Predicting the success of non-invasive ventilation in preventing intubation and re-intubation in the paediatric intensive care unit

Purpose To determine whether physiological parameters and underlying condition can be used to predict which patients can be managed successfully on non-invasive ventilation (NIV). Methods Review of case notes and computerised data of every paediatric intensive care unit (PICU) admission over 7 years where NIV was commenced. Data immediately prior to commencing NIV and 2 h after its establishment was collected. Univariable and multivariable statistical analysis was performed to compare variables. Results Eighty-three patients commenced NIV attempting to avoid intubation and 64% succeeded. Those who failed required a higher FiO2 (0.56 vs. 0.47, p = 0.038), had higher respiratory rates (53.3 vs. 40.3 breaths/min, p = 0.012) and lower pH (7.26 vs. 7.34, p = 0.032) before NIV and higher FiO2 after NIV commenced (0.54 vs. 0.43, p = 0.009). Those with a respiratory diagnosis were more likely to be successful. Patients with oncologic disease, particularly if septic, were less likely to avoid intubation using NIV. Multivariable models showed that after adjustment for mode of NIV and underlying diagnosis, respiratory rate before NIV was an independent predictor of success [adjusted odds ratio (OR) 0.95 (0.91, 0.99), p = 0.01]. Eighty patients were extubated to NIV but 15 required re-intubation. Those re-intubated had a higher systolic blood pressure (104 vs. 77.9 mmHg, p = 0.001) and diastolic blood pressure (64.5 vs. 54.1 mmHg, p = 0.0037) after extubation. Multivariable models showed that systolic blood pressure 2 h after extubation was independently associated with outcome [adjusted OR 0.96 (0.93, 0.99), p = 0.007]. Conclusions Parameters relating to respiratory and cardiovascular status can determine which patients will successfully avoid intubation or re-intubation when placed on NIV. Underlying disease and reason for admission should be considered when predicting the outcome of NIV.