Abstract
Background/Aims Radiation oncology is an integral component of high-quality cancer care, in both curative and palliative settings. Nationally, concern exists regarding access, quality, and diffusion of radiation oncology technologies, especially newer high-cost modalities. This study is exploring capture and quality of radiation oncology data at three CRN sites—GHC, KPCO, and KPNW. Our aims are to identify the magnitude and sources of data leaks and to expand capture of variables related to radiation therapy, (e.g. anatomic focus, radiation type, numbers and duration of sessions, and total radiation dose). Our ultimate goal is to conduct comparative effectiveness research using the full spectrum of cancer treatments.
Methods The current sites also participated in Burden of Cancer Care, an NCI-funded R01 study (PI: Mark C. Hornbrook) of cancer cases and matched, non-cancer controls. A comprehensive list of radiation oncology-related codes were developed using interviews of local billing experts, project lists, Internet searches, and review of CPT-4/HCPCC/ICD-9-CM code books. Radiation oncologists and business managers were interviewed to develop diffusion timelines and identify new data sources. A SAS program was written and distributed for use in Burden datasets to compare encounter- and claims-based evidence of radiation receipt among enrollees between 2000–2008, in the presence or absence of cancer diagnosis by tumor registry and/or claims/encounter data.
Results While use of radiation codes differ (e.g., facility revenue codes), the proportion of patients receiving radiation was similar across sites (N=27,210 of 627,682 cases and controls). About 85% (N=23,222) of patients receiving radiation therapies were identified as having cancer via tumor registry and encounter-/claims-based sources, with the remaining 15% identified by encounter or claims only. Of those identified by both methods, 99% (N=22,984) received radiation after cancer diagnosis. Overall 4.5% (N=1,228) of radiation recipients did not have a cancer diagnosis in the study period. Of those with data (N=985), the majority had either a cancer diagnosis outside the study period (23%, N=231) or a benign tumor (77%, N=754). Additional treatment data exists in network-based radiation departments and efforts are ongoing to integrate these data at two sites.
Discussion Further work is necessary to ensure radiation oncology data are comprehensive and complete across CRN sites.
