Assessment of quality of continuous intracranial pressure recordings in children

Abstract

Background: Bad quality of continuous intracranial pressure (ICP) recordings may potentially give wrong diagnostic information.

Method: This study examined the ICP recordings of 33 children undergoing ICP monitoring during a 12-month period. Their ICP recordings were stored as raw data files and analyzed retrospectively. An automatic algorithm applied to every subsequent 6-second time window of each recording assessed the quality of ICP recordings. Time windows containing 4-18 cardiac beat-induced single ICP waves were accepted; time windows containing artifact-induced pressure waves due to noise in the signal were rejected. For every ICP recording, the number of accepted and rejected time windows were determined. In addition, the raw data ICP signal was assessed by visual inspection to verify the automatic algorithm. Two methods of computing mean pressure of a time window are compared; mean ICP was either computed according to current practice independent of the presence or absence of single ICP waves or computed only for single ICP waves within the time window (mean(SW) ICP).

Results: In 4 of the 33 patients (12%), more than 70% of the time windows of their ICP recordings were rejected; in 3 patients 95-100% of time windows were rejected. Visual inspection of the raw ICP signals confirmed the absence of single ICP waves. In these cases, computing mean(SW) ICP, not by mean ICP, revealed the bad signal quality.

Conclusions: This retrospective study showed that in 3 of 33 patients (9%) the quality of the ICP recordings was so bad that they gave wrong diagnostic information. One way of obtaining quality control is to compute mean(SW) ICP, determined only for cardiac beat-induced single ICP waves within a time window.