Measurement of mutation-specific survival as a real-time cancer care quality indicator.

30 Background: As targeted therapies become widespread in the treatment of cancer, tracking clinical effectiveness as a function of mutation status, outside of the research setting, is needed. We evaluated whether mutation-specific survival statistics could be derived in a completely automated fashion from electronic medical record data sources, for quality assurance and purposes preparatory to research. Methods: Patients with cancer mutation analysis obtained for clinical care at Vanderbilt University up to June 2013 were included. Informatics algorithms were developed to automatically extract tumor mutation status, cancer type, date of diagnosis, and date of death or date of last contact using multiple structured and unstructured clinical data sources including billing codes, narrative pathology reports, and the Social Security Death Index. Survival was modeled using Cox proportional hazards, stratified by mutation type; mutations occurring less than 10 times were aggregated into an “Other” category. Re...