Observational Medical Outcomes Partnership

Task: Develop a process for evaluating data quality

It is well understood that analyses on any data source can only be informative if the underlying data is of sufficient quality to inform the question of interest. As one of its primary deliverables, OMOP will develop a structured process for assessing the characteristics of an observational database to determine its suitability for use in the identification and evaluation. This process will include an inventory of available data elements, a descriptive summary of occurrence of drugs and conditions, and an evaluation of the source’s ability to verify clinical observations through medical records or other means. OMOP will fund a person/organization to develop:

i. checks for completeness, accuracy, timeliness of the data files
ii. standard metrics – number of individuals covered overall and in specific time periods, types of data included, numbers of individuals with selected diagnoses, procedures, drug exposures.
iii. programs that perform these checks and create the measures.

This activity will allow uniform, routine, checking of any dataset that conforms to the common data model.

Deliverable: The Research Core will produce a guidance document that outlines a data quality evaluation process, along with the findings from applying the guidance to each of the Research Core databases. In addition, OMOP may develop specific diagnostic software programs that can be used to automate the readiness assessment of a variety of databases for identification and evaluation analyses. OMOP will put the guidance document and appropriately updated versions of any software in the public domain as early as feasible during the course of the pilot.