Seven publications, reporting six RCTs, were included in the review and meta-analysis (827 patients, range 34 to 300). All of the trials were assessed as having a low risk of bias for randomisation, and five were found to have a low risk of bias for allocation concealment. The results for all other bias domains were variable (breakdown provided in the review).
Pain: In meta-analysis of six trials, statistically significant improvements in pain, favouring vertebroplasty over control, were shown at early assessment (SMD 0.73, 95% CI 0.35 to 1.10) and late assessment (SMD 0.58, 95% CI 0.19 to 0.97). Further results were reported.
Spine-specific functional outcome: In six trials, statistically significant improvements, favouring vertebroplasty over control, were shown at early (SMD 1.08, 95% CI 0.33 to 1.82) and late assessment (SMD 1.16, 95% CI 0.14 to 2.18). Publication bias was evident; the classic fail-safe N was 131 and Orwin's fail-safe N was 29, indicating that missing studies were unlikely to change the conclusions.
Health-related quality of life: In five trials, compared with control, statistically significant improvements were observed with vertebroplasty, at early (SMD 0.39, 95% CI 0.16 to 0.62) and late assessment (SMD 0.33, 95% CI 0.16 to 0.51).
Adverse events: In six trials, no statistically significant difference in the incidence of adverse events was found between groups. No statistically significant difference in the incidence of secondary fractures between two weeks and two years was found.
Sensitivity analyses did not substantially change any of the results. No further evidence of publication bias was found (data not shown).