Temporal Statistical Relationship between Regional Cerebral Oxygen Saturation (rSO(2)) and Brain Tissue Oxygen Tension (PbtO(2)) in Moderate-to-Severe Traumatic Brain Injury: A Canadian High Resolution-TBI (CAHR-TBI) Cohort Study

Brain tissue oxygen tension (PbtO(2)) has emerged as a cerebral monitoring modality following traumatic brain injury (TBI). Near-infrared spectroscopy (NIRS)-based regional cerebral oxygen saturation (rSO(2)) can non-invasively examine cerebral oxygen content and has the potential for high spatial resolution. Past studies examining the relationship between PbtO(2) and NIRS-based parameters have had conflicting results with varying degrees of correlation. Understanding this relationship will help guide multimodal monitoring practices and impact patient care. The aim of this study is to examine the relationship between PbtO(2) and rSO(2) in a cohort of TBI patients by leveraging contemporary statistical methods. A multi-institutional retrospective cohort study of prospectively collected data was performed. Moderate-to-severe adult TBI patients were included with concurrent rSO(2) and PbtO(2) monitoring during their stay in the intensive care unit (ICU). The high-resolution data were analyzed utilizing time series techniques to examine signal stationarity as well as the cross-correlation relationship between the change in PbtO(2) and the change in rSO(2) signals. Finally, modeling of the change in PbtO(2) by the change in rSO(2) was attempted utilizing linear methods that account for the autocorrelative nature of the data signals. A total of 20 subjects were included in the study. Cross-correlative analysis found that changes in PbtO(2) were most significantly correlated with changes in rSO(2) one minute earlier. Through mixed-effects and time series modeling of parameters, changes in rSO(2) were found to often have a statistically significant linear relationship with changes in PbtO(2) that occurred a minute later. However, changes in rSO(2) were inadequate to predict changes in PbtO(2). In this study, changes in PbtO(2) were found to correlate most with changes in rSO(2) approximately one minute earlier. While changes in rSO(2) were found to contain information about future changes in PbtO(2), they were not found to adequately model them. This strengthens the body of literature indicating that NIRS-based rSO(2) is not an adequate substitute for PbtO(2) in the management of TBI.