A bedside score using arrival data to identify major hemorrhage after blunt pelvic ring fracture: Development and validation

Early recognition of major hemorrhage after blunt pelvic ring injury could guide mobilization of transfusion and hemorrhage-control resources. We developed an arrival-based bedside score and compared its performance with shock index.

We studied 239,736 adults with documented blunt pelvic ring fractures in the 2019-2024 National Trauma Data Bank. Major hemorrhage was defined as a hemorrhage-control procedure, hemorrhagic death, or transfusion of ≥ 1000 mL or ≥ 4 units of red blood cells within 4 h. Penalized logistic regression selected arrival variables for an integer score. We assessed performance in a held-out test set (n = 79,912) and at approximately matched alert volume. A model refitted on 2019-2022 admissions was evaluated in 2023-2024. Subgroup analyses were post hoc.

Major hemorrhage occurred in 26,479 patients (11.0%); 31.0% were aged ≥ 75 years, and 45.9% had fallen. In the test set, the score achieved an AUROC of 0.882 (95% CI, 0.879-0.886), exceeding shock index by 0.060 (95% CI, 0.055-0.064; P < 0.001). At approximately 15.6% of patients flagged, the score identified 678 additional major hemorrhage cases with 684 fewer false-positive alerts. The temporally validated model achieved an AUROC of 0.885 (95% CI, 0.881-0.888). Discrimination exceeded shock index for pelvic ring disruption (0.830 vs 0.778) and high-energy mechanisms (0.811 vs 0.764), but was lower after falls (0.779 vs 0.800).

In patients with documented blunt pelvic ring fractures, an arrival-based score identified major hemorrhage more accurately than shock index at a similar alert burden. Reduced performance after falls and uncertain applicability to suspected fractures warrant caution. External validation and prospective impact assessment are needed before clinical use.