The SemanticHEALTH report available at http://ec.europa.eu/information_society/activities/health/docs/publications/2009/2009semantic-health-report.pdf Gives a very critical but balanced overview of the current state of advance in Semantic interoperability.
This slide is from Nick Booth of BT Health and shows his view that computable clinical content can only be achieved by the marriage of terminologies such as Snomed with generic information models such as openEHR. In fact, such information models can be used to help control some of the more difficult aspects of Snomed such as post-coordination, where new Snomed concepts can effectively be created by joining together, existing concepts such as Left + joint replacement + Hip = Left Total Hip replacement
Barriers to sharing IAdapting EHR to cope with rapid changes in clinical requirements and varied clinical viewpoints
• Database design• Clinical software objects• Clinical content of Messages
Gathering and formalising computableclinical knowledge
• To inform application design• To define message content• To enable complex secondary uses analysis• To drive decision support and workflow
Presenter
Presentation Notes
There are many barriers to interoperability and some of the key high-level issues are being gradually solved by initiatives like IHE and CDA. Nevertheless, if we are to make further progress, we must resolve these 2 major problems.
The Clinical knowledge required to construct EHR systems is variably captured and expressed. Some of this is computable but is often locked away within vendor systems or inextricably locked in to particular message implementations.
This slide is from Nick Booth of BT Health and shows his view that computable clinical content can only be achieved by the marriage of terminologies such as Snomed with generic information models such as HL7 or openEHR. In fact, such information models can be used to help control some of the more difficult aspects of Snomed such as post-coordination, where new Snomed concepts can effectively be created by joining together, existing concepts such as Left + joint replacement + Hip = Left Total Hip replacement
Grew out of academic EHR projects e.g. GEHRNon-profit organisation based at UCL
Established by UCL and Ocean Informatics in 2000 to own the IP800+ Members from 71 countriesAll specifications & schemas publicly availableSoftware open source (GPL, LGPL, MPL)
Capture and representation of detailed computable clinical content
The clinical side of openEHR is primarily concerned with the definition of fine-grained clinical content, using archetypes, templates and termsets. The use of archetypes allows the key separation of clinical content from the more technical aspects of the information model – held in the lower ‘Reference layer’. This makes it much easier for clinicians to engage in the design and definition of computable clinical content.
Although the archetypes are distinct from the Reference model layer, they draw on the basic data types e.g. coded_text, Data/times/ quantities and re-integrate these into the Information model via traditional classes.
A peak expiratory flow rate displayed in the Ocean open-source Archetype Editor. This is understandable by clinicians but allows very rich modelling of data points. In this case the Expected peak expiratory flow rate is defined as a quantity ,as a volum
Although templates may be helpful in defining the data content of a data-entry screen, they do not define the appearance and layout of such screens as might be done by a GP system forms designer.
openEHR templates are used in several different ways: To aggregate together several archetypes for a particular use e.g to define data-entry requirements, content of a discharge letter or lab message. To ‘constrain the Maximal dataset approach of individual archetypes to reflect a minimal dataset for local or national use. To bind terminologies to the information model in an appropriate, safe and Agreed fashion. This approach allows relatively few archetypes ? 1000’s to be re-used within a large variety of clinical settings while maintaining a high degree of interoperability but allowing for controlled local variability .
The aim is to create an ‘ecosystem’ of reuseable EHR clinical components which are further constrained by templates used as message definitions, data-entry definitions or to produce code skeletons.
This is a fragment of a CDA discharge summary created by transforming an openEHR template built from archetypes. Although such interoperability transforms are not new, the difference here is that the transform logic can be defined per-archetype, as XSLT or code, and then re-used for every different message. It is also possible to do the same transform in the opposite direction. Similar techniques are also useful for integrating legacy e/g labs output with openEHR-enabled systems.
Barriers to sharing IIGaining consensus from a wide range of clinical groups across…
Professional boundariesOrganisational boundariesGeographical locationsDiffering disciplines
Aligning working methods from a wide range of clinical groups resolving…
TrainingProfessional issuesLegal issues
Presenter
Presentation Notes
There are many barriers to interoperability but the key high-level issues are being gradually solved by initiatives like IHE and CDA. Nevertheless, if we are to make further progress, we must resolve these 2 major diffuiculties
“Interoperability is of some importance … but getting systems right for individual patients is arguably more so”
eHealth Insider
“It must be kept in mind that semantic interoperability implementation also depends on social, cultural and human factors within each organisation, region and country, each system and each time period.”
EU Semantic Health report
Presenter
Presentation Notes
This is a recent quote from a GP computer system user in England. It is important to bear in mind that most requirements for interoperability are local, to support workflow and the healthcare process. Any drive towards widespread interoperability should be tempered by this reality, given that the process is known to be extremely difficult.
What are the causes of clinical diversity, which is a significant barrier to interoperability? Clinical ego and resistance to change is, of course, a significant issue. Senior clinicians continue to hold considerable power within their organisations and without their agreement, any significant progress in IT use can be very difficult, but this is often blamed for other more fundamental issues. Local innovation and research is a potent cause of diversity. It may not always be appropriate or successful but many of the major improvements in patient care, both in clinical management and improved workflow, have come from such local initiative. One of the significant drivers for the GCS project was the increasing numbers of departmental systems using MS Access or equivalents, arising from research activity but now running significant aspects of service delivery. Differences in organisational constraints play a role. Although front-line practitioners e.g in a social work/community health situation, can readily agree common working practices and data standards, there are often higher-level data or legal requirements that pull them in different directions. E.g data capture for audit or for legal reasons around child safety. Alan Hassey’s talk on clinical governance in shared record is of relevance, where it is very easy for different practitioners to ‘trip-over’ each others workflow data.
Operational constraintsHuman resources – training, staff contractsThe health service supply chain
Information Granularity“Family History of Breast Cancer”
GP electronic patient recordSpecialist Breast Cancer unitResearch Breast Cancer Genetics Unit
Presenter
Presentation Notes
Local operational differences may result in diversity. These may be technical, such as the need to support a legacy lab system, staffing and HR issues or local workflow arrangements e.g. when a particular service is delivered by nurse practitioners rather than doctors. The NHS is best thought of as a ‘supply chain’ of a number of communicating and interoperating enterprises, small and large, contributing to the health of individual patients and each with very different business drivers and operational requirements. Granularity - as we start to electronically communicate data across organisations, we become more aware of differences in the granularity of information that is regarded as ‘fit-for-purpose’, varying, in the case of Family History, from a simple yes/no in patient questionnaires to a detailed genetic history and DNA, since profile in a specialist genetics unit, but all falling under the issue of Family history of breast cancer. In the past this was rarely an issue, since at organisational boundaries, human beings ‘transformed’ the data from one format to another. whether more or less detailed but this is not an exercise that is easy to automate if we are sharing data between organisations.
Managing diversity
Golden Jubilee
Grampian
Lothian
eCardiology Record
DiagnosisBPECG
DiagnosisBPECG
DiagnosisDate of Diagnosis
BPSystolicDiastolic
Coded finding – “normal”Exertion level
Cuff sizePosition
ECGMultimedia
Automated report
Golden Jubilee
Grampian
Lothian
eCardiology Record
DiagnosisDate recorded
BPSystolicDiastolicCuff sizePosition
ECGAutomated report
DiagnosisEvent Date
BPSystolicDiastolic
ECGHeart ratePR intervalQRS interval
MAXIMAL DATASET
DiagnosisDate of Diagnosis (Event Date)
Date Recorded
BPSystolicDiastolic
Coded finding – “normal”Exertion level
Cuff sizePosition
ECGMultimedia
Automated reportHeart ratePR intervalQRS interval
eCARDIOLOGY TEMPLATE
DiagnosisDate of Diagnosis Date Recorded
BPSystolicDiastolicPositionCuff Size
ECGAutomated report
Heart ratePR intervalQRS interval
eCARDIOLOGY TEMPLATE
DiagnosisDate of Diagnosis Date Recorded
BPSystolic ‐163030003Diastolic ‐ 163031004
PositionCuff Size ‐ 246153002
ECGAutomated report
Heart ratePR intervalQRS interval
Snomed Term bindings
Snomed Query binding = Any
Cardiac condition
Presenter
Presentation Notes
Attempts to create a shared dataset commonly highlight inconsistencies between local representations of clinical concepts. The openEHR ‘maximal dataset’ approach affords a technically robust, neutral environment in which a range of common (though not necessarily shared) concepts can be negotiated and agreed by all parties. The final stage is to add Snomed term bindings to the eCardiology template.
Qld Health RepositorySA HealthConnectCancer Council of VictoriaDHS (Vic)Stat Health, McCauleyNEHTARCPA – cancer pathology
JP: openEHRsiteISO Standards
SG: openEHR based content modelling
ZA: Two companies working to provide a national solution
BR: National private health coordination using openEHRCL: openEHR site
National program involving openEHR
SE: Cambio + National trial of openEHR
DK: National pilot
CfH: Clinical content modelling
CA: Pilot EHR project considering openEHR
Microsoft’s internal openEHR site went live 2 weeks ago
Growing academic interest around the world
Slovenia: Vendor / national interestUK: “Sintero” –
Wellcome-fundedresearch data repository
SCO: considering openEHR
NL: Vendor using full openEHR back-end
US: College of Rhematologists –content modelling
US: College of Neurologists - content modelling
Presenter
Presentation Notes
This slide shows the current interest from vendors and jurisdictions around the world. There are also a large number of open source application initiatives and research efforts using openEHR which have not been shown.
The latest release of Microsoft’s Connected Health Framework acknowledges the value of archetypes and templates within an eHealth Enterprise Architecture.
SummaryThe ‘information model’ is a critical part of
the modelling of clinical contentTerminology is essential but not sufficientWe can only make scaleable progress if terminology and the structural information model work in harmony, which will require a coordinated EHR framework
openEHR offers particular advantagesAccessible to clinicians via archetypesReconciling interoperability with diversity will always remain a challengeThe openEHR ‘maximal dataset’ approach offers real advantage in managing diversity of shared information
Presenter
Presentation Notes
Interoperability will always require compromise and agreement between clinicians but recent advances in clinical information modelling such as openEHR, significantly simplify this process and reduce much of the burden on system suppliers, firmly placing the responsibility in clinical hands.