BACKGROUND: There are many challenges to conducting traditional post-marketing surveillance studies for rare cancer outcomes including precise case identification; exposure assessment; longitudinal follow-up; and sufficient study size. Using two large pharmacy claims databases to identify exposure and linking to cancer registries to determine outcome is one novel approach currently being implemented in an ongoing safety surveillance program.
OBJECTIVE: To describe methods using two pharmacy claims databases linked to state cancer registries to assess the incidence of osteosarcoma, a rare bone cancer, among patients taking teriparatide and a comparison group.
METHODS/APPROACH: Two population-based pharmacy claims databases, a Federal health insurance program and a commercial outpatient pharmacy database, are used to identify drug exposure and assemble the appropriate study cohorts. Cancer diagnosis information is obtained by linking the cohorts with state cancer registry databases who use the International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3) codes. Incidence of cancer among the exposed and unexposed cohorts will be estimated and compared.
RESULTS: Approximately 30 different state cancer registries initially agreed to participate in these linkage studies. The process for conducting the linkages between two different pharmacy claims databases and individual state cancer registries must be tailored to meet data privacy and other local requirements for all data sources. For one pharmacy claims database, a deterministic linkage using a direct identifier will be implemented either through a trusted third party or by the individual registry. For the other pharmacy claims database, linkage will be conducted using deidentification technology.
CONCLUSIONS/IMPLICATIONS: Using large pharmacy claims databases to obtain drug exposure information and linking with cancer registries to determine cancer outcomes minimizes the possibility of misclassification of the tumor type and has the potential to improve monitoring for patient safety without imposing burden on the patient.
