Health Terminologies: From Products to Process Pieter E Zanstra Radboud University Nijmegen Medical Center Co-ordinator EU SemanticHEALTH Specific Support Action: Semantic Interoperability Deployment and Research Roadmap

1
Health TerminologiesFrom Products to
ProcessPieter E ZanstraRadboud University
Nijmegen Medical CenterCo-ordinator EU
SemanticHEALTH Specific Support ActionSemantic
Interoperability Deployment and Research Roadmap
Health Terminologies Criteria for decision
makingeHealth Working Group Workshop, Brussels,
March 8, 2006
2
Presentation Overview

• About this workshop
• A bit history
• Clinical perspective
• Coding practice
• Assessment
• Next Steps from here

3
About this workshop

• Induced by Snomed SDO invitation
• To join or not to join, thats the question?
• Criteria for decision making
• Scoping the problem
• How to collect criteria
• National needs/experiences
• What is available
• How to organise
• Round table
• Follow-up workshops?
• In depth on topics arising from round table

4
A Typical Record
Reviewed. Primip FT No ANC probs F/D at
0845 Needed synto for ?contractions about
7cm Decent progress since. Epidural in-situ No
urges to push really according to
patient Contractions 410 but poor duration
strength (on 60 u/s /synto) PA Ceph 0/5 VE
Head at 1 Caput 1 No ??? OA position Cx
F/D NO DESCENT Tried pushing poor maternal
effort despite instruction Although epidural
working well, doesnt stop longstanding spasmodic
hip pain. Very problematic FHR? CTG ? No descent
with attempts at pushing poor effort. Plan a)
Push synto ? b) Need really try to push work
on technique c) reassess at interval 15 mins
D/W consultant Would not be happy to have to
exert traction with Forceps when no effort no
descent
5
Competing demands on recording
ICD, ICPM
DRG
Documentation
Categorise
6
What kind of business logic do we need?
ICD, ICPM
DRG
Documentation
7
CEN/ISSS eHealth FG Recommendation 13 (2005)

• The Member States, with the Commission, should
• ensure the Europe wide referencing and easy
  access to the content of existing health coding
  systems based on registration of such systems by
  the Eurorec Institute
• support the international convergence towards a
  common framework for formal representation, and
  eventually the development and maintenance of a
  multilingual clinical reference terminology. This
  effort should build on existing efforts in formal
  representation as GALEN, FMA and SNOMED, and be
  carried out in liaison with the WHO Family of
  International Classifications
• make the targeted reference clinical terminology
  publicly available free of charge
• support a common approach to link national
  classifications of procedures, to support
  cross-border reimbursement of health care.

8
Clinical terminology

• A clinical terminology is the collection of
  standard terms with their synonyms, which in the
  context of patient care support the recording of
  complaints, signs, symptoms, circumstances,
  process of illness, interventions, results,
  diagnoses, as well as the decision making of the
  care providers

9
What are the boundaries of Terminology

• Blood pressure
• Evaluation result (high, low)
• Measurement (diastolic, systolic) Archetype?
• Guideline (how to measure?)
• Rheumatoid arthritis
• Evaluation result
• Measurement (criteria)

10
What binds Records, Interface, Reporting,
Knowledge?
Interface Terminology
ReferenceTerminology
Health Record
Reporting Terminology (ICD)
Discharge Referral..
11
Example Interface terminology
PURKINJE
12
Record architecture terminology (mediation)
Name, Context and Content
physical examination
CONTEXT
breast
palpation
lump
NAME
present
CONTENT
13
Sound semantics underpinning continuity in care
14
Some Myths
Stage of development
Open Source
Open source will most definitely not be free of
cost A complex terminology is required, and we
should not expect this to appear without
significant investment The critical mass of
skilled dedicated contributors to a large
Health terminology is too small for a sustainable
system
Evidence Based Medicine
Only about 5-8 of present medical knowledge is
based on sound evidence. RCTs are very costly
and time-consuming. We can not longer ignore the
need for routinely collected observational data
of high quality. Problem of finding outliers in
the literature
International Standards
The national marketplace is to small for domestic
suppliers to survive. Early adoption of new
standards, and active participation in their
development will increase competitive edge. Today
developing countries are already strong
contenders in systems design. They are the
knowledge marketeers!
15
A bit of Archeology (1)

• SESAME Committee (1990)
• Medical Semantics Advisory Task (Mesat)
• Establish national standardisation organisations
  for semantical aspects in medicine
• European medical termbank (Eumet)
• European Model of Health Care (Euromodel)
• European Classification of Medical Procedures
  (Euclamep)
• Harmonization of validation and coding rules for
  ICD

16
A bit of Archeology (2)

• 2nd EU/CEN Workshop (1997)
• Central resources - public or private - are
  urgently needed for European Industry to address
  a single market
• Central reference resource of concepts
• Multilingual lexicons
• Language independent development tools
• A strategy to maximise European influence on
  international developments is needed
• New developments in HL7, CorbaMed, DICOM, UMLS
• Establishing evolutionary pathways from existing
  systems to future systems
• Communication architectures should address
  mediation conversion

17
Terminology generations and their roles

• 1st Generation
• Monoaxial, relatively short
• ICD, ICF, MESH,
• 2nd Generation
• Multiaxial, more detail
• Snomed 2, UMLS, ..
• 3rd Generation
• Network, Compositional, Formal, Software,
• Very detailed
• SNOMED CT, OpenGALEN, FMA,

18
Multiprofessional Records and terminology

• Differences in vocabulary
• Differences in the meaning of the vocabulary
• Differences in aspects/abstractions
• Nurse constipation is an elimination problem
• Doctor constipation is a bowel function
  disorder
• Difference in operational significance
• Reflecting differences in professional tasks

35
19
Whats in use? (2001)
20
Codes, codes everywhere...

• SNOMED-3
• READ-2
• MeSH
• ICD-9
• ICD-9-CM
• ICD-O
• NCSP
• ICPM
• OXMIS
• ECRI-UMDNS
• SNOP
• HCFA
• ACR-NEMA
• IUPAC-NPU
• LOINC
• DICOM-SDM
• MCTGE
• CDAM
• NGAP

AIDSTRIALS ChemID CHEMLINE GENE-TOX HISTLINE SDLIN
E TOXLINE TOXLINE65 TOXLIT PDQ AVLINE BIOETHICS CA
NCERLIT CATLINE DENTALPROJ MEDLINE POPLINE SERLINE
DOCUSER Dxplain AI/RHEUM Iliad GenBank OMS PSY TR
IFACTS NIOSH NPIRS NEDRES MED85
MED75 HSTAT HDA MED90 HealthSTAR ACR92 AIR93 BRMP9
6 NIC ULT BRMS96 COSTAR CPM CRISP COSTART DMD DSM
III IV DOR HHC INS LCH MCM MIM Neuronames WHOART
CTV3
CCHI (Canada) MBS-E (Australia) ICD-10-PCS
(USA) ICPM-NL (Netherlands) NCSP (Swedish
Version) NCSP (Finnish Version) ICPM-DE
(Germany) CCAM (France) SNOMED-CT
(USA-UK) OPCS-5 (UK) SKS (Denmark) ICIDH
(WHO) Digital Anatomist (UW) Nomina Anatomica
21
And there is more to come

• Official Inventory on IT systems
• 52 alien patient record systems
• 1 system used by 200 physicians of which 70 daily
  
• Inventory by students
• Estimated 200 alien systems
• At least 200 shortlists for diagnosis, treatment
• Only some terminologies properly
  documented/maintained

22
Variability in Coding Practice
23
Cultural differences?
Term
ReadV3
SNOMED-CT
Bounty bar Crème egg Kit Kat Mars Bar Milky
Way Smarties Twix Snickers

• C-F0811
• C-F0816
• C-F0817
• C-F0819
• C-F081A
• C-F081B
• C-F081C
• C-F0058

UbOVv UbOW2 UbOW3 UbOW4 UbOW5 UbOW6 UbOW7 Ub1pT
24
ICD9 8 Codes for cycling accidents
25
The exploding bicycle accidents

• ICD-9 (E826) 8
• READ-2 (T30..) 81
• READ-3 87
• ICD-10 587

26
Formalised coding rules for ICD ?
Cycling Accident
What you hit Your Role Activity Location
27
Unified Medical Language System

• US National Library of Medicine
• Defacto common registry for vocabularies
• Metathesaurus
• 1.8 million concepts
• categorised by semantic net types
• Semantic Net
• 135 Types
• 54 Links
• Specialist Lexicon

28
UMLS Pros and Cons

• An invaluable resource, but...
• No better than the vocabularies which are mapped
• Limited detail for patient care
• Unreliable for indexing or abstraction of
  knowledge
• Best for relating everything to MeSH for indexing
  literature
• Still limited by combinatorial explosion
• Still cant cope with fractal knowledge
• Not extensible - no help in building or extending
  terminologiese
• No help in reorganising existing terminologies to
  re-use for new purposes
• Top down
• Information still implicit
• Minimal help with software
• No help with data capture, user interfaces

29
GALEN Vision

• A means to cope with size and diversity
• Coherence without uniformity
• Patient information and clinical knowledge
• available, relevant, in your language,
• Specialised clinical component software
• interworking systems exchanging meaning
• Libraries of knowledge big enough to matter
• An open resource for Europe and beyond

30
GALEN designed to drive systems
Best Practice
Best Practice
31
Ontologies The invisible unifying foundation?
Decision support
Clinical Applications
Medical Records

Data entry
32
GALENs Approach Compositional
Pathological fracture of the neck of the right
femur caused by osteoporosis

• Fracture which
• lthasLocation-(AnatomicalNeck which
• 
  isDivisionOf-(Femur which
• 
  hasLaterality-right))
• hasCause-Osteoporosisgt

12
33
Simplicity for Users
34
Wrap-up
35
Still many open questions

• What level of detail is required
• Revolutionairy/evolutionairy implementation
• How do we cope technically
• How do we cope in the business process
• How do users cope cognitively
• How do we cope with version control in the new
  highly interdependent architectures

36
Maintenance dynamics
37
Exploitation Models

• Strong belief in Market forces in the 90s
• GALEN and Snomed failed to reach sustainable
  state
• Some more limited commercial solutions survived
• But market did not take it
• Was it ready?
• Are HL7v3 and SNOMED/GALEN part of Solution Gap
• Increasing belief in centrally funded Info
  structures
• Dissimilar structures to costly to maintain

38
Attention for

• Safety of patient data
• Guaranteed unrestricted access to patient data
• Pull-out strategies
• Multiple suppliers
• Purchasing rules
• Specification en Certification
• Ownership, Influence
• Extensions and updates
• Usage for all
• Availability (open source?)

39
Observations

• Development still driven by perceived need
• No yet hard evidence to prove added value
• For the next five year seen by industry as
  exhaustive list of fixed terms
• Growing evidence to reconsider the ontologies
  behind systems
• Both GALEN and SCT stem from early 90s
• Systems not plug and play
• Significant cost of translation
• Multitude of base cost goes into localisation

40
Clinical Terminology Summary

• Evolving use case
• Epidemiology
• Organising care
• Documenting Care
• Access to knowledge

• Evolving Technology
• Enumeration ICD
• Interlingua UMLS
• Formal Ontology SNOMED CT

• Technological Barriers
• Enumeration doesnt scale
• Computers cant read
• Humans cant organise
• Tower of Babel

• Unsolved Problems
• Quality Assurance
• Efficient data capture
• Interrater variability
• Cost Model

41
Establish the process

• Focus on real immediate needs, and realistic time
  scales
• Be prepared to throw away what you loved and
  cherised!
• Be aware and secure solutions for different
  cultures/ languages
• Do not just select a single product, but join
  forces to redesign with best of breed
• Involve and explain to those who have the burden
  of recording (registration dividend)
• Without a well managed network of compentent
  expertise centres, the process is very likely to
  fail

42
Acknowledgements
Radboud University Nijmegen Medical
Center Department of Medical Informatics
(Co-ordinator)
University College LondonCentre for Health
Informatics, UK
World Health Organisation Dept. Measurements
Health Information Systems, Switzerland
University of Manchester Health and
Bioinformatics Group, UK
Uppsala UniversityNordic Centre for
Classifications in Health Care, Sweden
National Institute for Strategic Health
ResearchHungary
University of St. EtienneDepartment of Public
Health Medical Informatics, France
Communication Technology Research, Germany
43
Thank you for your attention!

• Further information
• www.semanticHEALTH.org
• semantichealth_at_eski.hu
• Pieter E Zanstra
• Radboud University Nijmegen Medical Center
• P.Zanstra.mi_at_xs4all.nl