Augmenting state cancer registry data for quality-of-care assessment: A Georgia-based application to evaluate receipt of adjuvant therapies for breast cancer and colorectal cancer.

Abstract

6523 Background: The value of linking population-based cancer registry data with insurance claims files to assess quality of care has been demonstrated in numerous studies, including those using NCI’s linked SEER-Medicare database, covering patients age 65+ in fee-for-service plans, and studies linking registry data with Medicaid, private insurance, or managed care data covering the under-65 population. We describe a prototype program linking registry data with multiple data sources to assess quality of care for at-risk populations in a defined geographical area. Methods: Data exchange agreements were executed among the investigative site (Emory University), Georgia state government, and the claims data sources/vendors. We linked Georgia Cancer Registry (GCR) records for 1999-2005 incident cases of breast and colorectal cancer with enrollment and medical services records from Medicare, Medicaid, Kaiser Permanente of Georgia, and the State Health Benefit Plan (SHBP) which covers all state workers and dependents. Following data quality checks, algorithms based on National Quality Forum (NQF) endorsed breast and colorectal cancer quality measures were applied to each linked data set to assess performance. Results: The linked data sets included 60% of all breast and colorectal cancer cases in the GCR over the study period. Quality measure performance rates varied notably across payers. For example, the percent of Stage III colon cancer patients meeting the NQF standard for adjuvant chemotherapy in the linked GCR-Medicaid, GCR-Kaiser, and GCR-SHBP data were, respectively, 75%, 92%, and 92% (p<0.05). The rates for breast cancer patients meeting standards for adjuvant chemotherapy were 86%, 84%, and 87% (p=NS), respectively. Patients in the linked GCR-Medicare data (all age 65+) generally had lower performance rates for each NQF measure. Conclusions: Linking state cancer registry data with multiple public and private sources of administrative data is technically feasible, and may represent a viable strategy for building a national cancer data system for quality improvement, as recommended in 1999 by the Institute of Medicine.