Patients
This study was designed as a prospective, observational study in invasively mechanically ventilated children admitted to the paediatric intensive care unit (PICU) of the Beatrix Children’s Hospital between March 2017 and June 2018 who were identified by the attending physician to be ready for extubation. Our clinical algorithm describes weaning as follows: weaning starts when ventilator pressures and/or mandatory breath rate can be decreased. During this process, patients are assessed daily during morning rounds by the attending physician for extubation readiness (i.e. able to breathe spontaneously when on CPAP/PS with pressure support < 12 cm H2O, FiO2 < 0.4 and an adequate coughing reflex). Patients were eligible if they have been invasively ventilated for at least 24 h and the attending physician confirmed extubation readiness and extubation was expected within 8 h. For logistical reasons, patients were only studied on weekdays from 7 am to 5 pm if they had been intubated > 24 h prior to the ERT. Patients with depressed respiratory drive inherent to congenital or acquired central nervous system disorders, congenital or acquired injury to the phrenic nerve or diaphragmatic dysfunction, unstable haemodynamics (i.e. increase in vasoactive support or fluid boluses < 6 h before ERT), congenital or acquired neuro- and/or myopathy, continuous muscular paralysis 12 h before the ERT, patients who had a tracheostomy and patients with ETT leakage > 20% were not studied. Importantly, patients with clinically identified post-extubation upper airway obstruction were removed from analysis because we also wanted to explore the relationship between WOBimp and extubation outcome. The institutional review board (IRB) approved the study and waived the need for informed consent.
Measurement protocol
Patients were intubated with a cuffed ETT (KimVent, Microcuff Endotracheal Tube, Paediatrics, Roswell, USA) and ventilated with the AVEA® ventilator (CareFusion, Yorba Linda, CA, USA). Prior to the ERT, a 3.5 French (Fr) catheter for ETT < 4.5 mm and 5 Fr for ETT ≥ 4.5 mm (Argyle, Covidien, Mansfield, USA) with the tip of the catheter at the distal end of the ETT was inserted. The patient was then switched to CPAP/PS with the level of PS set similar to the added pressure above the level of PEEP during controlled MV, targeting an expiratory Vt of 5–7 mL/kg actual bodyweight (as there was no obesity in the patient cohort). Vt was measured at the Y-piece of the patient circuit using a self-calibrating pneumotachometer (VarFlex™, CareFusion, Yorba Linda, CA, USA). Flow trigger was set between 0.5 and 1.0 L/min. A heat moisture exchanger (Gibeck, Teleflex Medical, Vianen, The Netherlands) was in situ between the patient circuit and the ETT.
After a 5-min stabilisation period, data were recorded during 5 min of steady-state breathing. Subsequently, PS was turned down to zero and, after a 5-min stabilisation period, again data were recorded during a period of 5 min steady-state breathing.
Ventilator recordings were sampled at 100 Hz using the VOXP protocol and a custom-build software program (Polybench, Applied Biosignals, Weener, Germany).
Heart rate (HR), respiratory rate (RR), peripheral saturation (SpO
2) and fraction inspired oxygen (FiO
2) were recorded on case record forms at baseline (i.e. after the first 5-min stabilisation period), after 5 min of steady-state breathing on CPAP/PS, and after 5 min of steady-state breathing on CPAP. The Comfort B score was calculated at these same time points to assess patient comfort [
17]. Demographic and baseline clinical data were collected to characterise the studied population included gender, age, weight, 24-h paediatric RISK of mortality (PRISM) III score, admission diagnosis and ETT size [
18].
Extubation failure was defined as the need for reintubation within 48 h or use of non-invasive ventilation (NIV) post-extubation.
Data analysis
Ventilator recordings were analysed offline using a custom-build MatLab script (MATLAB 2018a, The Mathworks, Natick, USA). The median (IQR) of respiratory variables including peak inspiratory pressure (PIP), Ptrach, PEEP, mean airway pressure (mPaw), expiratory Vt number of breaths, RR, rapid shallow breathing index (RSBI), end-tidal CO
2 (ET-CO
2), peak inspiratory flow rate (PIFR) and WOBimp was calculated for the 5-min recordings after removal of artefacts. Peak inspiratory resistance (cmH
2O/L/S) was calculated using ETT size (3.0 mm–6.0 mm) and PIFR using formulae used by Khemani et al. [
15].
Statistical analysis
Data were assessed for normality using the Kolmogorov–Smirnov test. Descriptive data were expressed as median (25–75 interquartile range) or percentage (%) of total. For the univariate analysis, data recorded during CPAP/PS were compared with data recorded during CPAP alone using the Wilcoxon signed rank test. Subsequently, multivariate linear regression analysis using backward selection was performed to study the independent contribution of ETT size, VTe, inspiratory time (Tinsp) and PIFR to changes in WOBimp (∆WOBimp) because we presumed these variables to be related to WOBimp. Statistical analysis was performed using SPSS v23 (IBM, Armonk, NY, USA). P values < 0.05 were considered statistically significant.