Diagnostic accuracy of plain films in detection of thoracolumbar fractures in minor trauma patients: comparison with CT.

Abstract

Purpose: To evaluate the diagnostic value of X-ray in detecting acute thoracolumbar (TL) fractures in minor trauma patients, using computed tomography (CT) as the reference standard, and to assess the impact of rigid spine conditions and reader experience on performance.

Material and methods: This retrospective single-centre study included patients with acute TL fractures from minor trauma between July 2014 and December 2020, who underwent both X-ray and CT. On CT, the presence or absence of rigid spine conditions, location, and fracture morphology were assessed. Two independent readers (a radiology resident and an attending radiologist) evaluated the radiographs, blinded to CTs. Sensitivity, specificity, and accuracy were calculated, and interobserver agreement was assessed using Cohen's κ coefficient.

Results: Sixty-three patients (32 with rigid spines, 31 without) with 84 fractures were included. The resident radiologist showed lower diagnostic accuracy than the attending radiologist, with more false positives in the rigid-spine group. In both groups, unrecognised fractures were more common for the resident radiologist (61.2% in rigid-spine patients and 48.6% in non-rigid-spine patients) compared to the attending radiologist (51.0% and 40.0%, respectively). Thoracic fractures were more frequently missed than lumbar fractures. Interobserver agreement was moderate (κ = 0.44) in the rigid-spine group and substantial (κ = 0.67) in the non-rigid-spine group.

Conclusions: Radiographs cannot reliably exclude unstable TL fractures in minor trauma patients. Attention should be paid to the lower thoracic region when evaluating lumbar radiographs.