The relationship between radiologic union and clinical outcome in thoracoscopic scoliosis surgery is not clear, as apparent non-union does not always correspond to a poor clinical result. Our aim was to evaluate CT fusion rates 2yrs after thoracoscopic surgery, and explore the relationship between fusion scores and; (i) rod diameter, (ii) graft type, (iii) fusion level, (iv) implant failure, and (v) lateral position in disc space.

Between 2000 and 2006 a cohort of 44 patients had thoracoscopic scoliosis correction. Discectomies were performed and defect was packed with either autograft (n=14) or allograft (n=30). Instrumentation consisted of either 4.5mm (n=24) or 5.5mm (n=20) single titanium anterior rod and vertebral body screws. Fusion quality and implant integrity were evaluated 2yr following surgery using low-dose CT. At each disc space, left, right and mid-sagittal CT reconstructions were generated and graded using the Sucato 4-point scale (Sucato, 2004) which is based on calculated percentage of fusion across disc space.

Fusion scores were measured for 259 disc spaces in 44 patients. Rod diameter had a strong effect on fusion score, with a mean score of 2.12±0.74 for 4.5mm Ti rod, decreasing to 1.41+0.55 for 5.5mm Ti rod, and to 1.09+0.36 for 5.5mm Ti-alloy rod. Mean fusion scores for autograft and allograft subgroups were 2.13±0.72 and 2.14±0.74 respectively. Fusion scores were highest in the middle of implant construct, dropping off by 20–30% toward the ends. Fusion scores adjacent to the rod (2.19±0.72) were significantly higher than on the contralateral side of the disc (1.24±0.85). Levels where rod fracture occurred (n=11) had lower fusion scores than those without fracture (1.09±0.67 vs 1.76±0.80). Levels where top screw pullout occurred (n=6) had lower CT fusion scores than those without (1.25±0.60 vs 1.83±0.76).

Rod diameter (larger), intervertebral level (proximal or distal), lateral position in disc (further from rod) and rod fracture or screw pullout all reduce fusion scores, while graft type does not affect scores. The assumed link between higher fusion score and better clinical outcome must be treated with caution, because rod fractures did not necessarily occur in patients with lower fusion scores. It is possible that with a stiffer rod, less bony fusion mass is required for a stable construct. We propose that moderate fusion scores secure successful clinical outcomes in thoracoscopic scoliosis surgery.