Purpose: Persistent lower back pain after instrumental posterolateral desis may arise from incomplete fusion. We investigate the impact of experience on interobserver agreement in fusion estimation.
Methods: Four independent observers, two residents and two musculoskeletal radiologists, reviewed dedicated lumbar 64-MDCT scans and scored vertebral levels 1-5 after Glassman's grades, 1: solid bilateral fusion, 2: solid unilateral fusion, 3: partial bilateral fusion, 4: partial unilateral fusion, 5: non-fusion. We investigated two simplifying dichotomizations, solid bilateral fusion (Glassman 1) versus all others and uni- or bilateral fusion (Glassman 1-2) versus partial or non-fusion.
Results: Thirty-six patients with 61 operated lumbar levels were included. Interobserver agreement rates for four observers using Glassman's system were fair (kappa 0.32), either dichotomization showed moderate agreement (kappa 0.53 and 0.59). Observer pairs had comparable prevalence adjusted interobserver agreement rates (residents: PABAK 0.67 and 0.54; consultants: PABAK 0.57 and 0.71).
Conclusions: Difference in observer experience seems of minor impact.