Elsevier

Injury

Volume 43, Issue 4, April 2012, Pages 416-422
Injury

Inter- and intraobserver agreement on the Load Sharing Classification of thoracolumbar spine fractures

https://doi.org/10.1016/j.injury.2011.05.013Get rights and content

Abstract

The Load Sharing Classification (LSC) allocates one to three points to each of three different radiological characteristics of traumatic thoracolumbar fractures: the vertebral body involved in the fracture, the displacement of the fracture parts and the kyphotic deformity. Added up, a minimal score of three and a maximal score of nine can be obtained. When the LSC score is three to six, a short segment pedicle screw fixation suffices. When the LSC score is seven to nine, a high rate of failure in patients with a short segment pedicle screw fixation exists. In these cases an anterior stabilising procedure of the spine is advised.

The LSC has been examined by Dai and Jin, who claim an almost perfect inter- and intraobserver agreement, according to the Landis and Koch criteria. Dai and Jin only present results for the separate three items of the LSC and for the total LSC scores. Observer agreement for the two LSC score categories (three to six and seven to nine) have not been studied.

The aim of this study is to study the inter- and intraobserver agreement of the LSC for the total score, the three separate items and also for the two LSC score categories. Three observers determine twice the LSC scores of forty traumatic thoracolumbar fractures. The average standard Cohen's kappa values for the separate LSC items range between 0.06 and 0.48. For the total LSC score the average standard Cohen's kappa and weighted kappa values are 0.22 and 0.67 respectively. For the two LSC score categories, there is unanimous agreement in 55% of the cases and a majority agreement in 40%. In the remaining 5% of the fractures there is a split decision. Standard Cohen's kappa value for the two LSC score categories is 0.53. The standard Cohen's kappa values can be rated as fair to moderate. From these data it can be concluded that the inter- and intraobserver reliability of the Load Sharing Classification of Spinal fractures can be rated as fair.

Introduction

Since the introduction of transpedicular screws by Roy-Camille, short-segment posterior stabilization has become common practice in the operative treatment of traumatic vertebral fracture surgery.26 The possibilities for reduction of segmental kyphotic deformities are good and the posterior approach carries a low morbidity. However, many reports have revealed failures of short-segment pedicle screw instrumentation due to the lack of anterior support or rotational instability.22, 25, 28, 32, 34 These failures are non-fusion, loss of reduction of kyphotic deformities, pedicle screw migration or screw breakage.22, 23, 27, 31 Anterior stabilization offers superior anterior support and the loss of reduction of kyphotic deformities appears to be smaller than in short-segment posterior stabilization only.10, 15, 16 Disadvantages of this approach are a higher morbidity due to the more extensive, open approach i.e., thoracotomy, lumbotomy or thoraco-phrenico-lumbotomy at the thoracolumbar junction, and intraoperative reduction is more difficult. Although the recent application of thoracoscopic surgery has decreased this morbidity significantly.16

To support decisions about the best surgical approach for unstable thoracolumbar fractures McCormack et al. introduced the Load Sharing Classification (LSC) of spine fractures.21 This classification allocates one to three points to each of three different radiological characteristics of traumatic thoracolumbar fractures: the amount of comminution/involvement of the vertebral body, the amount of apposition/displacement of the fracture fragments and the amount of correction of kyphotic deformity. In this way, a minimal total LSC score of three and a maximal score of nine can be obtained. Using this system, McCormack et al. point out that spine fractures with low LSC scores (three to six points) can be managed with short-segment posterior stabilization only, whereas burst fractures with high LSC scores (seven to nine points) require anterior stabilization to prevent failure of the posterior implant.1, 13, 21, 24

In the retrospective study of McCormack et al. it was stated that the LSC can be used preoperatively to: (1) describe any spinal injury for retrospective studies, (2) predict pedicle screw failures in short segment posterior stabilizations, and (3) select spinal fractures for anterior reconstruction in fractures with LSC scores of seven and higher.1, 21, 24 This classification cannot be used to make decisions on surgical indications as it takes no account of ligamentous disruptions or the mechanisms of injury. Moreover, the LSC has been developed specifically to be applied preoperatively, after the indication for surgical treatment has been made, for an optimal assessment of the injured vertebral body and the subsequent optimal approach.1, 21, 24

For any classification to be useful, consistency between different observers is essential. The reliability of the LSC has been examined by Dai and Jin who report an almost perfect inter- and intraobserver agreement, according to the Landis and Koch criteria.11 This is remarkable as many other inter- and intraobserver studies on other fracture classifications can only claim a fair to moderate observer agreement.4, 8, 12, 14, 29 Furthermore, Dai and Jin only present results for the three separate LSC items and the total LSC scores, whereas the two LSC score categories, i.e., three to six and seven to nine, are not studied. Consequently Dai and Jin present no data on the reliability to use the LSC to discriminate between an anterior or posterior stabilising procedure of the spine, the discerning feature of the LSC.

The purpose of the present study is to determine intra- and interobserver agreement for the total LSC score, for the scores of each of the three radiological fracture characteristics, as well as for the two score categories, and to compare the findings, with the values reported in literature.11

Section snippets

Materials and methods

The LSC of spine fractures uses three radiological characteristics of the fracture and allocates one to three points to each of these items (Fig. 1).

Comminution/involvement (item A): according to McCormack et al., the comminution is assessed using the 2D sagittal CT-scan reconstruction images.21 Spine fractures with a comminution/involvement up to 30% of the vertebral body are allocated one point, two points are given for a comminution/involvement up to 60% and three points when the vertebra is

Inter- and intraobserver agreement on the scores of three separate items of the Load Sharing Classification

The standard Cohen's kappa and weighted kappa scores per item for three pairs of observers are presented in Table 2.

The standard Cohen's kappa values for the interobserver agreement of comminution/involvement (item A) range from 0.23 to 0.73 with an average value of 0.48; weighted kappa values for the interobserver agreement of item A: 0.47–0.85, average value 0.58. The standard Cohen's kappa values for the interobserver agreement of apposition/displacement (item B) range from 0.15 to 0.52 with

Discussion

An important aim of a fracture classification system is to assist in choosing the most appropriate treatment. Given sufficient reliability, such systems also enable surgeons to compare treatment outcomes of specific groups of patients. Employing classification systems before their reliability is determined can lead to conflicting and non-consistent clinical conclusions.9 An assessment of the reproducibility must take inter- and intraobserver variability into account.

Since LSC scores may vary

References (34)

  • A.H. Burstein

    Fracture classification systems: do they work and are they useful?

    J Bone Joint Surg Am

    (1993)
  • A.L. Carl et al.

    Pedicle screw instrumentation for thoracolumbar burst fractures and fracture–dislocations

    Spine

    (1992)
  • L.Y. Dai et al.

    Interobserver and intraobserver reliability in the load sharing classification of the assessment of thoracolumbar burst fractures

    Spine

    (2005)
  • T. Flikkila et al.

    Poor interobserver reliability of AO classification of fractures of the distal radius. Additional computed tomography is of minor value

    J Bone Joint Surg Br

    (1998)
  • R.W. Gaines et al.

    Experimental evaluation of seven different spinal fracture internal fixation devices using nonfailure stability testing. The load-sharing and unstable-mechanism concepts

    Spine

    (1991)
  • P.M. Gehrchen et al.

    Seinsheimer's classification of subtrochanteric fractures. Poor reproducibility of 4 observers’ evaluation of 50 cases

    Acta Orthop Scand

    (1997)
  • K. Kaneda et al.

    Anterior decompression and stabilization with the Kaneda device for thoracolumbar burst fractures associated with neurological deficits

    J Bone Joint Surg Am

    (1997)
  • Cited by (0)

    View full text