Evaluation of two novel thoracolumbar trauma classification systems

Background Despite numerous attempts at classifying thoracolumbar spinal injuries, there remains no consensus on a single unifying algorithm of management. The ideal system should provide diagnostic and prognostic information, exhibit adequate reliability and validity and be easily applicable to clinical practice. The purpose of this study is to assess the reliability and validity of two novel classification systems for thoracolumbar fractures - the Thoracolumbar Injury Severity Score (TLISS) and the Thoracolumbar Injury Classification and Severity Score (TLICS) - and also to discuss potential efforts towards research in the future. MATEREIALS AND METHODS: Seventy-one patients with thoracolumbar fractures were prospectively assessed by surgeons with different levels of training and experience (attending orthopedic surgeon, attending neurosurgeon, spine fellows, senior level and junior level residents) at a single institution. Plain radiographs, CT and MRI imaging were used to classify these injuries using the TLISS system. Seven months later, 25 consecutive injuries were prospectively assessed with the TLISS and TLICS systems. Unweighted Cohen's kappa coefficients and Spearman's correlation values were calculated to assess inter-observer reliability and validity at each point in time.Results For both the TLISS and TLICS algorithms, the inter-rater kappa statistics for all of the subgroups demonstrated moderate-to-substantial reliability (0.45-0.74), although there were no significant differences among the shared subgroups. The kappa score of the TLISS system was greater than that of the TLICS system for injury mechanism/ morphology. Correlation values were also greater across all subgroups (P ≤0.01). Statistically significant improvements in TLISS inter-observer reliability were observed across all TLISS fields (P Conclusion The TLISS and TLICS scales both exhibited substantial reliability and validity. However, the TLISS system displayed greater inter-observer correlation than did the TLICS and demonstrated significant improvements in reliability over time.