Multivariate autoregressive modelling combined with transcephalic electrical impedance: method to relate neonatal systemic circulation and respiration to cerebral circulation.

We studied the pulsatile component of cerebral circulation with transcephalic electrical impedance (delta Z) in six preterm newborns, three of whom had severe cerebral bleeding, peri-intraventricular haemorrhage (PIVH). The transcephalic electrical impedance delta Z signal, ECG, arterial blood pressure, (aBP) and respirogram were recorded on analogue magnetic tape for 30 min. Artefact-free stationary segments (lasting for 2 min) of the four signals were digitised. A digital multivariate autoregressive (MAR) model was used to study frequency-specific variability in the signals and to quantify interrelations between the variabilities of delta Z, HR, aBP and respiratory signals. MAR modelling describes a system where all the signals simultaneously explain each other. The inherent variability of delta Z was lower and the influences of respiration and aBP on delta Z significantly greater in infants with severe PIVH than in controls. These changes were observed at high frequencies corresponding to respiration and heart rate. This may be interpreted as a marker of pressure passivism in the cerebral circulation following PIVH. We conclude that in preterm babies the application of MAR modelling, together with transcephalic impedance, may be a new, helpful and quantitative method for the study of simultaneous interrelations between variables of cerebral and systemic circulations and respiration.