CHICAGO – The Marshall CT classification system is a better predictor of early death in patients with traumatic brain injury than is the Rotterdam score, an award-winning study found.
As observed in previous reports, higher scores on both computed tomography–based scoring systems were significantly associated with early death among patients with mild to severe traumatic brain injury, or TBI (both P values less than .0001).
In a logistic regression analysis, however, only the Marshall scoring system was significantly associated with early death, defined as death at hospital discharge (P less than .002), Dr. Daddy Mata Mbemba reported at the annual meeting of the Radiological Society of North America.
This is likely because the two strongest independent predictors of early death – the absence of basal cistern and positive midline shift – are included in both CT scoring systems, while the next strongest predictor, hemorrhagic mass, is included in the Marshall score only, he explained.
Marshall and Rotterdam are the two most commonly used radiologic scoring systems, according to Dr. Mbemba, but they group CT findings differently. The Marshall score includes the status of basal cisterns, midline shift greater than 5 mm, and hemorrhagic mass, whereas the Rotterdam score includes basal cisterns, midline shift greater than 5 mm, subarachnoid hemorrhage and/or intraventricular hemorrhage, and epidural hemorrhage, but not hemorrhagic mass, explained Dr. Mbemba, a radiologist at Tohoku University, Sendai City, Japan.
He noted that several studies have suggested that increased intracranial pressure may lead to early death, while vasospasm associated with subarachnoid hemorrhage and/or intraventricular hemorrhage that decrease intracerebral circulation is associated with a worsening clinical outcome.
To assess whether the Marshall or Rotterdam scores are related to death at hospital discharge, investigators in the current study reviewed the initial CT scans and status at hospital discharge of 245 consecutive patients with mild to severe TBI. Mild cases, defined by a Glasgow Coma Scale score of 13-15, were included if a CT examination was recommended according to New Orleans Criteria and/or the Canadian CT Head Rule. Patients’ mean age was 49 years (range, 15-93 years), and 67% were men.
At hospital discharge, 25 patients had died and 220 were alive. The median time to death was 3 days (range, 1-83 days), said Dr. Mbemba, who earned a trainee research award from RSNA for his study.
In a logistic regression analysis, CT findings independently related to early death were basal cistern status (odds ratio, 771.5; P less than .001), positive midline shift (OR, 56.2; P = .0011), hemorrhagic mass (OR, 12.9; P = .0065), and subarachnoid hemorrhage and/or intraventricular hemorrhage (OR 3.8; P = .0394), Dr. Mbemba said. The presence or absence of an epidural hemorrhage was not a significant predictor.
In a recently published study, initial Marshall and Rotterdam scores were significantly associated with mortality after severe TBI, while Glasgow Coma Scale scores on admission were not (Neurosurgery 2012;70:1095-105). No relationship was observed, however, between any of the three scoring systems and ICU intracranial pressure or brain tissue oxygen tension, suggesting that factors associated with outcome, may not always predict a patient’s ICU course, especially intracranial physiology.
Dr. Mbemba and his coauthors reported no relevant financial disclosures.