Introduction
The term interoperability can be sub-divided into levels, by the means of taxonomy. An issue
regarding taxonomy originates in their often qualitative or narrative nature, the lack of quantitative
measures for precise estimation of the level of interoperability. This is exemplified by the taxonomy
by Stroetmann et al. [2009] and the taxonomy utilised by Jochem [2010]. These taxonomies can
be used for classifying the level of interoperability between two or more interoperating information
systems. But in terms of expressing an exact level of interoperability, that can be used in for
instance the comparison of the results from studies on informatics, there is a need for a more
mathematical approach. Interoperability can be achieved through the application of informatics
standards. Some widely applied standards have been developed and published by HL7. The latest
interoperability standard from HL7, Fast Health Interoperability Ressources (FHIR), has received
great international attention1 from a vast community of FHIR supporters. In a Danish context FHIR
has recently been adopted as the basis for establishing a health data ecosystem. The ecosystem is
intended for gathering and utilising health data, from a variety of different clinical and social service
domains. The heterogeneity of the domains is affecting the need for domain specific data, to support
the clinical workflow. This kind of variety is the core competence in FHIR, which is developed to be
profiled for specific clinical or administrative purposes. However in an ecosystem including multiple
domains and use cases, with their related data models, variety can be expected to have an impact
on the level of interoperability. The aim of this study was to examine this issue through an analysis,
based on the problem statement: “What are the contents- and functions- related consequences for
clinical data, when individual levels of interoperability are meet in a FHIR ecosystem?”

Methods and materials
As a case for the analysis, a real life FHIR ecosystem was utilised, including models for two clinical
domains. The analysis behind the problem statement satisfaction, was based on the “Health
Informatics Service Architecture (HISA) specification procedure” of the standard “ISO 12967 ”. The
three formal viewpoints of ISO 12967: Enterprise-, Information-, and Computational Viewpoint,
was covered by means of formulated research questions. The two viewpoints: Engineering-, and Technolgy
Viewpoint, of the HISA specification procedure are not formally described in ISO 12967.
Those viewpoints where joined and formalised as “Developer Viewpoint”, and covered by an guided
interview with FHIR ecosystem experts. All viewpoints, except Enterprise Viewpoint, were analysed
at the three levels of interoperability defined by Stroetmann et al. [2009]: “Technical and syntactical
interoperability”, “Partial semantic interoperability”, and “Full semantic interoperability”. For
assessing the exact level of interoperability, a mathematical measure for the “effective semantic
interoperability” was utilised.

Results
Enterprise Viewpoint, resulted in a purpose description of the studied FHIR ecosystem along
with domain specific data models for the two domains: Home- care and nursery, and physiotherapeutic
training. Technical and syntactical interoperability, resulted in the following consequences:
Data can be exchanged, but only utilised by human interpretation; problematic patient identification,
due to the lack of machine interpretation; simple data viewers, with developer heavy free text
search engines. Partial semantic interoperability, resulted in the following consequences: Sharing
of common information, that can be machine processed; reuse of FHIR profiles and advanced
business transactions across ecosystem applications; application specific profiling is allowed. Full
semantic interoperability, resulted in the following consequences: Complete sharing of ecosystem
information, allowing for complete machine utilisation; No application specific profiling allowed.

Discussion and Conclusion
This study demonstrated the application of the suggested model for analysing a FHIR ecosystem,
along with the mathematical representation for the level of interoperability. The strictness of the
HISA specification procedure, complements the openness of FHIR, by contributing with a standardised
modelling framework. The suggested measure for interoperability is applicable as an expression
for the exact level of interoperability. The exact level of interoperability in single-layer information
architectures, like FHIR, is however criticised for being submitted to degeneration due to domain
changes. It can be concluded that the suggested model and the mathematical interoperability measure
are applicable as means to analyse interoperability in a FHIR ecosystem.