Seventeen studies were included (n=1,342 patients): nine RCTs (n=692), five prospective controlled studies (n=357) and three retrospective controlled studies (n=293). Sample size ranged from 14 to 279. In most studies follow-up was 24 months (range 12 to 48 months). In one RCT, the surgeon and patients was blinded to the intervention. In eight RCTs, the assessor of radiographic fusion was blinded. Six studies reported losses to follow-up. Jadad scores ranged from 1 to 3 out of a possible 5.
Radiographic non-union (RCTs only): RhBMP-2 was associated with a statistically significant reduction in risk of nonunion compared to autologous iliac crest bone graft at 12 to 24 months (RR 0.27, 95% CI 0.16 to 0.46, NNT=8, 95% CI 5 to 12; six RCTs). No significant heterogeneity was found. Similar results were found when studies were grouped by type of fusion.
RhBMP-2 plus collagen plus an interbody fusion cage was associated with a statistically significant reduction in risk of nonunion compared to autologous iliac crest bone graft plus a cage in patients who underwent anterior lumbar interbody fusion (RR 0.43, 95% CI 0.19 to 1.00; two RCTs). No significant heterogeneity was found.
Two RCTs that compared RhBMP-2 plus compression resistant matrix plus pedicle screw instrumentation with autologous iliac crest bone graft plus pedicle screw instrumentation in patients who underwent posterolateral lumbar fusion reported a significant reduction in non-union in rhBMP-2 groups; there was significant heterogeneity I2=51%.
There was no significant difference in risk of non-union between rhBMP-7 and autologous iliac crest bone graft or local autograft (three RCTs).
Secondary outcomes (RCTs): RhBMP-2 was associated with a significant reduction in operating time compared to autologous iliac crest bone graft in four out of five RCTs, a significant decrease in blood loss in two out of five RCTs and a significant decrease in length of stay in one out of five RCTs. There was no significant difference between rhBMP-2 and autologous iliac crest bone graft in the likelihood of improvement on the Oswestry Disability Index (four RCTs)
Adverse events: The authors stated that none of the adverse events appeared to be associated with the study intervention. Pain at the graft site was commonly reported for autologous iliac crest bone graft.
The funnel plot for the effect of rhBMP-2 on nonunion showed evidence of publication bias; results for Begg’s and Egger's tests were conflicting.
Results were reported for interventions evaluated in individual studies and non randomised studies.