IGF::OT::IGF EVALUATING FEASIBILITY OF AND POTENTIAL BIASES IN SUPPLEMENTING CANCER REGISTRY DATA USING EXTERNAL SOURCES

The proposed study is to examine how data linkage with claims based data, primarily Medicaid and private insurance data, will improve the comorbidity and treatment variables in the Kentucky Cancer Registry (KCR) data, and evaluate the accuracy and biases of data linkage through probabilistic linkage, and compare differences between statistical estimate in statistical analyses based on the original KCR data and the enhanced KCR data. Objectives: 1) Examine the probabilistic data linkage process and how cutoff values will introduce biases in identifying true matches. 2) Examine how linking with Medicaid data improves the registry data for variables such as comorbidity and treatment information. 3) Examine how linking with private insurance claims data, such as Humana, Anthem and state employee insurance data, improves the registry data for variables such as comorbidity and treatment information. 4) Examine how combinations of Medicare, Medicaid, and private insurance claims data improves the registry data for comorbidity, treatment information and certain quality of care measures. 5) Compare statistics estimates in modeling statistical analyses between the original registry data and augmented registry data, such as logistic regression models and Cox regression survival models.

Clin-STAR Database