Multicenter Validation of the Survival After Acute Civilian Penetrating Brain Injuries (SPIN) Score

Abstract

Background: Civilian penetrating traumatic brain injury (pTBI) is a serious public health problem in the United States, but predictors of outcome remain largely understudied. We previously developed the Survival After Acute Civilian Penetrating Brain Injuries (SPIN) score, a logistic, regression-based risk stratification scale for estimating in-hospital and 6-mo survival after civilian pTBI with excellent discrimination (area under the receiver operating curve [AUC-ROC = 0.96]) and calibration, but it has not been validated.

Objective: To validate the SPIN score in a multicenter cohort.

Methods: We identified pTBI patients from 3 United States level-1 trauma centers. The SPIN score variables (motor Glasgow Coma Scale [mGCS], sex, admission pupillary reactivity, self-inflicted pTBI, transfer status, injury severity score, and admission international normalized ratio [INR]) were retrospectively collected from local trauma registries and chart review. Using the original SPIN score multivariable logistic regression model, AUC-ROC analysis and Hosmer-Lemeshow goodness of fit testing were performed to determine discrimination and calibration.

Results: Of 362 pTBI patients available for analysis, 105 patients were lacking INR, leaving 257 patients for the full SPIN model validation. Discrimination (AUC-ROC = 0.88) and calibration (Hosmer-Lemeshow goodness of fit, P value = .58) were excellent. In a post hoc sensitivity analysis, we removed INR from the SPIN model to include all 362 patients (SPINNo-INR), still resulting in very good discrimination (AUC-ROC = 0.82), but reduced calibration (Hosmer-Lemeshow goodness of fit, P value = .04).

Conclusion: This multicenter pTBI study confirmed that the full SPIN score predicts survival after civilian pTBI with excellent discrimination and calibration. Admission INR significantly adds to the prediction model discrimination and should be routinely measured in pTBI patients.

Keywords: Gunshot wound; Outcome prediction; Penetrating brain injury; Prediction model; Traumatic brain injury.