Current EHR Developments: an Australian and International Perspective - Part 2

Click here for Part 1 of this paper

• GEHR and openEHR
• EHR Interoperability
• International EHR Standards
  • ISO/TC 215
  • CEN/TC 251
  • HL7 Inc.
• The Way Forward

GEHR and openEHR
The Good European Health Record (GEHR) project was a large European Union (EU)-funded project undertaken within eight European countries between 1992 and 1994. It was the first international research project to develop a specification for a "good" quality EHR based on detailed clinical, technical and medico-legal requirements. It was a landmark project and produced over 2,000 pages of public domain documentation and a formal object-oriented model. Unfortunately, it proved to be very difficult to implement for a variety of technical reasons but the requirements work still stands alone and was the major input into the first ISO EHR standard on requirements for an EHR Architecture.^{[ 7 ]}

In 1997, a small Australian group which included the co-founder of the GEHR project (Dr Sam Heard) and the senior technical architect who wrote the GEHR model (Thomas Beale), joined with the GEHR project leader, University College London (UCL), to overcome the implementation problems of GEHR. Further research and implementation trials were undertaken independently by UCL and by Ocean Informatics, a small Australian company formed to undertake this work. The UCL team undertook their work within two additional EU research projects called Synapses and SynEx. This resulted in an implementation solution based on federated databases and object dictionaries and a live clinical system based on this approach is still operating in the UCL Whittington hospital in London. Ocean Informatics developed a formal two-level modelling approach, including the breakthrough archetype methodology, to overcome the previous implementation problems. This new model was named the Good Electronic Health Record (sometimes referred to as "Oz" GEHR) and was tested in a series of four trials funded by the Australian Commonwealth Department of Health and Ageing (DoHA) and the General Practice Computing Group (GPCG).

In 2002 a decision was made to merge the work of UCL and Ocean to form a new model which was called openEHR. Further enhancements to the openEHR model occurred rapidly, both by the original developers and also from a variety of other contributors in Europe and the US via the newly formed openEHR Foundation. The Foundation was formed jointly by UCL and Ocean Informatics as an independent non-profit organisation. Its main roles are to promote the uptake of openEHR technologies globally; to maintain the openEHR specifications and control the change management process for the openEHR model; to protect the copyright of open source software components based on openEHR; and to act as a forum for discussion and contribution on openEHR and related technologies. The Foundation now has a membership of over 400 from 50 different countries, comprising clinicians, software engineers, academics and health administrators.

openEHR is currently undergoing its first major technical and clinical trial as part of the Australian HealthConnect trial projects. A preliminary requirements and design phase of this trial was undertaken in 2002-03 and the development of the software for the clinical trial in Brisbane commenced in early 2004. The trial is focussed on diabetes shared care involving two Divisions of General Practice, two hospitals and a range of community-based allied health professionals. The trial aims to test both the clinical usability of openEHR-based systems and also the technical suitability and scalability of openEHR as the underlying EHR architecture for HealthConnect. It should be noted that openEHR is not intended to be a standard in itself but rather an input to national and international standards such as CEN 13606.

Two other important openEHR projects underway at present are the DoHA/GPCG-funded development of an open source Archetype Editor by Ocean Informatics and a multi-partner US-based project to evaluate archetype interoperability. The Archetype Editor is based on the new emerging standard archetype language called Archetype Definition Language (ADL) which has been assigned to the openEHR Foundation. The Editor is an essential software tool for building clinical, demographic and other archetypes necessary to structure and validate the content of an EHR.

A detailed discussion of archetypes is beyond the scope of this paper but in simple terms, archetypes are models of clinical or other domain-specific concepts. They define the business rules (constraints) for valid values of a concept and they use terminology to identify components within the archetype. Archetypes may define simple concepts such as "blood pressure" or "address" or more complex concepts such as "biochemistry results" or "family history". Archetypes are an important component for semantic interoperability which is necessary for value-added EHR applications such as intelligent decision support and care planning. They are the key to ensuring future-proof EHRs since they are technology independent. They also enable future-proof EHR systems since the EHR software now contains only generic information structures (which change very infrequently). All clinical and other domain-specific knowledge (which changes frequently) now resides outside the software in archetypes which take the form of ADL or XML documents.

The other current archetype project involves the Mayo Clinic, UCSF (University of California San Francisco), the HL7 Templates SIG and Ocean Informatics. It aims to evaluate various aspects of ADL including its efficacy and the need for another new knowledge representation language. Existing alternatives being evaluated are OWL (Web Ontology Language) footnote and XML schema. Interoperability of archetypes between openEHR and HL7 CDA will also be tested. It should be noted that whilst archetypes were originally developed for the GEHR/openEHR models, they can also be developed for other Reference Models such as CDA and CEN 13606. This is extremely important for harmonisation and broad interoperability between different standards. This project is still in progress but early results are very positive. Dr Peter Elkin of Mayo Clinic has demonstrated interoperability between ADL and OWL (and therefore between archetypes and ontologies/terminologies) and has developed a meta-language called HEAL (Health Expression Archetype Language) which links ADL and OWL. This is a potentially very important development in the field of knowledge representation and interoperability.

EHR Interoperability
In order to support sharing and exchange of EHR information within EHRs, the main focus of EHR standardisation needs to be on interoperability. Functional interoperability is the ability of two or more systems to exchange information so that it is human readable by the receiver. This is the type of interoperability made possible through the use of HL7 V2.x messaging. However, to maximise the utility of shared information and to underpin applications like intelligent decision support, a higher level of interoperability is required. This is called semantic interoperability which is defined as the ability of information shared by systems to be understood at the level of formally defined domain concepts so that information is computer processable by the receiving system. Semantic interoperability is a multi-level concept with the degree of semantic interoperability dependent on the level of agreement on terminology and the content of archetypes and templates used by the sending and receiving systems.

At present, almost all EHRs are based on proprietary information models within EHR systems, with little or no interoperability between EHR systems and little or no ability to share EHR information beyond the immediate boundary of a single health organisation. In fact, it is often impossible to share EHR information between different disciplines within a single organisation (eg, between doctors and nurses) or between different applications within a single clinical information system (eg, a decision support or care planning application is unable to access the EHR which is tightly bound to the "EHR application").

One of the key requirements for shareability of the EHR is to break the nexus between the EHR and the EHR system - ie, the EHR should conform to an information model independent of both the physical database schema used for local storage and the applications which create, maintain and retrieve EHRs. This EHR information model should be independent of any particular implementation technology - ie, it should be a logical information model. Technology independence is also essential to make the EHR "future proof" to enable the possibility of lifetime EHRs.

In order to achieve semantic interoperability of EHR information, there are four prerequisites, with the first two of these also being required for functional interoperability:

1. a standardised EHR reference model - the semantics of EHR information structures
2. standardised service interface models - the semantics of interfaces between the EHR service and other services such as demographics, terminology, access control and security services
3. a standardised set of domain-specific concept models - archetypes and templates for clinical, demographic and other domain-specific concepts
4. standardised terminologies - the language of health which underpins the archetypes.

International EHR Standards
There are three main standards bodies currently active in international standards directly related to the EHR. These are ISO (International Standards Organisation), CEN (Committee European Normalisation - the European Standards Organisation) and HL7 (Health Level 7).

As with the definition of the EHR, there has been some confusion and disagreement about what constitutes the scope of EHR standards. Some take the view that EHR standards include standards for all of the "building blocks" for the EHR - ie, EHR structure, terminology, messaging, security, privacy, etc. This is not really a very useful classification since it essentially includes the whole content of health informatics. A more restricted view, which was adopted by the ISO/TC 215 Ad Hoc Group on Standards Requirements for the Electronic Health Record,^{[ 8 ]} is that EHR standards should be limited to the structure and function of the record per se and systems that process the record (ie, EHR systems). Using this view, EHR standards can be classified into several different groups:

• "Core" EHR standards - includes standards for the structure and content of EHRs and the functionality and interoperability of EHR systems
• Standards for EHR-related services - examples include demographic standards such as patient and provider identification, EHR access control and interface standards between the EHR and other services
• EHR standards for specific purposes - includes standards for EHRs related to specific technologies (eg, health cards, mobile computing); standards for the EHR in particular health sectors (eg, hospitals, primary care); and standards for the EHR for particular disciplines or stakeholders (eg, personal/consumer health records). A number of standards have already been developed in this category but these types of standards should be avoided wherever possible since there is a high risk of incompatibility with more generic EHR standards
• EHR meta-standards - these are high level standards such as the ISO Health Indicators Conceptual Framework.

ISO/TC 215
ISO/TC 215 is the peak international standards body for EHR and other health informatics standards. However, it is a relative newcomer to health informatics standards, having been established only five years ago. Some of the standards developed by TC 215 are produced de novo (eg, ISO/TS 18308 "Requirements for an EHR Reference Architecture") within the TC 215 working groups, but many others use existing standards from other national and international standards organisations as at least a starting point for an ISO standard. Examples of such organisations are IEEE, CEN, HL7, DICOM and Standards Australia.

In addition to the EHR Definition, Scope and Context Technical Report already discussed in this paper, there are a number of EHR standards, Technical Specifications and Technical Reports currently under development within TC 215. Examples covering each of the groups listed above are:

• Architectural Requirements for EHR Systems - this is a core EHR standard in early stage development as a de novo ISO standard
• Identification of Subjects of Care - This is an EHR-related service (demographic) standard which is being developed based on the Australian Health Client Identification standard
• Patient Healthcard Data: Limited Clinical Data - this is a specific technology standard which is one of eight health card standards which are being adapted by ISO from their CEN equivalents
• Framework for Emergency Data Sets - this is an EHR meta-standard being developed de novo within TC 215.

CEN/TC 251
CEN is the peak European standards organisation which transcends the national standards organisations of its member countries. It now has a membership of 28 countries which comprise all of the 25 EU states plus three other member countries which are not part of the EU (Norway, Switzerland and Iceland).

CEN currently has two main EHR standards under development which are both major revisions of previously published pre-standards^{[ a ]}. The first is CEN 13606 "Electronic Health Record Communication" which was originally published in 1999. The pre-standard had limited uptake due mainly to difficulties with implementation inherent in its single-level modelling approach but despite this, it is still the only comprehensive EHR interoperability standard in the world. In November 2001, a decision was taken by CEN to update 13606 and to adopt the openEHR archetype methodology. A Memorandum of Understanding was signed between CEN and the openEHR Foundation to enable the Australian members of openEHR to participate in the revision project.

CEN EN13606 will be a five-part standard consisting of the Reference Model, Archetype Interchange Specification, Reference Archetypes and Term Lists, Security Features and Exchange Models. The development is running behind its original target but is expected to be completed and balloted in the first half of 2005 when it will become a full de jure European standard. Australia is strongly involved in the development of this standard and a decision has been made to adopt 13606 as an Australian standard upon its completion. CEN has also foreshadowed the introduction of 13606 into ISO/TC 215 as the basis for the international EHR interoperability standard and this should be achieved by 2006.

The other main CEN EHR standard currently in development is CEN 12967 "Health Informatics Service Architecture" (HISA) which is a major revision of the earlier pre-standard titled "Health Information Systems Architecture" published in 1998. HISA is a high-level service-based architecture which is compatible with and "sits above" CEN 13606 and similar lower-level standards such as HL7 CDA. The revised standard has three parts which correspond to the first three viewpoints of the ISO 10746 RM-ODP (Reference Model for Open Distributed Processing) standard - Enterprise viewpoint, Information viewpoint and Computational viewpoint. The HISA standard will provide a reference model for health care IT services, facilitating the building and purchasing of interoperable systems.

HL7 Inc.
HL7 has in recent years become the pre-eminent US health informatics standards organisation. It was originally concerned only with messaging standards but has more recently become involved in standardisation of decision support and terminology. In late 2001 it announced its intention to also embrace the EHR and formed an EHR SIG (Special Interest Group) which has attracted considerable interest and involvement from within the HL7 community.

CEN and HL7 have signed an MOU (memorandum of understanding) to further co-operation between the two organisations, with a particular emphasis on harmonisation and, where possible, convergence. This is extremely important to future interoperability of health information systems between countries since CEN and HL7 are the two main standards organisations actually developing these standards. Current areas of harmonisation include data types, the HL7 CDA and CEN 13606 Reference Models and CEN/openEHR archetypes with HL7 Templates. Many members of CEN/TC 251 participate actively in HL7 developments and the Chair of HL7 is a member of the CEN 13606 revision Taskforce. ISO/TC 215 is also a forum for CEN-HL7 harmonisation and there are several such projects now underway.

The first EHR standard produced by HL7 is the EHR System Functional Model and Specification Draft Standard for Trial Use which has already been discussed. The HL7 Clinical Document Architecture (CDA) is not strictly an EHR standard since the CDA is not an EHR. However, the CDA forms an important sub-component of an EHR which has already been harmonised with the equivalent structures in CEN 13606 and openEHR. CDA is a subset of the CEN 13606 Reference Model and 13606 will be compliant with CDA Release 2. CEN 13606 is also not a complete EHR specification but rather, an EHR Extract for communication. It is therefore a subset of the openEHR model which contains the full semantics for a complete EHR. The other area of HL7 which is relevant to the EHR is the Templates SIG which is working with openEHR and CEN representatives on archetypes and templates but there are as yet no HL7 standards in this area.

The Way Forward
There has been a marked increase in interest and activity in the field of EHRs around the world in the past three to four years and it appears that this activity will accelerate significantly over the next five years. The drivers for adoption of EHRs are many but the main ones can be summarised as the need to improve the quality and safety of patient care whilst at the same time controlling and where possible, reducing health system costs.

There are also many barriers and difficulties to the widespread adoption of EHRs. The technical challenges inherent in building secure, robust and high quality EHR systems should not be underestimated and the lack of interoperability and other EHR standards until very recently has impeded progress. However, these difficulties are heavily outweighed by many human factors which have thus far not been adequately addressed. Perhaps the single most important of these is the development of consumer acceptance of the benefits of EHRs and trust that their personal health information held in EHRs will be private and secure. This requires the development of comprehensive but comprehensible policies, codes and legislation for privacy, consent and access to the EHR. Implementation of these policies and codes in usable software is a non-trivial issue but will be far easier than the non-technical challenges.

The second major area of non-technical challenge is the engagement, education and training of clinicians who are the main users of EHRs and EHR systems (although patients/consumers will hopefully also have access and input to their EHRs in the future). Broad clinician involvement in the design of end-user clinical applications is essential for success but vendors must also be convinced of the need to build more user-friendly and flexible systems. There is no doubt that EHR systems may significantly alter clinician work practices but changes need to be incremental and individual clinician workflow should be left as flexible as possible to accommodate legitimate variation in practice - we should never forget that medicine is an art as well as a science. This is not to belittle the need for evidence-based practice and the tremendous potential power of computerised applications like intelligent decision support and care planning. However, all too often in the past, vendors have built large and complex EHR and other clinical information systems which require significant and sudden changes to workflow patterns for no good reason other than "this is the way the software does it". These systems are usually based on the limited design input of one or a few technically-oriented doctors or nurses who often seem to lose sight of the realities of day-to-day practice once they join a design team.

The advent of archetypes and templates should play a major role in empowering clinicians through allowing them to directly specify the clinical structure and content of EHR systems. Archetypes will also help to improve the quality of EHR data through in-built data validation and can potentially help to improve the flexibility of data input and presentation in EHR applications. They will certainly make EHR systems more responsive to change as new clinical concepts become available or old ones need modification.

There are a number of big-ticket tasks for governments and standards organisations at a national level to lay a firm foundation for a ubiquitous EHR environment. These include:

• Development of a national health infostructure based on (harmonised) international health informatics standards
• Establishment of national governance structures for EHRs
• Introduction of national privacy legislation and codes
• Development of national repositories of archetypes and local repositories of templates
• Adoption of appropriate clinical terminologies at a national level.

There is much to be done but there are encouraging signs that major stakeholders are beginning to recognise that the very future of health systems depends on more efficient and effective information management. The EHR is arguably the most important foundation component in this pursuit.