Academic Facility Utilization and Survival Outcomes in Adult Head and Neck Sarcomas: An NCDB Analysis.

Objectives To investigate clinicopathologic and treatment factors associated with survival in adult head and neck sarcomas in the National Cancer Database (NCDB). To analyze whether treatment settings and therapies received influence survival outcomes and to compare trends in utilization via an aggregated national data set. Study Design Prospectively gathered data. Setting NCDB. Subjects and Methods The study comprised a total of 6944 adult patients treated for a head and neck sarcoma from January 2004 to December 2013. Overall survival (OS) was the primary outcome. Results Increased age and tumor size, nodal involvement, and poorly differentiated histology had significantly reduced OS (P < .001). Angiosarcoma, malignant nerve sheath tumor, malignant fibrous histiocytoma, osteosarcoma, and rhabdomyosarcoma histologic subtypes had significantly reduced OS, while liposarcoma, chondrosarcoma, and chordoma had improved OS (P < .001). Utilization of surgical therapy was associated with improved OS, while positive surgical margins were associated with treatment at a community-based cancer program and had reduced OS (P < .001). On multivariate analysis, treatment with radiation and/or chemotherapy was not significantly associated with OS; however, primary treatment with definitive chemoradiotherapy had significantly reduced OS. Patients treated at academic/research cancer programs (n = 3874) had significantly improved 5- and 10-year OS (65% and 54%, respectively) when compared with patients treated at community-based cancer programs (n = 3027; 49% and 29%; P < .001). The percentage utilization of these programs (56% vs 44%) did not change over the study period. Conclusion For adult head and neck sarcomas, treatment at an academic/research cancer program was associated with improved survival; however, despite increasing medical specialization, the percentage utilization of these programs for this rare tumor remains constant.