Retrospective validation of a risk stratification tool developed for the management of patients with blunt chest trauma (the <scp>STUMBL</scp> score)
Saptarshi Mukerji, Eunicia Tan, Charlotte May, Christina Micanovic, Paul Blakemore, Kelly J. Phelps, H. Melville, Peter Jones
Abstract
OBJECTIVE: To assess validity of the STUMBL score in New Zealand for complications of blunt chest trauma without multi-trauma and immediate life-threatening injuries. METHODS: A multi-centre, retrospective observational study was carried out in five EDs. Area under the receiver operating characteristic curve (AUROC) was calculated for all, early and late complications and ethnic sub-groups. Youden Index generated for each ROC was used to indicate cut scores for risks of complication, ICU admission, prolonged length of stay (LOS) and mortality. RESULTS: A total of 445 patients were included. AUROC for all complications composite were (0.73, 95% confidence interval [CI] 0.68-0.77), mortality (0.92, 95% CI 0.89-0.94), ICU admissions (0.78, 95% CI 0.73-0.81) and prolonged LOS (0.80, 95% CI 0.76-0.83) were calculated. The score performed better in the New Zealand European (Pākehā) sub-group compared to Māori and Pasifika (AUROC [95% CI]: 0.80 [0.73-0.85], 0.69 [0.56-0.79], 0.66 [0.46-0.82], respectively). Patients with scores >12 were at risk of complications from blunt chest trauma, >15 at risk of prolonged LOS and >18 at risk of ICU admission and mortality. CONCLUSIONS: The STUMBL score at a cut-off of <12 did not predict all complications sufficiently well to recommend for general use in our population. However, a score >15 predicted prolonged LOS and a score >18 predicted mortality sufficiently to be clinically useful for these outcomes. The score is more accurate in New Zealand Pākehā and needs to be used with caution in Māori and Pasifika populations. A larger prospective validation is required to further assess the score.