Terminologies, Data Standards, and other Informatics stuff'''

1
Terminologies, Data Standards, and other
Informatics stuff...

• Michael Hogarth, MDAssociate Professor of
  MedicineUC Davis School of Medicinehttp//www.ho
  garth.orgmike_at_hogarth.org

2
Summary

• Part I Introduction to informatics
• Part II HL-7 version 3, an information model
  and interoperability standard in healthcare
• Part III Terminological systems and why they are
  needed

2
3
What is informatics?

• It is primarily focused on information science
• Not primarily focused on technology, but
  technology and software are core tools used in
  the discipline.

The computer continues to be an exciting object,
it is increasingly present and it rarely fails to
attract attention. Unfortunately, the commodity
which is processed by the computer tends to be
overlooked.
Dr. Marsden Blois West. J. of
Medicine 1986
4
Information science

• Informatics experts study information and
  information systems rather than computer
  programming or computer design
• Information science is the study of information
  acquisition, storage, and use of information
  (information prescription)
• To think of an informatics person as a computer
  programmer is to think of a pathologist as a
  microscope engineer

5
Early beginnings...

• 1951 - UNIVAC, first commercial computer (only 46
  built from 51 - 54)
• 1959-60 - Early articles describing potential
  uses of computers in medicine

5
6
The birth of a discipline...

• Ledley and Lusted -- applying symbolic logic and
  mathematical methods to medical decision making

6
7
Dr. Robert S. Ledley

• MS in theoretical physics, Columbia DDS NYU
  Dental School
• 1950 -- Army post with the National Bureau of
  Statistics, Dental Division, where he had access
  to one of the first computers (SEAC)
• 1959 - first article outlining the use of
  mathematical and probabilistic methods in medical
  diagnosis - proposed computers could be employed
  for this
• 1973 - went on to develop the ACTA scanner, which
  introduced the notion of algorithm based image
  reconstruction and digital memory image storage
  -- the CT scanner!

7
more info --gt http//pir.georgetown.edu/nbrf/rslbi
o.html
8
Other innovations

• 1960 -- MEDLINE
• the first non-military modem-accessible
  database
• citations of medical journal articles
• became PubMed when it became web enabled in the
  mid 1990s
• 1966 - MUMPS -- Massachusetts General Hospital
  Utility Multi-Programming System)
• a programming language conceived by Niel
  Pappalardo while at Dr. Octo Barnetts
  cardiovascular research lab (MGH)
• 1970-90 the language of choice for healthcare and
  financial applications worldwide
• designed to make writing database-driven
  applications easy while simultaneously making
  efficient use of computing resources
• 1970s - Dendral, Mycin
• first expert systems
• MYCIN introduced the notion of accounting for
  uncertainty, first to be studied in a clinical
  context.

8
9
Informatics today
healthinformatics
clinicalinformatics
bioinformatics
9
10
The many faces of informatics
Pharmacy informatics
Public health informatics
Clinical informatics
Bioinformatics
11
Core themes in Health Informatics

• Knowledge representation
• Decision Support
• Medical Imaging
• Human interface design/usability research
• Information retrieval
• Decision Support Systems (DSS)
• Coding Systems/Terminology Systems
• Data Exchange Standards
• Telemedicine/Telehealth -- sensing devices, home
  care, geriatrics

12
Health informatics research
13
Relevant recent events

• 2000 National Academy of Sciences To Err is
  Human report
• 2001 National Academy of Sciences Crossing the
  Quality Chasm A New Health System for the 21st
  Century recommendation 9
• 2004 - Office of the National Coordinator for
  Health Information Technology (ONCHIT)
• 2005 California Regional Health Information
  Organization (CalRHIO)
• 2006 - Gov. Schwarzeneggers Executive Order
  S-12-06
• 2006 VaVISTA for Windows is being developed by
  the federal government for free use.
• 2007 ?

14
What is ONCHIT and what are RHIOs?

• Office of the National Coordinator for health
  Information technology
• Goal - improving healthcare through information
  technology.
• Established notion of regional healthcare
  information organizations (RHIOs)
• CCHIT - Certification Commission for Healthcare
  Information Technology
• Launched by ONCHIT as a voluntary, private-sector
  organization to certify HIT products.
• American Health Information Management
  Association (AHIMA), Healthcare Information and
  Management Systems Society (HIMSS), National
  Alliance for Health Information Technology
  (Alliance)

15
CalRHIO
16
Gov. Schwarzeneggers Executive Order
S-12-06

• where as the federal department of health and
  human services estimates that, in addition to
  improving the quality of chronic care management
  and reducing errors, increasing health
  information technology could reduce duplicative
  care and lower health care administrative costs,
  achieving a potential savings of 140 billion per
  year, or close to 10 of total health spending in
  the United States
• Pledged to allocate 240 million to facilitate
  rapid adoption and sustainability of health
  information technology for hospitals, physician
  groups, physicians, and other healthcare
  professionals

17
EMR - VaVISTA for Windows
18
personal health records -- not waiting for
healthcare to do it...
19
Google Health
http//videocast.nih.gov/podcast.asp?14196
19
20
20
21
2007 -- Open Source EMRs emerge
22
PART II -- Standards

• why have standards anyway?

22
23
The standards process

• Three ways a specification becomes standard
• Ad hoc group of interested organizations agree
  on the standard specification (ACR/NEMA)
• De facto A single vendor controls the market to
  the extent that its product is a standard (MS
  windows, ArcView, MatLab)
• Consensus organization formed by volunteers from
  interested parties (HL-7)

24
SDO

• Standards Development Organizations groups that
  exist to develop and promote a standard
• Examples with work/standards in healthcare
• Health Level - 7
• American National Standards Institute (ANSI)
• International Standards Organization TC 215
• American Society for Testing and Materials (ASTM)
• National Institute of Standards and Technology
  (NIST)
• Workgroup for Data Interchange (WEDI)
• International Standards Organization (ISO)
• European committee for standardization technical
  committee 251 (TC 251)
• CEN 13606

25
Data exchange standards

• Most information system infrastructures consist
  of a collection of systems rather than a single
  vendor system.
• System to system exchange of information is
  common healthcare systems
• LIS-EMR(results), EMR-LIS (orders), ADT-LIS,
  ADT-RIS, ADT-EMR, LIS-FIN (billing), etc...
• A standard method and format of exchange reduces
  the cost of developing interfaces between
  systems

26
Cost of clinical interfaces
courtesy NeoTools Inc.
27
HL-7

• Health Level 7
• Name refers to the highest level of the ISO
  communications model, which is focused on the
  definition of the data to be exchanged, the
  timing of the interchange, and semantic
  interoperability/portability

ISO Reference Model For Communication
28
HL-7 History

• 1987 an ad hoc standards group formed after an
  effort to develop an integrated health
  information system
• HL-7 versions
• Sep 87 HL-7 v1.0
• Sep 88 HL-7 v2.0
• Dec 94 HL-7 v2.2
• Mar 97 HL-7 v2.3
• 2003 HL-7 v2.5 (current standard)
• 2005 HL-7 v3.0 (in progress)

29
HL-7 v2.x example

• MSH\LABUCDHS200608151347ORMO01RESULTS
  .2.25404D2.1
• PID13340464XXTESTFLABSAHARA19751012F
  078000643078387483748
• ORCSN200608151347
• OBRL25407LABBG00001ARTERIAL BLOOD GAS
  COXL200608151345A.DS
• 20060815134500000STAFF DOCTOR0815BG00002
  R199.8000200608151347BGCOMP1R
• MSH\LABUCDHS200608151351ORMO01RESULTS
  .2.25409D2.1
• PID3340464XXTESTFLABSAHARA19751012F
  078000643078387483748ORCSC2006081513
  51
• OBR4391L25409LABBG00007HEMOGLOBIN, WHOLE
  BLOODL200608150500INTERF
  ACE20060815135100000STAFF
  DOCTOR0815BG00003R199.7200200608151351BG
  COMPR
• MSH\LABUCDHS200608161335ORMO01RESULTS
  .2.25456D2.1
• PID3342143XXTESTFHBCWILLS19720501F
  038001139088222334444
• ORCSC200608161335
• OBR4639L25452LABSC00051TSH
  (SENSITIVE)L200608161306IS.JM20060

30
HL-7 v2.x specifics

• Messages the file that has information about a
  healthcare event and which is sent from one
  system to another
• Segments
• Are specific to a particular aspect of the
  healthcare event
• Types of segments (identified by segment ID)
• MSH - (message header) - information about the
  type of the message, time sent, etc.
• ADT (admission/dc/xfer) information about an
  admission/discharge/transfer event
• PID - (patient id) patient demographics, name,
  MR number, address.
• PV1 - (patient visit) - information regarding the
  hospitalization such as doctor, referring doctor,
  location assigned, etc..
• OBR - (observation request) information about
  an observation order such as a laboratory test,
  provides information about the test requested
• OBX - (observation) information about an
  observation, in some cases results from a
  laboratory test
• EVN (event) information about the healthcare
  event such as date/time, type of event
• Field a data element within a segment

31
HL-7 Segments and Fields
Formatting characters used for the message

• MSH\LABUCDHS200608150807ORUR01RESULTS
  .2.25392D2.1
• PID3340745XXTESTJOSHUA19801015M
  010000121748572384083
• PV1ADM IN
• OBR16983L25359LABCF00003HIV RAPID
  TESTL2006081413502006081413
  5003310ANDERSONJOHNT0814CF00006R1010.01
  50200608141351CFFR
• OBX1STLABCF00003HIV RAPID TESTLSee EIA
  ResultNEGATIVEF1010.0150
• Y05D0615657IS.JMM
• NTE1
• NTE2
• NTE3
• NTE4
• NTE5HIV results phoned
• NTE6To JOHN M
• NTE7By MAHONEY,JOHN
• NTE8Date/time 08/14/06 1351

Observation segment
Observation display label
Observation value
31
32
The importance of data types

• What is a data type?
• Describes the kind of data you are representing
  with the value you have stored
• For example
• A stored value of 95.6 could mean
• 95.60 ninety five and 60/100ths of a second
  (time)
• 95.60 meters (unit of measure)
• 95.60 ICD-9-CM code (coded entry)
• 95.60 ninety five dollars and sixty cents
  (money)
• Data typing allows for the recipient system to
  also have the context of the data being provided
  so it can be appropriately processed

33
HL-7 Value Types Table 0125
33
34
Issues with HL-7 v2.x

• Structural Ambiguity - There are many ways to
  construct a valid message for the same
  observation, order, etc..
• Semantic Ambiguity - No requirements regarding
  concept encoding, so even if message is
  constructed the same, the meaning may be
  different CBC vs Complete Hemogram
• Model inconsistencies - Does not enforce a common
  model of care so different systems may model
  information differently, leading to
  incompatibilities (ie, panels vs. result)
• Frame based messages error prone - Difficult to
  debug, one omission makes entire message invalid
  due to frame shifting

35
Why HL-7 v3

• Develop a standard that
• Allows for system interoperability (rather than
  simply message exchange)
• Restricts options so interface building can truly
  be develop once, use many times
• Move towards a plug and play infrastructure
  (modular design)
• Take advantage of prevailing design architectures
  and methodologies

35
36
HL-7v3 basic design

• Uses an object oriented modeling approach
• Models are constructed of the clinical domain
  using a standard diagramming language (UML)
• UML fits well with an object oriented design
  perspective (UML was originally built for OO
  modeling)

36
37
HL-7 v3 - Reference Information Model
37
Andrew Hinchley. Understanding Version 3 A
Primer on the HL-7 Version 3 Communication
Standard Alexander Monch Publishing, 2005. ISBN
3-933819-19-9
38
The basic object model design
38
Andrew Hinchley. Understanding Version 3 A
Primer on the HL-7 Version 3 Communication
Standard Alexander Monch Publishing, 2005. ISBN
3-933819-19-9
39
HL-7v3 Example model
39
Andrew Hinchley. Understanding Version 3 A
Primer on the HL-7 Version 3 Communication
Standard Alexander Monch Publishing, 2005. ISBN
3-933819-19-9
40
HL-7 interfaces today

• s

• Relative proportions of use of different HL-7
  versions v1.0 - v3.0
• Most interfaces are v2.3.1 or v2.3 today

courtesy NeoTools Inc.
40
41
Part III Standardizing what we record

• Why is this important to standardize how we
  represent information?
• Key to ensuring correct interchange of
  information with another system
• Necessary for the systems to correctly label
  tests and file them in the appropriate places
  internally
• We now know it is not just a labeling issue
  since the label implies the meaning of what is
  being displayed
• It is really about concept representation, and
  making sure the conceptual information is being
  preserved among all users (readers) of the data

42
Coding vs. Terminology

• The distinction is largely arbitrary, but here is
  what I use...
• Coding systems
• goals
• coding function - assign an alphanumeric code
  to represent data so it can be represented
  consistently by different parties
• classification function - assign some
  categorization to the codes so the data they
  represent can be grouped (classified)
• Example -- ICD
• Terminology systems
• goals
• coding function
• classification function
• terminology function - allow for alternative
  representations (synonymy)
• knowledge function -- some include logic rules
  for constructing new concepts (compositional
  systems). These logic rules can be checked
  using software in order to ensure the
  classification is correct
• would allow a system to identify the following as
  not making sense broken brain, tear of the
  femur, etc..

42
43
Coding systems to know about

• the International Classification of Disease (ICD)
• evolved from the need to begin classifying
  deaths so as to better understand mortality
• Brief history
• England began collecting information on Births
  and Deaths through parishes as a way of
  understanding the health of the population (began
  largely due to the plague). Death reporting --
  Bills of Mortality (16th century)
• Was evident that allowing free text reporting, as
  done with Bills of Mortality, made aggregation
  very difficult
• Eventually, William Farr, a medical
  statistician in the General Register Office of
  England and Wales began proposing a uniform
  classification system
• The International Statistical Congress charged
  Jacques Bertillon, Chief of Statistical Services
  for Paris, with chairing a committee that
  included Farr. The committee was charged with
  developing a classification of death (1853)
• The first version was eventually adopted by the
  ISI in 1893 -- The ICD
• ICD, 9th revision (ICD-9) was published in 1975
• ICD, 10th revision (ICD-10) was published in 1990
  and is the international standard
• Mortality reporting by the US National Center for
  Health Statistics (NCVHS) is done using ICD-10

43
44
ICD-9-CM
4 digits original ICD-9

• 003 Other Salmonella Infections
• 003.0 Salmonella Gastroenteritis
• 003.1 Salmonella Septicemia
• 003.2 Localized Salmonella Infections
• 003.20 Localized Salmonella Infections, NOS
• 003.21 Salmonella Meningitis
• 003.22 Salmonella Pneumonia
• ..
• ..
• ..
• 003.8 Other specified salmonella infections
• 003.9 Salmonella infection, unspecified

5th digit added by HCFA (ICD-9-CM)
44
45
Issues with the ICD system

• Code specifies hierarchical position, so you can
  run out of numbers
• It is not combinatorial, so a modified type of
  disease requires a new code -- can dramatically
  increase the number of codes
• 47 codes for myocardial infarction
• It has rules for inclusion/exclusion which are
  not logically built into the structure, so cannot
  be computed
• Can be ambiguous -- NOS (not otherwise specified)
• Not comprehensive ICD-9-CM has 16,000 codes
• Originally designed for causes of death - has
  diseases, does not have other things we need to
  represent (rule out MI, headhache, etc..)

46
Terminology systems

• The Systematized Nomenclature of Medicine,
  Clinical Terms (SNOMED-CT)
• Evolved over many years (since 1965)
• Recently re-designed by the College of American
  Pathologists and merged with NHS-CT (formerly
  Read Codes) -- 1999
• Subsequently licensed by the National Library of
  Medicine for free use throughout the U.S.
  (2002-2007)
• Today (May 2007) -- purchased by International
  Health Terminology Standards Development
  Organization (IHTSDO - http//www.ihtsdo.org/),
  an organization formed by several nations for the
  purposes of standardizing terminologies
• countries participating can release SNOMED CT
  free for use in their jurisdictions
• current member countries Australia, Canada,
  Denmark, Lithuania, Netherlands, New Zealand,
  Sweden, UK, United States
• Logical Observation Identifiers Names and Codes
  (LOINC) http//www.loinc.org
• Developed by Regenstrief with funding from the
  National Library of Medicine
• The purpose of the LOINC database is to
  facilitate the exchange and pooling of results,
  such as blood hemoglobin, serum potassium, or
  vital signs, for clinical care, outcomes
  management, and research.

46
47
SNOMED CT
48
UMLS -- another system

• Project started by the National Library of
  Medicine in 1986
• Objective
• to solve what is the most fundamental barrier to
  the application of computers in medicine namely,
  the lack of a standard language in medicine
• Dr. Donald Lindberg, 1985

48
49
UMLS Structure

• 3 main components
• UMLS Metathesaurus
• Semantic Network
• SPECIALIST Lexicon and tools

50
UMLS Metathesaurus

• Built as a thesaurus a dictionary of concepts
  in which different strings are linked to
  concepts.
• Two strings pointing to the same concept are
  synonyms (same conceptual meaning)
• Strings (ie, phrases) come from the source
  vocabularies

50
51
UMLS Semantic Network

• Entity     Conceptual Entity         Idea or
  Concept             Functional Concept
              Qualitative Concept            
  Quantitative Concept             Spatial Concept
                  Body Location or Region
                  Body Space or Junction
                  Geographic Area                
  Molecular Sequence                     Amino
  Acid Sequence                     Carbohydrate
  Sequence                     Nucleotide Sequence
          Finding             Laboratory or Test
  Result             Sign or Symptom        
  Organism Attribute             Clinical
  Attribute         Intellectual Product
              Classification            
  Regulation or Law         Language         ..

51
52
UMLS Lexicon

• Developed to provide lexical information needed
  for the SPECIALIST natural language processing
  system
• Includes entries for each word or term
• Syntactic information
• Morphological information

53
Finally, some humor

• Even great tools cant protect against a lack of
  common sense....

53