Low-frequency sampling for PRx calculation does not reduce prognostication and produces similar CPPopt in intracerebral haemorrhage patients

Abstract

Background: The cerebral pressure reactivity index (PRx) correlates with the outcome in intracerebral haemorrhage (ICH) patients and has been used to define an autoregulation-oriented "optimal cerebral perfusion pressure" (CPPopt). PRx has been calculated as a moving correlation coefficient between mean arterial pressure (MAP) and intracranial pressure (ICP) averaged over 5-10 s, using a 2.5- to 5-min moving time window, in order to reflect changes in MAP and ICP within a time frame of 20 s to 2 min. We compared PRx with a different calculation method [low-frequency PRx (L-PRx)], where rapid fluctuations of MAP and ICP are cancelled (waves with frequencies greater than 0.01 Hz).

Methods: A total of 548.5 h of artefact-free data (sampling frequency 1 Hz) from 18 patients suffering from non-traumatic ICH were included in the analysis. L-PRx was calculated using minute averages, between both MAP and ICP, in 20-min moving correlation windows. CPPopt was calculated based on PRx and on L-PRx.

Results: The averaged PRx values for each patient correlated with L-PRx (P = 0.846, p < 0.001). CPPopt based on standard PRx was identified in eight patients. In contrast, a CPPopt value based on L-PRx could be found in 12 patients. CPPopt values by both methods correlated strongly with each other (P = 0.980, p < 0.001). L-PRx had a similar correlation with the National Institutes of Health Stroke Scale Score (NIHSS) (0.667, p = 0.002) as did PRx (0.563, p = 0.015).

Conclusions: L-PRx correlated with the outcome as good as PRx did. CPPopt could be identified in more patients using L-PRx. Slower MAP and ICP changes (in the range of 1-20 min) can be used for autoregulation assessment and contain important prognostic information.