De-identified data.

Introduction
Mortality prediction aids clinical decision-making and is necessary for trauma quality improvement initiatives. Conventional injury severity scores are often not feasible in low-resource settings. We hypothesize that clinician assessment will be more feasible and have comparable discrimination of mortality compared to conventional scores in low and middle-income countries (LMICs).
Methods
Between 2017 and 2019, injury data were collected from all injured patients as part of a prospective, four-hospital trauma registry in Cameroon. Clinicians used physical exam at presentation to assign a highest estimated abbreviated injury scale (HEAIS) for each patient. Discrimination of hospital mortality was evaluated using receiver operating characteristic curves. Discrimination of HEAIS was compared with conventional scores. Data missingness for each score was reported.
Results
Of 9,635 presenting with injuries, there were 206 in-hospital deaths (2.2%). Compared to 97.5% of patients with HEAIS scores, only 33.2% had sufficient data to calculate a Revised Trauma Score (RTS) and 24.8% had data to calculate a Kampala Trauma Score (KTS). Data from 2,328 patients with all scores was used to compare models. Although statistically inferior to the prediction generated by RTS (AUC 0.92–0.98) and KTS (AUC 0.93–0.99), HEAIS provided excellent overall discrimination of mortality (AUC 0.84–0.92). Among 9,269 patients with HEAIS scores was strongly predictive of mortality (AUC 0.93–0.96).
Conclusion
Clinical assessment of injury severity using HEAIS strongly predicts hospital mortality and far exceeds conventional scores in feasibility. In contexts where traditional scoring systems are not feasible, utilization of HEAIS could facilitate improved data quality and expand access to quality improvement programming.