How is the Marshall CT classification used to stage a traumatic brain injury?

Marshall CT Classification for Traumatic Brain Injury Staging

The Marshall CT classification is a six-category system that uses initial head CT findings to stratify traumatic brain injury severity and predict mortality, based primarily on the status of basal cisterns, midline shift, and presence/volume of mass lesions. 1, 2

Classification Categories

The Marshall system divides TBI patients into six distinct groups:

Diffuse Injury Categories (I-IV):

• Diffuse Injury I: No visible intracranial pathology on CT
• Diffuse Injury II: Cisterns present with midline shift 0-5 mm and/or high-density or mixed-density lesion ≤25 cm³
• Diffuse Injury III: Cisterns compressed or absent, with midline shift 0-5 mm, no high- or mixed-density lesion >25 cm³
• Diffuse Injury IV: Midline shift >5 mm, no high- or mixed-density lesion >25 cm³

Mass Lesion Categories (V-VI):

• Evacuated Mass Lesion (V): Any lesion surgically evacuated
• Non-Evacuated Mass Lesion (VI): High- or mixed-density lesion >25 cm³ not surgically evacuated

Clinical Application for Risk Stratification

Major criteria for suspected intracranial hypertension requiring treatment include Marshall Diffuse Injury III (compressed cisterns) or IV (midline shift >5 mm), or presence of non-evacuated mass lesions. 1

The classification demonstrates reasonable discrimination for mortality prediction (AUC 0.67-0.68), with a threshold score of ≥4 consistently associated with significantly higher mortality risk across multiple validation studies. 3, 4, 5

Prognostic Performance and Limitations

Strengths:

• Best performance for short-term outcomes: Marshall score shows relatively higher accuracy for in-hospital mortality (AUC 0.777) and acute-phase risk stratification 6
• Simple, widely adopted system requiring only basic CT interpretation
• Effective for identifying patients needing neurosurgical intervention 2

Important Limitations:

• Moderate overall discriminative ability compared to newer systems (Helsinki, Stockholm, Rotterdam) 6, 7
• Significantly reduced accuracy in older adults (≥65 years) compared to younger patients 7
• Does not incorporate traumatic subarachnoid hemorrhage or intraventricular hemorrhage, which independently predict mortality 8
• Provides minimal incremental prognostic value beyond clinical assessment (age + Glasgow Coma Scale) when used alone 9

Practical Recommendations

Use Marshall classification primarily for:

1. Acute emergency department triage and initial risk stratification 6, 9
2. Standardized communication between neurosurgery, critical care, and emergency teams 2
3. Identifying patients requiring immediate neurosurgical evaluation (categories III-VI) 1, 4

Critical pitfall to avoid: Do not rely solely on Marshall score for prognostic discussions with families or rehabilitation planning, as it makes no meaningful contribution to predicting functional outcomes at discharge or long-term recovery 10. For longer-term prognostication (6-12 months), consider Helsinki or Stockholm scores instead 6.

For comprehensive mortality prediction, combine Marshall findings with clinical variables (age, GCS, pupillary reactivity) and consider adding traumatic subarachnoid hemorrhage and intraventricular hemorrhage assessment, which can improve discrimination from AUC 0.67 to 0.77 8.