Toward Causal Relationships of Intracranial Pressure, Cerebrovascular Reactivity, Interventions, and Outcome in...

Most available evidence for treatment and outcome prediction in traumatic brain injuries (TBI) is derived from observational studies. Such association studies provide limited evidence for the causality of relationships. While randomized controlled trials are appropriately regarded as the gold standard for causal inference, there are many domains where such experimental approaches are not feasible. During the last decades however, mathematical methods have been developed to infer causal relationships from observational data. This study aims to investigate causal relations of systemic and cerebral physiological parameters, intracranial pressure (ICP)-lowering interventions, and outcome in patients with severe TBI.

Patients with severe TBI enrolled in the prospective observational Collaborative European Neurotrauma Effectiveness Research in TBI (CENTER-TBI) High-Resolution Substudy group were included (n = 201). We applied a causal inference algorithm to determine causal relations between the first week post-injury mean of values of ICP, cerebrovascular reactivity (PRx), cerebral perfusion pressure (CPP), heart rate, oxygen saturation, arterial partial pressure of carbon dioxide, ICP-lowering therapies [represented by the therapy intensity level (TIL) score], and 6-month outcome represented by Glasgow Outcome Scale-Extended (GOSE).

The median age was 51.5 years [interquartile range (IQR) 31 to 64], 158 were men (77.5%), and median GOSE at 6 months was 3 (IQR 3 to 5). PRx and TIL score were the only features identified as directly related to outcome in this analysis (r = - 0.19, p = 0.003; r = - 0.20, p = 0.002, respectively). Significant correlations between ICP and CPP (r = - 0.62, p < 0.0001); ICP and PRx (r = 0.31, p < 0.001); PRx and heart rate (r = 0.17, p = 0.007); and PaCO₂ and TIL score (r = - 0.19, p = 0.004) were identified after conditioning on triples of the other features.

The results suggest that cerebrovascular autoregulation status (PRx) and ICP-lowering therapies may be more directly associated with outcomes than ICP itself and that these factors should be considered in ICP monitoring and management strategies.