Post on 23-Feb-2016
description
The Challenge and SNOMED CT
Karen Gibson
The Challenge Significant investment in eHealth is underway Clinical records:
◦ Not only a record for the author◦ Essential to inform the next person in the care
team Clinical safety risks of poor quality, ambiguous
communication Desire to:
◦ make systems more interoperable◦ improve data quality ◦ improve ability to re-use information for reporting,
management etc.
Clinical Terminology is complex
Humans spend 4-10 years learning medical terminology at University!
We need to make their language computable
No silver bullets
Clinicians say things in many different ways◦ Sometimes legibly◦ Often in shorthand
Terminology needs to maintain fidelity of information – be true to what clinician is trying to say
EHR’s need to source information from many different systems◦ Legacy systems with legacy data◦ Legacy terms and ways of coding (if coded at
all)
How do we begin to bring this together?
And do so in a way which ensures stakeholders can be confident that the information is accurate and capable of being aggregated and reused.
What is the answer? SNOMED CT
◦ Most comprehensive clinical terminology available
~ 350,000 concepts~ 1,000,000 terms
◦ Purchased and maintained by a group of collaborating nations for use in their eHealth initiatives (IHTSDO)
Only part of the answer:◦ Supplemented by other terminologies – eg.
medicines and administrative◦ Knowledge of the information model (context)◦ Other emerging technologies (eg. NLP)
What is the problem? SNOMED CT
◦ Complexity ~ 350,000 concepts~ 1,000,000 terms
Only part of the problem◦ Lack of implementation knowledge◦ Lack of tools to assist◦ Lack of funding to meet costs of
implementation◦ ? Lack of will
Some problems have been addressed
IHTSDO has addressed (or is working to address): International Governance Open Standard Intellectual Property Quality ? Mapping to other standard terminologies/ classifications
Others are being tackled by NEHTA: Cost – free to use in Australia (as member of IHTSDO) ‘Australianisation’ National reference sets Medicines component
Key principles/ Traps for new players
Do look to SNOMED CT-AU first
◦It is endorsed by COAG◦It is the most comprehensive clinical
terminology available◦It is supported by NEHTA and IHTSDO
SCTID: 22298006
SNOMED CT-AU A concept and its descriptions
Myocardialinfarction
Synonym MI - Myocardial infarctionSCTID: 1784872019Synonym Infarction of heart
SCTID: 37441018Synonym Cardiac infarctionSCTID: 37442013Synonym Heart attack
SCTID: 37443015
Fully Specified Name
Myocardial infarction (disorder)SCTID: 751689013
Preferred term Myocardial infarctionSCTID: 37436014
MyocardialinfarctionSCTID: 22298006
Injury ofanatomical site
SCTID: 123397009
Structural disorder of heart
SCTID: 128599005
Myocardialdisease
SCTID: 57809008
Is a
Is a
Is a
MyocardiumstructureSCTID: 74281007
Finding siteInfarctSCTID: 55641003
Associated morphology
Relationships• Links concepts within SNOMED CT• Ensures unambiguous meaning • Create hierarchies which aid navigation and retrieval
Key principles/ Traps for new players
Consider the user interface carefully:◦Don’t show Fully Specified Names to
users They’re intended to provide a
unambiguous reference point for computability
They are not worded in a way clinicians speak
◦Do choose a preferred term
Fully specified name
Preferred term (Australia)
Amebic appendicitis (disorder)
Amoebic appendicitis
US Spelling
Semantic tag:• indicates hierarchy• not needed at
clinical level
Unambiguous Reference Point
Key principles/ Traps for new players
Consider the user interface carefully:◦Don’t show all of SNOMED CT in a drop
down list (too many terms!)◦Unless you have tools to assist
searching ◦Do use Reference sets to assist
implementation: Reduce the complexity for the user Speed identification of the correct
term
Problem/diagnosis : Select term
SNOMED CT in Drop down list without any parameters implemented
Problem/diagnosis : Appendi
Improved searching – limited to clinical finding hierarchy
Could be further improved through Refset development
Key principles/ Traps for new players
Reference sets Do require maintenance Therefore:
◦Do use NEHTA reference sets wherever possible (because NEHTA maintain them!)
◦Do use the hierarchies of SNOMED CT to guide creation of RefSets wherever possible
◦ Recognise that if you pick ad hoc terms across hierarchies you will need to manually maintain the list
◦ Sometimes there is no choice – eg. allergies – but there is a cost
Key principles/ Traps for new players
Minimise mapping and data translation:◦There is a safety risk introduced every
time the clinician’s language is translated (Chinese whispers…)
If you do need to map or translate:◦Do keep the original wording/ data entry
as well as the mapped equivalent
Key principles/ Traps for new players
Trap for new players:◦Synonyms may be found in the wrong
hierarchy (different meaning)◦This is why when translating SNOMED CT
translators look at the words within the hierarchy to establish true meaning
◦However, this trap is not just for translators, but also when mapping or creating reference sets.
Is it worth the effort? Even simple use of SNOMED as a flat code
list can add value:◦Allows meaningful exchange of data◦Both end-points can cross-reference to a
standard unambiguous definition◦Simple decision support can be enabled
For example – US Centre for Disease Control, HITSP
and NHS all publish simple lists
Is it worth the effort?
But for those up to the challenge, more advanced use of SNOMED CT offers further potential value
Is it worth the effort?
Ability to Exchange data knowing it can be explicitly and accurately interpreted
Ability to improve data quality:◦More structured data entry◦Agreed constraints can be applied
Is it worth the effort?
Ability to run externally developed queries:◦ For example:
Automatically run mandatory reporting Identify at-risk populations Identify cohorts for clinical trials Trigger presentation of evidence based guidelines when
first releasedNote Kaiser Permanente have a central area which
develop queries/ scripts which are then distributed throughout the organisation
Is it worth the effort?
Ability to utilise external decision support engines:◦Already happening in medicines area◦Opportunity for improved decision
support applications in other areas
Ability to contribute to PCEHR
Is it worth the effort?
Ultimate aim is improving health outcomes and patient safety:◦Through better sharing information◦Ensuring accuracy of information◦Identifying those at risk
Clinical Terminology is essential
Perhaps speaking to the converted, but unless we can agree and implement consistent terminology we will never achieve the goal of better information sharing….
We’ll just be sharing data….