Primary Central Nervous System Tumors: Comparing Two National Cancer Registries

Abstract

The National Cancer Database (NCDB) and the SEER (Surveillance Epidemiology and End Results) program are the 2 largest cancer registries in the United States. However, considerable differences exist between them regarding the sampling frame as well as the participating facility characteristics. In this study, NCDB and SEER are compared for primary central nervous system (CNS) tumors with the aims of discussing the implications for researchers and evaluating the generalizability of both databases. Patients with primary CNS (brain and spinal cord) tumors from 2004 to 2013 were identified using the International Classification of Diseases for Oncology, Third Edition topographic codes. Patient demographics, tumor characteristics, and treatment-related variables were compared between the 2 databases using standardized differences, with values >0.1 denoting statistical significance. A total of 588,534 primary brain tumor (NCDB, 416,162; SEER, 172,372) and 34,827 primary spinal cord tumor (NCDB, 26,602; SEER, 8225) cases were identified. We found significant differences regarding age (sd. diff., 0.16 for brain; 0.13 for spinal cord), and Hispanic origin (sd. diff., 0.34 for brain; 0.61 for spinal cord). For NCDB and SEER patients, estimated 5-year overall survival was 59% and 64%, respectively. Analysis of 623,361 patients with primary CNS tumors, which are identified using both the NCDB and SEER databases, showed significant differences in age, histopathologic classification of tumors, tumor behavior, and treatment of tumors between 2 databases. Overall, the differences observed between 2 databases provide helpful points for the researchers who would like to use NCDB or SEER. These observations should be taken into account when researchers design studies using these databases and discuss the generalizability of their findings.