RESUMO
Background: Early diagnosis and fixation of fractures unlikely to unite can prevent months of morbidity. The Radiographic Union Score for Humeral fractures (RUSHU) is a summative scoring system developed to aid identification of patients at higher risk of developing humeral shaft non-union. Plain radiographs taken six weeks after injury are given a score between four and 12 based on signs of union. Our aim was to assess the validity of the RUSHU prognostic model in an external population. Methods: The radiographs of fifty-seven patients were scored independently according to RUSHU methodology by three reviewers (blinded to patient outcome). Interobserver intraclass correlation (ICC) was calculated. Results: Of the cohort, six (10.5%) progressed to non-union after six months. We observed an interobserver ICC co-efficient of 0.89 (95%CI0.84,0.93) in RUSHU score at six weeks. Median score was significantly higher in the union cohort (10v5 p < 0.001). Using the score of < 8 to predict non-union gave an area under the ROC curve of 0.87 (95%CI 0.83,0.90). Conclusions: In this retrospective single-centre study, we have demonstrated good inter-rater reliability. We would suggest that the RUSHU model be assessed in further external validation studies. RUSHU has the potential to reduce morbidity of delayed treatment of non-union.