Interobserver variability in grading lumbar fusions

Inter- and intraobserver variability in grading lumbar fusion status radiographically was assessed. The objective was to determine the interobserver variability and intraobserver reproducibility in the assessment of two level noninstrumented lumbar fusions. Fifty sets of radiographs with anteroposterior, left and right bending, and flexion-extension lateral views were assessed by six observers of varying experience and background, with fusion status graded. Kappa statistical analysis revealed only fair interobserver agreement in grading lumbar fusion status. Intraobserver reproducibility was higher in more experienced observers. The results indicate only fair reliability in terms of interobserver agreement to grading of lumbar fusion status. Variability in assessing lumbar fusion radiographically may explain some of the variability in fusion rates reported in the literature and poor correlation that can be seen between clinical outcome and radiologic outcome.