Performance of the MOST Score for Severe Traumatic Brain Injury: First External Validation in Resource-Limited Neurosurgical Systems

Background: Prognostic models can assist neurosurgeons in making early decisions for patients with severe traumatic brain injury (TBI), yet most have been developed in high-resource trauma systems. The MOrtality Score for TBI (MOST) simplifies prediction using readily available clinical variables. Its performance in resource-limited settings remains unknown. Our objective was to externally validate the MOST score in a resource-limited neurosurgical environment and to compare its performance with the CRASH and IMPACT models. Methods: We performed a multicentric retrospective cohort study of adults with severe TBI (Glasgow Coma Scale [GCS] < 9) admitted to two neurosurgical referral centers. Demographics, injury characteristics, incident-to-presentation time (minutes), imaging, and corrected MOST/CRASH/IMPACT scores were collected. Primary outcome was in-hospital mortality; secondary outcomes included Glasgow Outcome Scale (GOS) at discharge. Discrimination was assessed with ROC AUCs and pairwise DeLong tests; calibration with Brier scores. Multivariable logistic regression identified independent predictors of mortality and poor functional outcome (GOS ≤ 3). Results: Among 1,105 patients, overall mortality was 12.1% and 94.8% had GOS ≤ 3 at discharge. Median incident-to-presentation time was 360 minutes (IQR 180–540); rural patients arrived later than urban patients (480 vs 180 minutes, p < 0.0001). In adjusted models, higher MOST scores were associated with increased mortality (OR 1.60, 95% CI 1.37–1.86), while higher GCS was protective (OR 0.59, 95% CI 0.42–0.83); IMPACT showed a modest inverse association (OR 0.87, 95% CI 0.78–0.97). MOST showed the best discrimination for mortality (AUC 0.85) versus CRASH (0.79) and IMPACT (0.82) (all pairwise p ≤ 0.026) and the lowest Brier score (0.187). Conclusions: The MOST maintained robust performance under the system constraints typical of low-resource neurosurgical care, confirming its generalizability beyond high-income trauma networks. These findings highlight the potential role of validated prognostic models in guiding triage and communication in severe TBI.