Continuous Cerebral Autoregulation Monitoring by Cross-Correlation Analysis

Publisher Website
Google Scholar
Abstract
In order to validate cross-correlation analysis between spontaneous slow oscillations of arterial blood pressure (aBP) and intracranial pressure (ICP) or flow velocity as a means to assess the status of cerebral autoregulation continuously, we compared its results with different autoregulation bedside tests. The second aim was to check the method's stability over longer time periods. aBP, ICP, and flow velocity in the middle cerebral artery (FVMCA) was measured continuously in 13 critically ill comatose patients. Cross-correlation analysis was performed online and offline between aBP and ICP (CC [aBP → ICP]) and aBP/FVMCA (CC [aBP → FVMCA]). Three different autoregulation bedside tests (cuff deflation, transient hyperemic response, orthostatic hypotension) were performed immediately before a 29-min cross-correlation test period. In addition, continuous cross-correlation autoregulation monitoring was performed over multiple hours (in order to analyze for stability and to assess the influence of other factors). Cluster analysis revealed two main clusters. Cluster 1 (indicative for disturbed autoregulation) showed a centroid at t = -0.21 ± 3.32 sec, r = 0.43 ± 0.18 for CC [aBP → ICP], and t = 0 ± 3.14 sec, r = 0.44 ± 0.18 for CC [aBP → FVMCA]. Cluster 2 (indicative for normal autoregulation) revealed a centroid at t = 4.94 ± 3.74 sec, r =- 0.4 ± 0.16 for CC [aBP → ICP], and t = 3.38 ± 4.44 sec, r = -0.38 ± 0.18 for CC [aBP → FVMCA]. Comparison between the cross-correlation test results and the bedside tests showed a sensitivity of 44-73% for CC [aBP → FVMCA], whereas CC [aBP → ICP] was more specific (60-80%). Long-term monitoring revealed stable cross-correlation tests in about 45% of the measurement time. It is concluded that cross-correlation between aBP, ICP, and FVMCA is a valid means to monitor the autoregulation status continuously, although further improvement of sensitivity and specificity is needed to make it reliable for clinical decision making.