CERVICAL DISC REPLACEMENT: THE RELIABILITY OF RADIOLOGICAL MEASURES

Abstract

Introduction Cervical disc replacement has been advocated as a technique for retaining segmental motion following discectomy. The in vivo kinematics of the prosthesis are predicted to influence outcome, but remain poorly understood. The aim of this study was to determine whether factors that might possibly influence outcome can be reliably measured.

Methods Anteroposterior and lateral xrays were performed pre-, inter- and post-operatively on 67 patients undergoing cervical prosthetic disc replacement. The xrays were reviewed by 3 observers. Measurements were performed manually by two observers and using digital image analysis software by a third observer.

The inter-rater reliability of categorical measurements such as overall cervical alignment, degree of disc degeneration and length of spinous processes was assessed using unweighted kappa scores. Intra-class correlation coefficients (ICCs) were calculated using a two-way random effects model to assess inter-rater agreement in the observation of continuous variables such as intra-operative disc angles, post-operative shell angles and change in focal lordosis. The intra-rater reliability of measurements of disc space angulation was calculated on a subset of 17 sets of xrays measured by three observers on two occassions, five months apart.

Kappa and ICC values were interpreted as recommended by Altman.

Results The inter-rater reliability of measuring the degree of disc degeneration (κ= 0.29 p< 0.0001) was fair and the spinous process length (κ = 0.79 p< 0.0001) and overall spinal alignment (κ = 0.69 p< 0.0001) good.

The inter-rater reliability of measuring pre-operative focal lordosis (ICC 0.88 (95%CI 0.82–0.92 p< 0.0001)), intra-operative disc angle (ICC 0.86 (95%CI 0.79–0.92) p< 0.0001) and post-operative shell angle (ICC 0.99 (95%CI 0.98–1.00) p< 0.0001) were excellent.

ICCs were higher when the average of the rater scores was considered. The ICCs were substantially reduced when agreement between the observers and values obtained using digital imaging was assessed.

The intra-rater reliability of measurements of focal lordosis however revealed good agreement when measured manually (ICC 0.68 (95%CI 0.06–0.89) p=0.02) but very good agreement when measured using digital imaging software (ICC 0.82 (95%CI 0.54–0.93) p< 0.0001). The inter-rater reliability of average disc space height when measured using digital imaging software was excellent (ICC 0.83 (95%CI 0.58–0.94) p< 0.0001).

Conclusions Whilst the reliability of determining the degree of disc degeneration in the cervical spine is fair, the measurement of focal lordosis, intra-operative disc angles and post-operative shell angle is good or excellent. As these radiological measures can be reliably assessed, they may be further evaluated as predictors of outcome following cervical disc replacement.