Assessment of the Marshall and Rotterdam Scores in Predicting Elevations in Intracranial Pressure and Inpatient Mortality in Neurotrauma

Introduction: Intracranial pressure (ICP) monitoring is a key focus in management of neurotrauma patients. However, the decision to place ICP monitors relies on clinical judgment. This study assesses the utility of the Marshall and Rotterdam classification systems to predict inpatient mortality and ICP elevations in neurotrauma.

Methods: This retrospective cohort study identified 393 patients with severe TBI that underwent ICP monitoring at Indiana University Methodist Hospital between 2015-2023. ICPs were recorded and Marshall and Rotterdam scores were derived from admission head CT scans. Binary logistic regression models with odds ratios (OR) and 95% confidence intervals (CI) assessed the predictive ability of both scoring systems for inpatient mortality, initial intracranial hypertension (i-ICHTN), and delayed-onset intracranial hypertension (d-ICHTN) characterized by normal initial ICP with future elevations. ICP analysis was separated by whether patients underwent upfront surgery before monitoring.

Results: Of 393 patients, 19.3% had i-ICHTN and 51.9% had d-ICHTN. Rotterdam scores predicted inpatient mortality (OR 1.660, CI 1.335–2.064), while Marshall scores did not (OR 1.105, CI 0.951–1.285). In patients that did not have upfront surgery (n=303), Rotterdam scores predicted i-ICHTN (OR 1.466, CI 1.074–2.000) while Marshall scores did not (OR 1.085, CI 0.821–1.435). Neither Marshall scores (OR 1.178, CI 0.874–1.586) nor Rotterdam scores (OR 1.154, CI 0.831–1.602) predicted d-ICHTN in these patients. Marshall scores were not calculated for patients that had upfront surgery; Rotterdam scores did not predict i-ICHTN (OR 1.233, CI 0.773–1.967) or d-ICHTN in these patients (OR 0.772, CI 0.502–1.187).

Conclusion : The Rotterdam system was superior to the Marshall system in predicting inpatient mortality and i-ICHTN in medically managed patients, but neither can effectively predict initial or delayed ICP elevations in patients initially managed surgically. This highlights the need for a robust system to predict ICP derangements in surgically managed neurotrauma patients.