Distribution of tidal ventilation during volume targeted ventilation is variable and influenced by age in the preterm lung

ELECTRONIC SUPPLEMENTARY MATERIAL

DETAILED METHODS

Study population

This study was performed on the Neonatal Intensive Care Unit of The Royal Women’s Hospital, Melbourne, Australia between August 2008 and July 2009. The study was approved by the institution’s Human Research and Ethics Committee (HREC 08/04). Signed, informed parental consent was obtained for each infant.

Stable preterm infants less than 32 weeks’ completed gestational age at birth and between 24 hours and 10 weeks of life were eligible for study if receiving SIPPV with VTV, using a maximum PIP of 30 cm H2O via the Dräger Babylog 8000+ ventilator (Drägerwerk, Lübeck, Germany). Ventilator settings, including the targeted VT, were determined by the treating clinician and not altered during the study. Infants were excluded if they were too unwell for routine handling, had fragile skin, required a fractional inspired oxygen concentration greater than 0.9, had hypotension refractory to treatment, an established air leak syndrome, evolving abdominal pathology, congenital cardiac disease or a chromosomal anomaly.

Measurements and Method

The gestational age, birth weight, age and weight at time of study, respiratory diagnoses, ventilator settings, ductus arteriosus status and relevant concomitant illnesses and medications were recorded from each infant’s medical records.

Relative impedance change (ΔZ) during tidal inflations was measured in the supine position using EIT (GeoMF II EIT system, Cardinal Health, Hoechberg, Germany) sampling at 44Hz and using the proprietary back-projection algorithm [27, 29]. This involved placing 16 neonatal electrocardiogram electrodes (Kendall Puppydog™ 1042PTS, Tyco Healthcare, Mansfield, MA), at equidistant intervals, circumferentially around the thorax at nipple level. Electrodes were pre-trimmed to avoid contact between adjacent electrodes and signal quality was confirmed prior to study. Following electrode application the infant was left undisturbed in the supine position for at least twenty minutes prior to data collection. Then, during a period of quiet rest, three individual recordings of physiological parameters, each of two minute duration, were taken over approximately 15 minutes. This triplet sequence was repeated twenty minutes later. During each recording the infant’s chest was videoed to enable identification of movement artefact or unusual breathing patterns during analysis of the EIT data. Airway pressure was simultaneously recorded by the EIT unit via its single analogue input channel.Flow at the airway opening was recorded from the ventilator using separate acquisition software (LabVIEW™, National Instruments, Austin TX) operating on the same computer. Tidal volume (VT) at the airway opening was then integrated from the flow signal.

Oxygen saturation, respiratory rate and heart rate were measured throughout by the bedside monitor (Philips IntelliVue™ MP80 Monitor, Eindhoven, Netherlands) and manually recorded at 12-second intervals.

Data analysis

The quality of electrode conductance improves with time therefore the unfilteredEIT recordings for each infant were reviewed offline in reverse order using an analysis program custom-built by one of the investigators (AS) and running in MatLAB (Mathworks Inc, Natick, MA). From the six available recordings, a total of three, 30-second artefact free periods were selected, each from separate recordings. The first encountered artefact free period was chosen from the most recent recording. If there was no such period, the previous recording was analysed.To isolate the ΔZ due to tidal ventilation (ΔZVT) a low-pass filter at a frequency of 10 breaths above each infant’s respiratory rate was applied to each selected data set.

Distribution of tidal ventilation

A functional EIT image was generated using the standard deviation image of the ΔZVT time course of each individual pixel in the 32 x 32 matrix [27, 29]. For the purpose of analysing the variation of spatial ventilation, distribution profiles of the 32 x 32 matrix were generated from the anterior to the posterior aspects of the chest. Each bar of the EIT profiles representing one of 32 slices of the chest arranged from anterior to posterior chest, and expressed as a percentage distance from the sternum (most anterior; 0%) to the spine (most posterior; 100%) [26, 40]. To simplify interpretation, the magnitude of ΔZVT within the area under the curve (AUC) for the anterior, middle and posterior thirds of the chest, and the geometric centre of ventilation within the chest, was calculated [26].

Regional variability of tidal ventilation

In each 30-sec period, the PEEP to PIP amplitude (ΔP) for all inflations, and the peak to trough tidal difference in regional impedance in the anterior (ΔZVTANT) and posterior (ΔZVTPOST) hemithoraces (expressed in countless units; cu) that coincided with the ΔPwere determined. To determine the variability of ΔP, ΔZVTANT and ΔZVTPOST the coefficient of variation (CV) in each parameter was calculated from each 30-sec recording. The differences in the CV of each parameter were then compared. in addition, the CV of VT at the airway opening within each 30-sec recording was also determined if synchrony of the EIT and flow signals could be confirmed for all inflations.

The study population was analysed in entirety and then divided, for subgroup comparison, into those infants aged ≤7 and >7 days (representing the acute and early chronic phases of preterm respiratory illness respectively), by need for any supplemental oxygen versus ventilation in air (as a proxy for magnitude of lung disease) and also by VTV (≤3.0 mL and >3.0 mL, with 3.0 mL being the median value). The true difference in regional VT was not known. A convenience sample of 30 infants was chosen, based on our admission rates and demographics. After testing for normality, data was analysed with either t-test or repeated-measures ANOVA with Tukey post-test as appropriate. Statistical analysis was performed using GraphPad Prism version 4.02 for Windows (Graphpad software, San Diego, CA), and a p value <0.05 considered significant.

1