Cancer claims and registry data: A California linkage for quality measurement.

Abstract

296 Background: Breast and colon are 2 of the most prevalent cancers in California according to the California Cancer Registry (CCR). Health claims data contains a wealth of information on diagnoses and treatment but lacks clinical information critical for evaluating care. Registry data contains key clinical information, but often does not contain the complete treatment regimen. This project aims to assess the feasibility of linking commercial claims data to population-based cancer registry data, and to use the linked data to examine variation in quality measures at the regional and physician organization (PO) levels. Methods: Nine NQF endorsed breast and colon cancer measures were selected and measure specifications with code sets were developed. Data on members identified with a diagnosis of breast and/or colon cancer in claims (2009-2012) from 4 commercial HMO health plans was linked with CCR data. Results were generated at the regional and PO levels. Results: The feasibility test was a success; CCR and claims data was linked for 8,757 individuals, and the data sets proved complementary. Rates relying on the linked dataset were typically higher than when either data source was used alone. For example, see the Table. Performance was strong across 8 of 9 measures with scores ranging from 80-98%. The 9th measure, which relied only on CCR data, was 51%. PO level measurement was limited by small sample sizes, but where sample size was adequate, a significant amount of variation in performance across POs was found. Conclusions: The project showed that linkage of claims and registry data is feasible and that the linked dataset supports more robust assessment of the quality of cancer care. Further, the data dictionary and programmable code sets that were developed are available to other entities interested in creating linkages. With larger sample size, robust benchmarks could be established. Linked registry and claims data could be used for additional research in areas such as compliance with clinical guidelines and examining patterns of treatment. [Table: see text]