DELIVERABLE - eHGI Deliverable TN D7.3 (formerD2.3) Draft... · COVER PAGE Project Acronym ... 2012...

eHealth Governance Initiative Joint Action JA-EHGov & Thematic Network SEHGovIA

DELIVERABLE

TN D7.3 (former D2.3)

Draft thematic Roadmap

on semantic interoperability

TN WP 7 – Roadmap

Project co-funded by the European Commission within the ICT Policy Support Programme and the Health Framework Programme

Version: 2

Date: 17-12-2013

eHealth Governance Initiative WP 7

__________________________________________________________________________________

2

COVER PAGE

Project Acronym eHGI – SeHGovIA Thematic Network

Grant Agreement number 270941

Status* Final

Dissemination level** CO

Author(s) & Organization(s)

Michèle Thonnet – Ministry of Social Affairs and Health - France

Contact [email protected]

* Status: Draft / In Progress / Final

** Dissemination level: PU = Public or

CO = Confidential, only for members of the consortium and the Commission Services

ABSTRACT

Semantic interoperability is critical for cross-border access to health information. However, the concept has to be clarified, as it refers to a model, to the structure of information, to terminology. Moreover, semantic components and tools in the health domain correspond to a great diversity of objects, from diseases nomenclatures to medicines or to reimbursement factors. There is now a huge number of different terminologies, nomenclatures and coding systems, most often different between Member States. This makes all exchanges difficult or impossible, even inside one country.

Indeed, clinical usage of nomenclatures is rare. This is why the document focuses on a possible roadmap towards progressive and practical semantic interoperability. At the same time, it is necessary to follow the fundamental ongoing works currently in progress in the scientific community. To that end, it is possible to exploit the resources accumulated through European Research programmes since 1992, until the present SemanticHealthNet project.

The proposed principle is to follow a modest approach and to build progressively an European infostructure, largely based on the achievements of epSOS Large Scale Pilot.

The document presents recommendations for governance and organization, for practical developments and for coherence with the building of the fundamental semantic ecosystem.

Change History


__________________________________________________________________________________

3

Version Date Status Author Details Review

000 16.05.2012 PreDraft M Thonnet outline EXCO

001 16.07.2012 PreDraft M Thonnet, N Paquel

002 09.08;2012 Draft M Thonnet WP LPPD, T&A

01 24.05.2013 Draft M Thonnet WP LPPD, T&A,

02 29.07.2013 Draft M Thonnet Final structure

review

EXCO

03 24.08.2013 Draft M Thonnet, N Paquel

04 07.10.2013 Draft M Thonnet SHN, WP P&S,

WP COORD,

PSC

05 26.11.2013 Draft M Thonnet, N Paquel

06 17.12.2013 Final M Thonnet

Statement of originality:

This deliverable contains original unpublished work except where clearly indicated otherwise. Acknowledgement of previously published material and of the work of others has been made through appropriate citation, quotation or both.


__________________________________________________________________________________

4


__________________________________________________________________________________

5

ABBREVIATIONS

eHGI eHealth Governance Initiative

eHN eHealth Network

EU European Union

EXCO Executive Committee

ISM Interoperability, Standards and Market

JA Joint Action

LPPD Legal & Protection of Personal Data

MS Member State

PSC Project Steering Committee

T&A Trust & Acceptability

TN Thematic Network

Basic terms and acronyms used in the document are in Annex 1

About this document

The present document is focused on semantic interoperability. This priority has been recognized as an area for MS and EC collaborative action in several policy documents and it was reaffirmed in the Multiannual Work Plan of the eHealth Network in November 2012.

It has been produced and delivered by the eHGI Roadmap working group (WP7) after interactions with:

- eHGI working groups and discussions in the Executive Committee and in the Project Steering Committee

- members of the SemanticHealthNet workshop in September 2012

This work was initially not intended to be designed and developed by the WP7, but supposed to be part of the WP ISM : Interoperability (technical and semantic) , Standards and Market), as indicated in the DoW of the SEHGovIA project. Nevertheless for practical and timeline reasons, WP7 offered to the EXCO and PSC who agreed upon to deliver a first Deliverable on this matter.


__________________________________________________________________________________

6

Consensus Statement

The point of departure for this document is the CALLIOPE recommendations “Towards a european-wide infostructure : semantic interoperability"

The Roadmap Work Package collected information from eHGI workshops and working groups – notably the Ghent workshop that the eHGI organized with the SemanticHealthNet in September 2012.

The document drafts were circulated in the eHGI and discussed in EXCO and PSC meetings. Comments and corrections from eHGI WGs have been integrated in the document.

The document will become the basis for the corresponding elements of the eHGI Roadmap.


__________________________________________________________________________________

7

TABLE OF CONTENTS

1. Key concepts and objectives ................................................................................................. 8

1.1. A foreword: necessary clarification in a complex domain ........................................................... 8 1.2. Fundamental objective and vision ....................................................................................................... 11 1.3. A Roadmap towards progressive and practical semantic interoperability ........................ 13 1.4. Perspectives .................................................................................................................................................. 15

2. Principles for action .............................................................................................................. 17

2.1. Reminding the basic objectives and actions .................................................................................... 17 2.2. Pursue a successful track ......................................................................................................................... 17 2.3. A practical approach: from use cases definition to generalised application ...................... 18 2.4. Mutualisation, cooperation and national policies ......................................................................... 20 2.5. The key semantic resources debates: monitor and act together ............................................ 22 2.6. Take into account social and organisational evolution of healthcare ................................... 23 2.7. Trust and acceptability: involve all stakeholders .......................................................................... 23 2.8. Legal aspects ................................................................................................................................................. 24 2.9. The critical interface problem : foster software innovation ..................................................... 25

3. Recommendations ................................................................................................................. 27

3.1. Recommendations: governance and organisation ........................................................................ 27 3.2. Recommendations: start practical developments......................................................................... 35 3.3. Recommendations: towards fundamental semantic ecosystem ............................................. 39

Annex 1: Terms and acronyms ............................................................................................... 47

Annex 2: Background and achievements ............................................................................ 51

Basic semantic resources: a growing profusion ......................................................................................... 51 A fundamental driver: EU directives, regulations and programs ........................................................ 56 An objective for a growing number of Member States ............................................................................ 59 Selection and adoption of terminologies and nomenclatures: an ongoing question .................. 60 Towards pivotal systems ..................................................................................................................................... 63 Linking fundamental terminologies and nomenclatures with practical cases: eH multiannual work plan .......................................................................................................................................... 63

Annex 3: About choice of medical nomenclatures .......................................................... 64

Annex 4: References ................................................................................................................... 71


__________________________________________________________________________________

8

1. Key concepts and objectives

1.1. A foreword: necessary clarification in a complex domain

Much confusion may obfuscate the debates about semantic interoperability. It is useful to remind some key concepts.

What is Semantic interoperability

• In the perspective of eHealth, semantic interoperability concerns both computer interpretable data and human interpretable information and knowledge. Exchanged information must be understandable by humans and simultaneously can be acquired, treated, used by information systems, applications or services in or by which the data were not originally captured. Also, such systems can be, from an IT point of view, heterogeneous.

• Semantic interoperability must be ensured at all geographic ranges, from local interoperability inside a healthcare institution to regional, national, cross-border interoperability. Difference of languages add only one more difficulty, and it is by and large not the most difficult to address, albeit it is costly. Differences between specialities in medicine or between different professions create often more obstacles.

• Semantic interoperability concerns different usages, even inside the clinical domain (i.e. not in public health or epidemiology and research). Indeed, the goal is not only to allow for a health professional to access an EHR and be able to understand the content and to integrate it and exploit it in/through an information system. It is also to be able to refer to knowledge bases, to reference repositories such as drugs dictionaries. It also concerns the patient, which should be able "to require access to their own health records, to suitable educational materials, and also to specialised medical content such as scientific literature and clinical practice guidelines" (Semantic Health Roadmap).

Three "levels" to consider

• As pointed out by the Calliope Roadmap, semantic interoperability is only possible if a clear sharing of roles between three health informatics elements is in place:

1/ A reference model.

Terminologies and tools correspond always to an underlying model. For instance, any classification of diagnoses is related to the present state of medical knowledge and medical tools (this can change – as, in another domain, classification of species, taxonomy has changed from morphology/anatomy to genetics); nomenclatures of medical acts refer to classification of diseases but also to the organisation of healthcare activities. Problem is that some terminologies correspond only implicitly to a model, which is the implicit view of the community / organisation that developed the terminology, and this model is not always itself coherent, or is not coherent with the underlying model of another actor – which leads


__________________________________________________________________________________

9

to misunderstandings. Others are built on a conceptual model, which relates to a specific domain of medicine or to a specific activity – for instance cardiology or surgical activity or public health…. In that case, the model gives a clear definition of all data needed in the considered “domain”, i.e. specific activity, and of their relations.

2/ Archetypes / templates structure

For every action, event or document, there is a way to assemble information, and this is what archetypes/templates structure are about. The agreement on common definitions for such structures should rely on scientific or technical basis: the way, even the protocol, to perform a measurement or to organise a surgical intervention, or the structure of a laboratory result documents are examples. EN 13606 archetypes, HL7 CDA structure and corresponding templates address this type of need.

Types of documents, type of data, information on an information or document (for instance localisation, date…) are designated and manipulated as "metadata"

3/ Terminology.

Once the data are defined and organised, they have to be poured into the structure, using classifications, terminologies, nomenclatures and often coding systems. Advantages of coding systems are the capability of identifying a piece of information (a disease, a diagnosis, an anatomy part etc.) with a code, which can be translated for humans in diverse national languages. Structured and coded information can be more easily understood, treated, exchanged between machines and automats (and then, when needed, expressed for humans in more usable format using intelligent interfaces).

To enter information, health professionals (as most humans) favour often free text, but, when it comes to communication between different parties, in the same country or with foreign country, coding is still the most valuable way, in the present state of technologies and information systems. Codes will be used for understanding what the information is about (metadata) as well as to understand critical information elements. The problem of entering information and coding – which must be easy in every day work -, will be addressed later in this document.

Multiple information objects to be identified and understood

Exchange of medical information refers to multiple domains, involving also data that are not medical/clinical. In most situations, exchanges will include:

- Identification

- Administrative information

- Medical "scientific" information (such as described in medical ontologies and nomenclatures) - diseases, anatomic descriptions, medical actions, biology etc.

- Non textual or numeric information (sound, pictures, video, medical images)

- Nursing information

- Information on drugs, on other medical products (as chemicals for laboratory reactions), on medical devices (including implantable ones)


__________________________________________________________________________________

10

- Healthcare professions identification

- Healthcare organisations identification

- Information for payments by insurance, based on specific nomenclatures

- Information on environment or disease causes (accidents, poisons, pollutants…)

- Population characteristics (for public health)

- Dates

- …

In any practical approach and in any use case, the whole information has to be understood and manipulated, referring to different reference resources. Some of the exchanged data are not healthcare specific, and, in this case, it is important to comply with general standards (as for date, or measure units).

Theoretically and practically, the most critical information is the medical one, and this is why numerous works, and also numerous debates, have been devoted to it. However, one must not forget the other ones, and the fact some are very difficult to transfer easily between countries: this is the case for healthcare organisations and units nomenclatures and for payment systems, notably. Contrary to these data, medical information is based on a wide international common scientific and technical basis.

Rich background and achievements

See Annex 2 for developments and a first list of existing tools and past and present attempts to build semantic interoperability.

Semantic analysis, development of nomenclatures and codes for health and healthcare has been an important scientific and practical work since long. One difficulty is already the abundance of semantic reference databases, while it is also a rich basis for future developments.

As seen in Annex 2, a myriad of languages and systems exist, there is however nothing to be afraid of: semantic interoperability in the health domain is a complex problem because the domain is huge and has multiple facets. It is also complex because of the richness and diversity of human languages, national and professional ones.

Since the GEHR (Good European Health Record) research program in 1992, the EU has devoted great efforts to health semantics, presently focused in the SemanticHealthNet programme.

Meanwhile, an important international work has been launched to address the problem of interoperability, based first on US and UK existing nomenclatures, and aiming to develop a general, ontology based, standard. This led to SNOMED CT, managed by the International Health Terminology Standards Development Organisation (IHTSDO).

There are still questions about the ambition of SNOMED CT and about its development model and also business model. In any case, it is an important milestone in the building of semantic interoperability.


__________________________________________________________________________________

11

However, it must be recognised that only very mature systems, apart from some research activities can profit presently of SNOMED CT. Distance with real world applications and professional practice is huge, and the gap could be widening as healthcare systems evolve towards coordination and integration with social and environmental factors and systems.

Accordingly, from a practical, political and financial point of view, it may seem that the roadmap towards semantic interoperability should not focus mainly on this question of adopting SNOMED CT, paying or not licences for it etc. This debate should not slow down the process towards a limited, realistic and progressive approach to semantics. As for the scientific ambition of SNOMED CT, it has to be deeply studied and monitored by the SemanticHealthNet. For the eHealth Network, it is necessary to consider closely the governance of the IHTSDO and its business model. Indeed, the fact that semantic interoperability in eHealth should depend on a closed and costly tool has to be questioned, while the EU and Member States policy and legal frame tend to define essential reference tools as free and freely reusable.

Due to existing nomenclatures and languages as well as diversity of needs and organisations, fact is that future developments of eHealth will need semantic interoperability and that this need will be fulfilled by intermediary, pivotal servers able to translate as well as possible a growing list of thesauri, nomenclatures, ontologies, from a specialty to another one, from a domain to another one, from a country to another one.

SNOMED CT, or the future developments of an international SNOMED like tool will certainly play an important role in building such middleware resources.

As many more language and domains are to be covered in the middle and long range, it will be necessary to rely on high-level tools able to define interactions between nomenclatures and concepts. Such an effort is already started and even ISO standards are defined (see Annex 2). One should not neglect the huge advances under development in the Semantic Web, from such high level developments to immediate, practical tools as those developed jointly by Google, Microsoft and Yahoo! European R&D programs should address this strategic domain, and development of eHealth semantic interoperability is a great opportunity also from this point of view.

1.2. Fundamental objective and vision

A pillar for cross-border eHealth

As Calliope has already stated, and as this was confirmed by the e-Health Network, progress toward semantic interoperability is a priority in the development of e-Health. It is a fundamental condition to ensure circulation and management of healthcare data inside countries as well as in cross-border European and worldwide relations. In fact, working on cross-border semantic interoperability adds a level of complexity but is closely related to every country problem in the matter.

A specific domain which cannot be isolated


__________________________________________________________________________________

12

There are two aspects when addressing the development of semantic Interoperability, be it for cross-border or for national exchanges of medical information.

- Identifying and/or building basic reference models, concepts, terminologies, vocabularies, codes. This is a fundamental and evolving domain. Developments and choices are long-range tasks, which can have important impacts in the healthcare systems and their capability to analyse and organise data. Moreover, such impacts are often difficult to foresee. For instance, developments based on incompatible models can lead to apparent, while confusing, similarities between terminologies, which may be dangerous when researching and accessing data in a medical record. This is why it is very important, for a safe development of e-Health, even in a restricted and controlled domain in the short term, to monitor the evolution in the general semantic interoperability world and to establish a close link with other aspects of e-Health so as to select the appropriate semantic references: there is a need for governance by the eHealth Network.

- The other aspect of semantic interoperability relates to services and infrastructure allowing for practical use of the semantic resources. As studied notably by epSOS, by the and now by the SemanticHealthNet, semantic interoperability is not possible without a network of European and national servers (to identify reference sources, know precisely what is the origin and version of a value set in a given document, access to multiple reference sources…). Moreover, European and Member States Information Systems devoted to semantic interoperability need to support the use of expert systems and to support use of clinical data for multiple purposes (including knowledge development, feeding registers and administrative simplification).

- Here, the governance at European level is a critical need and must be coordinated with the responsible institutions and policies in Member States. In all cases, semantic interoperability requirements must also be coordinated with advances in other components of technical interoperability, while taking into account legal and organisational issues and integrating semantic interoperability needs in a common European infrastructure.

No unification is needed but simplification is necessary

The Semantic Health project had pointed out that sharing clinical meaning does not automatically imply (and cannot require) identical terms and data structures. The goal of semantic interoperability is to be able to recognise and process semantically equivalent information homogeneously, even if instances are heterogeneously represented, i.e. if they are differently structured, and/or using different terminology systems, and/or using different natural languages. This equivalence needs to be robustly computable, and not just human readable – to be able to assist humans in precise understanding of the information as well as allowing possible integration inside heterogeneous information systems.

Inside countries, it is inevitable to develop multi-terminology approaches – and corresponding techniques to support interaction and correspondences - because of existing systems already in use. Moreover, all countries, in the eHGI survey on the matter, pointed out resistance from professionals to change an accepted solution. At the European level, it is important to conform to the principles of the Directive 2011/24/EU: the proposals to obtain better semantic interoperability will not imply important modifications of the


__________________________________________________________________________________

13

national information systems, especially when important advances have been realised using a particular solution. Strategy has to propose architecture able to perform and maintain conversions and translations between systems.

This does not mean that fostering simplification and coherence is not, simultaneously, a necessary orientation. The uncoordinated coexistence of multiple reference basis and processes, proprietary or not, creates important unnecessary costs, imposes additional burden on the final users and raises new essential interoperability issues. Therefore, it is of utmost importance to limit as much as possible the number of systems and to guarantee a quality mapping between them.

Financing and cooperation

It must be emphasised that semantic interoperability needs permanent and important financial resources.

The most important problem is not acquiring licenses for some terminologies (as SNOMED CT for example – this will be discussed later in the document). The most important issue is that semantic interoperability is a moving target. Medical knowledge evolves at a very fast pace, due to international research and evolution of technical tools but also to evolution of diseases themselves. Scientific progress diffuses faster and faster. Accordingly, semantic interoperability has to be supported by flexible tools and reference servers updated in real time. Semantic interoperability needs not only repositories and powerful search engines but also an efficient and resilient technical infrastructure. Moreover, in a cross-border perspective, translation and conversions will be permanently needed.

As explained above, even if unification is not mandatory, cooperation and mutualisation must be fostered.

1.3. A Roadmap towards progressive and practical semantic interoperability

The main objective and main basis is exchange of clinical information

Objective is to define a practical evolution allowing cross-border secured access to essential personal medical information by a healthcare professional entering in a medical relation with a foreign patient. Indeed, this "on the spot" situation is the most frequent and important one in e-Health, where access to information is mediated by a third party. Although semantic interoperability is critical for public health and for research, the more immediate objective concerns clinical relation.

To that end, semantic interoperability is necessary from formulating requests and finding documents (using "metadata" dictionaries) to understanding unambiguously the content and integrating it in heterogeneous information systems.

Many difficulties come from the fact that, although many nomenclatures and coding systems exist, today codification for clinical purpose is embryonic. People code for other purposes (reimbursement, research etc.). What is necessary is a semantic resource that speaks the Health Professional language. Secondary coding for purposes other than


__________________________________________________________________________________

14

clinical (eg knowledge development, public health, epidemiology, feeding registers and administrative simplification) needs as much as possible to be derived from primary clinical coding.

Foster and organise convergence and mutualisation while respecting existing differences

As this has been specified above, it is possible to obtain interoperability between heterogeneous systems but it is also essential for the eHN to coordinate Member States in order to reduce differences and engage in common programs for new developments.

As participants noted at the Brussels eHN meeting in 2012, "Difficulties in semantic interoperability were revealed by the Member States' responses to the eHGI questionnaire last summer, namely a huge complexity due to the use of an exceptional number of standards and coding systems in Members States."

However, it is also very difficult to evolve in that domain, notably when it comes to terminologies currently used by professionals and healthcare organisations and inside their information systems. Indeed, the objective is not to agree on common terminologies that would stay unused for a long time and could become obsolete, but to define a basis for efficient sharing of information between large numbers of actors.

This is why it is recommended:

- To engage in progressive actions – taking every opportunity to promote convergence and avoiding the development of new incompatible terminologies and coding systems if not strongly justified.

- Simultaneously to develop conversion and translation systems using the European infrastructure (as seen above: network of servers).

Start with concrete applications inside a limited domain

In the first phase, it is necessary to reduce complexity through referring only to constrained common documents. As this is reminded later, the diverse projects supported at the European level (SemanticHealth, Calliope, epSOS) have all recommended a limited approach, with a precisely defined type of exchanges or document, a limited medical domain, a limited number of situations. They have also emphasised the fact that this limited domain must correspond to choices based on common medical priorities defined by Member States.

The e-Health Network Members, in the Brussels meeting held in November 2012, approved the Multiannual Work Plan, which includes:

- Priority 2: Addressing Semantic and technical barriers to interoperability

- Priority 4: Guidelines on patient's summary set of data for cross border electronic exchange

- Priority 5: Guidelines on interoperability of e-Prescriptions


__________________________________________________________________________________

15

These priorities are closely related. Patient Summary and e-Prescription terminologies, nomenclatures and data coding give a first opportunity to link practical use of semantic reference tools and conversion systems with organisational, legal, trust and acceptability and technical issues. Moreover, epSOS results have already delivered proposals for data sets and technical infrastructure.

In its May 2013 meeting in Dublin, the eHealth Network has confirmed that definition of the Patient Summary minimum data set was a priority and that the epSOS proposed components was to be studied to that end. For e-Prescription, a first priority is to work on the practical use of a common pharmaceutical and medicinal products nomenclature and coding system, at least as intermediary common reference database.

A typical convergence between European and National interoperability

One cannot expect that Healthcare Professionals will use a national Patient Summary and will be able to interpret as many as Summaries that there are Member States having defined one for internal use. The development of a European Patient Summary is a powerful way to foster the development of Summaries in national systems. Healthcare Professional should be able to interpret rapidly and simply a Summary for a foreign patient. Therefore, even if some information may slightly differ or if some items are missing, the general structure and main value sets (groupings of data) should be the same throughout Europe, allowing a translation / transcoding server to present it to the Health Professional or Health entity information system, this server having only to adapt to specific national presentation modes. These structure elements and corresponding data sets are the basis for defining a limited domain for semantic interoperability throughout Europe.

1.4. Perspectives

As many observations have proven, and notably the recent eHGI survey, there is a great difference between national adoption of a given terminology or coding system and practical use. The most used nomenclatures and coding systems are national ones, for specific use (most often payment). The more ambitious ones are used in a very limited number of information systems and most often known only of researchers.

In fact, three steps must be considered for a given semantic standard (as for all standards):

- Adoption by a competent public authority in a country, or by a healthcare institution, or a professional association…

- Concrete implementation in information systems, in institutions

- Real use by healthcare professionals

One characteristic of epSOS pilot has been to consider real use cases and experiment in daily life situations. This is the way to follow. Work on Patient Summary value sets, national presentations and usage, will necessitate analysing and refining progressively real world conditions. It will also be necessary to define processes for maintenance, as evolution and correction will impact the document as well as the underlying information system.


__________________________________________________________________________________

16

Following this progressive approach, the objective is not to choose immediately a set of standard reference tools, but to define possibilities after a precise analysis of a chosen domain and situation. The general landscape will be progressively clarified, but officially supported tools should not be selected before definition of what are the practical needs and processes.


__________________________________________________________________________________

17

2. Principles for action

2.1. Reminding the basic objectives and actions

The eHGI Interoperability, Semantic and Technical team reminded general principles for infrastructure and infostructure development (presented in the 3rd eHealth Network meeting in Dublin – 14th May 2013:

1. Encourage a greater cooperation among Member State

2. Encourage a greater cooperation between national authorities

3. Enable the recommendation of standards and (harmonised) profiles

4. Use purchasing power of public sector

5. Foster data portability

6. Link and harmonise coding systems

7. Facilitate access to existing standards and medical vocabularies

8. Stimulate usability engineering for structured and encoded data

9. Consider motivation of healthcare providers

2.2. Pursue a successful track

As reminded above, the European Union has followed a systematic track at a quite fast pace and has strong assets, among which high level researchers and numerous industry specialists.

A modest approach on solid basis

Already in 2008, the SemanticHEALTH project, in its roadmap for research and deployment, recommended a policy of incremental steps and a focused, modest approach to terminology development, using limited subsets of terminologies relevant to the health improvement priorities of Member States.

" A stepwise approach is the cost efficient1, quality oriented and clinically relevant path forward in order to balance efforts and benefits. It can be progressed by (a) increasing the amount of semantically structured information, and (b) increasing the number of use cases."

This is being applied by the Semantic Health Net, which has chosen as example cardiovascular disease.

1 A study by Semantic Health suggested that the cost of full semantic interoperability would spiral out of control and that a partial interoperability level would deliver the best cost benefit ratio.


__________________________________________________________________________________

18

Calliope confirmed: "The focus of efforts on terminologies should be on sound, ontology-based2, user centred development of small, limited in scope and focused sets of terminologies related to particular use cases".

Agreed information models

"Alongside lexical and linguistic support in different European languages, respective agreed information models along with bindings to terminologies are important requirements." [CALLIOPE]

Pursue CALLIOPE conclusions and exploit epSOS findings

Calliope has proposed a Roadmap, based on a consensus of health authorities, stakeholders and standardisation organisations. Supported also by a wide majority of actors, epSOS LSP attained the operational stage with concrete proposals: build pivot documents on a commonly agreed model, group data in value sets, map to existing code systems, set up a logical infostructure (Catalogues, Terminology Access Service – see Annex 2). In 2012, epSOS has experimentally proven a process ensuring semantic interoperability for Patient Summary and e-Prescription.

It is now urgent to enter the formal development phase on real day-to-day exchanges. Principles set up by projects and recommendations are to be followed – with a flexible approach: adapt as necessary to practical conditions and continue learning trough concrete applications.

2.3. A practical approach: from use cases definition to generalised application

Towards wide development

In this eHGI Roadmap, the use cases step corresponds to the choice of precise situations for real applications. This is no more an experiment project: generalised development is the objective.

This is not contradictory with the need for a long-range vision of convergence and bridging of semantic resources. Action in this fundamental matter will be largely based on observations in real world application.

A global approach of semantic elements`

In this approach, the semantic problems should be studied at all levels. As CALLIOPE Roadmap recommended, " a tight linkage between data structure to be shared and the used terminology that provides the content for these structures (value sets) is critical. This should

2 A same ontology can be the source of different nomenclatures, for instance for diverse users. Moreover, it is doubtful that a common unique ontology will ever be capable to describe the whole medical activity – and this is even truer as the perimeter of the activity evolves and the relations with other domains are more and more important.


__________________________________________________________________________________

19

be closely linked to the development of respective care pathways (agreed workflow(s) by clinician representatives."

As specific diseases and situations will be chosen as priorities, it will be necessary to devise precise structure for value sets, grouping data that bring globally a transferable information, defining archetypes or templates (see Annex 2.1/tools/containers)

Take into account the available infrastructure

The eHealth semantic infostructure will have to adapt to existing infrastructure resources. In 2014-2020, this means a permanent liaison with the Connecting European Facility. The logical architecture for a network of servers depends on it. As for other aspects of eHealth, this means bidirectional influence.

To translate a document in real time, the infostructure will need to find information on the authors and their reference resources, locate the needed basis and perform translation and presentation to the user. This means a high level of performance and availability, when the application will be widely deployed. In this domain also, eHealth constraints are high, and they have to be measured for every new extension. In many cases, it will be realistic to limit specifications as long as the needed infrastructure is not powerful enough to support them.

Apply a systematic process when launching a practical phase

The progressive approach should start with operational, exchanges with semantic interoperability, on a limited domain and in a limited number of voluntary Member States as the first kernel.

Prerequisite

For a Member State to participate (and in general for all eHealth Network Member State), it is necessary to have already defined a strategy and created a reference base of resources, a thesaurus and a coding system in the concerned domain. When a country is less advanced in this type of semantic services, it would be preferable to foster a grouping, a convergence with existing solutions in more advanced countries. When the databases and processes exist already, the double movement of building intermediary conversion services and fostering convergence should be started.

Preliminary study

An understanding of what is possible and what should be done necessitates a deep study, supported by the EC and the eHN and under responsibility of the "Standing Coordination Group".

• Identify what exists in the Member States and what they want to achieve.

• Identify problems & goals through use case method.

• Consider or identify obstacles that have prevented the development of interoperability using available semantic tools (if no nomenclature or classification exists for a specific type of information, evaluation of the reasons, and probable underlying structural difficulties).


__________________________________________________________________________________

20

The fact that existing semantic terminologies have been proposed but are not used has to be scrutinised. As identified in many studies (see the EU Semantic Interoperability Research and Deployment Roadmap - European project – 2006-2009), a main difficulty is the preference of healthcare professionals to use their specific nomenclatures when they exist or resistance to use a structured classification in general. This is a matter of trust and acceptability and depends on the advantages the use of generic classifications may bring to the professionals as well as the quality and simplicity of tools allowing to use them.

• Define practical possibilities, at different levels: the highest being automated transmission of the meaning, lower ones being limited to an assistance for human intervention.

• Propose a roadmap to attain a better situation

Stakeholders

From start to launch of the first implementations, it is critical to validate real usage with stakeholders and especially with a panel of healthcare professionals and healthcare institutions. The key criterion for success and for ulterior diffusion is trust, acceptability and confirmed usability.

2.4. Mutualisation, cooperation and national policies

As in other domains, multiplicity of languages, and also of cultures and organisation, should be considered richness, and also a challenge for Europe and an asset on the worldwide market.

To exploit it, cooperation between all Member States and European stakeholders is mandatory.

National differences and domains of convergence

As reminded above, even the use of the same standardised tool or terminology or nomenclature does not prevent from incompatibilities – which are in fact the rule without intervention of an intermediary platform. Apart from language problems, there are differences in the versions of standards, the way they are implemented, the extensions that have been added, due to local specificities.

For fundamental standards in the medical area, as diagnosis, coordination between Member States has to be strongly reinforced, as there is a scientific basis. Also, for healthcare as well as for digital single market development, nomenclatures for medical products and devices evolve towards more and more European or international ones.

However, conforming to the principles of the Directive 2011/24/EU, the proposals to obtain better semantic interoperability should not imply important modifications of the national information systems, especially when important advances have been realised using a particular solution.


__________________________________________________________________________________

21

From an economic point of view, it will be necessary to compare, in the short and long range, the cost of maintenance and mapping between systems and the cost of evolving towards common ones.

Foster development and cooperation of national bodies

A governance process is needed both at national and European levels in order to avoid further dispersion of solutions adopted in different countries and sectors and allow a progressive agreement on the international infostructure to be maintained and mapped.

The Semantic HEALTH had recommended (2008) the establishment of sustainable national bodies (e.g., national centres for multilingual, multicultural adaptation of international classifications and terminologies), linked in respective European networks. More generally, it is important to strengthen the relations between Member States institutions in charge of health semantic domain to foster a common political understanding of the stakes and of the measures to be taken in the diverse concerned domains.

Multinational cooperation of all actors

Calliope stated "Consider the area of semantic interoperability as an area largely catering to multinational collaboration and empower a collaborative governance framework that will facilitate collaboration of the various stakeholders, incl. International SDOs and relevant industry bodies at all three layers: steering, strategic and empirical."

Common rules at European level

"Encourage the definition of appropriate quality standards for data in medical records and other electronic medical data/documents that are to be shared across borders."

"Evaluation, quality assurance (QA), and maintenance of terminology resources should also be addressed at EU level. Wide scale evaluation is needed, for instance, for the use of terminologies with re-usable clinical models of content and process (such as archetypes)." [CALLIOPE]

A key contribution to national interoperability

Cross-border semantic interoperability and in general eHealth data flows interoperability first and foremost objective is to allow a European patient to access personal medical information in his own country.

However, it is also a powerful way to help Member States ensure interoperability inside national systems. Aiming to cross-border interoperability increases the complexity and the difficulties but allows for mutualisation and implies definition and application of solid common models and tools. It is also the only way to create and maintain a complex and costly infostructure.

As many countries are implementing standards – with difficulties for effective diffusion -, mutualisation of developments and, when possible, unification around common solutions, should benefit all parties.


__________________________________________________________________________________

22

This is a strong argument towards cooperation. On the contrary, the risk of underestimating cross-border interoperability – which is real – is for countries to develop isolated semantic ecosystem and for Europe to let the split widen between more or less advanced countries.

2.5. The key semantic resources debates: monitor and act together

Dealing with existing and new terminologies

There are already many terminologies (see Annex 2), which represent together a huge investment. No one today will invest in a new terminology, at least in already known domains and with traditional approach. Apparently, and surely in the short term, choice is thus to select (and possibly adapt) what does exist.

However, as reminded under, adopting a technology is not using it – a first task is usually to limit the perimeter (see for example Annex 3: About choice of medical nomenclatures). Also, existing national – even regional – nomenclatures and coding systems have to be taken into account, although it is always better to try and adopt an international resource. Moreover, new terminologies are already developing, presently for genetics.

In fact, the problem is not only to choose terminologies but also to participate in their development. This is where cooperation between Member States and with the European Commission and competent institutions is mandatory. This will allow for an evolution towards modern way of developing reference tools and standards through "specialists and users crowdsourcing", open worldwide cooperation. This is how future terminologies will be developed.

Monitor

The main principle is here to monitor closely what's happening in the world of terminologies – medical ones and others – while maintaining the practical approach. This means a regular and efficient liaison between the eHealth Network and the terminologies and ontologies specialists and SDOs in Europe but also at the international level.

Reciprocally, difficulties arising in the domain of semantic interoperability when implementing a practical eHealth development must not be solved by local specific "practical" solutions. This would be the risk, frequent in information systems, of accumulating specific patch-ups. The solutions have to be discussed with terminology experts of the concerned domain.

For any important question about usage of an international resource, or creation of a common resource, the experts' advice should lead to a common position of eHealth Network members, supported by the EC.

Standards

The same process is to be followed for the standardisation phase. As in other aspects of eHealth, and other sectors, the whole standardisation process is now under question. Notably, the relation between the formal standardisation mechanisms of ECS-ISO and the


__________________________________________________________________________________

23

Standard Development Organisations are to be defined – the discussions with IHTSDO are a typical example (see Annex 2).

Clarify usage

Semantic tools are related to a model and a usage. The eHealth Network should make it clear that nomenclatures related to payment and reimbursement procedures should not be mixed up with medical purposes.

On the contrary, while the semantic interoperability objective is now addressed primarily for continuity of care, public health and research are closely linked and are based on the same scientific basis. Accordingly, the experts groups should take them into account.

2.6. Take into account social and organisational evolution of healthcare

As reminded above, medicine as well as healthcare systems objectives and organisations evolve fast. In particular, healthcare evolves from a pyramidal system towards a more transversal one, where more importance is given to patients empowerment, to ambulatory care, to small hospitals and care structures, while previously the University Hospital needs were prominent in the definition of semantic tools, with a strong orientation towards research and, sometimes, public health. This evolution goes at different paces in Member States in relation with healthcare system organisation and with national economic and social conditions and culture. However, it is a common worldwide trend as it follows the growing level of education, the wide adoption of the Internet for information acquisition as well as new technologies in medicine.

When defining priorities and use cases, all stakeholders have to be associated, and the solutions must be able to adapt to all categories. Notably, as nurses' role increase, nursing and care information will have more and more importance.

Also, everyday life conditions will be taken into account. European programs such as, in 2014, Ambiant assisted living and European Innovation Partnership on Active and Healthy Ageing should be considered.

2.7. Trust and acceptability: involve all stakeholders

Initial production of data is a key moment to guarantee future possibilities for translation and conversion. This implies that healthcare professionals and organisations invest in the production of structured and standardised data – which necessitates modifications in the information systems, professional training, human efforts and financial costs.

Any program will be a failure if professionals do not enter the correct information in the system, in everyday work. That will be the case if the systems interfaces are not efficient and if usability is low. These weaknesses have handicapped many products and services, notably in the healthcare domain.


__________________________________________________________________________________

24

This is why it is necessary to involve professional associations in the semantic interoperability discussion, in the general vision as well as in concrete developments. To that end, Calliope proposed also to "support co-ordination of work of national and European professional associations to engage into the development of terminology and translation services developed together with data structures and linked to the development of respective care pathways."

All new aspect in the infostructure that may impact the process and the ergonomics of data capture by professionals will have to be tested against existing systems. This is also why cooperation in training should be built with professional associations.

2.8. Legal aspects

Intellectual property

A main objective is that terminologies and semantic standards must become freely available at the point of use.

Most terminologies, nomenclatures and coding systems have intellectual property rights attached to them. Some of them are not free, but the movement towards free usage under open licenses is expanding3, even while many countries health authorities acquire licenced terminologies, mainly SNOMED CT. For this type of general tool, scientifically defined, this is not only an economic obstacle for many applications but also an obstacle for a wide communication and exchange of information, for public health information and medical research and finally for free access to information by the citizens.

Indeed, these types of information can be considered, at least most of them, as fundamental referential tools, which should be widely accessible to allow for the development of the information society and building of applications inside the open digital market.

A great proportion of semantic reference bases come from publicly funded research or programs and contribution of public actors. Researchers, universities and laboratories support more and more collaborative work to produce and maintain them. In conformance to the EU policy about Public Sector Information, such resources should be accessible and reusable freely (Directive 2013/37/EU).

Even if a resource is not free, as there are important costs attached to production, storage, versioning & uninterrupted 7/7 delivery services and maintenance, these reference tools should be made accessible at minimal costs and with very few conditions.

It is once again a common problem that Europe Member States should address together, in order to have a stronger position in the international debate and define rules for participation in the production and usage of semantic reference tools.

3 It has to be noted that Health Level Seven International has made its standards for the electronic exchange of health information free in 2013.


__________________________________________________________________________________

25

Information content

Understanding another country's documents, notably prescriptions, does not imply that the information can be used in the country where the citizen / patient / professional has requested it. Notably, some drugs are not authorised in some countries or have no real/exact equivalent.

From a legal point of view, this leads to the difficult question of substitution. In most cases, the professional in the country of treatment has to substitute (he can be confronted to different situations: sometimes there is a generic substitute, but sometimes there is no equivalent authorised in the country). Even the "simple substitution" (the basic molecule being the same) is an element to be considered, as legal framework (professional qualification, authorised products, processes...) differs between countries. epSOS addressed this problem in its experiment on e-Prescription.

For semantic interoperability, this situation has to be taken into account. Information on availability of a given molecule, usual solutions for substitution for a given medicine should be included in the system. It will not be managed by every country but by the intermediary platform, which will have to indicate at least the situation and, if possible, propose the equivalent or at least best substitution drug. This will be possible starting with key medicines used for specific domains (for instance cardiovascular disease selected by the Semantic Health Net) or for chronic diseases chosen as priorities.

2.9. The critical interface problem : foster software innovation

Ensure acceptability and usage of semantic tools by the professionals is today most often impossible, due to the absence of powerful new generation software.

Every human thinks, acts and exchanges using his own national and professional language. Every system that hampers the practical activity of a healthcare professional is and always will be rejected and, moreover, is not effective, or even may cause dangerous mistakes (patient safety risk).

Indeed, a good system is a system where the final user, when entering the primary information, enters structures and codes without having the feeling that he does so. This puts most of the challenge on the side of the industry and not on the user. Indeed, we know that there is a general resistance to enter primary information in a structured way, at least using present software. Aside from the availability of terminological reference tools, the user interfaces are thus of considerable importance. These interfaces have to adapt to different professional activities even inside an organisation. Moreover, interfaces should be developed for tools allowing access to information for patients (who are to be more and more empowered).

This needs high-level technical advances in user interfaces. In other sectors, the software industry has invested heavily on such problems, but it has not be a critical objective for healthcare software, because of difficulties but also because many systems are conceived and installed without enough studies of professional user needs and corresponding


__________________________________________________________________________________

26

technical developments. Given the costs implied, it is not realistic to believe that industry will alone invest in it without external incentives.

This question of interfaces is one to be dealt with in the Roadmap.


__________________________________________________________________________________

27

3. Recommendations

3.1. Recommendations: governance and organisation

[These recommendations concern mostly all "technical" aspects and are part of the Governance recommendations in the eHGI Roadmap - in particular, this is the case for the Semantic and technical Standing Coordination Group. They are reminded here and specific aspects concerning semantic interoperability are pointed out.]

Establish a Standing Coordination Group (SCG)

Reminding: Object

Interoperability in eHealth represents a long-term objective of the eHealth Network and is subsequently anchored in article 14 of the Directive 2011/24/EU. Despite the long-term character of the task, most European efforts to tackle semantic and technical interoperability have focused on short-term projects. The creation and administration of chains of projects – often with the same participants – lead to a bureaucratic overhead. Moreover, many projects were not sustainable beyond the lifetime of the project and failed to connect with the strategies of Member States and business cases. epSOS has highlighted such sustainability challenges for Member States and has pointed to the need for EU level governance and co-ordination of priorities and the development of common assets.

To overcome these limitations, the eHGI recommends establishing a long-term committee for semantic and technical interoperability.

This "Standing Coordination Group" would not replace the eHGI but would take over operational

aspects of its interoperability agenda. It should promote and coordinate the development,

maintenance, standardisation and adaptation of European semantic and technical interoperability

assets. Key tasks should be:

- Sharing information between Member States about best practices and learning experiences in

eHealth interoperability

- Fostering extension of existing bilateral cooperation projects

- Playing a leading role in the process of coordinating and recommending eHealth standards in

Europe.

To this end, the SCG should ensure permanent liaisons between eHealth competent authorities and competence centres4 in EU/eHN Member States, scientific groups, stakeholders, EU

4 These are public institutions in charge of defining nationally eHealth standards and technical rules. Usually, there is only one competence centre for all aspects of eHealth. If it is not the case (for instance one for ID and cards and one for data),, both should be concerned).


__________________________________________________________________________________

28

DGs, so as to present propositions to the eHN, to alert on risks and opportunities, to pilot studies if deemed necessary par the eHN.

The Standing Coordination Group should consist of a coordination team and a stable group of experts. In order to reconcile national and European strategies, the group of experts5 shall be preferably drawn from national eHealth competence centres. Additional experts from academia should be invited to contribute to specific topics if needed. Stakeholders shall be consulted via the eHealth Stakeholder Group.

Key role in the semantic interoperability area

Semantic interoperability is a moving and distant goal. Its realisation demands a long-range strategy and sustained actions. It demands also a continuous liaison between academic and research world, learned societies, healthcare systems actors.

This SCG will continue cooperation established between the eHGI and the SemanticHealtNet and other relevant groups and projects.

The SCG will have to work on short term practical progresses while maintaining a long-term ambitious vision built on evolution of social behaviour, empowerment of the citizen, progress of technologies, evolution of medical science and techniques.

EU and Member States tasks

Members of the Standing Coordination Group should be designated by the National Authorities responsible of online healthcare that are members of eHealth Network. The standing coordination group will rely on financial and technical support by EC programs – notably CEF – and on strong links with the policy makers and competence centres in Member States members or the eHealth Network.

In the case where there are several competence centres in a country, National Authorities

responsible of online healthcare will define which one is in charge of semantic interoperability policy or of a specific domain (medical coding, medicine…).

To present practical proposals to the eHealth Network, the SCG will monitor continuously the strategies for eHealth Semantics implemented by Member States and concrete evolutions in the matter. This will be obtained through regular liaisons with the National Authorities responsible of online healthcare. Member States will need to accept methodology and content for all diffusions of a semantic standard or of a tool.

Strength, weaknesses, opportunities, threats

Time has come for such a structure, and members of the eHealth Network have expressed their approval for existence of an SCG.

In the semantic area, a specific risk would be to embark on recurrent discussions on standards. Acting fast on concrete developments mandated by the eHN should reinforce the credibility of the SCG.

5 Member States may choose to select distinct experts for technical and semantic interoperability.


__________________________________________________________________________________

29

Dependence with other actions

All actions are dependant of the existence of the SCG.

Planning and sequencing

The SCG should be created as fast as possible – no later than November 2014, with structure and mandate defined on period 2014-2020 (as CEF).

In its 4th meeting (2013 November 19th), the eHealth Network members agreed to form a group of experts working on the elaboration of a sound proposal aiming to get funding from CEF on eHealth for semantic and technical interoperability.

Schema: the standing coordination group

(see general Roadmap)

Adopt specifications for a Semantic and technical operational mechanism

Object

Define the general framework, notably referring to epSOS findings, drawing from and building on proposals of the SemanticHealthNet Virtual Organisation6.


Political agreement on the framework, analysis of distance between different proposals and existing infostructures in Member States.

To that end, it is urgent to perform inventories of resources and registers, documenting past choices, supported by national awareness involving all current code users


This is a first step, necessary to establish a reference framework. It could be expected for May 2014.

Establish a financial mechanism for the development of the semantic infostructure

Object

Apart from the regular functioning of the Standing Coordination Group, it is necessary to establish a common action mechanism and budget for financing the necessary permanent infostructure and infrastructure such as a portal to access to different resources,

6 Tbe SemanticHealthNet WP8 builds a virtual organisation, comprising main groups and stakeholders concerned by Health semantics. This body would "assert what semantic interoperability services need to exist, how they should be established, co-ordinated, overseen and governed. It will advise funding bodies on priorities for research, for education and professional standards, and will advise national eHealth programmes on the maturity of various semantic interoperability approaches and where investments should optimally be made. This WP will draw especially on the European eHealth Governance initiative, and deliberately includes these representatives from several countries.


__________________________________________________________________________________

30

development, maintenance and updates of the Master Value Set Catalogue and Master Translation/Transcoding Catalogue defined by epSOS.


The semantic infostructure relies on common services and national ones. Constraints, tasks and financing need to be evaluated and shared at political level within each Member State. The common action mechanism depends on an agreement for contribution of Member States and CEF.

It will be necessary to define a mechanism to transfer specific projects results and experimental solutions towards regular financing.


The semantic infostructure development and regular operation is particularly dependant on a common and stable financing. Actions can start with specific projects or research programs but they will then stay as experiments.


The standing coordination group will not coordinate the common action mechanism. It will be established by EC and the eHN. However its functioning depends on the existence of the SCG: this one will consult its network and deliver advices for integrating new elements coming from projects in the stable semantic infostructure.

Therefore, the financial mechanism will rely on SCG advices.

It depends also of the general relation of eHealth with the CEF.


The mechanism has to be defined by the EC and the eHN. Some time has to be devoted for implication and negotiation by the ministries. Also, mechanisms for evolution of the functioning have to be defined.

The process of definition of the mechanism should start as fast as possible and should not last more than one year.

Integrate the management of the European common semantic infostructure in an eHealth specific institution

Object

The semantic interoperability infostructure will comprehend a network of services that will be ensured as a special layer in the general eHealth infrastructure, which itself is supported by a common European infrastructure for basic services.

It is important to establish an operational management for European common eHealth services as those mentioned above (Master Value set…) and, if necessary, physical servers. Under the mandate of the eHN, a special entity should be in charge of this mission, taking


__________________________________________________________________________________

31

responsibility for the services and/or components, independently of the location of the bases, portals and servers (but will negotiate and control the IT providers).

As epSOS has observed: "The physical location of the storage place was deemed politically

sensitive requiring a political decision on whether or not a particular country can be a host for

content concerning the whole European community. It was concluded that, in the longer run, the

maintenance of the epSOS MVC, epSOS MTC on which is based the epSOS Ontology needs to be

officially performed by a European entity."


The building of a competent entity to manage common European service will follow rules defined by the EC and Member States for allowing development of the open market and of public services.

Member States will have to define a national competence centre for managing semantic resources (or more if this is the case).


The building of such an entity will be politically complex and it will need a stable financing resource. If it is not set up when the first applications move beyond the experimental phase, it will be necessary to define specific agreements between Member States. Development will be hampered by recurrent negotiations on roles and responsibilities.


The creation of the managing entity is dependant of the proposal for a common action mechanism and budget.


The eHealth Network should start negotiations with EC and Member States, grouping the common action mechanism and the creation of the management entity.

Provisional functioning is possible for a short period but all management responsibilities should be transferred as fast as possible.

Define a mechanism for progressive extension of practical advances

Object

The eHN should define precise rules to organise, among its members, kernels of voluntary Member States agreeing to implement together components of the envisioned infostructure. New developments may be a specific document, integration of new templates or value sets or tools as a specific server. This will allow starting usage rapidly, then expand the use of the chosen semantic element and be able to gain more and more participation.

Then, Member States who share a common strategy are to be strongly encouraged to collaborate in order to avoid redundancy and minimise costs and show what is possible by doing. This collaboration can take the form of common investment in expensive tools


__________________________________________________________________________________

32

(terminology server), exchange of tenders, exchange of methodology, cross validation of subsets, cross-validation of translations, training of experts, evaluation of pilots etc.

All eHN participants must agree on the mechanism, especially on the financial aspects. Indeed, it is necessary to define the financial conditions for the Member States wishing to join and to adopt the solution – which will be generalised eventually.


Circulation of information is critical. Concerned Member States have to provide information on strategy and practical implementation.Strength, weaknesses, opportunities, threats

There are two risks to avoid:

- Creation of competing kernels, which would be dangerous for the whole eHN action. - Attempt to obtain from start a participation of all or a majority of Member States,

which would slow down or block the process.

However, it is not possible to start a development if Member States declare a strong opposition. This is why rules for launching an action have to be agreed.


Dependant of existence of SCG and, rapidly, from the financial mechanism (which may be replaced by specific agreement while not yet stabilised).


The progressive definition of this type of organisation can be beneficial for eHealth Network and SCG organisation.

Define and apply a standardisation policy

Object

There will not be in the future an everyday semantic interoperability without adoption of standards7 for a large number of components, from terminologies to software tools interfaces and application programming interfaces.

This is not first a technical matter but a politic one, as all the history of standards has shown. Defining and making applicable a standardisation policy is a strategic component and it is a key aspect of eHealth governance and a critical task for the eHealth Network.

This policy must of course be a component of the general standardisation policy conducted by the EC.

This policy should also be applied for pre-standard applications ("practical developments" in this Roadmap): a standard exists only if used (which is not the case for a lot of them – a cause for non credibility). When a consensus has been reached for a specific value set or document, a set of rules should apply to all concerned actors.

7 Characteristics of any standard are: stable, consensual, precise and non proprietary


__________________________________________________________________________________

33

In the eHealth semantic domain, the eHN has to foster a common policy for all Member States and define a precise role for the European level (i.e. European Committee for Standardization). After analysis by the experts group in the SCG, the eHN should propose a reorganisation of the relations with Standards Development Organisations (c, IHTSDO, HL7…). Close relations must also be organised with other European institutions in the field, such as EMA.


The semantic interoperability cannot be separated from other aspects of eHealth and of Information society standards in general. The EC needs to conduct a general reorganisation so as to participate on an equal basis to production and maintenance of semantic reference standards, bases and tools, in close accordance with stakeholders and also with European industries in the domain of ontologies, terminology, text analysis and translation.

This is a common task, shared by the Member States: in this complex and difficult domain, they should join forces and propose special mechanisms for coordination of national standardisation institutions. These institutions should limit separate relations at the national level with international SDOs and reinforce specific cooperation in the area, inside ECS.

As for all other eHealth standards (see general eHealth governance), the eHN should propose Member States on a voluntary basis to enact regulatory rules of conformance for suppliers: capability to provide the defined value set or document in a standardised and open way (no closed format, proprietary protocol, closed interfaces, as either the industry can provide the format, either publicised interfaces and formats able to be used by common conversion mechanism), publicly known interfaces (Web Service or messaging system). Defined with the industry associations and taking into account the reality of the market, these rules should preferably if adopted be enforced for all public sector bodies' suppliers. Indeed, the public sector – in most countries the major purchaser of eHealth applications (sometimes in an extended sense- see under) - should require vendors to use recommended standards, profiles and coding systems in public tenders. This rule should apply not only to the public sector in a restricted sense but to all actors in the healthcare sector – professionals and institutions – that operate with public national financing, or mandatory financing by a social security system. The rule should also apply to all European programs supported projects.


Europe has important assets but its role in standardisation processes is hampered by its fragmentation and the diversity of the situation in Member States. This is especially true in the semantic domain, where precisely diversity has be a powerful motivation for the development of multilingual, multicultural tools and where previous actions have produced important findings.

Without a strong common movement towards open semantic standards, European actors and European vision of health and healthcare would not be taken into account from start in many standards.


__________________________________________________________________________________

34

These risks are to be considered for medical terminologies and codes and for medicines and medical devices.


• Dependence on the creation of the standing coordination group.

• Dependence on the general eHealth standardisation policy conducted by the eHN.


This should be a priority task for the eHN, which should propose an action to DG Sanco and DG Connect.

Establish a strong liaison with the eHealth reinforced stakeholders group

Object

Stakeholders participation in definition and development of eHealth semantic is key, especially for trust and acceptability by all of them (Health professionals, healthcare institutions as hospital associations, care sector associations, patients, industry).

Reinforcing the stakeholders group implies to define procedures for a continuous cooperation – and financing.


Health ministries representatives participating in the eHN have to inform stakeholders in their country and to validate the status of participants.

The EC and the eHN will define stable liaisons and financing mechanism.


The risk is to enter in too many political debates inside a reinforced stakeholders group – notably between competing associations in a profession and a country. However, the situation would be worse if these debates were multiplied inside many Member States. Once again, the best way to avoid too much politics is to concentrate, through requests of the SCG, on practical application of a given, limited, semantic area.


The financing mechanism should be included in the SCG budget, as the stakeholders group is considered a permanent expert group.


This is a progressive task, which will be prepared by the eHN and will induce political discussions inside Member States.


__________________________________________________________________________________

35

3.2. Recommendations: start practical developments

Launch the European Patient Summary program

Basic objective

(This paragraph will be linked to the Patient Summary Roadmap)

Objective is to allow the patient and the healthcare provider in clinical relation with him to access fundamental information on his health. To that end, define a common Patient Summary and set up an infrastructure to allow cross-border access. Such a summary is a concise clinical document that provides electronic patient health data applicable both for unexpected, as well as scheduled, healthcare contacts. Here, semantic interoperability assures the meaningful interpretation of medical data and provides the basis for an automatic translation of coded information from one language into another.

Directive 2011/24/EU Article 14 specifies "The objectives of the eHealth network shall be to: … (b) draw up guidelines on: (i) a non-exhaustive list of data that are to be included in patients’ summaries and that can be shared between health professionals to enable continuity of care and patient safety across borders;"

It must be noted that the Summary must also be accessible to the citizen himself for his own information or control, in this case in his own natural language. There is no semantic difficulty here, except for the medication history (see under the medication history problem).

Cooperation for a progressive approach

Building the Patient Summary implies to start with cooperation between voluntary Member States in the eHealth Network to develop concretely a first instance of a Patient Summary, based on Guideline on Patient Summary minimum/non exhaustive dataset for electronic exchange under the cross-border directive 2011/24/EU.

This Patient Summary objective has received a strong support by eHealth Network members at the May 2013 meeting in Dublin. No Member State opposed the use of the set of data from epSOS to construct a guideline. 20 Member States specifically supported it.

Following the "modest" approach, the development of the Patient Summary will be progressive and long-range. Summaries data set content may differ between diseases.

For example, action could start with the specific diseases selected by the eHealth Network.. In that case, the Summary would exist only for those people suffering from these chronic diseases. The summary would concentrate detailed information only for those diseases. The Healthcare professionals would be able to access this important information even in case of emergency/unexpected contact for another motive.

Indeed, if a specific disease is selected, or if sub categories are distinguished (notably inside the cardiovascular disease analysed by the SemanticHealthNet), it is possible to deepen the semantic analysis and to convey a richer set of information, going further in automated conversions).


__________________________________________________________________________________

36


The Patient Summary contains the mandatory Medication Summary. Its full realisation depends on the existence of the common medicine database and capability of transferring understandable information on the medicines.

Cross-border access to Patient Summary with conversion/translation services needs the whole semantic infostructure to be in place, as well as the general eHealth Infrastructure, notably for security and authorisation of access.

In the semantic domain area, the problem is not only to find all the services necessary for converting codes and/or translate languages. It will be also necessary to know precisely the origin of the information and to be able to guarantee its quality (see under).

Accordingly, the full objective, even for a limited set of information, will not be accessible before the implementation of these basic components.

Also, the development of the Summary is costly and depends on permanent financial resource (but provisional solution could be used).

However, the Patient Summary is clearly a way to stimulate the development of eHealth, as its objective and usage are obvious for all concerned parties. It will be the first document allowing refining and testing the infostructure. This is why the process should be started rapidly.

Planning and sequencing, EU and Member States tasks

The first phase is the definition of guidelines to produce the summary in all voluntary countries. This needs to define a minimum common dataset. This is a critical task. According to present evolution of the healthcare system, it would be important to validate nursing and care information dataset – coordination with nurses is to be developed.

Some countries have already defined national EHR or Summary. There, it will be necessary to study conversion problems. It is possible to allow for optional items clearly distinguished from the minimum common ones.

The second phase should probably be the use of the common Summary inside participating Member States, using their own technical architecture, and without the translation problem, except for countries that have more than one official language. Those ones will have to start testing multi-lingual Summary.

The third phase – cross-border access - will occur when the security infrastructure will be available and when the necessary tools (Value set and Translation catologues notably) will be built.

In a fourth phase, the Summary could become a master document, containing general information, with more specific components attached, for instance a special diabetes component or a cardiac disease one. Also, a special synthesis may be defined for urgency.


__________________________________________________________________________________

37

Tools to build in parallel

Indeed, parallel to the work on guidelines and use of Patient Summary, the necessary tools have to be built at EU level, coherently with epSOS architecture: Master Value Sets Catalogue, Master Translation/Transcoding Catalogue.

Accompanying measures

• Definition with healthcare professionals of a training program and cooperation at the European level.

• Communication program on the European Patient Summary. Creation of an information site presenting results and advances, allowing for groups to share successes and difficulties (this site would be later integrated in eHealth one for wider perimeter, but it is useful to identify the Summary and not wait for more).

• Vendors and health care providers should be required to provide the facility for patient access to their Patient Summary.

Strength, weaknesses, opportunities, threats, EU and Member States tasks

The main problem, when it comes to Summaries and EHRs is the difficulty to define content – as there is never an agreement on what is minimum fundamental information.

A clear rule should be defined: in the first phases, the Summary corresponds to a minimum kernel of information, agreed on by all actors.

Another difficulty is the production and maintenance of the Summary. It has to be addressed at national level, as it depends on organisation of the healthcare system and on national laws.

However, the concrete development of the Summary could be considered a project. Eventually, when the common action budget has been defined (see above governance), it could be used.

The main load is here on participating Member States, as they will develop the first national instance, they will establish the national production mechanism and they will follow the progress, analysing difficulties and acceptability and informing others.

Prepare Medication Summary and cross-border access to e-Prescription

Object

Create the necessary eco-system allowing interpretation of medication summary and e-prescriptions by doctors, pharmacists and patients across borders and to know which drug is functionally identical or which is the closest equivalent

According to epSOS constatations, "We have neither a common data model nor a common

vocabulary for medicinal products or pharmaceutical products throughout Europe. This means that

we basically cannot exchange the knowledge on a medicine from one nation to another."…" the only

actual solution for resolving the problem of the EU cross-border medicine identification seems to be a

centralized / federated European Medicines database".


__________________________________________________________________________________

38

This objective implies defining a close cooperation with the European Medicine Agency, which is building a database and should have the main responsibility in this matter.

The tasks are:

- Build a model for presentation of information on medicine, according to international standards (notably ISO IDMP). The model must include substances, proportions, pharmaceutical dose form, regulatory information. For medication history and e-prescription, the model must also include specifications by the medical doctor (or other health professional).

- Build the medicine database – i.e. coordinating with EMA which has to stay responsible of IT related with medicinal products at EU level. The base can be replicated in participating countries, as epSOS has recommended.

- Build a Master Translation/Transcoding Catalogue (which may be managed by the EMA – even if mirrored at the national level).


It is necessary first to study more closely the situation in all EU/eHN countries.

A prerequisite is the existence of e-prescription or equivalent in the country or of a strategy or a project in the domain.

Definition of the model and building of the database and mechanisms will last long. Moreover, the trend towards open market will perhaps impact the medicinal product markets, reducing the diversity between commercial products.

Medication history and e-Prescription themselves will not be cross-border accessible as long as a trustable mechanism for conversion does not exist.

However, in the first phase, it is possible to build more rapidly the system through limitation of the perimeter. This can be done if it is decided to start with some specific diseases for patient summary. Then, a limitative list of medicines, critical for the selected diseases, might be established (for cardiovascular disease or diabetes for instance).


Medicine database and middleware services can be developed independently of other components of eHealth but medication history and e-Prescription need the existence of the global eHealth infrastructure and rules.


At the European level: coordinate the study, build the database and the diverse components for management of e-prescriptions.

At the Member States level, under recommendations of the eHN, reduce diversity of pharmaceutical products nomenclatures and codes, support a national control point able to dialog with the European database and with translation tools.


__________________________________________________________________________________

39

Member States should also incentive professionals for entering coded information – or facilitate this entry (see accompanying measures).


Work at EMA and epSOS operational test have given strong elements for progress, while they have also shown the difficulties due to absence of common model and base.

The main risk lies in the possible growing of gap between different European countries. Indeed, it is much easier to develop national e-prescription and this has already been successfully started in countries or regions. This is an asset but the problem comes from the diversity of models and coding systems. This diversity will probably increase, as prescription is an important element for an economic control of healthcare and this may cause diverse options according to every nation' policy.

Accompanying measures

• Facilitate coded data entry: this is a key step for the creation of e-prescription. Moreover, modifying the codes systems and the rules will probably encounter some resistance from professionals as well as IT managers and industry. Therefore, it is necessary to work with those actors, at the European and national level, to facilitate data entry. As industry will not invest much on a long-range action, a special European R&D project could be launched (a similar problem is encountered for all eHealth documents and data sets – see under in Recommendations for the eco-system).

• Study legal aspect of substitution from a product to another, as rules differ between countries – about the possible role of each professional (this can probably follow the general rule "in Rome, do as Romans do") but also on authorised drugs.

3.3. Recommendations: towards fundamental semantic ecosystem

Foster the creation of a network of competence centres.

Object

In response to a survey of stakeholders, CALLIOPE recommended to establish a (virtual) European Competence Network or Centre for guidance, evaluation, QA, maintenance, exchange of experiences, best practices in development and use of ontologies and terminologies. The urgent need for one or more reference centres where the “clients” (users, projects, industry, etc.) could find the necessary knowledge, advice and support to solving their concrete problems and avoid divergent efforts and duplication was stressed repeatedly. Multi-lingual terminology and classification / coding services should also constitute part of these activities.


Member States will define a responsible institution for eHealth semantics and specify its responsibilities. There are many institutions in charge in the country – for medical devices, drugs, medical treatment etc. but every country needs (and frequently has) a unique


__________________________________________________________________________________

40

institution responsible for eHealth applications. Common objectives and priorities at EU level could be designed and worked out through close involvement of Member States habilited competence centres participating in SCG.


Regular meetings and permanent liaisons will help prevent increase in the dispersion of choices and techniques. It will allow for better cooperation and mutualisation of developments, development of open source software and contribution to European and international standardisation.


This action can start as soon as possible – Competence centres are already exchanging much information, albeit informally. Once the possible political problems solved, the network will set up a collaboration on the Web to organise discussion, production of documents, coordination in European and international groups.

Build the coherent components of an infostructure

Object

To progress towards semantic interoperability, a coherent package of services has to be created and maintained in Europe. Some critical elements are known.

It is necessary to decide about the basic components and about the general architecture.

The chosen solutions must be coherent with other eHealth needs and with the European underlying architecture.

Services are to be defined and refined progressively, usage starting with the first practical actions, notably in the Patient Summary and e-Prescription domains.

• A “Convergence Service” concept has been proposed by many projects, including CALLIOPE, and is being established. It will provide a directory of usable assets, links to terminologies used, a wiki for all definitions developed and used in projects, a "cartography" of data models in use and possibility for convergence, common building blocks, guidance and recommendations for specific use cases. This “Convergence Service”, established through an European portal or equivalent technology, could be an important element for a future European Infostructure. This is the task of the SemanticHealthNet Virtual Organisation being built.

• epSOS has defined the Master Value Sets Catalogue and the Master Translation/Transcoding Catalogue (already deemed necessary for the Patient Summary). epSOS results must be analysed and the eHN should propose an architecture and the model of the services after a study led by the standing coordination group.


Independently of the location and management of physical servers, the semantic infostructure will comprehend national services and middleware ones at the European level.


__________________________________________________________________________________

41

Member States will need to:

- establish an inventory of resources and registers, documenting past choices, national awareness involving all current code users

- clarify and express their strategy in the matter;

- create national equivalent of convergence service; this will allow actors to be informed and to find the nationally used nomenclatures and codes – in the whole system, this service should be the only contact point for the European semantic eco-system; as explained above, it will be possible for these national services to physically duplicate the central databases.

- When possible (no service existing), apply commonly defined specifications for local translation/transcoding catalogues and value sets Catalogue; these are necessary as unification inside countries is not attainable, at least not rapidly for a given domain; moreover, a great diversity of types of data will correspond to multiple terminology services (as for diseases, drugs…). If solutions exist already, the concerned Member State will develop interfaces and protocols according to the specifications of European tools.

At the European level, it will be necessary to:

- coordinate the general specifications of the semantic services

- foster standardisation of interfaces and messages inside the general infostructure

- publish specifications for interfaces and protocols for national entities.

These tasks could be part of the mission of the SCG.


As for all eHealth developments, the diversity of approaches for the semantic infostructure is a reality, corresponding notably to the organisation of the healthcare system and to the legal rules and administrations in charge of different domains. This can be considered richness, and it is important to exchange about successful mechanisms and tools. The risk is that the development of national infrastructures, after some years of operations, will have stabilised heterogeneous technical procedures, which would make circulation of information in the European eco-system more difficult.


Specification is dependent of the creation of the network of competence centres.

Effective operation is dependent of:

- the eHealth infrastructure development above European communicating facilities;

- the creation of common financing mechanism and of an European managing entity (see above in Governance);


__________________________________________________________________________________

42

- the launch of the Patient Summary action, as nothing is possible without a concrete project to use the infrastructure/infostructure.


A first phase is preparation of the action and political decision by Member States in the eHN, after proposition of precise specifications by the SCG working with the network of competence centres.

For Member States, this is the most urgent task: it is necessary to invest now on the infostructure. This is why, at the same time, synergies should be established between Member States, for instance with common public procurements.

The second phase is the creation of the Convergence service and of the network of corresponding services in the Member States

The third phase will be the adaptation or convergence of the national contact points and the development of the common tools (MVC, MTC).

The fourth phase will have to be simultaneous with the Patient Summary development. It will allow for validation and fine tuning of the specifications.

Define rules and mechanism for quality and certification

Object

Inside the general system for quality assessment and certification of sources, there is a need to develop special requirements for semantic interoperability.

An important issue is that of liability when professionals in one country make clinical decisions and take responsibilities on the basis of information captured and encoded in another MS. Issues of quality of information in records will need to be addressed at the level of agreeing criteria, exchanging good practices and audit and monitoring information to maintain trust. Similarly, the nationally performed actions of transcoding/conversions and translations to the national language introduce a level of processing which may affect the quality of the outcome. It will be necessary to define and follow common principles of quality assurance, in accordance with concerned organisations, notably SDOs. Such common rules will be applied to define the Circle of trust envisioned by epSOS and included in the framework agreement between participating Member States.

As Calliope stated: " common approaches to testing, evaluation, quality assurance, maintenance of semantic resources are typical areas for joint efforts".

A constant monitoring is necessary to avoid deviations in the use of terminologies, nomenclatures and coding, due to specific problems in the different countries. There is also a necessity to ascertain the origin of an information or a document. This last point is not related specifically to semantic interoperability but to general security and trust conditions inside the eHealth infostructure and technical infrastructure.


__________________________________________________________________________________

43

Quality here depends on national contact points such as those proposed by epSOS8 and on the European middleware. They must be able to guarantee which version of a reference base they use and to alert on any modification.


At European level: manage a process to verify and validate coherence of all the references at the finest level (for instance conditioning, posology, units…).

At Member State level: there is a requisite for participation: having defined a strategy for terminologies and codes, if possible referring to international standards.

Then, Member States will have to provide complete information on specificities and modifications in their national system of codes and semantic references, verify and validate entry systems if there is automatic coding.

An efficient system should be set up to gather informations on difficulties and errors directly from citizens and professional users.


Credibility and acceptability of access to health information should be seriously hampered by errors and would foster the return to local, well-known, data, making cross-border access an utopy.


Quality can be assessed only if regular procedures have been defined for the whole eHealth development.

Quality assessment depends on the existence of the network of competence centres.


First, competence centres should start an exchange on their strategy for quality assessment of semantic resources and of coding tools. Then, they would propose the eHN a set of common procedures and a charter of good practices.

Elaborate and maintain a strategy and practical actions on intellectual property rights

Object

Allow for a free access to all essential semantic resources at the point and moment of use.

Allow for open collaboration on essential semantic resources.

Extend the application of the European open data policy to dictionaries and nomenclatures that are from public administrations origin and from programs financed by governments and mandatory social security systems (see Directive 2013/37/EU).

8 "An epSOS NCP is an organisation legally mandated by the appropriate authority of each PN (Participating Nation – ie Member State) to act as a bidirectional technical, organisational and legal interface between the existing different national functions and infrastructures".


__________________________________________________________________________________

44

Define and protect intellectual property rights on the conception of semantic bases and tools, while fostering the free and open access to content, subject to licenses derived from the work of the open data movement, of Creative Commons, of EU Public License for software – according to the type of resource.


This is a common task, where the joint forces of the EC and of Member States are necessary to support actions in the field. This has to be carried out by the SCG institutional network of health authorities and competence centres, in collaboration with EMA and other concerned European institutions. Also, EMA and other concerned European institutions should participate. Moreover the EC and the eHN should organise exchanges at the international level with other countries and with concerned SDOs, and groups "controlling" the diverse international resources.


Europe has strong assets and the strength to promote openness in this strategic domain. It is time for that: the movement to openness is now general and strong in all scientific domains.

Already, for software, epSOS has favoured the development of open source tools.

If a common position is not defined and applied, weakness may come from fragmented discussions with international entities.


This action depends on the definition of a standardisation policy.


It is time to start the legal study and to take into account intellectual property rights in the standardisation strategy.

Ensure capability of IT providers to export and import data

Object

Data are produced, exported and imported by Information systems, notably electronic health record systems.

Objective is that IT providers:

- at least can export data in a known structure and through interfaces that allow exchanges with national and European semantic services;

- integrate progressively the chosen common standards inside their system (terminologies, coding systems, archetypes and templates, data models).

A middleware will always be mandatory, due to evolutions, management of versions, new value sets etc. but the more the software use common standards, the less are the strain on conversion systems and the risk of errors.


__________________________________________________________________________________

45


To that end, a close cooperation has to be maintained with the industry associations and with users associations (professional and healthcare institutions) at the European and national levels.

It is also necessary to incentive IT providers:

- support through specific programs; - use the purchasing power of the public (or publicly financed) sector


If there is no incentive and control, risk is that powerful companies could progressively lock the system in their own version of a standard.

If IT providers make few efforts, the load will be heavy for the conversion services.

It has to be noted that industry representatives have already participated widely in epSOS.

A condition is to associate IT industry along the whole process of development of the middleware.


Adaptation of interfaces9 and integration of standards are possible only when the middleware and the main standards are specified.

Progress depends also of standardisation policy.

For Small and Medium enterprises especially, integration and adaptation are much easier and stimulating if the open approach is applied.


Technical discussions can start rapidly with the IT industry – and industry representatives have already participated in epSOS.

Support R&D European programs for innovation in user interfaces

Object

As explained in the principles, advanced European R&D programs are necessary to develop intelligent software allowing capture of data in natural language or even vocal recognition and automated structuration and coding.


Projects should be launched inside the CORDIS 7th program:

EC issues a call for a study to develop an RD strategy (in this field), followed by a call for a RD project to data entry of encoded and structured data

9 The interface problem is addressed in the following recommendation


__________________________________________________________________________________

46

Cooperation with user representatives and with a panel of professionals (following epSOS experiment) is mandatory.


Natural language interpretation, text analysis and translation are a strong asset of European research and Small and Medium enterprises.

If effort is not made, it will be very difficult to generalise eHealth and especially structured information capture. Resistance will always be strong due to the tight agenda of health professionals.

The risk is to lack ergonomic software interfaces when the generalisation phase of the Patient Summary will start, as it is an important factor for acceptability.


This action is closely associated with the general effort on capability of IT providers to export and import data, but it concerns especially the final user – ie Healthcare Professionals and at a later stage the citizens/patients.

This action is long-term one and is not immediately dependant of other actions. In the middle range, interfaces will be practically adapted to current state of data and documents exchange.

However, it is necessary to evolve as well as possible to capture information for Patient Summary.

It is possible to do it with simpler algorithms


A research project takes time. However, it is perhaps possible to support shorter research on a more limited perimeter so as to propose more ergonomic interface for the Summary. Once again, if a reduced number of diseases is considered, provisional solutions, assistance to the user can be developed and will bring useful information.


__________________________________________________________________________________

47

Annex 1: Terms and acronyms

Terms

Archetype

(Eurorec)

A clinical archetype is an agreed, formal and interoperable specification for representing a given clinical entity such as a clinical observation, a finding, a plan or a treatment within an electronic health record.

Catalogue

A complete list of items

Classification

(Wiktionary)

A distribution into groups, as classes, orders, families, etc., according to some common relations or attributes.

Synonym : Taxonomy

Coding system

A correspondence between any list of items (a classification, a catalogue, a nomenclature etc. with a numeric list)

Data Dictionary

In an IT "information system", the data dictionary gives noun of a data, definition, possible reference to a given classification or nomenclature, possible corresponding code, conditions for usage etc.

Data model

In an IT "information system", the data model gives relations between the data

Dictionary

(Wiktionary)

A reference work with a list of words from one or more languages, normally ordered alphabetically and explaining each word's meaning and sometimes containing information on its etymology, usage, translations and other data. By extension, any work that has a list of material organized alphabetically.

Infostructure


__________________________________________________________________________________

48

Usually, the technical infrastructure supporting an information system. As this includes semantic tools and servers, Infostructure specifies, in eHealth, that the infrastructure includes the level above general technical systems as the Internet et other telecom and IT multi-sector tools.

Interoperability

(Semantic Health Report – January 2009)

Health system interoperability is the ability, facilitated by ICT applications and systems,

• to exchange, understand and act on citizens/patients and other health-related information and knowledge

• among linguistically and culturally disparate health

professionals, patients and other actors and organisations

• within and across health system juridictions in a

collaborative manner.

Metadata

Data that describes other data, serving as an informative label

Nomenclature

(Wiktionary)

A set of names or terms.

Synonym : Terminology

Ontology

(Adapted from Wikipedia)

In computer science and information science, an ontology formally represents knowledge as a set of concepts within a domain, using a shared vocabulary to denote the types, properties and interrelationships of those concepts.

Remark : an ontology corresponds to a model of a domain. It is a robust basis for classifications and nomenclatures, but it reflects existing knowledge. Relations can change as the model does, for instance in scientific knowledge in medicine.

Semantic interoperability

(Semantic Health Report – January 2009)

Semantic interoperability addresses issues of how to best facilitate the coding, transmission and use of meaning across seamless health services, between providers, patients, citizens and authorities, research and training. Its geographic scope ranges from local interoperability (within, e.g., hospitals or hospital networks) to regional, national and crossborder interoperability. The information transferred may be at the level of individual


__________________________________________________________________________________

49

patients, but also aggregated information for quality assurance, policy, remuneration, or research.

Taxonomy

(Wiktionary)

A classification; especially, a classification in a hierarchical system.

Synonym : classification (A distribution into groups, as classes, orders, families, etc., according to some common relations or attributes.)

Template

A model for documents, messages or other objects.

Terminology

(Wiktionary)

A set of names or terms.

Synonym : Nomenclature

Acronyms

(Annex 2 gives a list of main international semantic resources with their acronym)

ATC

Anatomical Therapeutic Chemical, WHO drug classification including INN

CDA

Clinical Document Architecture (HL7 V3 document standard)

CDISC

Clinical Data Interchange Standards Consortium – "develop and support global, platform-independent data standards that enable information system interoperability to improve medical research and related areas of healthcare"

CEF

Connecting Europe Facility – building the Digital Services Infrastructures, led by EU DG Connect

CEN

European Committee for Standardization – Comité Européen de Normalisation

eHN

eHealth Network – EU Member States voluntary network of national authorities responsible for eHealth


__________________________________________________________________________________

50

EMA

European Medicine Agency

EN 13606

Electronic Health Record Communication European Standard

HL7

Health Level 7 - non-profit organization involved in the development of international healthcare informatics interoperability standards

ICD (2 meanings)

• WHO International Classification of Diseases

• before adoption of INN, WHO drug classification was ICD (International Common Denomination)

IHTSDO

International Health Terminology Standard Development Organization

INN

WHO drugs International Nonproprietary Name

MTC

epSOS Master Translation/Transcoding Catalogue

MVC

epSOS Master Value Set Catalogue

SCG

Standing Coordination Group (eHGI proposal)

SDO

Standard Development Organization

SNOMED CT

Systematized Nomenclature of Human Medicine – Clinical Terms

UML

Unified Modeling Language

WHO

World Health Organization


__________________________________________________________________________________

51

Annex 2: Background and achievements

Semantic interoperability has been addressed through Research programs in Europe, by the EC and by Member States since long (see below). More recently, many countries have launched strategies associated with the development of information systems and especially of Electronic Health Records. They take into account international SDOs nomenclatures, terminologies, coding systems and tools (WHO's ICD, SNOMED (SNOMED CT and older versions), LOINC, HL7, IHE…) although often in different versions) as well as existing national ones.

One difficulty is already the abundance of semantic reference databases, while it is also a rich basis for future developments.

As eHGI first document to the e-Health Network stated: "Based on the foundations laid by recent research, Europe has a unique window of opportunity to make faster progress in this area because many of its health system stakeholders are very well aware of the cultural and linguistic challenges the new model of collaborative health care involves, but also of the opportunities arising to meet the growing needs for the sustainability of our health systems."

Basic semantic resources: a growing profusion

This paragraph is a reminder of the abundance of resources. There is however nothing to be afraid of: semantic interoperability in the health domain is a complex problem because the domain is huge and has multiple facets. It is also complex because of the richness and diversity of human languages, national and professional ones.

A myriad of languages and systems

This impressive abundance will continue to grow:

- Health definition is widening, to cover notably "external" causes (environment, personal way of life) or new perception of health (for example for psychiatry)

- Medicine in a wide sense, biology, genomics, chemistry, all related scientific domains are evolving at an accelerated rate (also due to IT tools usage in research and data analysis)

- There are many components in the healthcare sector, closely related, from types of treatments to associated equipment and to organisation of healthcare entities.

- There are diverse medical specialities, and also diverse professions, whose underlying model is different – notably doctors and nurses.

- Nomenclatures and codes are used for different purposes – although related -, clinical treatment, payment and reimbursement, public health, research; they also address different populations (and more and more are needed for general public, or researchers of other domains)

It has to be noted that the scientific character of medical activities provides theoretical basis for common data structures and dictionaries, classifications, terminologies, ontologies.


__________________________________________________________________________________

52

However, medical practice itself, healthcare organisation, legal framework, cultural differences introduce many differences between professions and situations and between countries. In fact, many different terminologies, classifications, nomenclatures and coding systems exist, which correspond to different history, needs and different usages.

Some are cited here below (main ones and typical examples).

Medical data international resources

Due to the importance of the problem, many international works have addressed the fundamentals of semantic interoperability, leading to high level languages, classifications, nomenclatures or even ontologies, concerning diseases and acts.

ICD9/ICD10 (International Classification of Disease)

The World Health Organisation's leading reference tool, including nuanced classifications of signs, symptoms, abnormal findings, complaints, social circumstances, and external causes. ICD is available in 42 languages. Origin of ICD and main uses are statistical analysis, epidemiology, reimbursement and resource allocation.

SNOMED CT (Systemised Nomenclature of Medicine-Clinical Terms)

(see Annex 1 and Annex

Following the SNOMED first version, SNOMED CT is oriented towards clinical reporting. It is a terminology based on an ontology (logical system of concepts). SNOMED CT is maintained, developed and distributed by the International Health Terminology Standards Development Organisation, IHTSDO. It is available in 4 languages, other translations being developed. SNOMED CT ambition is to crossmap to all other important terminologies and coding systems (for instance LOINC). IHTSDO has started working with WHO to crossmap ICD and SNOMED CT.

ICPC (International Classification of Primary Care)

Developed by the world organisation of general practitioners (WONCA), this classification addresses the needs of primary care, organising reasons for encounter, general problems of the patient, primary care activities to try and find diagnosis etc. ICPC has been translated in more than 20 languages.

LOINC (Logical Observation Identifiers Names and Codes)

LOINC is a database for identifying medical laboratory observations and allow for interchange of this observation between heterogeneous systems. It has been extended to more types of observations and measures (notably nursing). It is proposed or used as a standard in 150 countries.

Bioinformatics and Molecular Medicine and the influence of Web open solutions

GO (Gene Ontology) and the Open Biomedical Ontologies consortium

Initiated by the molecular biology community which developed the Gene Ontology, OBO

develops an effort for collaborative ontology development. The family of proposed methods


__________________________________________________________________________________

53

and languages are born in the Internet and semantic Web world and follow the trends of

openness and collaborative development.

DO (Disease Ontology)

Based on these principles, the Disease Ontology is "an open source ontology designed to link disparate datasets through disease concepts", providing "a computable structure of inheritable, environmental and infectious origins of human disease to facilitate the connection of genetic data, clinical data, and symptoms through the lens of human disease." Terms will be linked to existing terminologies (including SNOMED if possible through a licensed mechanism)

Nursing

ICNP (The International Classification for Nursing Practice)

…"provides a formal terminology for nursing practice and a framework into which existing vocabularies and classifications can be cross-mapped to enable comparison of nursing data".

The ICNP is being cross-mapped with existing nursing taxonomies, among them Nanda International (formerly the North American Nursing Diagnosis Association, now international, with partner organisations for different linguistic areas).

Medicinal products and devices

Other critical classifications and nomenclatures concern drugs and medical devices. Corresponding classifications can be related to the scientific definition or function of the product (while others (sometimes for the same elements) or refer to the industrial product identification. In many cases, such product identification systems are currently non significant from a medical information system point of view, being linked to the production process (typically the medical devices classifications). Moreover, they are nationally defined, as names and sometimes excipients presentation may vary.

ATC (Anatomical Therapeutic Chemical), including ICD/INN (International Common Denomination/International Non-proprietary Name)

ATC is the drug WHO classification, based on function and including INN, a fundamental sub level designing the chemical substance (before called ICD medicine). Each INN is a unique name that is globally recognized and is public property. A nonproprietary name is also known as a generic name.

IDMP (Identification of Medicinal Products)

ISO standard for data elements and structures for the unique identification and exchange of regulated medicinal product information.

GMDN (The Global Medical Device Nomenclature)

Supported internationally, the GMDN is a system of recognised coded descriptors in the format of preferred terms with definitions used to generically identify medical devices and related health care products. It is the only nomenclature system by which medical devices


__________________________________________________________________________________

54

can be officially identified within the European Economic Area (EEA) for the European Vigilance System (EUDAMED) and it is being endorsed by many of the worlds’ National legislators.

Libraries resources

They are often at the origin and the basis of clinical terminologies, and they attempt now to cross-reference the major ones.

UMLS (Unified Medical Language System)

Developed by the US National Library of Medicine. Its “Meta-thesaurus” includes the Medical Subject Headings (MeSH) terms. The UMLS identifiers, known as “Concept Unique Identifiers” (CUIs) and “Lexical Unique Identifiers” (LUIs) have become the de facto standard identifiers for medical concepts and terms.

The Web

Schema.org

An attempt by four US search engines (Bing, Google, Yahoo!, Yandex) to propose tags that developers can use to markup their pages in order to help users find them according to their content. Schema.org wants to accelerate the Semantic Web process. Due to importance of health and healthcare on the Web, the group develops a Health and Medical vocabulary to "help patients, physicians, and generally health-interested consumers find relevant health information via search." This vocabulary identifies entity types (such as signs and symptoms, risk factors, devices, anatomy, diet, cardiovascular, dentistry….) but does not create new nomenclatures, providing a way to annotate them with codes that refer to existing controlled medical vocabularies (MeSH, SNOMED, ICD, etc) when they are available."

Such an initiative must not be underestimated for future development of metadata – information about information that must be understood to access documents and information.

Dominance of national nomenclatures

National nomenclatures dominate in every country, most often developed for financial and administrative purposes. Some have been created locally; others use national adaptations of medical nomenclatures, especially ICD. Such classifications tend to evolve according to economics and organisation of the healthcare system, and they are often biased, even when compared to their original model. Examples are ICD9/CM in the US, Read/CTV-2 in the UK, CCAM in France…

A world of diverse nomenclatures and terminologies

No matter how strong will be the efforts to foster convergence between resources corresponding to the same object, multiplicity and diversity of semantic resources will continue, for all fundamental reasons explained above.

Moreover, cooperation and gateways are necessary between domains - for example between diseases and drugs or medical devices, between genetic information and treatment etc. In fact, multiple agreements exist between the developers and maintainers


__________________________________________________________________________________

55

of terminologies, nomenclatures and coding systems – notably when a resource needs to use SNOMED CT terms, as the business model of IHTSDO places severe constraints for usage.

From a practical point of view, nearly all documents interchange needs to refer to multiple resources (disease, treatment, drugs…).

The need for powerful tools

Semantic interoperability is not possible without a powerful infrastructure. As this is necessary inside a country as well as for cross-border exchanges, many tools and resources have been developed or are under construction.

Containers: documents or blocks of organised data

To be manageable, Information and data are organised in structured documents or messages. Either the data is transferred inside a known structure, either at least metadata are added which describes the contained information and allow looking for it. An important document organisation in health is HL7' Clinical Document Architecture (CDA), adopted as a standard by many countries and professional groups. Specific document types are created as CDA templates (for instance for Biology result). Another approach (not contradictory) is to organise information in an Archetype, according to the fundamental structure of set of data being described – for instance a blood pressure measurement. The OpenEHR consortium notably conducts this effort. OpenEHR produces also templates for given activities, as a particular surgery for instance.

Translation: where the data comes from and what it means exactly

In fact, it is possible to use different structures and standards, if a validated translation system exists. What is mandatory is to know which one has been used to create information. Diverse nomenclatures can be used for a given data, but it is necessary to know the precise version and who is responsible for the maintenance.

If this is possible, intermediary platforms will be able to translate. However the risks lie in situations where a same word is in fact used for different meanings, or covers a different perimeter in the [two] countries between which interchange is requested – one has to be sure that the items of the different nomenclatures have been mapped and the correspondences have been validated for all participating countries -.

Where are which resources

Every resource has to be maintained and needs management of versions, as all actors using it do not upgrade their systems simultaneously – and changes are numerous and fast. This is why a unique reference source is necessary. When receiving a document or a request, it is very important to localise the necessary resources. That is being done by special services in countries which develop semantic tools.

Access to resources

Common tools are necessary to access the semantic reference bases once identified. Standardisation organisations devise such tools. In health domain, HL7 has developed the


__________________________________________________________________________________

56

"Common terminology service" (CTS). CTS define an Application Programming Interface (API) that can be used by HL7 Version 3 software when accessing terminological content (independently of the architecture to access the services). HL7 and OMG (Object Management Group – responsible for general object modelling and access on the Internet) cooperate for a generic terminology service on the Web.

Multi-terminologies servers

These are complex and powerful tools in charge to translate information from a system to another one as well as possible. After checking precisely the necessary resources, they ensure the translation whenever possible between different terminologies and nomenclatures. In a multilingual environment, they have to refer to terminologies translations when available

A fundamental driver: EU directives, regulations and programs

The sustained effort on semantic resources

The first programs towards semantic interoperability were launched in 1992: GEHR (Good European Health Record), which laid the basis for a succession of works and the GALEN project (Generalised Architecture for Languages, Encyclopaedias and Nomenclatures in medicine).

In 2006, the DG Information Society and Media, bringing together Universities and experts as well as key ontologies, classifications, terminologies and nomenclature developers, launched the Semantic Health Project. In 2009, the Project published a report: "Semantic Interoperability for Better Health and Safer Healthcare - Deployment and Research Roadmap for Europe". In December 2011, the Semantic Health Network was launched, funded by the Seventh Framework Program for Research & Development.

Semantic interoperability related programs have evolved from Research projects towards eHealth policies for continuity of care, access to medical information, while research activities are still supported in a wide scale.

From the Open eHealth Initiative to the eHealth Network: the road to an European eHealth coordination and governance

• In 2007, Member States with the EU Commission created the Open eHealth Initiative, which is driven by Member States health administrations.

• In 2008, as part of the initiative, the Calliope Thematic Network project was launched, to create a European coordination network for eHealth interoperability implementation. Initiated by 17 health authorities and 10 organisations representing networks of physicians, community pharmacists, patients, industry and health insurers. It has represented a targeted effort aiming to establish an appropriately governed, composed and structured open forum, with the focal goal to support Member States to implement interoperable eHealth solutions, in close collaboration with the key stakeholders, including users, industry and payers.


__________________________________________________________________________________

57

• Simultaneously, in 2008, the Initiative launched the epSOS Large-scale pilot. Expanded in 2011, epSOS will end in June 2014.

epSOS approach has been to study and experiment practical exchange of precise documents, especially when a professional, authorised by a patient, will ask for information in another country. Two key documents were considered: patient summary and e-prescription.

In April 2012, the large-scale pilot entered into operational mode. In this phase, the developed solutions are tested in a real-life environment.

This practical approach by epSOS is a fundamental step.

• In 2008 also, the Commission presented a "Proposal for a directive of the European Parliament and of the Council on the application of patients' rights in cross-border healthcare"

• In 2011, the eHealth Governance Initiative project was launched, bringing together Member States at Political Governance levels and eHealth Stakeholders. The project is supported by a Joint Action under the EU Public Health Programme and a Thematic Network under the Competitiveness and Innovation Programme-ICT strand.

The eHGI program included the creation of a High Level eHealth Governance Group, comprising state secretaries and director generals from the national ministries of health.

• In March 2011, the directive on the application of patients’ rights in cross-border healthcare was adopted. It entered into application on October 25th 2013.

In its Article 14, the directive states: "The Union shall support and facilitate cooperation and the exchange of information among Member States working within a voluntary network connecting national authorities responsible for eHealth designated by the Member States ".

• In December 2011, the High Level eHealth Governance Group disappeared and was replaced by the eHealth Network.

The SemanticHealthNet consortium project

(http://www.semantichealthnet.eu/)

From 1/12/2011 to 30/11/2014, the SemanticHealthNet will develop "a scalable and sustainable pan-European organisational and governance process for the semantic interoperability of clinical and biomedical knowledge, to help ensure that EHR systems are optimised for patient care, public health and clinical research across healthcare systems and institutions."

The consortium, through a clinically driven work plan, exemplified in cardiovascular medicine, brings together experts in the medical semantics domains and studies relations with general technical infrastructure as well as with stakeholders as industry, public authorities, SDOs. Objective is to "define an infostructure for Semantic Interoperability Resource – comprising a knowledge management architecture and specifying concerned

http://www.semantichealthnet.eu/


__________________________________________________________________________________

58

systems and their requirements as well as certification processes. The main objective is however to define an European Virtual Organisation (Governance, mission, activities, budget…) together with a virtual environment to manage collections of semantic interoperability resources and a network of accreditation bodies."

As it was planned, SHN and eHGI have set up close relations. SHN results will allow defining what will be the relations between academic and theoretical work in the semantic domain and the general eHealth governance by the eHealth Network.

Directive 2011/24/EU

…of the European Parliament and of the Council of 9 March 2011 “on the application of patients’ rights in cross-border healthcare”.

Article 14 implies in general semantic interoperability but precisions are given on the first steps:

2. The objectives of the eHealth network shall be to:

…

(b) draw up guidelines on:

(i) a non-exhaustive list of data that are to be included in patients’ summaries and that can be shared

between health professionals to enable continuity of care and patient safety across borders; and

(ii) effective methods for enabling the use of medical information for public health and research.

The directive is also very clear on the necessity for mutual prescription recognition, even if it is not an e-prescription. In that case, information systems are still necessary, to allow for the pharmacist or the doctor to understand the prescription delivered by the country of affiliation of the patient. Therefore, the directive addresses specific domains of semantic: medicinal products and devices.

In the general definition of principles, Article 11 states that:

Where medicinal products are authorised within a Member State and have been prescribed in that

Member State by a member of a regulated health profession within the meaning of Directive

2005/36/EC for an individual named patient, it should, in principle, be possible for such prescriptions

to be medically recognised and for the medicinal products to be dispensed in another Member State

in which the medicinal products are authorised.

…

The implementation of the principle of recognition should be facilitated by the adoption of measures

necessary for safeguarding the safety of a patient, and avoiding the misuse or confusion of medicinal

products. These measures should include the adoption of a non-exhaustive list of elements to be

included in drug prescriptions. … The recognition of prescriptions should also apply for medical

devices that are legally placed on the market in the Member State where the device will be

dispensed.

… the Commission…shall adopt:

- c/ measures to facilitate the correct identification of medicinal products or medical devices

prescribed in one Member State and dispensed in another, …The Commission shall consider, inter

alia, using the International Non-proprietary Name and the dosage of medicinal products;


__________________________________________________________________________________

59

- d/ measures to facilitate the comprehensibility of the information to patients concerning the

prescription and the instructions included on the use of the product, including an indication of active

substance and dosage.

Here, semantic interoperability is directly addressed and includes comprehensibility of the transmitted information by the patient. This implies, for cross-border healthcare, a rigorous nomenclature that must be translated in national natural (not specialised) language for the patient – a problem of interface and natural language, which is addressed later in this document.

Calliope perspective

Calliope presented its conclusions in 2010, among which the eHealth Interoperability Roadmap. In Calliope vision, semantic interoperability is a key component of a global system. Objective is to provide common eHealth services, as Patient summaries, Electronic prescription, Chronic Care Management, relying on a foundation of "eHealth infostructure" whih itself depends on the ICT infrastructure foundation. The infostructure comprises notably Clinical terminologies and classifications and codifications, Data structures and value sets EHR, EMR…, Data and knowledge management tools, Data interoperability and accessibility, Data bases and Registries.

Calliope proposed fundamental recommendations for semantic interoperability (see under).

epSOS: Semantic services

(from http://www.epsos.eu)

"Semantic interoperability requires the following elements:

A common data structure of the three documents to be exchanged (Pivot Documents).

A commonly understood medical terminology based on value sets extracted from official code

systems used in these documents, namely the epSOS Master Value Sets Catalogue (epSOS MVC). The

content of the epSOS Master Value Sets Catalogue serves as a basis for translating language and

cross-references between different code systems among the Member States, resulting in the epSOS

Master Translation/ Transcoding Catalogue (epSOS MTC). The content of the epSOS MVC will also be

provided in an ontology (coded in OWL) to foster semantic interoperability.

A way of accessing and maintaining the content in the epSOS Master Translation/Transcoding

Catalogue, which is transparent for the user: the Terminology Access Services Interface.

All of these components are necessary to support the health professional in making competent and

informed decisions, fostering the continuity of care."

An objective for a growing number of Member States

A difficult evolution from a fragmented world

In 2012, the eHGI launched a survey on semantic interoperability in Member States. The results show a growing concern but also great difficulties.


__________________________________________________________________________________

60

A clear strategy towards semantic interoperability is either already official, either under definition in most countries, but all insist on barriers for adoption.

Most countries are developing Electronic Health Record systems, with diverse repartition between structured and unstructured information. Accordingly, there is a movement towards usage of standards, at least national.

From the survey, it appears that nearly all countries declare that they have rules and even obligations for coding, but this is a different matter than semantic interoperability and reference to international standards. Indeed, analysis shows that the most used coding system is a national one. In fact, these most used national coding systems correspond to payment and reimbursement mechanism (this will be commented later).

Fragmentation appears to be important inside countries and between countries and, however, the demand for new coding systems remains important. The survey shows that there is not, practically, a movement towards convergence of diverse classifications and terminologies.

A still limited number of countries has embarked in the creation of terminology centres and the development of terminology servers, but there are discussions in many others.

Many advances

Already, leading countries are addressing the diversity of terminologies. For instance, Belgium analyses the feasibility of a federal “terminology service” to deal with all terminologies and classifications used in the country through a federal CMV (Controlled Medical Vocabulary). In France, ASIP Santé – Agency for shared health information systems – has started building semantic interoperability services inside the general national Reference Frameworks Repository.

ASIP' services would include:

Multi-terminologies and value sets management server, with a web-service interface based on the “Common Terminology Services release 2” (CTS2) standard, providing services for the collaborative build, localization and maintenance of reference terminologies and value sets.

Terminologies and value sets query server, offering a CTS2 web-service interface for querying existing reference terminologies and value sets, as well as the mappings between these.

The eHGI survey on semantic interoperability, in 2012, showed that 6 countries had already invested in terminology servers.

Selection and adoption of terminologies and nomenclatures: an ongoing question

Many reports and many countries representatives insist on the necessity to gather around one terminology, SNOMED Clinical Terms. Moreover, multiple members of the eHealth Network, in the Dublin meeting have supported "the proposal in the report that the Commission should look at the possibilities of the SNOMED CT terminology and that covering the licencing costs is a strong first signal." After the discussion, "The Chairs agree on follow-up work needed on SNOMED CT, and to table a proposal during the meeting in November.


__________________________________________________________________________________

61

They agree that the membership fee is not the issue but EU financing would send out a clear message to national institutions."

Therefore, it is necessary to discuss the importance of SNOMED CT and the agenda to follow.

The research for one solution

SNOMED CT as a whole has a very ambitious objective as it aims to encompass all aspects of health, medicine and healthcare. In this vast domain, SNOMED CT wants to base firmly the terminology on a ontology for health. Consequently, the number of concepts and terms is already huge, and growing. Moreover, IHTSDO (International Health Terminology Standards Development Organisation), who manages the terminology, establishes relations and works on gateways with all other vocabularies and nomenclatures, specialised (LOINC in biology for instance) or general, as WHO's ICD.

SNOMED CT is certainly an important effort and a key component of the semantic interoperability landscape.

The question: is participation in SNOMED CT a priority, what is the practical impact for middle-range period and what are the real issues and difficulties.

It has to be reminded that, although SNOMED CT is the second semantic standard made available among Member States having answered the eHGI survey (after ICD10), it is not much installed inside information systems and it is very rarely used or known by Healthcare Professionals. Admittedly, they should not know how to use it and they should not have to code the information themselves, except for scientific research. Whatever they are or will be, advanced nomenclatures and coding systems have to be used with minimum effort by the professionals, through highly intelligent software interfaces (see a corresponding recommendation in 3.3). However, Professionals should know which tool is being used and trust it. Moreover, existence of diverse, better-known, often national, nomenclatures hampers the practical adoption of any new semantic tool.

SNOMED's limits

SNOMED CT usage is still being debated; barriers had been already described by the SemanticHEALTH project 10

SNOMED CT size is huge and growing (which generates heavy management and translation costs for national implementation plans)

One has to remember that licensing costs, even if paid by the EU budget, are just a small part of the implementing costs, which need to be covered by Member States.

The governance and economic model of IHTSDO are by some looked at as contradictory with a strong trend towards openness not only of public information but also of essential infostructures and standards (HL7 for instance has made all its standards free in 2013).

10 Semantic Interoperability Deployment and Research Roadmap - D 6.1- Barriers, approaches and research priorities for integrating biomedical ontologies


__________________________________________________________________________________

62

It has to be noted that – in the US notably – new important developments continue to flourish, with a modern approach: open and collaborative. This is the case with "Open, Biological and Biomedical Ontologies" (OBO), which aims to allow coexistence of multiple ontologies, including the Gene Ontology (See Disease Ontology statement about open and collaborative process - under Creative Commons license -, following OBO principles).

Such projects are developed in the scientific domain but they should constitute the basis for practical semantic interoperability. SNOMED CT development, started in US and expands in UK and other EU countries now but still closely controlled internationally through a costly licensing system. Albeit ambitious and aiming to be the general ontology of medicine, SNOMED CT, might be unsettled later, not being connected to new domains and not exploiting the international scientific and professional health community.

In any case, contribution to a common terminology is always mostly financed by public money, and the European open data policy should be followed, conforming to the principles of Directive on the re-use of public sector information (2003/98/CE, under revision in 2013). Moreover, the IHTSDO model excludes a really cooperative building of a common terminology, open to all contributors ie non members or licensees. As it was said during the eHN meeting in Dublin, the licensing of SNOMED CT by the Commission would send a clear message to national institutions. It would also reinforce the governance and the economic model of IHTSDO.

Perspective

SNOMED CT will certainly play an important role as a translator tool between diverse terminologies, in countries and at the European and international level. Precise usage and governance remain to be defined.

For SNOMED CT as for other key semantic tools, the eHealth Network will have to take decisions when it comes to intermediary translation tools for cross-border exchange of medical information. It should also foster convergence between countries whenever it is possible and to coordinate relations with SDOs and amongst them IHTSDO. To that end, the eHN should rely on a permanent technical coordination structure such as the Standing Coordination Group which himself will maintain the liaison with academic organisations (as today SemanticHealthNet) and national competence centres.

The eHealth Network, for its part, has to devise and apply a strategy towards practical development. As this will be presented later, the SHN will see to it that the future semantic cross-border infostructure is progressing in a coherent manner and it will follow the orientations devised by the SemanticHealthNet. However, the key objective is the real and practical development of eHealth on a wide scale in Europe. Then, SNOMED CT licensing is probably not presently the most critical point in the roadmap towards semantic interoperability. Before deciding about relations with IHTSDO and potential official SNOMED CT adoption, it should be clear what the use is, why it is needed, and if and when other coding systems are possible and useful, and how to link any choice made with the mandatory national reimbursement coding system.


__________________________________________________________________________________

63

Towards pivotal systems

Due to existing nomenclatures and languages being really used as well as diversity of needs and organisations, fact is that future developments of eHealth will need semantic interoperability and that this need will be fulfilled by intermediary, pivotal servers able to translate as well as possible a growing list of thesauri, nomenclatures, ontologies, from a specialty to another one, from a domain to another one, from a country to another one.

As many more language and domains are to be covered in the middle and long range, it will be necessary to rely on high-level tools able to define interactions between nomenclatures and concepts. Such an effort is already started, as proven by the works around the ISO standard ISO 25964-1 allowing exchange of thesauri data using RDF, SKOS and SKOS-XL ("Semantic Web" formal languages). One should not neglect the huge advances under development in the Semantic Web, from such high level developments to immediate, practical tools as those developed jointly by Google, Microsoft and Yahoo! European R&D programs should address this strategic domain, and development of eHealth semantic interoperability is a great opportunity also from this point of view.

Linking fundamental terminologies and nomenclatures with practical cases: eHN multiannual work plan

Article 14 of the 2011 directive on the application of patients’ rights in cross-border healthcare was devoted to the cooperation and the exchange of information among Member States working within a voluntary network connecting national authorities responsible for eHealth designated by the Member States. A key objective was to draw up guidelines (1.b.i) on a non-exhaustive list of data that are to be included in patients’ summaries and that can be shared between health professionals to enable continuity of care and patient safety across borders.

The creation of the eHealth Network by EU Member States followed rapidly and, as reminded above, three priorities of the eHealth Network concern semantic interoperability or are strongly dependant on it – Priority2, Adressing semantic and technical barriers to interoperability – Priority 4, Guidelines on patients summary set of data for cross border electronic exchange – Priority 5, Guidelines on interoperability of ePrescriptions.


__________________________________________________________________________________

64

Annex 3: About choice of medical nomenclatures

Reasons of Member States for choosing a semantic medical nomenclature and how they intend to use it: the case for SNOMED CT

This document is an eHGI Information paper. An associated document gives a description of choices and current status in each MS. It is available in eHGI documentation.

MAKING USE OF SNOMED CT:

KEY QUESTIONS and STATUS as of SEPTEMBER 2013

1. Introduction

This document aims at explaining in a synthetic way why certain Member States (MS) have decided

to use SNOMED CT (Clinical Terms) and what role that terminology plays in their overall strategy for

semantic interoperability.

The MS which have contributed to this paper are those that are currently members of the IHTSDO

(International health Terminology Standard Development Organization), the organization

responsible for producing SNOMED CT.

2. Rationale for using SNOMED-CT

A number of European MS, 13 (including Belgium), have -to date- decided to join the IHTSDO. Joining

the IHTSDO allows the MS to get direct access to the terminology and information about its use and

IHTSDO terminology resources, as well as being able to influence the development of the

terminology.

The rationale for deciding to join the organization and to make use of SNOMED CT is usually based

on one or all of the following considerations:

- The Electronic Health Record (EHR) and clinical information must be as much as possible the starting point for any medical data sharing. Re-creating data for multiple purposes must be avoided as much as possible. This demands structured data to support care and EHRs.

Narrative text is not adequate as a starting point; structured (including encoded) text is needed

for cross-border interoperability, data re-use and “processability”, decision support and

portability of citizen’s health data in an open eHealth environment.


__________________________________________________________________________________

65

- The role of standardized terminologies: the path to structured, encoded and interoperable health data needs support from a general-purpose terminology system with a specific set of fundamental features11

Although currently imperfect, there is no real alternative to SNOMED CT at this time if one

wants to provide a fine grained and multi-disciplinary care terminology which, despite primarily

being meant for clinical records and communications, is enabled to cover the needs of all

related sectors (administration, social care, research, secondary data use). SNOMED CT provides

a very comprehensive set of clinical concepts together with an ontology which organizes these

concepts. This set serves as a global conceptual resource for the SNOMED CT user community

but can be adapted by each MS provided that a few key principles are respected.

- The increasing complexity of patient care requires content in the EHR to be unambiguous and understandable to multiple care providers, to be comprehensive and to provide relevant and rich detail; this has not always been the case in paper records. The use of a national/international set of structured terms for fast (and legible) selection in care record keeping supports this aim and clinical specialties that have adopted SNOMED CT are working towards structured records with comprehensive structured (i.e. machine processable) data content.

- The objective is to effectively integrate appropriate semantic resources formally used in a given territory to address a variety of information requirements (clinical, financial, epidemiological, research etc.). IHTSDO has as an objective to achieve collectively, within the limit of resources made available by its members , the integration (mapping) of key available international semantic resources such as LOINC, ICD, ICPC -only to name a few- with SNOMED CT. The mapping with national legacy systems remains the responsibility of individual MS. Special interest and projects groups have been setup to support specific needs and harmonization work. This work is needed for transition purposes but in the long run convergence is highly desirable because of the amount of resources needed to obtain the integration.

The decision to use SNOMED CT does not thus exclude the use of other semantic resources but

it supports the idea that the codification of data needs to start from clinical concepts usable at

the point of care. It is then the role of national “terminology servers” (or translation engines) to

allow automated conversions between coding systems. In some countries this will be done with

decentralized terminology servers, which is a decision the terminology system allows.

- SNOMED CT is thus seen as an essential but partial element of a semantic interoperability strategy; this strategy ultimately aims at allowing clinicians to code data without ever seeing any code and preserving the quality and precision of the data. Physicians do not need to understand SNOMED CT, the EHR makes every day work more safe and efficient for staff. Professionals use SNOMED CT by recording their preferred clinical terms or expressions, allowing the real-time encoding of information at the point of care.

11 Should be concept-oriented and designed to be multilingual - Should have a methodology to adapt to upcoming needs of healthcare - Should be fit for left-side and also right-side semantic standardization - Should have the potential to be used in knowledge-based systems - Should have a strong potential to be an international standard


__________________________________________________________________________________

66

3. Preparing SNOMED-CT for national use

3.1: Full set, sub-set, use case set: Translating what and by whom?

One of the very first questions an MS has to answer when entering IHTSDO is to decide if it will use

SNOMED-CT as such or if it will decide to use only a part of it. In the latter case, the choice can be to

work on a use case approach, like Lithuania, or on a national refset approach, like Canada or

Belgium. One has then to first proceed with a selection of the SNOMED CT concepts which requires

adoption of a selection methodology, competent and trained human resources and the elaboration

of implementation or user guidelines. There is much availability of material, guidelines and

selections of refsets already made within IHTSDO or in one or more of its member countries which

can support this work through not having to start from scratch.

As English is not the official language of most EU MS, SNOMED CT needs to be translated in the

official MS language(s) of each MS. For MS which have several official languages, this burden of

translation is even more important although it can potentially be shared with other MS. An example

is Spain accessing the IHTSDO Spanish translation being able to do adjustments and review of this

into Spanish rather than translating all on its own.

Those choices have of course important implications.

- The decision to translate SNOMED CT as such is costly but guarantees a full international linguistic interoperability. A full translation provides a thorough quality review of the content but also allows for knowledge to be built within the country. The question of the “usability” of SNOMED CT “as such” remains however intact.

- The decision to build a broad national refset based on SNOMED CT which is translated in national language(s) is also costly as it requires important resources to deal with the selection of concepts. The arguments for such a decision are the following: o Not all SNOMED CT concepts are intended for use as data within the clinical record. Some

concepts only have navigational purposes for browsing or locating related concepts in the terminology.

o A smaller terminology with distinctive concepts is easier to use (more efficient for supporting fast and preferred choices in electronic clinical record-keeping) than a vast terminology with many rarely used quasi-synonyms.

o The costs of maintenance and translation are more limited. o The data model of the electronic medical record makes some concepts superfluous. E.g.

concepts with a time indication (old, first trimester,...) can be eliminated because the birth date, date encounter, start and stop date of clinical findings and procedures are already known and should not be covered by extra terms.

o A combination of primary concepts in one term is to be avoided because it leads to an explosion of concepts. Moreover, primary concept may need a separate registration because they have called for specific care paths and care takers.

o Cross-mapping between systems and International interoperability resources remains preserved as the whole SNOMED-CT concept set remains available. Imported SNOMED-CT codes not present in the national refset are however not (necessarily) available in national language.


__________________________________________________________________________________

67

After a review of SNOMED CT, certain concepts can be considered as obsolete, ambiguous,

unspecific, unusable or not adapted to national context. Consequently the national refset is

considered more efficient and responsive to use while preserving a high level of granularity. The

result of the work should ideally be validated by other countries contributing to a new

international SCT ‘subset’.

- The decision to select only the concepts needed to support a specific use case is of limited cost but is not compatible with a global “terminology sever” strategy aiming at integrating all semantic resources at national level. It will not provide full semantic interoperability but the ability of identifying European refsets like in epSOS or “professional specialty refsets” provides selected interoperability.

Member States also have different strategies when addressing the question of how to proceed with

translation: Using translators, using healthcare professionals or a mix of both. When a strategy of

pre-selection of concepts is chosen, use of healthcare professionals is usually the rule. There is a

recommended process of translation identified within IHTSDO which is clearly described in

documentation.

3.2 Pre and Post-coordination of concepts:

When deciding a strategy, the link between the SNOMED CT thesaurus and other cross-mapped

thesauri or databanks such as the drugs database and the possibility to associate/combine SNOMED

concepts need also to be considered as it can lead to an important reduction of distinct concepts.

Pre-coordinated concepts are the de facto standard in most electronic health records. Therefore,

recorded concepts tend to be very much pre-coordinated at present.

However, the combination of primary concepts and their attributes duplicate pre-coordinated

terms. This duplication in encoded representation may cause difficulties in interpreting content as

equivalent in certain conditions.

Typical instances of combined concepts are all the axis modifiers, such as ‘absence of’ ‘family history

of’, etc.

Encoding concepts with their causing agent (biological, chemical, physical) is another source of a

never-ending number of combined concepts.

Furthermore, some attributes can be coded in an alternative coding system to SNOMED CT. E.g.

when the medication is coded in ATC, post-coordination of adverse events should be post-

coordinated with ATC to avoid redundant coding.

Such a choice of “post-coordinated concepts” however relies on the willingness of the software

industry to implement a certain number of functional specifications and rules. A users’ ability to

post-coordinate clinical terms (i.e. construct multi-faceted encoded statements) in care records thus


__________________________________________________________________________________

68

depends on support from IT developers. This is thus at best a mid-term strategy and in the short

term it is usually accepted to keep more pre-coordinated concepts than ultimately needed.

4. Dealing with National Legacy Systems, Standards and International standards development organizations.

The development of a national reference terminology is critical in order to support the overall “one

entry- multiple use”12 policy. As mentioned above, using SNOMED CT as a cross-mapping reference

terminology is one of the reasons for opting for it. The standardization work at national level has

often been developed in an ad hoc, silo based and incremental way. The political decision to align

standards and legacy systems is thus essential but is a time and resource consuming process.

For some legacy systems, the mapping exercise is pretty limited but even in this case, it requires a

very close collaboration with all users in order to come to an agreement on a “convergence”

calendar. It is however important to start at one point in time and to immediately align any new

development.

One has in particular to revisit and map all the concepts embedded in standard messages and

registries. Mapping with systems used for reimbursement is particularly challenging but needs to be

considered as early as possible as it usually requires a lot of human input and is of critical

importance for users and therefore critical for the successful introduction of SNOMED CT. This is all

depending on how far along the implementation of EHR a country is, ie starting from “scratch” or

making changes to an existing platform.

In order to map between systems and to allow an integrated maintenance of both concepts and

rules, the investment in a TERMINOLOGY MANAGEMENT SYSTEM13 is a must. The market is today

quickly evolving and available products are equipped with enriched functionalities such as fuzzy

mapping which makes possible (pre)automated conversion support. Although the added value of

such a tool looks obvious, the decision to make the investment is not easy to take given its relatively

high cost and the fact that it remains for now a “niche” market. It’s a long term investment which is

not always easy to make but one which will provide large benefits in the end.

5. User Interfaces (Run time components)

Investing in a national reference terminology and creating the conditions for its national

maintenance is an absolute prerequisite but is far from sufficient to make it “usable” by final users.

12 A single entry allows registration in multiple systems intended for different uses.

13 IHTSDO has developed a system both for maintenance of the core (international) terminology and for use by MSs who want to translate, maintain and distribute the terminology. However available functionalities do not always match with MS needs.


__________________________________________________________________________________

69

The goal is to have healthcare professionals coding without having the feeling that they are forced to

assign codes: in a clinical setting, the matching between a lexical expression in a given language and

the concept in the national reference terminology needs thus to be as fast and intuitive as possible.

In order to achieve this, state of the art code conversion, syntactic and semantic functionalities need

to be embedded in the EHR. Given the high fragmentation of the EHR market, the level of

investment required, and the language dependency, it is probably not realistic to expect that each

EHR provider will be able to deliver those functionalities at an acceptable level of price and

performance. Public authorities could leave the development and implementation of terminology

interfaces entirely to the industry. However the Health Authorities could test key providers of

semantic technology making results public (pre-procurement public assessment). They could

eventually even negotiate conditions in case of a contract for a larger group of users.

In many instances, providers of terminology management systems do also provide run time

components.

In a highly connected country, direct communication of the EHR with a national terminology server

or a network of collaborative terminology services is probably the best solution to guarantee an

optimal universal service and, at the same time, to allow an interactive maintenance of the

resources with the input of the final users. Observing end-user patterns of terminology use and

identifying the most frequent combinations and trends of terminology search and selection is a

source of potential economies of scale and usability enhancements.

6. Testing and validation

Before putting the SNOMED CT based national refset to life, preliminary tests in different

environments are necessary in order to guarantee sufficient robustness and its usability. As

mentioned above, this also requires specific agreements with the industry.

However the work and expertise present in the several MS implementing SNOMED CT can also be

used to support validation work and/or diminish costs:

- Cross-comparison of methodologies used for selection of SNOMED CT between MS (provided that scope is identical)

- Cross-validation of translated set or sub-set in one specific language or reuse of those translations (provided that SNOMED translation rules have been applied). One example is epSOS.

Furthermore, exchange of information on quality, performance and price of existing tools

(terminology management system, run time components) would also be highly beneficial. Common

tender could also have a significant impact on costs.


__________________________________________________________________________________

70

7. Costs

Aside from the “entry” and license costs to IHTSDO (to cover joint priorities and work such as cross-

mappings to other standards and maintenance and update of the international core of SNOMED CT)

which is calculated according to national GDP, the development of a national terminology server

requires substantial additional resources.

Here is a non-exhaustive list of those associated costs:

- Selection and validation of concepts: qualified healthcare professionals

- Translation of concepts: translators and/or qualified healthcare professionals

- Participation in IHSTDO Special Interest Groups and projects: Human resources

- Permanent coordination structure: “national terminology (release) center”: Qualified Human resources

- Mapping with other (legacy) systems: terminologists

- Terminology management system: licence/usage costs

- Terminology server infrastructure costs

- Run time components: licence/usage costs

- Testing costs

8. Market Development

The adoption of an international terminology has a number of benefits both to the ability of EU

healthcare vendors to compete in other markets, but also to improve the quality of healthcares

systems in the EU as it reduces the burden of development through reducing the different country

specific approaches. The adoption of a single international terminology will enable:

Specialist niche suppliers to emerge that can lever the terminology and its modeling to improve current functionality

Suppliers to deliver products across the EU and beyond as they can develop to a single terminology and coding system

Reduced cost of healthcare systems in the long terms as maintenance of the terminology, subsets, query and report specifications, business rules for decision support and medical alerts etc is not hospital/country specific

Improvement of vendor healthcare systems as development can focus on functionality and less on terminology and mapping maintenance


__________________________________________________________________________________

71

Annex 4: References

Many informations come from documents and pages of Web sites of the European Commission, of Standard Development Organisations, of European Programmes as eHealthNet, epSOS, of international Organisations as WHO and IHTSDO, of all organisations and projects which are listed in Annex 2. Hereafter are basic milestones.

SemanticHEALTH Semantic Interoperability Deployment and Research Roadmap December 31, 2007 D 1.1 Conceptual Framework for eHealth Interoperability D 3.1 Comparative Analysis and Initial Socio-Economic Recommendations for Improving Semantic Interoperability D 7.1 Draft Semantic Interoperability Deployment and Research Roadmap

SemanticHEALTH Report January 2009 Semantic Interoperability for Better Health and Safer Healthcare RESEARCH AND DEPLOYMENT ROADMAP FOR EUROPE

epSOS Semantic Services Definition D3.5.2 May 31st, 2010

Calliope EU eHealth Interoperability Roadmap - December 2010

DIRECTIVE 2011/62/EU OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 8 June 2011 amending Directive 2001/83/EC on the Community code relating to medicinal products for human use, as regards the prevention of the entry into the legal supply chain of falsified medicinal products

DIRECTIVE 2011/24/EU OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL of 9 March 2011 on the application of patients’ rights in cross-border healthcare

COMMUNICATION FROM THE COMMISSION TO THE EUROPEAN PARLIAMENT, THE COUNCIL, THE EUROPEAN ECONOMIC AND SOCIAL COMMITTEE AND THE COMMITTEE OF THE REGIONS eHealth Action Plan 2012-2020 - Innovative healthcare for the 21st century

epSOS : Semantic Sustainability- 12 September 2012

POLICY PAPER ON SEMANTIC AND TECHNICAL INTEROPERABILITY Proposed by the eHealth Governance Initiative - 18th April 2013

3rd eHealth Network 14th May 2013 Cover Note by the Secretariat Topic 3 - Report on progress

DELIVERABLE - eHGI Deliverable TN D7.3 (formerD2.3) Draft... · COVER PAGE Project Acronym ... 2012...

Documents

Transcript of DELIVERABLE - eHGI Deliverable TN D7.3 (formerD2.3) Draft... · COVER PAGE Project Acronym ... 2012...