1st Annual EUBIROD meeting Dasman Center for Research and Treatment of Diabetes, Kuwait City Kuwait City, Kuwait, 2nd-4th May 2009

Standardized Diabetes Data Definitions

Scott Cunningham, U.Dundee, Scotland, UK

First BIRO Academy Residential Course, Kuwait City, Kuwait, 2nd May 2009

Slides

Objectives of standardized diabetes data definitons were to define the BIRO Common Dataset and to explain ‘data about the data’ through a data dictionary and annexed series of XML documentation that would be widely available to authenticate the contributing information.

In his speech, Scott Cunningham, a technical consultant from the University of Dundee, Scotland, UK, reports his direct experience in the BIRO project as leader of the Common Dataset and Data Dictionary workpackages. He reports how an initial review of the clinical datasets used by all partner institutions was seen as fundamental to understand the level of consistency in terms of recording, data definition and units of measurement across participating centers. Based upon these characteristics, high priority items were selected, and, where necessary, data mappings and algorithms were specified in order to allow partners to meet the standard definitions. Data was then assigned a “validity” weight (high, medium or low) describing how well it could meet the agreed definitions. Based on such key requirements, a series of descriptive elements were created to document the ‘metadata’ for each of the records gathered. The first set of metadata describe the standardised elements, documenting units of measurement, data type, range values, definition, validity, mapping criteria etc. The second set of metadata must be provided by each partner in order to document how consistent their local systems are with the standard definition. Information gathered at this stage includes consistency with definition, completeness of local data and an overall quality score.

In the presentation, Scott describes the common dataset, including 82 data items, of which 53 describe clinical data e.g. clinical and process outcomes, 21 were created to specify information regarding the local clinical site and 8 to include geographical references. The classification allowed to gain a detailed understanding of clinical processes and local geography. These mappings and standardised definitions have been used to create an electronic directory for diabetes care, providing the foundation for the BIRO data repository. Clinical and non-clinical information has been gathered using a standardised XML data structure containing supplementary comments required to fully document the data analyses.

Scott Cunningham highlights how the development of data dictionaries and data standards can be used to improve the quality, relevance, consistency and comparability of national information about health. In BIRO, a European minimum common dataset and data dictionary have been developed to allow wide applicability in diabetes registers. The process will continue as the project expands for the EUBIROD project.