Classification of Civilian Craniocerebral Gunshot Wounds: A Multivariate Analysis Predictive of Mortality

Abstract

Management of cerebral gunshot injuries has changed considerably since Cushing's (1916) and Matson's (1948) classification schemes, developed during World War I and World War II, respectively. These military injuries are characterized by either very high mass, low-velocity shrapnel wounds or by high muzzle velocity missiles causing extensive destruction of tissue. The preponderance of low muzzle velocity weapons seen in clinical practice and the availability of computed tomographic (CT) evaluation within minutes after presentation has altered the range of prognostic indicators available to the neurosurgeon and the amount of relative importance placed on each factor. Raimondi and Samuelson (1970) noted this difference in wound ballistics and offered a classification scheme based on initial neurologic assessment. No well-defined classification system for civilian craniocerebral gunshot wounds has been proposed that evaluates and integrates clinical, laboratory, and neuroradiologic data. A retrospective study was performed on all 62 civilians with gunshot wounds to the head admitted to the University of Virginia Hospital between December, 1984, and November, 1990. The patient population consists of 86% males and 14% females, with an age range of 10–72 years; 60% self-inflicted wounds and 32% patients who died en route or immediately upon arrival at the hospital. The overall mortality rate was 55% at 1 week postinjury. Although we have demonstrated an association between some previously defined factors and prognosis in civilian injury, such as admission Glasgow Coma Scale (GCS) (p = 0.001) and initial pupillary response (p < 0.001), we have also defined other significant predictors of outcome including abnormal coagulation states on admission (p < 0.001) and the neuroradiologic examination. Parameters for heads assessed using CT for statistical significance included number and type of intracerebral planes crossed (midsagittal, p < 0.004; midcoronal, p = 0.859; horizontal midventricular, p < 0.002; all three planes crossed, p = 0.017), ventricular involvement (p = 0.024), presence of intracerebral hemorrhage (p = 0.013), and estimation of volume of contused brain on initial evaluation (p < 0.001).