gms | German Medical Science

Introduction: The use of FHIR resources within IHE profiles has proven to be increasing over time, in particular in the last year. However, the approach on how FHIR is being used within IHE profiles varies. It is the objective of this paper to identify different approaches and apply these for an analysis of FHIR based profiles.

Methods: Based on the list of IHE profiles [1], [2] 33 FHIR based profiles were identified. They were reviewed regarding the approach taken for introducing FHIR into existing IHE profile or newly established IHE profiles. The following criteria were used: (i) providing FHIR functionality by means of wrapping IHE actors with FHIR based actors, (ii) adding new FHIR based actors and /or transactions to an existing IHE profile covering the full or only a part of the original profile's functionality, (iii) changes imposed on basic FHIR resource properties and function, (iv) amendments for specific resource operations related e.g. to queries, (v) FHIR based content profiles, (vi) profiles supporting transitions from XDS based data object content to FHIR based granular information representation,(vii) the IHE domain assignment of the FHIR based profile and (viii) the version of the FHIR resource, the IHE profile state (trial or final) and its test record in connectathons.

Results: Overall, establishing FHIR based profiles has started with one profile in 2016 and led to 18 profiles in 2019. So far only one profile achieved the status “final” and the majority (26 of 33) is in the status “trial”. Regarding the approach on how FHIR is being used in IHE profiles, four classes could be derived: (i) FHIR based content profiles were found in 3 (9%), (ii) using wrapping of existing IHE based actors and transaction with FHIR based actors became apparent only in 2 (6%) profiles, (iii) extensions to resource functions (e.g. $ihe-pix) were required for 4 (12%) profiles, (iv) adding or newly establishing FHIR based actors and transaction received the vast majority 21 (64%), and finally for 3 (9%) profiles an assignment was not possible mainly due missing information (status “public comment”). The analysis shows, that FHIR functionality is clearly embedded in the profiles ((iii) and (iv)), however in most cases following the Pareto principle of FHIR when identifying FHIR based actors and transactions.

Discussion: Besides deriving four different classes, this analysis also reveals, that the agile nature of FHIR leads to a version distribution of the FHIR resources used (9% R2, 33% R3 and 58% R4) and thus requires profile updates. As expected ITI (33%), PCC (30%) and QRPH (18%) are the most active domains related to FHIR based resources [1], [2]. The uptake of these profiles based on connectathon results is low [3]. So far only 11 of the 33 profiles have been tested, however they gained good acceptance with tests by on average 15 vendors. A significant increase is expected for upcoming connectathons and also motivated by the ongoing specification of the MHDS profile [3], which as a kind of whitepaper demonstrates effective FHIR profile usage.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.