SNOMEDCT

1
SNOMED-CT

• Dr Chris Wroe
• Integration design team London LSP
• BT Health

BCS ASSIST Joint Meeting on SNOMED-CT 28/09/2007
2
Overview

• London background
• Advances in SNOMED-CT and the opportunities they
  provide
• Supplier uptake
• Challenges in implementing advances
• Approach to tackling challenges

3
London background
Messages e.g. assessment, referral discharge
core clinical systems
Acute care systems
Primary care systems
Mental health systems
Community health systems
4
London background
Messages to additional departmental systems e.g.
orders and results
Clinical coding
core clinical systems
Acute care systems
Primary care systems
Pathology
Theatres
Mental health systems
Community health systems
Radiology
Pharmacy
5
Message contents

• Some message content needs to be captured using
  a controlled terminology to allow for
  interoperation.
• Candidates
• Allergies, current medication, immunisations,
  diagnoses, risk factors including family history,
  tests and procedures ordered and performed,
  certain aspects of test result, adverse events
• Controlled terminology needs to be shared across
  systems
• Message content will form an important part of
  the patients record. Therefore terms must allow
  for required detail.

6
Information management
Data feeds for analysis and reporting
core clinical systems
Acute care systems
Primary care systems
Data warehouse (Information management system)
Mental health systems
Community health systems
7
Data feed contents

• Controlled terminology needs to be shared across
  systems
• Analysis often doesnt require same granularity
  as patient record.

8
Alternatives to SNOMED-CT
9
Pros and Cons of existing vocabularies ICD
Used to class events in to finite number of
categories

• Useful for analysis
• Stable over time
• Detailed guidance on use of specific codes
  supporting consistency
• Straightforward to implement (although
  supporting coding rules is a specialist task)

• Less useful for use in a patient record
• Not elsewhere classified Not otherwise
  specified
• Not enough detail
• Combinatorial explosion when people try

10
ProblemsThe exploding bicycle(codes for
injuries involving cyclists)

• 1972 ICD-9 (E826) 8
• 1999 ICD-10

Courtesy Alan Rector, Jeremy Rogers, University
of Manchester
11
1999 ICD10 (Australian edition) 587 codes

• V31.22 Occupant of three-wheeled motor vehicle
  injured in collision with pedal cycle, person on
  outside of vehicle, nontraffic accident, while
  working for income
• W65.40 Drowning and submersion while in bath-tub,
  street and highway, while engaged in sports
  activity
• X35.44 Victim of volcanic eruption, street and
  highway, while resting, sleeping, eating or
  engaging in other vital activities

Courtesy Alan Rector, Jeremy Rogers, University
of Manchester
12
SNOMED-CT advances
13
Pros and Cons of existing vocabularies Read 2
Used in primary care for patient records

• Tailored to primary care needs
• Lots of experience in making it usable during a
  consultation (upto 15 codes per consultation)
• Mnemonics, macros, memorable codes
• Straightforward hierarchies to browse
• Straightforward for analysis hierarchy in
  the code

• Can be high degree of variability in use of
  terms (PRIMIS)
• Difficult to generalise structure to other care
  settings
• Cannot support different ways of organising
  terms
• Cannot support more than five levels of detail
• Same combinatorial explosion
• Therefore not an option for integration

14
SNOMED-CT advances More content

• A reference terminology designed to cross care
  settings
• 400,000 concepts drawn from predecessors
• Now been added to directly
• Examples
• Orthopaedics
• Primary open reduction of fracture of patella and
  fixation with tension band wiring (procedure) -
  310865004
• Mental Health
• Acute exacerbation of subchronic hebephrenic
  schizophrenia (disorder) - 191538001
• Community health
• Referral to community podiatrist (procedure) -
  306351008
• Primary care
• Well man health examination (procedure) -
  281029006
• Radiology
• Scaphoid X-ray (procedure) - 241071006
• Pathology
• Hepatitis delta agent antibody assay (procedure)
  - 104320001
• Supports integration by covering different areas
  in the same terminology

15
SNOMED-CT advances Flexible, more detailed
structure

• Not linked to the code anymore
• Pneumococcal pneumonia - ... (disorder) -
  233607000
• Two parents reflecting at least two ways of
  organising the concept
• A lung disease Pneumonia due to Streptococcus
  - ... (disorder) 34020007
• An infectious disease Pneumococcal infectious
  disease - ... (disorder) - 16814004
• Relevant parents in hierarchy actually worked out
  by looking at its defining relationships
• Causative agent Streptococcus pneumoniae 9861002
• Associated morphology Consolidation 9656002
• Associated morphology Inflammation 23583003
• Finding site Structure of interstitial tissue of
  lung 113254000

16
SNOMED-CT advances Separation of terms from
concepts

• The same medical concept may be represented by
  many different terms.
• Whipple procedure lt-gt Pancreaticoduodenectomy lt-gt
  Whipple operation concept id 116241004
• A Term can represent more than one concept
• Cold - ... sensation quality (qualifier value)
  84162001
• Cold - Common cold (disorder) - 82272006

17
SNOMED-CT advances Term Composition Conceptual
Lego
gene
protein
polysaccharide
cell
expression
chronic
acute
Lung
infection
abnormal
inflammation
normal
bacterium
deletion
polymorphism
ischaemic
mucus
virus
Courtesy Alan Rector, Jeremy Rogers, University
of Manchester
18
Logic as the clips for Conceptual Lego
Primary open reduction of fracture of patella and
fixation with tension band wiring
Hand which isanatomically normal
Courtesy Alan Rector, Jeremy Rogers, University
of Manchester
19
2. Uptake with London suppliers

• Additional text / description if required

20
Acute care

• Experience of using SNOMED-CT within the Cerner
  Millennium product.
• Use SNOMED-CT to capture patient diagnoses and
  procedures performed (expected live in LC1)
• Focus on taking advantage of increased content

21
Millennium screenshot
22
Mental and community health

• RiO version 5 onwards supports SNOMED-CT in
  several different areas of the application.
• Use of SNOMED-CT configurable to some extent
• InPractice systems developing new version of
  Vision with SNOMED-CT support.

Primary care
23
RiO candidate screen design (subject to change)
24
3. Challenges
25
Challenges

• SNOMED-CT is simply not a plug in replacement
  for predecessor terminologies
• More complex data model that needs to be
  incorporated at the heart of the application
• Concepts versus terms
• Multiple types of relationships attached to a
  concept
• More data, longer fields.
• Analysis procedures more complex due to increased
  size and complex data model

26
Increase in content available

• Large content size good for integration less
  good for any one setting. How does the user find
  the term they need?

Examples taken from BT demonstration SNOMED-CT
browser.
27
Filtering searches

• When entering a diagnosis only one of the
  following terms is appropriate. These terms are
  from completely different parts of SNOMED-CT.
• Eccrine ductal carcinoma 400208002
• Eccrine ductal carcinoma of skin 403939009

28
Increase in complexity of structure

• Now a rich relationship structure but can be
  difficult to navigate through

Browsing to appendicectomy using a CTV3
navigation subset
Browsing to appendicectomy using SNOMED-CT
relationships
Examples taken from BT demonstration SNOMED-CT
browser.
29
Challenges Supporting compositionality

• Its not as easy as building Lego.
• How does it relate to the user?
• How does it relate to the application data
  model?
• How do you display a composition?
• How do you perform analysis on data using
  composed terms?
• How do you cope with being able to say the same
  thing in different ways?

30
Multiple different ways of saying the same thing
Laparoscopic appendicectomy
Emergency laparoscopic appendicectomy
emergency
laparoscopic
appendicectomy
operation
excision
appendix
31
Overlapping with data models
Quick search for penicillin allergy term
(precoordinated)
32
Selecting a precoordinated term
Selecting the appropriate term (precoordinated)
33
Relating to a data model
Normal form generation and analysis has been used
to extract the causative substance
34
4. Addressing the challenges
35
Two complementary approaches

• Solve software implementation issues once in a
  common terminology service
• Rather than solve challenges in the general
  case, attack challenges on a case by case basis
  in London.

36
Terminology services

• Provide a common service that applications can
  integrate with in order to manipulate SNOMED-CT
  data.
• Term search
• SNOMED-CT Subset management
• Relationship access
• Selection statistics
• Normal form generation
• Hierarchy calculation
• BT has a terminology service live since June 2006
  focussing on supporting SNOMED-CT term search.
  CaTTS (coding and taxonomy / terminology service)
• Open question as to how much can be done in a
  common service and how much needs to be done on a
  context specific way that only the clinical
  application can provide.

37
SNOMED CT Integration

• Design approaches and problems
• Dr. Robert Challen
• Integration design
• BT

38
(No Transcript)