Citation

Buchholz CJ, Jia L, Manea C, Petersen T, Wang H, Stright A, Young J, Calland JF. Am. Surg. 2022; ePub(ePub): ePub.

Copyright

(Copyright © 2022, Southeastern Surgical Congress)

DOI

10.1177/00031348221123087

PMID

36120831

Abstract

BACKGROUND: Patients with rib fractures have variable clinical courses and it is difficult to predict which patients will do poorly. Ideally this prediction would happen at the time of admission to facilitate effective triage. One scoring system devised to this end, is the Battle score. This study aims to evaluate the efficacy of the Battle score as triage tool, and to re-tool it for performance in an inpatient trauma setting.

METHODS: A multivariate logistic regression model was trained on patients admitted to a level one trauma center with at least one rib fracture. A composite outcome was used to classify those who had poor outcomes. Eighteen candidate predictors were analyzed in univariate analysis, then the most promising fed into the logistic model until a triage score was built and internally validated by bootstrapping.

RESULTS: Of the 838 patients who met the inclusion criteria, 145 (17.3%) patients had a defined poor outcome. The relevant predictors included in the final scoring system were number of ribs fractured, chest tube, pulmonary contusions, chronic obstructive pulmonary disease, and Glasgow coma score. Age was not found to be predictive. This score was found to have higher fidelity in predicting poor outcomes than the original Battle score (AUROC.858 vs.649.).

DISCUSSION: An easy to calculate clinical scoring system was created to triage patients with rib fractures at the time of admission. Age may be of less importance than previously thought, while injury burden and history of lung disease may play a larger role.

Language: en

Keywords

trauma; regression modeling; rib fractures