METHODS:  Consecutive patients who were hospitalized (2008-2013) with non-penetrating moderate or severe TBI were studied. The primary outcome was HPC, defined by both a relative increase in contusion volume by ≥30 % and an absolute increase by ≥10 mL on serial imaging. Logistic regression models were created to identify independent risk factors for HPC. The HPC Score was then derived based on the final model.

RESULTS:  Among a total of 286 eligible patients, 61 (21 %) patients developed HPC. On univariate analyses, HPC was associated with older age, higher initial blood pressure, antiplatelet medications, anticoagulants, subarachnoid hemorrhage (SAH) subdural hematoma (SDH), skull fracture, frontal contusion, larger contusion volume, and shorter interval from injury to initial CT. In the final model, SAH (OR 6.33, 95 % CI, 1.80-22.23), SDH (OR 3.46, 95 % CI, 1.39-8.63), and skull fracture (OR 2.67, 95 % CI, 1.28-5.58) were associated with HPC. Based on these factors, the HPC Score was derived (SAH = 2 points, SDH = 1 point, and skull fracture = 1 point). This score had an area under the receiver operating curve of 0.77. Patients with a score of 0-2 had a 4.0 % incidence of HPC, while patients with a score of 3-4 had a 34.6 % incidence of HPC.

CONCLUSION:  A simple HPC Score was developed for early risk stratification of HPC in patients with moderate or severe TBI.