Study Design: Prospective randomized study.

Objectives: To compare the strategy of spinal fusion with that of rehabilitation for patients with chronic low back pain.

Methods: A multicentre trial of 349 candidates for spinal fusion (where both patient and surgeon were uncertain of the outcome) were randomised to either an operation that the surgeon considered was most appropriate for that patient or to an intensive rehabilitation programme. Rehabilitation was based on a 3 week (15 day) model of exercise therapy, spine stabilisation techniques and education using cognitive behavioural principles. Follow-up was at least 2 years from randomisation. The trial was 90% powered to show a 4 point difference between groups at α= 0.05. A full economic analysis is available.

Outcomes Measures: Oswestry Disability Index (ODI); the Shuttle Walking Test (SWT); SF-36 and EuroQol EQ-5D.

Results: 176 patients were randomized to surgery and 173 to rehabilitation. Demographic features including sex, age, diagnosis (spondylolisthesis, post-laminectomy syndrome, others) duration of back pain, smoking history, litigation, employment status, planned numbers of fused levels and baseline ODI were similar for both groups.

Results: For the surgery group the mean ODI improved from 46.6 (SD 14.5) to 34.9 (SD 21) at two years. For the rehabilitation group mean ODI improved from 44.8 (SD 14.8) to 36.2 (SD 20.6) at two years. For the surgery group the mean SWT improved from 254 (SD 209) to 350 (SD 244.8) at two years. For the rehabilitation group mean SWT improved from 247 (SD 185) to 310 (SD 203) at two years. For the surgery group the mean SF-36 Physical component score improved from 22.2 (SD 18) to 43.6 (SD 32.1) at two years. For the rehabilitation group the mean SF-36 Physical component score improved from 24.0 (SD 20.6) to 40.5 (SD 31.1) at two years.

Conclusions: This is a comparison of treatment strategies: There was no clinical or statistical difference in outcome between the strategy of spinal fusion and that of rehabilitation. Patients randomised to surgery (spinal stabilisation) and patients randomised to rehabilitation have indicated a treatment effect, but this may be due to natural history. The surgery results parallel those reported in other trials. At two years the treatment costs of the surgery arm were approximately twice those of the rehabilitation arm. The costs of rehabilitation depend on how many patients opt for surgery (22% in this trial). “Failed” non-operative treatment is commonly listed as an indication for surgery. “Failed” non-operative treatment should include intensive rehabilitation appropriately supported by the treating surgeon.