Inter-observational analysis of computed tomography parameters to predict nonobvious posterior ligament complex injury in neurologically intact patients with thoracolumbar trauma

Abstract

Introduction

Assessing the integrity of the posterior ligament complex (PLC), as a key element in the characterization of an unstable Thoracolumbar fracture (TLF), is challenging, but crucial in the choice of treatment.

Research question

How to create a reproducible score using combined parameters of Computed Tomography (CT) to predict nonobvious PLC injury. How CT parameters relate with PLC status.

Material and methods

Retrospective analysis of neurologically intact patients with an acute traumatic TLF, who underwent CT and Magnetic Resonance Imaging (MRI) within 72 h, in the Emergency Department of a single institution between January 2016 and 2022. Four investigators rated independently 11 parameters on CT and PLC integrity on MRI. The interrater reliability of the CT parameters was evaluated, and two risk scores were created to predict PLC injury on CT using the coefficients of the multivariate logistic regression.

Results

154 patients were included, of which 62 with PLC injury. All CT measurements had excellent or good interrater reliability. Patients with Horizontal Fracture of the lamina or pedicle (HLPF), Spinous process fracture (SPF) and Interspinous Distance Widening (IDW) were positively associated with PLC injury (p < 0.001, p < 0.001 and p = 0.045, respectively). Risk Score 2 (RS2), which included only statistically significant variables, had a total of 75.9% of correct classifications (p < 0.001), with a sensitivity of 71.0% and specificity of 78.3% to estimate PLC injury detected in the MRI.

Discussion and conclusion

Standardized procedures pre-established in the CT measurement protocol were effective. Identically to early findings, those three CT measurements showed a positive relation to PLC injury, thus enhancing the conclusions of previous studies. Comparing to the reliability of the CT findings above mentioned, the score was less precise.