ABSTRACT
BACKGROUND: Thoracolumbar injury classification and severity score (TLICS) is not very reliable for assessment of injury to the posterior ligament complex, even when scored by experts. It is not reasonable to score every compression fracture or burst fracture the same as there is great variety in the severity of compression fractures and burst fractures. PURPOSE: To propose a modified TLICS (mTLICS) and evaluate the performance of the mTLICS system by measuring the agreement between scores determined by radiologists using both systems and actual treatment procedure delivered. MATERIAL AND METHODS: We retrospectively evaluated 134 patients with acute lumbar and thoracic spinal trauma after undergoing magnetic resonance imaging (MRI) using new mTLICS and conventional TLICS system. Inter-observer agreements of TLICS and mTLICS scores were analyzed using the kappa statistic. Nonparametric correlation analysis was used to determine correlation (R) among each score and the surgical intervention. RESULTS: The mTLICS system showed slightly higher correlation than TLICS (Rs, TLICS, 0.592 and 0.613 vs. mTLICS, 0.628 and 0.639). If we consider a total maximal score of 4 to be a negative surgical indication, mTLICS showed significantly higher sensitivities than TLICS, and if we consider a total minimal score of 4 to be a positive surgical indication, mTLICS showed significantly higher specificities than TLICS. CONCLUSION: The mTLICS score corrects deficiencies in the TLICS system that lead to ambiguity in the radiological diagnostic criteria. mTLICS is a more suitable scoring system than TLICS for predicting surgical management accurately, especially for morphological injuries.
