Adoption of Reference Datasets

Data collection practices have been explicitly defined by National Service Framework (NSF) for Diabetes. We have used available domain dataset standards; in particular, the Diabetes Continuing Care Reference (DCCR) Dataset, to harmonise the multiple types of data stored in our system.

Datasets like the DCCR are comprised of structured lists of individual data items, each with a clear label, definition and set of permissible values, codes and classifications.

Adoption of a single reference dataset definition has been a cornerstone for producing a single source of data definitions produced by different system inputs (medical sensor devices, electronic form-based history pick-lists, manually entered data, etc.). These definitions provide a means to metadata tag and aggregate data outputs. From this, secondary uses of information can be derived or compiled or both in support of specific health and social care purposes.

Data collected and integrated within the overall structure of a standardised dataset context provides the starting point for building an individualised monitoring capability and risk analysis framework.

Datasets like DCCR are also important for building consistent end-user combinatorial queries (i.e., ‘declarative’ queries that are scriptable). The choice of the DCCR still does not solve inconsistencies in data definitions on a ‘global’ basis (e.g., non-standardised units, test reporting formats) and so these remain sources of potential error in development of an ‘any-to-any’ interoperable risk framework. The investment in datasets is fundamental to improving content labelling, accessibility, structured search that underpin planning, delivery and monitoring of services within health and social care. Fundamentally, they can assist same standard of information to be generated from care records, independent of the organisation or system that captures the base data.

The adoption of DCCR definitions at each participating clinical site (giving a semantic equivalence and therefore interoperability for data being collected) was also considered essential for the success of any subsequent multi-site integration steps, e.g., in forming an interoperable information service family dependent on local and national healthcare messaging fabrics.