Background context: Current imaging techniques used to evaluate fusion status after a posterolateral fusion such as radiographs, computed axial tomography (CT) scans, and tomograms are known to be inaccurate, with error rates estimated from 20% to 40%. Previous studies evaluated CT scans using 2-4-mm thick slices with limited reconstructions.
Purpose: The purpose of this study is to determine the intraobserver and interobserver agreement of plain radiographs and fine-cut (1-mm) CT scans with sagittal and coronal reconstructions in evaluating fusion status after instrumented posterolateral fusions. The correlation between radiographic evaluations and CT scan evaluations was also analyzed.
Study design/setting: Cross-sectional, blinded.
Patient sample: One-year radiographs and CT scans of 86 patients who had single-level instrumented posterolateral fusions.
Outcome measures: Fusion grades based on previously published criteria were determined.
Methods: Three spine surgeons graded the fusions of 86 patients who had single-level instrumented posterolateral fusions using 1-year postoperative flexion/extension lateral and anteroposterior radiographs, and fine-cut CT scans with sagittal and coronal reconstructions. The technique used to obtain the radiographs and the CT scans was the same in all cases. Two separate readings, 2 weeks apart, were done on each patient by each surgeon. The kappa coefficients for interobserver and intraobserver variability were determined.
Results: The intraobserver agreement using CT scans to assess fusion status was moderate for both classification systems (Molinari=0.48, Glassman 0.47). The intraobserver agreement using X-rays to assess fusion status was fair for the Molinari classification (kappa=0.37) and moderate for the Glassman classification (kappa=0.43). The interobserver agreement using CT scans to assess fusion status was moderate for both classification systems (Molinari=0.48, Glassman 0.48). The interobserver agreement using X-rays to assess fusion status was fair for both classification systems (Molinari=0.24, Glassman 0.26). Observers agreed most often when the fusion was assessed as solid (Molinari k=0.61, Glassman k=0.63). The rating on the radiographs and CT scans agreed only 46% to 59% of the time.
Conclusions: Fine-cut CT scans with reconstructions have a considerably greater degree of interobserver and intraobserver agreement compared with flexion/extension and anteroposterior radiographs. Observers agree most often when the fusion is assessed as solid. Fusion evaluation based on radiographs agrees with CT scans only half the time. Future studies are needed to correlate the findings on fine-cut CT scans with surgical exploration.