This page is part of the FHIR for FAIR - FHIR Implementation Guide (v1.0.0: STU 1) based on FHIR v4.3.0. This is the current published version in its permanent home (it will always be available at this URL). For a full list of available versions, see the Directory of published versions

General Recommendations

1. FAIRness is an ongoing aspiration driven by application rather a pursuit of perfection for its own sake. Therefore, an incremental approach for continuous improvement should be followed.

   1. Do not pretend to implement everything from the beginning i.e. Do not try to “boil the ocean”!

      1. in terms of principles to be realized

      2. in terms of maturity level

      3. in terms of coverage across communities. For example, a solution applicable within a well-defined community may not be fit for purpose for a wider target (e.g., global search)

   2. You need to determine where you are with the selected data and metadata (“as is” FAIR assessment), what you want to achieve and the resources you have available.

   3. Metrics, such as the RDA FAIR data maturity indicators, are essential. These are used to evaluate improvements made in an iterative manner, to realize successful objectives.

2. The HL7 FHIR standard is a widely recognized and used standard in the health space providing value in the implementation of FAIR health data and services

   1. Even if HL7 FHIR can provide support to data FAIRification, implementing HL7 FHIR is not a sufficient requirement for being FAIR. This guide provides suggestions on how to use FHIR to improve FAIRness.

   2. The HL7 FHIR standard provides its highest value in realizing Interoperability and Reusability principles. Hybrid solutions, i.e by using combined FHIR and non-FHIR technologies, can be used as well to realize the other FAIR principles.

   3. This would not exclude the possibility to realize also Findability and Accessibility principles by using FHIR. However, a cost benefit evaluation, considering the level of adoption by that community, should be accomplished.

3. As with the FAIR principles in general, the FHIR implementation of the FAIR principles should follow an incremental approach.

   1. in this sense, depending on the context of usage, such as standards adopted by the community, the cost-effectiveness and the community goals, an architectural migration path with a consistent set of intermediate objectives should be defined. Objectives might be related to:

      1. the RDA indicators to be realized;

      2. the maturity of their realization;

      3. how much is realized by using HL7 FHIR.

   2. A minimal set of FAIR data maturity indicators should be prioritized by local application or business needs. RDA has given an essential minimal set as part of their canonical specification. Some pragmatic considerations:

      1. earliest progress is likely to be made with the principles for Findability, Accessibility and Reusability.

      2. Much tougher and longer term is the implementation of Interoperability when data are not FHIR by design. Interoperability is one the strengths of the FHIR standard. 

      3. FHIR provides a rich and expressive source of metadata elements which are critical for the implementation of Reusability. They also include metadata elements to describe data provenance and license for data usage.

4. FAIR / FHIR by design is the future shape of success. In fact (a) it is hard to make data FAIR when “source” data are poor (even if potentially findable and accessible…) or not designed for that; (b) the FHIR conversion of data may be not easy or not sufficient if data have not been designed from the beginning for that purpose. Thus:

   1. the availability of natively conformant FHIR data is an added value

   2. the specification/adoption of FHIR profiles against which to assess can help in raising the quality and the FAIRness of data.

Please refer to the HL7 FHIR and FAIR principles and HL7 FHIR and RDA FAIR Data Maturity Indicators for more details about the implementation of the FAIR principles and the RDA indicators.