Sonographic bridging callus: An early predictor of fracture union

BACKGROUND: There is currently a lack of agreed criteria for sonographic assessment of callus and reliability between reviewers. The primary aim of this study was to determine criteria and reviewer agreement for sonographic bridging callus (SBC) on ultrasound. The secondary aim was to analyse the use of ultrasound to detect bridging callus in a prospective cohort of patients with a conservatively managed clavicle fracture.

METHODS: A prospective cohort of conservatively managed displaced midshaft clavicle fractures underwent ultrasound scanning at three-, six- and 12-weeks post-injury. The main outcome was nonunion confirmed at six months on CT scanning. Five patients with confirmed nonunion were compared against a control group of 15 patients with timely union at three months. The ultrasound scans were interpreted by two blinded reviewers to evaluate sonographic callus features with agreement determined by weighted kappa. A further validation study was undertaken by four blinded reviewers using the intraclass-correlation-coefficient (ICC) using the most clinically relevant findings of the pilot work.

RESULTS: At three weeks post-injury fibrocartilaginous material was present in 80% of patients (16/20). When detected this was associated with union (sensitivity 93%, specificity 60%, p = 0.03) with the inter-observer agreement rated 'fair' on kappa (0.44). At six weeks only 10% (2/20) of patients had bridging callus on radiograph but 60% (12/20) had sonographic bridging callus (SBC) and when present all united (sensitivity 80%, specificity 100%, p = 0.002). At 12 weeks, bridging callus was present on both radiographs and ultrasound in all patients that united (n = 15, sensitivity 100%, specificity 100%, p < 0.001). No patient that developed a nonunion at six months post-injury had SBC at any time point. At six-weeks the absence of SBC had a positive predictive value for nonunion of 63% of patients (5/8) and by 12 weeks it was 100% (5/5). The SBC detection rated 'very strong' for intra- (kappa 0.92) and inter-observer agreement (kappa 0.84). The ICC of SBC at six-weeks with four blinded reviewers was 0.82 (95% confidence interval 0.68-0.91).

CONCLUSIONS: This is the first study to establish time specific ultrasound fracture findings with a repeatable technique and assess the agreement between blinded reviewers.