LARGE-SCALE DATA ANALYSIS TO CHARACTERIZE VARIATIONS IN CANCER TREATMENTS ACROSS THE UNITED STATES

Data collected in the course of clinical care in electronic health records (EHRs) and in administrative claims databases may be used to assess the process and outcomes of care. Observational Health Data Sciences and Informatics (OHDSI) is a large-scale open-science initiative to generate evidence from an international network of data sets by a community of scientists. With over 300 formal member researchers and 3000 participants on its forums, OHDSI covers all aspects of evidence generation including a common data model known as the Observational Medical Outcomes Partnership (OMOP) Common Data Model, a comprehensive clinical vocabulary, software to aid data conversion and quality assurance, innovative statistical methods in causal inference along with highly efficient implementations, user interfaces to help researchers convert data and run studies, tools for visualizing analytic results, and clinical expertise to ask and answer important clinical questions. OHDSI has previously conducted a study on treatment pathways in cancer patients with three common chronic diseases: hypertension, depression, and type-2 diabetes. The study found that in cancer patients, similar to non-cancer patients, there is greatest agreement in first-line treatment for diabetes and, to a much lesser extent, treatment for hypertension. There is greater variability in second-line treatment for diabetes and for all treatments for depression.