FDA funds chart validation activities to support development of high-performing claims-based algorithms to improve identification of health outcomes of interest (HOIs) for use in Sentinel’s active post market risk identification and analysis system (ARIA). Adjudicated information from medical charts remains the gold standard for validating claims-based algorithms. Although medical chart review is resource intensive and cannot be done for every outcome or every study, it adds rich individual level detail to summary data provided by routine queries. It enables investigators to capture validated information that is otherwise difficult or impossible to ascertain in large claims-based databases alone. Therefore, identifying ways to preserve the results of prior validation studies extends the benefits of those studies beyond the original study results and potentially enables the development and testing of predictive analytics.
This project seeks to add a new table to the Sentinel Common Data Model (SCDM) to capture results from chart validation studies and add data elements to allow for the preservation of medical records that were digitally translated using optical character recognition (OCR) technology.
The SCDM changes enable capture of variables including the algorithm used, outcome or clinical condition of interest, case status, and case data source.
The new SCDM table enables Sentinel to capture vital information about case status and enables investigators to preserve and traceback digitalized health records in a privacy preserving manner where Data Partners have full operational control over the health information.
Judith C. Maro, PhD, MS; Department of Population Medicine, Harvard Pilgrim Health Care Institute and Harvard Medical School, Boston, MA
Suzanne Carter, PhD, MS, MBA; Katherine Chiu; Robert Jin, MS; Adee Kennedy, MS, MPH; Jennifer Lyons, PhD; Meighan Rogers Driscoll, MPH; Justin Vigeant; Department of Population Medicine, Harvard Pilgrim Health Care Institute and Harvard Medical School, Boston, MA