4 - Select Test Data Resources

Task: Select test data resources

The OMOP Research Core will focus its tools development and methods testing activities against a limited set of observational data sources, which will be chosen for their practical value in efficiently testing methods as well as their potential longer term strategic value for future pharmacovigilance systems such as the Sentinel network. Both administrative claims databases and electronic health records will be represented. Some priority will be given to data sources that allow OMOP researchers to go back to the original source medical records, although that characteristic is more applicable to evaluation (hypothesis testing) than to identification (hypothesis generating). Hypothesis generation, inclusive of both monitoring for pre-specified events (e.g., DMEs/HOIs) and ‘data mining’ for non-specified conditions will not actually require access to source medical records. Deriving or confirming the definitions for the Health Outcomes of Interest, which is a separate exercise may require source record verification should the validation literature fail to be convincing or conclusive.

Choice of which data sources to work with will be constrained by availability and ease of use for research purposes and their specific willingness to work with OMOP. Most government and non-commercial payer/provider databases will only be able to be accessed remotely—i.e., behind each institution’s firewall, with only results data returned to OMOP. Commercial data sources will allow for a more centralized approach. Commercial sources will be evaluated and several chosen primarily on the basis of analyses that do not require access to source medical records, unless one or more of those sources proves capable of supplying such access. Additional data sources may be accessed and analyzed by Extended Research Consortium members using the process described above. Sources will be researched and chosen by the OMOP core with input from and oversight by the OMOP Advisory Boards.

Deliverable: OMOP will establish contracts or MOUs with several core data sources as part of the pilot and will seek to develop data models and data evaluation tools to support those specific data sources at a minimum. The list of data sources will be made public via press releases and the OMOP web site as they are finalized.