metadata standards for health portals
TRANSCRIPT
Metadata Requirements for Health Portals
Tim BensonAbies Ltd, R-Outcomes Ltd, UCL
[email protected] 2015 Madrid, 28 May 2015
Outline
• Why metadata matters• XDS Cross Enterprise Document Sharing• Clinical documents and clinical
statements• Specific requirements and items needed• Stringent standards that are fit for
purpose
Background
• Wellcome Trust Sintero Project project leader Dr E Conleyhttp://sintero.vargula.co.uk/
• IHE UK Affinity Domainhttp://wiki.ihe-uk.org/XDS_Metadata_specifications
• Principles of health interoperability HL7 and SNOMED Springer 2012http://www.springer.com/gb/book/9781447128007
What is Metadata?
• Metadata describes other data • Used for indexing, discovery and
search• Provides information about what
sort of things an item contains– Metadata is not an abstract– It says an entry contains a diagnosis,
but does not say what the diagnosis is– This helps in information governance
Dublin Core Metadata Element Set
• All elements are optional and repeatable
• Mostly free text• Designed by
librarians for librarians
• Mainly for paper archives
1. Title2. Creator3. Subject4. Description5. Publisher6. Contributor7. Date8. Type9. Format10. Identifier11. Source12. Language13. Relation14. Coverage15. Rights
6
Acute Care (Inpatient)
GPs and Clinics (Ambulatory)
Long Term Care
Other Specialized Care(incl. Diagnostics Services)
Patient Longitudinal Record
Typically, a patient goes through a sequence of encounters in different Care Settings
Source: IHE
XDS
• IHE XDS Cross-enterprise Document Sharing
• Based on ebXML Registry Standard (ISO 15000-4, 2004)
1. Sources post document packages to the Repository
3. Consumers search for documents with specific information
2. Repository registers the documents metadata and pointer with the Registry
4. Consumers retrieve documents from Repository (-ies)
XDS Document (Metadata):
TypePatientAuthorFacilityAuthenticator…
Repository
Origin of Documents Package
Registry
How XDS WorksConsum
er
Source: IHE
XDS Metadata handling
Patient Identity Source
Document Registry
Document Repository
Document Source
Document Consumer
Patient Identity Feed
Query Documents
Retrieve Document
Provide and Register
Document Set
Register Document Set
Generates
Stores
Interprets
Adds
Source: IHE
Documents and Statements 1/2
• Clinical Document is discrete electronic composition about one patient, intended to be read or used by a human.
• Clinical Statement is the smallest meaningful category of stand-alone medical information. A single data entry.– Statement is a very small structured
document
Documents and Statements 2/2
• Different uses and granularity• Both share properties of
– Persistence, wholeness, coherence, human readability, stewardship and authentication
– CDA, XDS and FHIR share these ideas• Both require search, sort and display
by type, date, author, source
Clinical Document
• Traditional document metaphor• Human-to-human communication of
discrete documents • e.g. pdf report or letter, image, video
• Sharing across organisation boundaries• Computer-processable data is hard to
extract – e.g. HL7 CDA Level 3 does not work as
well as expected
Clinical Statement
• Computer-processable data entries– Data that can be counted, analysed and
charted• Content is unambiguously coded and
tightly specified– e.g. diagnosis, medication entry, test
result– select and sort statements from multiple
sources on the fly for lists, charts and graphs
Health Metadata
• IHE XDS– Affinity domain– Based on ebXML Registry
• HL7 CDA header– Templates
Basic Requirements
• What• Who by• Who about• When• Where• Form
I keep six honest serving-men (They taught me all I knew);Their names are What and Why and When And How and Where and Who.
Rudyard Kipling. The Elephant’s Child. Just So Stories. 1902
Optionality
• Completeness • If search fails to find something, big safety
risk• Mandatory elements
– One patient– One creator– One creation time– One purpose
• In metadata optionality and repeats are bad
We need Standards
• Every source feed must provide exactly the same metadata– Else data won’t be found
• Stringent computer-processable data– Unique identifiers for things, people and
places– Codes for concepts + content (SNOMED CT)– Date/time formats
What 1
• Title (free text)• Class – general category, course
granularity – Note: HL7 CDA does not have a slot for this– Equivalent to HL7 RIM classCode +
moodCode• Type – specific category, fine granularity
– Equivalent to HL7 RIM code• Specialty – clinical service • CareType – care setting
Examples
Example Document Statement
Title Urology discharge
summary Weight entry
Class Correspondence Finding
Type Discharge summary Weight
Specialty Urology General Practice
CareType Inpatient Primary care
What 2
• UUID• OriginalID• Confidentiality• Language• Format
• Class, Type, Specialty, CareType, Confidentiality, Language and Format use SNOMED CT RefSets
Other
• When– CreationTime– EventTime
• Who– PatientId– CreatorId
• Where– LocationId
• Extensions– Tag (optional, repeatable) using SNOMED CT RefSet
Discussion
• Every source system must provide standard metadata– Allow default when not known
• All items (except Title) are structured and computer-processable– Codes use SNOMED CT RefSets
• Same structure for documents and statements• Lack of metadata standards means that portal
projects are – (1) expensive – (2) incompatible
Summary
• Why metadata is crucial in portals• Documents and statements have
similarities• XDS and CDA• Specific requirements• Need stringent metadata
standards• Low hanging fruit
Questions