Objective Impaired cerebral pressure autoregulation (CPA) is associated with brain damage and mortality in preterm infants. CPA can be calculated by means of spectral analysis of spontaneous fluctuations in mean arterial blood pressure (MAP) and a surrogate of cerebral blood flow obtained by near-infrared spectroscopy using coherence (a frequency-domain measure of correlation) and gain (the ratio of magnitude of flow to pressure). The Autoregulatory Index (ARI) describes the degree of CPA, where 1 indicates no change in flow with MAP and 0 indicates proportional changes, i.e. complete loss of CPA.

Methods Sampling and analysis of the oxygenation index (Hamamatsu, NIRO-300) and invasive MAP were performed simultaneously using in-house written software in Labview. 110 minutes of stable signal from one extremely preterm infant was divided into epochs of 10 to 40 minutes and subdivided into segments of 5 minutes with 50% overlap. We applied a Hanning-window before FFT-transformation in two frequency bands (0.003–0.04 Hz and 0.04–0.1 Hz). Thresholds of statistical significance were calculated for coherence (Taylor JA et al, 1998) and gain was used to estimate ARI.

Results Prolonging the epochs increased statistical significance. Coherence for epochs longer than 20 minutes were all statistically significant yet ARI was 0.80 and 0.67 for low and high frequencies, respectively.

Conclusions In spite of highly significant coherence this extremely preterm infant had near-normal ARI. The conventional threshold is >0.4. Also, as there is a time-lag for change in vascular smooth muscle cell tone it was expected that CPA was slightly less complete in the high frequencies.