Compensatory-reserve-weighted intracranial pressure versus intracranial pressure for outcome association in adult traumatic brain injury: a CENTER-TBI validation study

Background: Compensatory-reserve-weighted intracranial pressure (wICP) has recently been suggested as a supplementary measure of intracranial pressure (ICP) in adult traumatic brain injury (TBI), with a single-center study suggesting an association with mortality at 6 months. No multi-center studies exist to validate this relationship. The goal was to compare wICP to ICP for association with outcome in a multi-center TBI cohort. Methods: Using the Collaborative European Neuro Trauma Effectiveness Research in TBI (CENTER-TBI) high-resolution intensive care unit (ICU) cohort, we derived ICP and wICP (calculated as wICP = (1 − RAP) × ICP; where RAP is the compensatory reserve index derived from the moving correlation between pulse amplitude of ICP and ICP). Various univariate logistic regression models were created comparing ICP and wICP to dichotomized outcome at 6 to 12 months, based on Glasgow Outcome Score—Extended (GOSE) (alive/dead—GOSE ≥ 2/GOSE = 1; favorable/unfavorable—GOSE 5 to 8/GOSE 1 to 4, respectively). Models were compared using area under the receiver operating curves (AUC) and p values. Results: wICP displayed higher AUC compared to ICP on univariate regression for alive/dead outcome compared to mean ICP (AUC 0.712, 95% CI 0.615–0.810, p = 0.0002, and AUC 0.642, 95% CI 0.538–746, p < 0.0001, respectively; no significant difference on Delong’s test), and for favorable/unfavorable outcome (AUC 0.627, 95% CI 0.548–0.705, p = 0.015, and AUC 0.495, 95% CI 0.413–0.577, p = 0.059; significantly different using Delong’s test p = 0.002), with lower wICP values associated with improved outcomes (p < 0.05 for both). These relationships on univariate analysis held true even when comparing the wICP models with those containing both ICP and RAP integrated area under the curve over time (p < 0.05 for all via Delong’s test). Conclusions: Compensatory-reserve-weighted ICP displays superior outcome association for both alive/dead and favorable/unfavorable dichotomized o