Comparison of Fusion Rates Based on Graft Material Following Occipitocervical and Atlantoaxial Arthrodesis in Adults and Children.

Background: Fusion rates following rigid internal instrumentation for occipitocervical and atlantoaxial instability approach 100% in many reports. Based on this success and the morbidity that can be associated with obtaining autograft for fusion, surgeons increasingly select alternative graft materials.

Objective: To examine fusion failure using various graft materials in a retrospective observational study.

Methods: Insurance claims databases (Truven Health MarketScan® [Truven Health Analytics, Ann Arbor, Michigan] and IMS Health Lifelink/PHARMetrics [IMS Health, Danbury, Connecticut]) were used to identify patients with CPT codes 22590 and 22595. Patients were divided by age (≥18 yr = adult) and arthrodesis code, establishing 4 populations. Each population was further separated by graft code: group 1 = 20938 (structural autograft); group 2 = 20931 (structural allograft); group 3 = other graft code (nonstructural); group 4 = no graft code. Fusion failure was assigned when ≥1 predetermined codes presented in the record ≥90 d following the last surgical procedure.

Results: Of 522 patients identified, 419 were adult and 103 were pediatric. Fusion failure occurred in 10.9% (57/522) of the population. There was no statistically significant difference in fusion failure based on graft material. Fusion failure occurred in 18.9% of pediatric occipitocervical fusions, but in 9.2% to 11.1% in the other groups.

Conclusion: Administrative data regarding patients who underwent instrumented occipitocervical or atlantoaxial arthrodesis do not demonstrate differences in fusion rates based on the graft material selected. When compared to many contemporary primary datasets, fusion failure was more frequent; however, several recent studies have shown higher failure rates than previously reported. This may be influenced by broad patient selection and fusion failure criteria that were selected in order to maximize the generalizability of the findings.