To Örebro University

PURPOSE: Rib fractures, common in blunt chest trauma, affect 10% of trauma patients and are linked to increased pulmonary morbidity and mortality. This study applies machine learning to identify predictors of complications in conservatively managed rib fracture patients.

METHODS: Data from the 2013-2021 American College of Surgeons' Trauma Quality Improvement Program included adults (≥ 18 years) with isolated thoracic injury from blunt trauma and conservatively managed rib fractures. Variables included demographics, comorbidities, injury severity, injury patterns, admission vitals, and complications. The permutation importance method identified top predictors of in-hospital complications.

RESULTS: Of 321,355 rib fracture patients, 183,303 (57.0%) had isolated rib fractures. The five primary predictors of complications in all rib fracture patients were age, Glasgow Coma Scale (GCS) on admission, Revised Cardiac Risk Index (RCRI), chronic obstructive pulmonary disease (COPD), and alcohol use disorder. For isolated rib fracture patients, the same predictors applied but in the order: age, RCRI, GCS, COPD, and alcohol use disorder. A logistic regression model using these predictors showed acceptable discriminative capacity for complications in the full cohort [AUC (95% CI): 0.72 (0.71-0.72)] and isolated rib fracture patients [AUC (95% CI): 0.72 (0.71-0.73)].

CONCLUSION: Cardiovascular risk, age, and level of consciousness on admission are key predictors of complications in conservatively managed rib fracture patients. Though complication rates remain low overall, elderly patients with multiple cardiovascular risk factors face a heightened risk of deterioration.