Study Of Reliability And Reproducibility Of Four Kinds Of Classification Systems For Thoracolumbar Fracture

BACKGROUND CONTEXT：Thoracolumbar fracture is a disease which has high incidence rate, high costand high disability rate,and it is the most common disease of spinal fracture. As a medicalproblem its classification has always been concerned by experts both at home and abroad.The main causes of thoracolumbar fracture are the high falling injury, traffic accidentinjury and osteoporosis pathological fracture. In recent years, due to the developmentof industry, the increase of traffic accidents, and the aging of the population, itsincidence has been increasing, so the exploring the classification method further of newthoracolumbar fractures, to guide clinical treatment more efficiently, it has a importantsocial and medical value. Because of the complexity of thoracolumbar fracture, famousscholars of various countries has done different researches for classification ofthoracolumbar fracture, after nearly a century of development, classification ofthoracolumbar fractures is increasing, the concept of traditional conservative treatmenthas changed, and the positive operation treatment has become the main trend at present,andits curative effect has been proved by a lot of cases. But complications arise, whichrequires clinicians’ cognition on therapy getting deepening, new classification method ofthoracolumbar fracture speed up the development of the field. Several classifcationsystems have been proposed, the most frequently used are the Denis, AO, load sharingclassification (LSC) and thoracolumbar injury classification and severity core (TLICS)classifcation systems.PURPOSE:To compare the interobserver reliability and intraobserver reproducibility of the Denis,AO, LSC and TLICS classifcation systems and identify the clinical decision-making in thetreatment of thoracolumbar injury．METHODS:Sixty five consecutive patients in our hospital from June2012to June2013with adiagnosis of a thoracolumbar fracture were enrolled in this study. All patients had atraumatic mechanism of injury and were initially seen in our emergency department as either the primary or tertiary treatment facility. Patients had detailed evaluations providedby the emergency room physicians, the general surgery trauma service, and the orthopedicspine service. All patients had been initially evaluated by plain film radiography and axialcomputed tomography with biplanar reconstructions in the sagittal and coronal planes.Treatment of their spinal fractures was then undertaken under the direction of the spineAttending. The clinical and radiographic findings of65patients with thoracolumbarfracture were prospectively presented to3groups of orthopedic residents. These injurieswere consecutively scored, using the Denis, AO, LSC and TLICS classif cation systems.On the basis of these classif cation systems, the decision of treatment was made. Theclinical and radiographic data for each of the cases in this series were randomly reorderedand reevaluated with the same algorithm after three months. Cohen s k coefficient wascalculated for the key parameters of each classification system to analysis the reliabilityand reproducibility of the Denis, AO, LSC and TLICS classif cation systems.RESULTS:Each successive system played an important role in advancing contemporaryunderstanding of thoracolumbar injuries. Most classif cations were, however, based on asingle individual s, or a comparatively small group s, retrospective review of a case series.In most instances, these grading systems were never validated or modif ed by theiroriginal developers, a shortcoming that prevented their continued evolution. The AO classif cation attempted to design a comprehensive system in which every fracture f t that hisclassif cation system was capable of predicting neurologic def cits. The interrater kappastatistics of all subgroups in TLICS (morphology, status of the posterior ligaments, totalscore, predicted management) were within the range of moderate to substantial reliability.There were no statistically significant differences noted between the respective kappavalues in two evaluations. In the TLICS, subgroups of evaluators, showed variation inreliability as expected with the highest reliability occurring in the senior resident group andattending spine surgeon group. The lowest reliabilities were in the nonspine attendingorthopedists and junior residents. In each group, the neurologic status was consistently thecategory with the highest interobserver and intraobserver reliability. In the Denis and AOclassifications, the highest reliabilities were again in the senior residents and spineattendings. The lowest were again in the nonspine attendings and junior residents. CONCLUSIONS:With the advent of new surgical technologies, classif cation systems continue todevelop, ultimately embracing the importance of spinal columns and injury mechanism.Each successive system played an important role in advancing contemporaryunderstanding of thoracolumbar injuries. As a management tool, the TLICS seems to be anacceptably reliable system when compared with the Denis and AO systems. The TLICShas good reliability and reproducibility designed to simplify injury classification, evaluateexactly and facilitate treatment decision making of thoracolumbar spinal columninjures．Further study would be benef cial to elucidate the validity of the scoringtechniques used in the TLICS as it relates to patient outcome for def nitive managementof their spine fracture.