Four trials (n=634 patients) were included in the review. Sample sizes ranged from 61 to 289 patients. Three trials were used for the primary analysis and one was included solely for sensitivity analysis. Study quality was good for treatment allocation, randomisation and use of intention-to-treat analysis. Patient/caregiver blinding was not possible due to the nature of the intervention. Blinding of outcomes was poor due to the use of patient administered questionnaires. Loss to follow-up ranged from 2% to 19%.
Surgery led to a slight improvement in ODI score (mean difference 4.13, 95% CI -0.82 to 9.08, p=0.10); there was moderate heterogeneity (I2=44.4%). The change in ODI using the study with imputed data was minimal, and remained non significant. Pooled analysis showed that the early complication rate associated with surgery was 16% (95% CI 12 to 20); no heterogeneity was reported. The inclusion of the study focused on patients with isthmic spondylolisthesis reduced the overall complication rate to 13% (95% CI 6 to 20); there was substantial heterogeneity (I2 = 66.9%).
Further analysis including the study focused on patients with isthmic spondylolisthesis (using available or imputed results of another trial), showed a statistically significant reduced mean difference in ODI of 3.9 (95% CI 0.17 to 7.62, p=0.04, I2=21.4%).