Initial cranial CT scan for the assessment of prognosis in head injury

Summary A total of 208 multiple trauma patients with head injury (HI) were investigated who had been treated in the period from 1990 to 1995. The average age was 35.2 ± 17.7 years; the injury severity according to ISS was 30.2 ± 8.6 points; 20.5 % died as a result of the HI; the mortality of all patients was 26.5 %. The Glasgow Coma Scale (GCS) was determined at an average of 22 min after trauma (8.0 ± 4.3 points) at the scene of accident. The patients were classified according to GCS into minor HI (group 1: 14–15 points), moderate HI (group 2: 9–13 points) and severe HI (group 3: 3–8 points). Patient outcome was assessed by the Glasgow Outcome Scale (GOS) and was classified as good (GOS 4 and 5) and poor (GOS 1, 2 and 3) outcome. At the latest, 2 h after trauma, a CT scan of the head (CCT) was done. The HI groups are compared regarding frequency of types of injury. In all HI groups the fractures of the bony face occurred at the same frequency (36.0–38.9 %). The frequency of calotte fractures (Kal-Fx) increased from group 1 (8.0 %) to 2 (19.2 %) and 3 (25.6 %); fractures of the skull base significantly differed between group 1 (16.0 %), 2 (7.8 %) and 3 (33.4 %). Epidural hemorrhage (EDB) appeared only in group 2 (7.8 %) and 3 (6.7); subdural hemorrhage was found in group 1 (2.7 %), 2 (7.8 %) and 3 (10.0 %). Subarachnoid hemorrhage (SAB) was significantly more frequently seen, dependent on HI severity, in group 3 (26.7 %) compared to group 2 (11.7 %) and 1 (8.0 %). Intracerebral contusion (ICK) significantly increased from group 1 (12.0 %) to 2 (27.3) and 3 (45.6 %). Brain swelling (BS) also significantly increased from group 1 (8.0 %) to 2 (19.5 %) and 3 (49.0 %) and lesions of ventricles (VL) from group 1 (2.7 %) to 2 (11.7 %) and 3 (20.0 %). Midline shift (13.4 %) and signs of herniation (4.5 %) only occurred in group 3. The analysis of correlation/regression and receiver operating characteristics was able to predict 79 % of patients' outcome accurately using GCS (r 0.54; P < 0.0001) alone, using CCT (r 0.65; P < 0.0001) 87 % were correctly predicted with significant variables Cal-Fx, EDB, SAB and BS. CCT with GCS (r 0.74; P < 0.0001) were able to predict 88 % accurately with significant variables Cal-Fx, EDB, BS and GCS. The combination of CCT with GCS, age and ISS (r 0.78; P < 0.0001) was able to predict only 87 % correctly, although the r value was the highest; significant variables were Kal-Fx, EDB, BS, VL, GCS, age and ISS.