The Reliability and Validity of the Thoraco-Lumbar Injury Classification and Severity score

Abstract

Abstract Background The TLICS/TLISS is a recently introduced classification system for thoracolumbar spinal column injures designed to simplify injury classification and facilitate treatment decision making. Before being widely adopted, the reliability and validity of the TLICS/TLISS must be studied. Aim of the Work To determine the interrater and intrarater reliability and the validity of the TLICS score and its predecessor the TLISS scoring system in the clinical practice. Patients and Methods A total of 7 articles with 10 datasets were used to test the inter- and intrarater reliability and validity of the TLICS/TLISS score for thoracolumbar spine trauma. Included studies presented Thoracolumbar trauma cases’ details (including clinical data, plain radiographs, CT scans and MRI) to spine surgeons allowing them to score them using the TLICS/TLISS score in each of its components (neurologic status, PLC integrity and fracture morphology/mechanism), the final score and surgeons’ agreement with the scores treatment recommendations; as well as comparing the treatment recommendations (surgical vs non-surgical management) with the treatment the patients actually received. At a later time the surgeons were assigned the task with the cases reordered. The interrater reliability, as well as the intrarater reliability of the score for each component and sum, were evaluated by Cohen’s unweighted k-value and Spearman’s rank order correlation. In addition, the sensitivity and specificity of the score (validity) were evaluated by the percent of correct treatment recommendations according to the sum of the TLICS/TLISS and the treatment actually received by the patients. Results Interrater reliability assessed by generalized kappa coefficients was 0.45 ±0.17 for injury morphology/mechanism, 0.91 ±0.03 for neurologic status, 0.42 ±0.13 for posterior ligamentous complex status, 0.36 ±0.14 for TLICS/TLISS total, and 0.59 ±0.10 for treatment recommendation. Respective results using the Spearman correlation were 0.52 ±0.18, 0.95 ±0.05, 0.57 ±0.13, 0.75 ±0.10, and 0.64 ±0.20. Intrarater kappa coefficients were 0.53 ±0.14 for injury morphology/mechanism, 0.89 ±0.07 for neurologic status, 0.53 ±0.15 for posterior ligamentous complex status, 0.46 ±0.16 for TLICS/TLISS total, and 0.61 ±0.02 for treatment recommendation. Respective results using the Spearman correlation were 0.63 ±0.08, 0.90 ±0.03, 0.64 ±0.10, 0.77 ±0.03, and 0.60 ±0.02. The percent of correct treatment recommendation by the score and the treatment actually received by the patients was 94.4% ±1.5, with sensitivity of 0.91 ±0.06, specificity of 0.94 ±0.01, PPV of 0.94 ±0.01, NPV of 0.93 ±0.02. Conclusions The TLICS/TLISS score has good reliability and validity and it compares favorably to other contemporary and old thoracolumbar fracture classification systems.