Does Bone Morphogenetic Protein Change the Operative Nonunion Rates in Spine Fusions?

Study Design. A retrospective cohort study with medical record review. Objective. To determine if there is difference in the operative nonunion rates with and without the use of bone morphogenetic protein (BMP) after spinal fusions by analyzing data from an integrated health care system's spine registry. Summary of Background Data. BMP was first approved in 2002 for use in single-level anterior lumbar fusions. Follow-up studies have advocated its use in reducing the need for reoperations for nonunions. Recent studies, however, have questioned these conclusions and the usefulness of BMP in spinal fusions has been highly debated. Methods. We identified 9425 spinal fusion cases between 2009 and 2011 from a spine registry in a large integrated health care organization. Patient characteristics, diagnosis, operative times, length of stay, and reoperations were extracted from the registry. Reoperations for nonunions were adjudicated via medical record review. Cox regression models were used to evaluate the risk of reoperation while adjusting for confounders. Results. In our cohort, there were 5456 BMP cases and 3969 non-BMP cases. The mean age was 60.4 years (standard deviation: 12.9 yr), with the majority being females (53%). The median follow-up time was 1.2 years (interquartile range: 0.6-2.0 yr). Reoperation rates for BMP versus non-BMP nonunions for all fusion cases with follow-up of 1 year or more (1.9% vs. 2.2%) and follow-up of 2 years or more (2.3% vs. 2.6%) were not statistically significantly different. Operative nonunion rates did not reach statistical significance for different spine regions and for different fused columns (anterior only, posterior only, or combined). After controlling for differences in patient characteristics, operative times, levels fused, and spinal regions, the risk of reoperation in the BMP versus non-BMP groups was 0.67 (95% CI: 0.42-1.06). Conclusion. In this large cohort of spinal fusions at all spine regions involving all fused columns with and without BMP, we found no statistically significant difference in operative nonunion rates.