Plain film measurement error in acute displaced midshaft clavicle fractures

• Published in: Canadian Journal of Surgery 2016-10, Vol.59 (5), p.311-316
• Main Authors: Archer, Lori Anne, MD, Hunt, Stephen, MD, Squire, Daniel, MD, Moores, Carl, MD, Stone, Craig, MD, O’Dea, Frank, MD, Furey, Andrew, MD, MSc
• Format: Article
• Language: English
• Subjects: Adult; Aged; Aged, 80 and over; Clavicle; Clavicle - diagnostic imaging; CT imaging; Diagnostic errors; Estimates; Female; Fracture Fixation - standards; Fractures, Bone - diagnostic imaging; Humans; Injuries; Male; Medical imaging; Middle Aged; Orthopedics; Patients; Prevention; Radiography - standards; Reproducibility of Results; Software; Statistical analysis; Studies; Surgery; Tomography, X-Ray Computed - standards; Trauma; Young Adult
• ISSN: 0008-428X; 1488-2310
• DOI: 10.1503/cjs.003016

Background Clavicle fractures are common and optimal treatment remains controversial. Recent literature suggests operative fixation of acute displaced mid-shaft clavicle fractures (DMCFs) shortened more than 2 cm improves outcomes. We aimed to identify correlation between plain film and computed tomography (CT) measurement of displacement and the inter- and intraobserver reliability of repeated radiographic measurements. Methods We obtained radiographs and CT scans of patients with acute DMCFs. Three orthopedic staff and 3 residents measured radiographic displacement at time zero and 2 weeks later. The CT measurements identified absolute shortening in 3 dimensions (by subtracting the length of the fractured from the intact clavicle). We then compared shortening measured on radiographs and shortening measured in 3 dimensions on CT. Interobserver and intraobserver reliability were calculated. Results We reviewed the fractures of 22 patients. Bland–Altman repeatability coefficient calculations indicated that radiograph and CT measurements of shortening could not be correlated owing to an unacceptable amount of measurement error (6 cm). Interobserver reliability for plain radiograph measurements was excellent (Cronbach α = 0.90). Likewise, intraobserver reliabilities for plain radiograph measurements as calculated with paired t tests indicated excellent correlation ( p > 0.05 in all but 1 observer [ p = 0.04]). Conclusion To establish shortening as an indication for DMCF fixation, reliable measurement tools are required. The low correlation between plain film and CT measurements we observed suggests further research is necessary to establish what imaging modality reliably predicts shortening. Our results indicate weak correlation between radiograph and CT measurement of acute DMCF shortening.