Prediction of Cerebral Contusion Volume by Admission Data

Introduction: Brain contusions are an intraparenchymal brain bleed of traumatic origin that are a leading cause of morbidity and mortality in neurosurgery. Few studies have characterized factors that affect contusion volume. Here we aim to characterize objective measurements and clinical exam findings that predict contusion volume change after admission.

Methods: Admission and in-hospital data was collected on a total of 285 patients with cerebral contusions between 2014 and 2019 at our institution. Contusion volume was categorized as a binary measurement – less than 15 mL vs 15 mL or more. Outcome measures were compared utilizing Chi-square contingency table analysis and Student t-tests. Stepwise logistic regression was used to determine significant predictors of outcome measures. Analyses were carried out using SAS, version 9.4 (SAS Institute, Cary, NC). Statistical significance was selected as p = 0.05. No adjustments were made for multiple comparisons.

Results: Patients with contusion >15 ml were found to present with significantly lower Glasgow Coma Scale (GCS) eye (2.54 vs. 2.98, p=0.0170), verbal (2.61 vs. 3.38, p=0.0014), motor (4.00 vs. 4.68, p=0.0112) and total scores (9.10 vs. 11.04, p = 0.0041). These patients also had longer ICU stays (7.89 vs. 6.84 days, p=0.0467). No labs or objective admission data were found to vary between the two groups including serum sodium levels. Logistic regression showed that GCS verbal scores were predictive of contusion volume sizes demonstrating a significant negative relationship (OR: 0.779, 95% CI: 0.667-0.911). Hosmer and Lemeshow Goodness-of-Fit Test showed an appropriate fit of the model (p=0.7172).

Conclusion : GCS scores and ICU length of stay vary between patients based on contusion size. We found that a low admission GCS verbal score was a significant predictor of increased contusion volume.