1 HL7 RIM: An Incoherent Standard Barry Smith and Werner Ceusters ORG Ontology Research Group University at Buffalo

2 Severe problems with HL7 RIM –FMA –GO + other OBO Ontologies –FuGO –SNOMED CT –UMLS Semantic Network –NCI Thesaurus –ICF (International Classification of Functioning, Disability and Health) –ISO Terminology Standards –SCoP –IPAP Algorithm –SNODENT

3 Not e.g. with HL7 Clinical Document Architecture –FMA –GO + other OBO Ontologies –FuGO –SNOMED CT –UMLS Semantic Network –NCI Thesaurus –ICF (International Classification of Functioning, Disability and Health) –ISO Terminology Standards –SCoP –IPAP Algorithm –SNODENT

4 The hype “HL7 V3 is the standard of choice for countries and their initiatives to create national EHR and EHR data exchange standards... Significant V3 national implementations exist in many countries, e.g. in the UK (e.g. the English NHS), the Netherlands, Canada, Mexico, Germany and Croatia.” (R. Spronk)

5 The reality “None of the implementations have a national scope” (R. Spronk) Even local working implementations are hard to find (in spite of claims, e.g. by Oracle) Standard response: ‘set to go live in 6 months’

6 The hype The RIM is “credible, clear, comprehensive, concise, and consistent” It is “universally applicable” and “extremely stable”

7 The reality HL7 V3 documentation is 542,458 KB, divided into 7,573 files remains subject to frequent revisions The RIM documentation is especially difficult to understand is marked by sloppy and unexplained technical uses of terms such as ‘act’, ‘Act’, ‘Acts’, ‘action’, ‘ActClass’ ‘Act-instance’, ‘Act-object’

8 from HL7 forum “I am... frightened when I contemplate the number of potential V3ers who... simply are turned away by the difficulty of accessing the product. “Some of them attend V3 tutorials which explain V3 as the hugely complex process of creating a message and are turned off. [They] simply do not have the stamina, patience, endurance, time, or brain-cells to understand enough for them to feel comfortable contributing to debates / listservs, etc., so they remain silent.”

9 Gunther: HL7 is “the data standard for biomedical informatics” HL7TheDataStandardForBiomedicalInformatics.ppt

10 Are definitions like this a good basis for achieving semantic interoperability in the biomedical domain? LivingSubject Definition: A subtype of Entity representing an organism or complex animal, alive or not.

11 Person (from HL7 Glossary) Definition: A Living Subject representing single human being [sic] who is uniquely identifiable through one or more legal documents

12 The RIM’s normative specifications are supposed to guarantee consistent messaging across all health-care institutions – solving the problems of inconsistency in HL7 V2 yet the HL7 organization has not even succeeded in making its own V3 Glossary conform to the RIM’s normative specifications after 10+ years

13 Gunther (p. 151) On the one hand, the RIM is to ‘establish semantic interoperability across a vast and growing number of subject domains (e.g. laboratory, clinical health record data,... public health, clinical research, etc.)’

14 Gunther (p. 153f.) On the other hand, ‘Imperfections and inconsistencies in the HL7 documentation exist’ ‘inconsistency and ambiguity are deeply unavoidable in constructing a collaborative volunteer-based standard, which brings together a wide range of people from different backgrounds....’

15 Gunther (p. 154) ‘As different people edit parts of the specification, inconsistencies in form and quality may emerge; as some ambiguities are clarified, other previous systematic ideas may be corrupted; and well-meant glossary entities may cause confusion.’

16 So: On the one hand the RIM is to facilitate agreement on consistent meanings across the entire range of clinical domains. On the other hand the RIM’s own collaborating authors cannot reach agreement even amongst themselves. Yet still the resultant inconsistent ‘normative’ recommendations are approved by HL7 ballot

17 Gunther (p. 151) The RIM ‘defines the grammar of a language for information in healthcare’. ‘All data is in a form in which Entities (people, places, things: NOUNS) are related in Roles (RELATORS) to other Entities, and through their participations (PREPOSITIONS) interact in Acts (VERBS).’

18 Problems of scope Act = intentional action No processes (verby items) outside Act How can the RIM deal with disease processes, drug interactions, traffic accidents?

19 Problems of scope Entity = persons, places, organizations, material No things (nouny items) outside Entity How can the RIM deal with wounds, fractures, bodily organs, biomolecules? How can the RIM deal with diseases?

20 Diseases in the RIM... are not Acts... are not Entities... are not Roles, Participations, Role-Links... So what are they? Answer: they are Acts of Observation A case of pneumonia is an Act of Observation of a case of pneumonia

21 Mayo on ‘Act’ as “intentional action” Is a snake bite or bee sting an intentional action? Is a knife stabbing an intentional action? Is a car accident an intentional action? When a child swallows the contents of a bottle of poison is that an intentional action? of_Observations

22 The RIM has no coherent criteria for deciding For this reason, HL7 dialects are formed – and the RIM does not do its job. One health information system might conceive snakebites and gunshots as Procedures. Another might classify them with diseases, and so treat them as Observations. If basic categories cannot be agreed upon for common phenomena like snakebites, then the RIM is in serious trouble.

23 The methodology does not impede the formation of local dialects In the UK, the $ 35 bn. NHS National Program “Connecting for Health” has applied the RIM rigorously, using all the normative elements, and it discovered that it needed to create dialects of its own to make the V3-based system work for its purposes (it still does not work)

24 HL7’s Clinical Genomics Standard Specifications an individual allele as an Act of Observation Why is Act of Observation used for biomolecules but not for pharmaceutical products? Gunther: because biomolecules ‘are in the process of being discovered.’

25 HL7: Items in reality about which we have information that is known a priori are identified as Entities Items in reality about which we are still discovering information are identified as Acts of Observation. Gunther: ‘ What seems like an arbitrary split... is in fact the practical thing to do’ – it helps us to avoid ‘frequent modifications’

26 Lessons for the Future Establish standards only after thorough pilot testing Encourage criticism and open discussion Create clear documentation under expert supervision Avoid attempting to impose untested systems like the RIM across entire nations from the top down

27 What’s gone wrong? People of good will are making mistakes because of insufficient concern for clarity and consistency International standards are built in uncontrolled fashion Money is wasted on megasystems that cannot be used

28 What should be done? Extend the RIM to allow adequate treatment of all the types of entity relevant to information exchange in biomedicine, including: non-intentional processes, diseases, infections, biomolecules, etc.

29 Gunther: ‘A complete and integrated ontology of everything would certainly be nice to have; however, we think it is impractical and dangerous to force such a model into being independently of the RIM. The moment such a model gained traction people would then expect that the RIM reflect that other model. Why should there be two models, if in the end one is to reflect the other? Instead, a single model of real world objects should suffice, but must contain well-defined features for information- management functions.’

30 Gunther (paraphrased): HL7 has created a clunky ontology which is full of gaps which force strange and arbitrary seeming choices; however, it would be impractical and dangerous to complete it to create a more coherent framework. The single HL7 model must suffice, even with its strange rules for information-management functions which make the documentation so difficult to understand.

31 Conclusion: Are there practical consequences? Are there consequences for health IT? The hours (days, months) of software developers attempting to create HL7 V3- based applications work not only cost money (sometimes large amounts of money). They also reinforce scepticism on the part of potential users of health IT systems

32 Conclusion: Are there practical consequences? Are there consequences of clunky standards for the quality of patient care? Let us hope not