Rebecca Crowley, MD, MS

1
Data Standards and Coding in Pathology

• Rebecca Crowley, MD, MS
• APIII 2007

2
Outline

• Need for standards
• Data Content Standards definition
• Desirable features of a coding scheme
• ICD-9
• SNOMED
• LOINC
• UMLS
• NCI Thesaurus
• Common Data elements

3
Standards Need

• A prescribed set of rules, conditions or
  requirements concerning definition of terms
  classification of components, specification of
  material, ..
• unambiguous communication
• shared meaning
• definition of terms

4
Definition for Clinical Terminology
Standardized terms and their synonyms which
record patient findings, circumstances, events,
and interventions with sufficient detail to
support clinical care, decision support, outcomes
research , and quality improvement and can be
efficiently mapped to broader classifications for
administrative, regulatory, oversight, and fiscal
requirements.
5
The Importance of Standard Terminology

• Diagnosis and classification of cases or
  procedures
• Unambiguous communication of patient
  characteristics
• Application of guidelines, choice of therapy,
  consultation
• Expert systems and decision support
• Epidemiological and outcomes research
• Exchange of data between systems
• Exchange of data into billing systems
• Over 100 special purpose, medically-related
  vocabulary systems exist
• Institutions regularly deal with at least 5-10 of
  these

6
Desirable Features for Medical Coding Systems

• Complete, comprehensive, appropriate detail
• Clear (non-redundant, synonyms identified)
• Mapping to other commonly-used systems
• Multiaxial, compositional scheme
• Standard qualification of terms
• Representation of uncertainty and time
• Hierarchies and inheritance of meaning (aggregate
  or separate)
• Language independence
• Definitions of terms
• Unique, context-free identifiers (codes)
• Syntax and grammar for combining identifiers
• Non-proprietary standard

7
Multi-axial Coding Schemes
Anatomy
Etiology
Disease
S. pneumoniae
Lung
Infection
Composition A D E Pneumococcal pneumonia
8
Semantic Relationships
Gas exchange
Body
part-of
function-is
Thorax
Diaphragm
contained-in
Lungs
spacially- related-to
part-of
Right lung
part-of
Lower lobe

• Graphical representation (nodes and arcs)
  semantic network
• Nomenclature definitions relationships
  Ontology

9
Relationship Hierarchies
Affects
General relationship
Manages
Treats
Disrupts
Specific Relationships
Complicates
Interacts-with
Prevents
10
Implementation

• If relationships are defined in a terminology
  system, databases can be designed to take
  advantage of them
• Query results will be consistent and reproducible
  across systems
• If relationships are not defined
• Queries related to relationships may not be
  possible
• Local implementations of relationships will be
  idiosyncratic, incomplete and not reproducible in
  other systems

11
Standard Coding Schemes

• ICD-9
• NCI Thesaurus
• SNOMED
• LOINC
• UMLS

12
Standard Coding Schemes ICD-9

• International Classification of Disease
• ICD-9-CM most widely used today
• Heritage to the London Bills of Mortality in 1662
• First edition in 1900, maintained by WHO, revised
  at 10-year intervals
• Originally for classifying cause of death now
  for coding diagnoses used in hospital (technical
  component) billing
• Used with CPT for some professional component
  billing
• Extensions to other special-purpose
  classifications
• ICD-10 released in 1992, substantially larger
  with a different coding scheme migration mandated

13
ICD-9 in Epidemiology

• Most large scale epidemiology and outcomes
  studies using health plan, insurance company are
  based on ICD-9
• ICD are gross categories of diagnoses
• No notion of severity
• No notion of variability in natural history
• Studies based on ICD have substantial imprecision

14
ICD-9-CM
240 Simple and unspecified goiter
240.0 Goiter, specified as simple
Any condition classifiable to 240.9,
specified as simple 240.9 Goiter,
unspecified Enlargement of
thyroid Goiter or struma
NOS
diffuse colloid
endemic hyperplastic
nontoxic (diffuse)
parenchymatous
sporadic
Excludes congenital (dyshormonogenic) goiter
(246.1) 241 Nontoxic nodular goiter
Excludes adenoma of thyroid (226)
cystadenoma of thyroid (226)
241.0 Nontoxic uninodular goiter
Thyroid nodule
Uninodular goiter (nontoxic) 241.1
Nontoxic multinodular goiter
Multinodular goiter (nontoxic)
241.9 Unspecified nontoxic nodular goiter
Adenomatous goiter
Nodular goiter (nontoxic) NOS
Struma nodosa (simplex)

• One axis
• 3-digit number plus decimal
• Some hierarchy

15
NCI Thesaurus

• Maintained by the National Cancer Institute
• ontology-like vocabulary
• broad coverage of the cancer domain, including
  cancer related diseases, findings and
  abnormalities anatomy agents, drugs and
  chemicals genes and gene products and so on
• cancer diseases and combination chemotherapies,
  it provides the most granular and consistent
  terminology available
• currently contains over 34,000 concepts,
  structured into 20 taxonomic trees
• published under an open content license in a
  number of formats including OWL

16
SNOMED

• Systematized Nomenclature of Human and Veterinary
  Medicine, maintained by a branch of CAP
• Initially Standard Nomenclature of Diseases and
  Operations (1928) then Standard Nomenclature of
  Pathology (SNOP)
• SNOMED-RT released in 2000, added relationships
• SNOMED-CT released in 2002, added Clinical Terms
  v. 3

17
SNOMED Features

• Multiaxial hierarchical
• Topography, Morphology, Living organisms,
  Chemical, Function, Occupation, Diagnosis,
  Procedure, Physical agents, Social context,
  General
• 357,000 concepts, 400,000 coded terms
• Multiple hierarchies with 920,000 explicit
  relationships
• Defined syntax for code combination (composition)
• LOINC, ICD-9, Bethesda system and other mappings

18
SNOMED Codes
Pneumococcal Pneumonia (D-13510)

• T-28000 Lung NOS
• M-40000 Inflammation NOS
• L-25116 Strep pneumoniae

19
SNOMED Concepts and Terms
Eclass preferred term vs. synonym
20
SNOMED Relationships
21
SNOMED Relationships
22
SNOMED Expressions
23
SNOMED Maintenance and Licensing
SNOMED CT was acquired in April 2007 by the
International Health Terminology Standards
Organization (IHTSDO) NLM is the U.S. Member
of the IHTSDO - one of nine Charter Members. Via
a contract with the IHTSDO, NLM will pay an
annual fee (approximately 5.5 million for 2008)
to make use of updated English and Spanish
versions of SNOMED CT free of charge to anyone in
the U.S UMLS licensees have broader rights to
create and distribute SNOMED CT extensions and
derivative works under the uniform international
IHTSDO Affiliate Licence that is now part of the
UMLS License. The license covers all types of use
in all countries. Fees apply to some types of
distribution or use in countries that are not yet
Members of the IHTSDO. Source
http//www.nlm.nih.gov/research/umls/Snomed/snomed
_faq
24
LOINC

• Logical Observation Identifiers, Names and Codes
• Clinical observations and laboratory tests (more
  than 10,000 codes)
• Specifies item measured, challenge (if challenge
  test), property measured, measurement timing,
  type of specimen (or system), precision, method
• Avoids "mapping" internal codes every time two
  systems are interfaced

25
LOINC Example
LOINC Code 5306-6
Component Property Time System Scale Method Ricket
tsia sp AB Titre Point Serum Quant Comp. Fix.
Identifies lab tests
26
Other Coding Schemes

• CPT
• MeSH medical subject headings
• Micromedex (Drugs)
• Nursing codes NANDA, NIC and NOC

27
Performance of Coding Schemes in Expressing
Clinical Concepts
3061 clinical concepts from chart review scale
0 to 2
Scheme Diagnoses Findings Modifiers Treatment Proc
edures Overall ICD-10 1.60 1.08 0.27 0.46 0.26 0.
62 ICD-9-CM 1.61 1.23 0.36 0.51 1.00 0.77 CPT 0.
00 0.13 0.07 0.58 0.52 0.17 SNOMED 1.90 1.82 1.69
1.52 1.78 1.74
Chute, et al. 1996.
28
Appropriate Uses for Coding Schemes
29
Potentials

• Physicians and other healthcare workers may apply
  codes directly
• Potential for automation
• Most codes are applied retrospectively by
  "coders" working with the medical record
• Auto-encoding
• Natural language recognition
• Coding organization is responsible for assuring
  accuracy must review

30
UMLSUnified Medical Language System

• Development supported by the National Library of
  Medicine
• Beginning around 1990
• Four components Metathesaurus, Semantic Network,
  Information Sources Map , Specialist Lexicon
• Medical Concept List
• Synonyms, multiple languages
• Relationships between concepts
• Linkage of standard codes to medical concepts
• Supports translation between coding schemes
• Supports conversion from synonyms to canonical
  names to codes
• Can run as a vocabulary server with appropriate
  software

31
1. UMLS Metathesaurus

• Contains biomedical concepts and terms from many
  controlled vocabularies
• Concept name, concept unique identifier,
  co-occurring concepts, context, definition,
  related concepts, part of speech, source
  vocabulary, semantic type
• Preserves names, meanings, hierarchy, attributes,
  and inter-term relationships from the sources
• Adds some basic information to each concept
• Establishes new relationships between terms from
  different vocabularies

32
UMLS Metathesaurus

• gt1M biomedical concepts
• 5M concept names
• 100 source source vocabularies

33
2. UMLS Semantic Network

• Categorizes UMLS concepts
• Provides relationships between UMLS concepts
• Semantic types are network nodes
• 135 total types
• Organisms, anatomical structures, biologic
  function, chemicals, events, physical objects,
  and concepts
• Relationships are links
• 54 link types
• Is-a, part-of, process-of, physically-related-to,
  spatially- related-to, temporally-related-to,
  functionally-related-to, and conceptually-related-
  to

34
3. Information Sources Map

• A database of biomedical information sources
• A set of WWW applications that manage and use the
  ISM
• Registration of information sources via the WWW
• Accepts English language queries and returns
  lists of sources and links

35
4.UMLS Specialist Lexicon

• 200K single and multi-word terms
• Lexical records
• Used in NLMs SPECIALIST natural language
  processing system

baseanaesthetic spelling_variantanesthetic entr
yE0008769 catnoun variantsreg entryE0008770
catadj
36
Common Data Elements

• Objective
• To identify a set of common data elements that
  will serve as a
• foundation for national comparisons.

37
Common Data Elements

• Anatomic pathology consultation reports are
  largely unstructured and are thus resistant to
  automated analysis. This limits the current
  usefulness of these reports and isolates
  pathology data from emerging data-intensive
  disciplines such as bioinformatics.
• Standardized Common Data Elements (CDE's) for
  anatomic pathology are needed to permit sharing
  of these data by standardizing the metadata of
  these reports.

38
Complicated
39
Easy

• Mitotic Index
• ___ Less than 1 mitotic figure per mm2
• ___ 1 or more mitotic figure per mm2

40
Summary

• Need of standards
• Different coding schemes
• Common Data elements
• Data representation
• Data communication