.45 vs 0.95, 0.54-1.86) and postextubation epoch-1 fold changes (1.25, 0.94-1.55 vs 0.94, 0.82-1.11) were higher in failed extubations compared with successful extubations. Multivariable linear mixed-effects regression was used to create prediction models for success of extubation, using relevant variables.

The baseline and postextubation HRCi were significantly higher in neonates with extubation failure compared with those who succeeded. Models using HRCi and clinical variables to predict extubation success may add to the confidence of clinicians considering extubation.