module i unit 02 en v01.3
TRANSCRIPT
-
8/4/2019 Module i Unit 02 en v01.3
1/32
1
HL7 E-LEARNING COURSE
I n t roduc t ion
In t roduc t ion t o Voc abu lar ies in
Hea l thcare
-
8/4/2019 Module i Unit 02 en v01.3
2/32
2
HL7 E-LEARNING COURSE
MODULE I - INTRODUCTION
UNIT I.1 INTRODUCTION TO HEALTHCARE INTEROPERABILITY
UNIT I.2 INTRODUCTION TO VOCABULARIES IN HEALTHCARE
UNIT I.3 INTRODUCTION TO UNIFIED MODELING LANGUAGE (UML)
UNIT I.4 INTRODUCTION TO EXTENDED MARKUP LANG UAG E (XML)MODULE V HL7 V2.x
UNIT V.1 INTRODUCTION TO HL7 VERSION 2.X, DATA TYPES, ACK
UNIT V.2 HL7 V2.X: PATIENT ADMINISTRATION, ORDERS AND RESULTS
UNIT V.3 HL7 V2.X: Z-SEGMENTS / IMPLEMENTATION / PROFILES
UNIT V.4 HL7 V2X.XML: XML IMPLEMENTATION OF V2.X MESSAGING
MODULE T HL7 V3
UNIT T.1 INTRODUCTION TO HL7 V3
UNIT T.2 REFERENCE INFORMATION MODEL RIM / DERIVED MODELS
UNIT T.3 HL7 V3 DATA TYPES AND THEIR XML REPRESENTATION
UNIT T.4 HL7 V3: FROM THE MODEL TO THE MESSAGEMODULE C HL7 CDA R2
UNIT C.1 INTRODUCTION TO HL7 CDA R2
UNIT C.2 CDA R2 ARCHITECTURE: HEADER, BODY AND ENTRIES
UNIT C.3 CDA R2 IMPLEMENTATION GUIDES
UNIT C.4 CDA R2 ENTRIES: CLINIC AL STATEMENT
Language: EN (ENGLISH) Version: 1.3 Year: 2010
-
8/4/2019 Module i Unit 02 en v01.3
3/32
3
TABLEOFCONTENTS
Introduc tion ____________________________________________________________________1
Introduc tion to Voc abula ries in Hea lthc are _____________________________________1
PART 1- INTRODUCTION TO VOCABULARIES _______________________________________6
INTRODUCTION _________________________________________________________________6
Some definit ions ______________________________________________________________6
Na tura l Language Prob lems___________________________________________________7
KNOWLEDGE REPRESENTATION _________________________________________________ 11
NEED FOR CONTROLLED VOCABULARIES________________________________________ 15
EXAMPLES OF CONTROLLED VOCABULARIES____________________________________ 16
Cont rolled vo cabula ries for d iagno sis ________________________________________ 16
SNOMED CLINICAL TERMS (SNO MED CT) ____________________________________ 16Interna tiona l Classifica tion o f Diseases ICD (10 9-CM) _______________________ 19
ICPC-2 (Interna tiona l Classifica tion of Primary Care) ________________________ 19
DRG (Dia gnosis Rela ted G roups) ___________________________________________ 20
Vo c abula ries for Drugs ______________________________________________________ 20
ATC (WORLD HEALTH O RGANIZATION)______________________________________ 20
NATIONAL DRUG CODES (NDC) ____________________________________________ 20
Cont rolled Voc abula ry fo r Laboratory________________________________________ 20
LOGICAL OBSERVATION IDENTIFIERS, NAMES AND C ODES (LOINC) ___________ 20
Cont rolled Voc abula ries for Proc edures______________________________________ 21Current Proc edura l Termino logy (CPT-4) ____________________________________ 21
HCFA COMMON PROCEDURE CODING SYSTEM (HCPCS) ____________________ 21
Controlled Voc abula ries fo r Nursing __________________________________________ 22
OTHER CONSIDERATIONS ______________________________________________________ 23
Unified Med ic a l Langua ge System (UMLS) ____________________________________ 23
PART 2 - HL7 VO CABULARY DEFINITIONS ________________________________________ 25
HL7 VOCABULARY WORKGROUP_______________________________________________ 25
Mission ___________________________________________________________________ 25Stra tegy __________________________________________________________________ 25
HL7 VOCABULARY MODEL _____________________________________________________ 25
Conc ept Domain _________________________________________________________ 25
Code System _____________________________________________________________ 26
Concept _________________________________________________________________ 26
Externa l Code Systems ____________________________________________________ 26
Interna l Code Systems_____________________________________________________ 27
Coded Conc ept __________________________________________________________ 27
Va lue Sets ________________________________________________________________ 28
Bind ing Rea lms____________________________________________________________ 29
-
8/4/2019 Module i Unit 02 en v01.3
4/32
4
COMMON TERMINOLOGY SERVIC ES____________________________________________ 29
CTS ARCHITECTURE________________________________________________________ 30
Some C TS Runtime M essage API Examples__________________________________ 31
CTS Imp lem enta tions and rec om me nded (non-manda tory) rea d ings________ 31
-
8/4/2019 Module i Unit 02 en v01.3
5/32
5
-
8/4/2019 Module i Unit 02 en v01.3
6/32
6
PART 1- INTRODUCTION TO VOCABULARIESIf we c annot na me it, we c annot c ontrol it, financ e
it, resea rch it, tea c h it or develop pub lic polic ies...
Norma Lang , nursing p rofessor
INTRODUCTION
In Part 1 of the introduction to the world of health information standards, you
learned that systems interoperability has two components: functional
interoperability and semantic interoperability. In this part, we will discuss the
semantic component, i.e., how to ensure common understanding of information
sent using the sta ndards d isc ussed in Part 1 by using c ontrolled voc abularies.
The o b jec tive of this sec ond part is for you to understa nd the imp lica tions of using a
c ontrolled voc ab ula ry, wha t c ommon mistakes are ma de , and how b est to a c hievesemantic interoperability.
You will see some examples of controlled vocabularies and the way they relate to
HL7.
La ter in this c ourse, we will discuss how to ac hieve func tiona l interop erab ility throug h
imp lementa tion guides.
Reme mb er that semantic interoperability is the way in which, once da ta has be en
collected, information can be meaningfully interpreted and incorporated into the
receiving system. In order to achieve this type of interoperability for any aspect ofthe hea lthca re reco rd, we need to use the sam e voc abulary.
The use o f com puters in ma naging pa tient administration informa tion has served to
highlight the complexity of language handling, especially medical vocabularies.
Therefore, it is nec essary to design voc abulary co ntrol strateg ies so tha t the c linica l
informa tion stored in Hea lth Informa tion Systems c an b e shared , either for
administrative purposes or in making clinical decisions (perhaps incorporating the
use of automated decision-support tools) in ways that maximize the quality and
safety o f pa tient c are.
In order to understand the problems of language, we re going to imagine a screen,
but we ll c ome ba c k to that later.
Let s review som e d efinitions befo re m oving o n.
Some definitions
Word: Set o f cha rac ters expressing a c onc ep t tha t can be found in a d ic tionary.
Term: One or more w ords used as a unit.
Concept: An idea , ac tion or thing w ith one single me aning.
-
8/4/2019 Module i Unit 02 en v01.3
7/32
7
Synonyms: Two or more te rms for the same c onc ep t.
Homonym: Words or terms tha t a re spelled and p ronounced a like but have d ifferent
meanings.
Code: An expression of a term or conc ep t in a fo rma l nota tion or classific a tion.
Natural Lang uag e Problem s
A few pa ragraphs ag o w e a sked you to imag ine a sc reen. Did you?
You have probably all come up with different screens since many interpretations
are possible (as we see in Figure 1); there are window screens, movie screens, flat-
sc reen TVs, CRT sc reens, LCD sc reens on c ameras, even sunsc reens.
Figure 1: Thinking about a sc ree n
Why is it that everybody does not understand the same thing by screen? Maybe
bec ause more co ntextual informa tion is needed spec ifying w hat type of sc reen we
wa nted you to imagine.
The rea lity is tha t pe op le c om munica te a nd und erstand eac h other not only
because they use the same language and words, but also because there is a
c om mo nly understood c ontext surround ing the ir me ssages.
Figure 2 represents the communication process among people. Unlike people,
c om puters cannot b e imp lic itly relied on to infer the c ontext in which message s a re
delivered. Thus the imp ortanc e both of explic itly sta ting (and ma inta ining) c ontext
and of c rea ting a nd ma intaining c ontrolled voc ab ularies.
Send er and receiver must be fam ilia r with both the langua ge (enc od ing) and the
c ontext, and they must be identica l for both p arties.
-
8/4/2019 Module i Unit 02 en v01.3
8/32
8
Figure 2: Comm unica tion p roc ess betw een peop le
Knowing the environment in which we communicate, or the audience we
c om munic a te w ith, is essent ial in the c om munic a tion proc ess.
Bac k to our example: If the c ontext is a d rugstore a nd we should ha ppen to ask for
a sc ree n, the sa lesperson will likely offer us a sunsc reen (Figure 3) and not a liquid
c rysta l disp lay. How ever, in a c om puter sto re, the sa lesperson will probab ly d irec t us
to the latest line of LCD monitors, and in an appliance store we are likely to be
presented with a sta te-of-the-art TV.
Figure 3: Sunsc reen
When the communication channel cannot fully depend on external context, as is
the case with computerized information systems, the message issuer must be as
specific as possible with regard to both data and context so that the recipient can
unde rstand the m essage without a mb iguity.
For exam ple, a first a nd last na me suc h a s John Smith m ay refe r to tho usands of
different people, but if we include a standardized identification number and the
date of birth we should be able to identify a single person. Even so, when trying to
be spe c ific in hea lth-c are c ommunic ations, we c an enc ounter amb iguity problems
inherent to natural language that we must also resolve when dealing with medical
language.
Ambiguity occurs when a language or vocabulary allows for more than one
meaning for a single word or expression of a concept. Often, resolution is only
possible when the c ontext or situa tion is known.
-
8/4/2019 Module i Unit 02 en v01.3
9/32
9
This ambiguity a rises from :
Synonymy : Relationship of similarity in the meaning of certain words called
synonyms; for examp le, feverand pyrexia.
Polysemy: The c apac ity of a single te rm to express ma ny d ifferent m ea nings. Forexample, Paget s d isease refers to two c om plete ly d ifferent a ilme nts, one a ffec ting
the b rea st and the o ther the bone s (and has add itiona l me anings besides). Simila rly,
mouth ma y refer to the orifice o n one s fac e o r the op ening o f a c ave or river.
Homonymy: Two terms with the same pronunc ia tion a nd / or spelling tha t mea n
d ifferenc e things. The d ifferenc e b etw een hom onymy a nd polysemy is tha t in
polysemy there is a single source word, whereas in homonymy there are two or
mo re unrela ted source wo rds. Thus, in o rder to identify a hom onym we m ust stud y its
etymology. For example, the verb rose (past tense of rise) has a different source
than the noun rose(flower).
Let's look at a more concrete example of synonymy. If a doctor wants to
c om munica te to a nurse tha t a pa tient is in a febrile sta te, the d oc tor may use a ny
of the fo llowing expressions:
The pa tient ha s hyperthermia", is pyret ic , is in a febrile sta te , or has a
temp erature a bove 100 F .
-
8/4/2019 Module i Unit 02 en v01.3
10/32
10
Despite the synonymy , the two professiona ls understand each other.
Rep resenting the same c onc ep t with different w ords ma y not b e a p rob lem , as long
as there is a context facilitating the communication process. As we have already
said, communication between professionals and computers requires specifications
that may be omitted in c omm unica tion amo ng peo ple.
Lets look at several examples of possible HL7 segments that could transmit the
blood type of a p atient:
OBX| 1| CE| ABO^ABO GROUP| | O^Type O|
OBX| 1| CE| BLDTYP ABO GROUP| | TYPEO^ Typ e O|
OBX| 1| CE| ABOTYPE ABO GROUP| | OPOS Type O|
Even w ithout a know led ge of the HL7 standard , we c an infer tha t ea c h of the threesegments above expresses a patients O blood type, but what is obvious for a
human be ing is not so for the computer.
ENCODING is the mec hanism tha t allows com puters to understand languag e.
-
8/4/2019 Module i Unit 02 en v01.3
11/32
11
KNOWLEDGE REPRESENTATIONFigure 4 presents the so-called knowledge pyramid, which represents actors
c om munica ting using a n enc od ing system and the func tion relating them.
Campbell KE et al., Representing thoughts, words, and things in the UMLS. JAMA
1998 Sep-Oct;5(5):421-31.
Figure 4: Knowledge Pyramid
Objects are the rea lity unit, bo th c onc rete and abstrac t. These o b jec ts a re
described through conceptsor ideas, which are units of thought made up by
means of c om mo n a nd p rom inent p rop erties of a set of o b jec tives. To rep resent
these ideas or thoughts we use symbolsor terms that are, in short, the linguistic
expressions of a conc ep t.
One of the biggest challenges of medical information is the representation of
medical knowledge in such a way that it can be reliably manipulated byinformation systems.
Decision support modules are essential components of many health information
systems. These a pp lic a tions interpret pa tient d a ta entered through the e lec tronic
clinical record and access knowledge bases (pharmacologist databases,
electronic books, any type of information with scientific evidence levels) in order to
gene ra te rem inders and wa rnings. They need stric tly controlled voc abula ries,
because just one natural language ambiguity could generate false positives or
otherwise imp eril pa tient sa fety.
To a rrive a t a p rop er understand ing of the com ponents involved in the use o f ac ont rolled voc abula ry, it is nec essary to c lea rly de fine som e rela ted aspec ts.
-
8/4/2019 Module i Unit 02 en v01.3
12/32
12
Let s review som e more d efinitions in the c ontext o f Figure 5:
Figure 5: Voc abulary Co ntrol
Natural Language: The universe o f expressions used to c om munica te ideas. It may
inc lude words from na tive languages (Spanish, Portug uese, Eng lish, etc ).
Generally, the na rra tive text of m ed ica l rec ords is expressed in na tural language.
Vocabulary: A set of terms used or available for use by an individual or group, or
within a spec ific type o f work or know led ge field (d om ain). The voc abula ry of aphysician is d ifferent from the voc abula ry of a n a rc hitec t.
Simila rly, we see ma ny different typ es of voc abula ries used in a me d ica l rec ord;
from laboratory data to medication administration reports, to natural language
narra tive text. Bec ause pa tients do not usually c om munica te using a m ed ic a l
vocabulary, the scope of the medical vocabulary must extend to the natural
lang uag e sphere.
Controlled Vocabulary: An approved set of terms constrained for use in a specific
environment. Within an electronic user interface, a controlled vocabulary may
prov ide a list o f terms from whic h the user choo ses a term to rep resent a fac t or rea l-
wo rld situa tion. Numerous controlled voc abula ries have been d eve lop ed for use in
c linica l ap p lic a tions.
Lets see how controlled medical vocabularies are classified according to their
c harac teristics or orga niza tion:
Terminology or thesaurus: The set of a ll the words or word groups with
spec ific m ea nings in a d om ain. These sets of w ords a re c a lled terms .
Taxonomy: A terminology ordered according to the logical relations ofterms regarding a particular point of view. For example, if we consider the
terminology of a diagnosis and order the relevant terms according to their
etymology instea d of the ir ma nifesta tions, we a re c rea ting a taxonomy.
Nomenclature: A sub-set of a terminology for a given domain. It is made up
of the terms (or group of terms) of a terminology and their relationships. A
nomenclature does not occur naturally as a result of use or custom, but is
created by some official body that organizes (and/or standardizes) the
terminology.
Classifications: A c lassifica tion is an orderly system of c onc ep ts belong ing toa d om ain, with imp lic it and e xp licit order princ iples. Their definition dep end s
-
8/4/2019 Module i Unit 02 en v01.3
13/32
13
on their expected use. A classification does not try to be extensive, that is,
does not include all the concepts of a domain. Its representation does not
have to b e unequivoc al and c an b e a mb iguous. The content of both a
terminology and a nomenclature c an b e orde red unde r a classification.
Depend ing o n its app lica tion in an informa tion system , a c ontrolled voc abula ry ma ybe c harac terized as one o f the follow ing:
Interface voc ab ulary or Terminology: A controlled vocabulary from which
the user can choose a term from a list in order to enter information into a
system. Suc h a group ma y include a ll the lexic a l varieties, ac ronym s,
abbreviations and ja rgo n tha t a re employed by the system s users, eac h w ith
an implicit meaning given by the context they are used in. Within the same
system several Interface terminologies can be used, having been adjusted to
a spec ific type of user.
Referenc e vocab ulary o r Terminology: A controlled vocabulary used for a
mo re deta iled rep resenta tion of da ta in a n informa tion system . The referenc e
terminology is used to store data in the data base. Multiple interface
terminologies converge into a single reference terminology. Generally, a
nomenc la ture is used tha t must inco rpo ra te a ll the dom ains in whic h a set of
interfac e te rminolog ies app lies.
Output vocab ulary: Terminolog ies or c lassifica tions used for informa tion
ana lysis. Terminolog y informa tion systems must p rov ide tools to rec over
informa tion rep resented with the referenc e terminology and transform it into
the d esired output voc ab ulary.
-
8/4/2019 Module i Unit 02 en v01.3
14/32
14
Voca bulary c ontrolis the strategy by whic h automated systems imp lem ent solutions
to the na tural language p rob lem . This strategy is app lied by limiting voc abularies to
a strict set of terms (thus the phrase controlled vocabulary) and involves agreeing
on the exclusive use of those terms included for information expression.
A c ontrolled voc ab ulary is the information-systems counterpart of natural languag eand consists of a com bina tion of restric ted terms and gram ma r rules. (Tab le 1)
The m a in c harac teristics of a c ontrolled voc abula ry inc lude the fo llow ing:
There is a stric t set of terms which a re unam biguous and ac c ura te.
The set o f te rms is sta nd ardized .
New c onc ep ts req uire integ ration into the voc ab ula ry, and ma y not be
introduced ad hoc.
Users req uire tra ining before e mp loying the voc abula ry.
Controlled vocabulary is a key component in achieving the interoperability of
health information systems.
Within the hea lthc are informa tion te c hnology field , controlled voc abula ries not only
facilitate system interoperability, but also enable statistical and epidemiological
analysis, reports for the decision-making process, planning of care and follow-up
strate g ies, etc .
Natural lang uag e Controlled voc ab ulary
Ambiguous, filled with synonymy and
polysemy.
Highly c ontext dep endent
Very expressive a nd flexib le
New c onc ep ts a re e asy to express
NOT sta nd a rdized Does not require specialized training
Sub-group of na tura l langua ge
c onsisting of:
Restricted terms
Gramma r rules
With restriction in the different
levels Rig id , non-ambiguous, ac c ura te
Sta nda rd ized
New concepts require
integ ra tion in the voc abula ry
Requires training before being
used
Tab le 1: Natural langua ge vs. Cont rolled voc abula ry
-
8/4/2019 Module i Unit 02 en v01.3
15/32
15
NEED FOR CONTROLLED VOCABULARIESWe have established that controlled vocabularies are essential to system
interoperability. What else makes these vocabularies essential?
They c an be used for:
> Standard izing free text or struc tured c ontent o f the med ica l rec ord
> Rep resent ing c linic a l ob servat ions and eva lua tions.
> Coding tests and results.
> Identifying drugs.
> Intercha nging c linic a l da ta in rea l time .
> Rep resenting syntac tic a nd semantic a spec ts of med ic a l c onc ep ts.
> Rec ove ring and ana lyzing da ta , and supp orting the dec ision-ma king p roc ess.
Having discussed the importance of using controlled vocabularies, lets examine
va rious exam ples of terminolog y sta ndards (see Tab le 2 be low ).
The c hoic e o f one voc ab ulary over ano ther will be influenced by ea ch
voc abula rys c harac teristic s. It is imp ortant to bea r in mind tha t ea c h voc abula ry is
c rea ted with a particula r purpose. For example, it is gene ra lly not a dvisab le to use a
vocabulary that was created for medical practice billing for epidemiology
purposes.
Coding system Objec tive
ICD-9; ICD-9-CM
ICD-10 (ICD-10-CM, ICD-10-PCS)
For abstraction and classification for
epidemiological purposes or
administra tive reimb ursem ent.
DRGs
AP-DRGs, APR-DRGs
In case mix analysis and grouping of
med ic a l d iag noses.
SNOMED CT For c linica l and d irec t ca re of
patients.
MESH, ULMS For literature searches.
LOINC To exchange c linica l ob servat ions in
HL7 messages.
Tab le 2: Cod ing system ob jec tives
-
8/4/2019 Module i Unit 02 en v01.3
16/32
16
EXAMPLES OF CONTROLLED VOCABULARIESIn order to understand vocabularies better, lets examine some vocabulary systems
in mo re d eta il.
Controlled vocabularies for diagnosis
SNOMED CLINICAL TERMS (SNOMED CT)
SNOMED is a m ed ic a l nomenc la ture d eve lop ed by the Colleg e o f Americ an
Pathologists (CAP). It includes terms of all medical domains, including veterinary
medicine.
Its c urrent version, SNOMED CT, results from its c ombinat ion w ith the Read Co des,
the U.K. clinical use nomenclature, to create an extensive and detailed
nom enc la ture tha t is strong ly c linica lly oriented with an inte rna tiona l base.
In 2007 SNOM ED was transfo rmed into the IHTSDO (Internationa l Hea lth Termino logyStandards Developm ent Orga niza tion) and is no longe r dep end ent on CAP
(Colleg e o f Americ an Patholog ists). An interna tiona l boa rd with the p articipa tion o f
multip le c ountries with extensive experienc e in med ic a l informa tion tec hnology no w
ma intains this sta ndard terminolog y.
SNOMED CT is the richest voc abulary ava ilab le to d esc ribe c linic a l findings,
d isea ses, p roc ed ures, etc . Its ma in fea tures a re:
Its com positiona l foc us
Its interfac e a nd referenc e voc abula ry func tionalities
The c om positiona l foc us refe rs to the fa c t tha t it a llow s the c om bination of simp le
terms (lung + inflammation), or the addition of modifications to a concept (severe,
mild , sudden onset , etc ).
SNOMED CT is an excellent nom enc la ture to be used as a refe renc e vo c abulary,
due to its leve l of d eta il and qua lity of sem antic rela tions. Neverthe less, SNOMED CT
also includes interface functionalities, that is to say, for each concept there are
several possible descriptions that can be used as elements of an interface
voc ab ula ry in an imp lementation.
SNOMED CT c onta ins
More than 365,000 conc ep ts
Almost one million de sc riptions
Nearly one and a half million relationships
SNOMED CT is not me rely a d iagnostic nome nc la ture; it a ttemp ts to e mbrac e the
whole spectrum of controlled vocabularies within the healthcare domain. Its
365,000+ co nc ep ts a re group ed in hierarchies. Tab le 3 show s exam ples of ea c h:
-
8/4/2019 Module i Unit 02 en v01.3
17/32
17
Conc ep ts hierarchies Examples
Find ings Swelling of a rm
Disease Pneumonia
Proc ed ure/ intervention Biopsy of lung
Ob servab le ent ity Tumo r stage
Body struc ture Struc ture of thyroidOrganism DNA virus
Substanc e Ga stric ac id
Pharma c eutica l/ b iologic p rod uc t Tamo xifen
Spec imen Urine spec ime n
Qua lifier va lue Bila te ra l
Physica l ob jec t Suture nee d le
Physica l force Fric tion
Events Flash flood
Environments/geographical
locations
Intensive care unit
Soc ial c ontext Orga n do nor
Tab le 3: SNOM ED CT hierarchies and Exam ple te rms
One of the ma in c harac teristics of Snom ed CT is tha t not only does it provide a va lid
list of terms, but each term is also defined according to its relationship with the
others, and these relationships c an be und erstood by an informa tion system .
Here are some examples of the possible relationships used to describe diseases or
clinical findings.
place of finding
assoc iate d with
after its
ca usal agent
due to
assoc iate d morpho log y
seve rity
course
ep isod ic ity
pa thologica l proc ess
etiology
finding
followed by
For exam ple, w ithin SNOMED CT there is a rela tionship b etween the d iabetes
concept and the diabetic foot concept, through a due to attribute (diabetic
foo t - due to - d iabetes ). These sem antic relationships p rovide informa tion
whic h d ec ision support systems c an use to app ly rules.
SNOMED CT aims at transmitting ALL the c onc ep ts that have been expressed
throughout history in the health-care d omain unam biguously.
-
8/4/2019 Module i Unit 02 en v01.3
18/32
18
Thus, it is used as a refe renc e c linic a l terminology to rec ord a ll releva nt a spec ts of
med ic a l ca re in standard ized terms.
A user or interface terminology may incorporate jargons, localisms and lexical
va riants of e ac h institution. Therefore, a n interfac e terminology must a lways be
mapped to a reference terminology in order to be able to represent medicalknowled ge in de ta il. This refe renc e voc abula ry must me et the fo llow ing
characteristics:
Domain width: To a ntic ipa te a ll possib le o b jec ts or events tha t may be c ollec ted as
da ta the voc abula ry must rep resent the ent ire dom ain.
Non-redundancy: Mechanisms must be established that prevent multiple terms for
the sam e c onc ep t from b eing a dd ed to the voc ab ulary as different conc ep ts.
Synonymy : Non-unique multiple terms must be accommodated for the same
concept.
Non-vagueness: Incomplete meanings must be avoided. E.g.: ventriclecannot be
considered a completely meaningful concept, and not even a generic class of
anatomical term: it means nothing by itself unless it is clarified that it is a cardiac
ventricleor bra in ventric le.
Non-ambiguity: Each concept must have a single meaning; in case of homonymy,
each concept must be stripped of its ambiguity. E.g., Paget s d isea se" must be
divided into two concepts: Pagets disease of bone, Pagets disease of breast,
etc.
Multi-classification or polyhierarchy: The system must no t b e restric ted in suc h a way
that the same concept cannot be assigned to as many classes as necessary. E.g.,
pneumoniais a de sc endant o f bo th pulmonaryand infec tious d isea se.
Context consistency: Conc ep ts tha t exist in seve ra l classes must b e a llow ed to have
the same meaning in each of them. Inconsistencies must be avoided. E.g.,
c ortic osteroid, in the c ontext o f both hormones and anti-inflam ma tory drugs ,
must have ident ica l a ttributes, both for itself and its desc end ants.
Relationship explicitness: The mea ning of the relationships amo ng c onc ep ts must bec lea rly sta ted . E.g., the rela tionship b etw een staphyloc oc c i pneumonia and
pneumonia must be differentiated from the relationship between pneumonia
and sta phyloc oc c i: the fo rmer is a c lass relat ionship wherea s the la tte r is an
etiological relationship. Pneumonia "is a" staphylococci pneumonia, pneumonia "is
c aused by" stap hyloc oc c i.
Concept permanenc e: Old c onc ep ts must not b e d eleted . Rep lac ement o f existing
c onc ep ts by be tter conc ep ts must a lso b e supp orted : E.g., HIV infec tion rep laces
AIDS. There must b e e xp lic it links betw een the rep lac ed c onc ep t a nd the
replac ing conc ept.
-
8/4/2019 Module i Unit 02 en v01.3
19/32
19
The use o f not c lassified som ew here e lse must not be a llowe d; suc h terms c annot
be used as referenc e voc abula ry.
International Classifica tion o f Disea ses ICD (10 9-CM)
The World Hea lth Organiza tion (WHO) c rea ted the Interna tiona l Classifica tion of
Diseases (ICD) and has maintained it for more than 100 years.
Remember that a classification is an orderly system of concepts belonging to a
dom ain with imp lic it a nd explic it p rinc ip les of order and c lass definition, dep end ing
upon expe c ted use.
ICD originated as a c lassifica tion of c auses of d ea ths and is still used today to rep ort
morbidity. It was first edited in 1892; its most recent revision in 1992 was the tenth
(ICD-10).
In som e c ountries ICD wa s adapted to b ette r me et loc a l need s. For examp le, in the
U.S. co des we re a dded to ICD to a llow its use fo r b illing , and the C M (C linica l
Mod ifica tion) variant was c rea ted , c urrent ly in its 9th iteration (ICD-9-CM).
Version 10 (ICD-10) has no procedures, only diagnoses, and its categories are not
compatible with former versions, since it progressed from a numeric organization of
c hap ters to a n alpha numeric organiza tion.
ICD-10 is considered the w orldwide sta ndard for morta lity and m orbid ity rep orting.
ICPC-2 (Internationa l Classifica tion of Primary Care)
ICPC (Inte rna tiona l Classifica tion o f Primary Care) is a c lassifica tion c rea ted by theWorld Organization of Family Doctors (specifically the WONCA International
Classifica tion Com mittee) to be used exc lusively in p rimary c a re. It is the evo lution o f
mo re than 20 yea rs of WONCA e xperienc e in p rimary c are c lassifica tions.
It is made up of about 700 codes, with a sufficient granularity for use by general
practitioners in the outpatient environment. It allows representation of reasons for
c onsulta tion, diag nosis, diag nostic p roc ed ures, administrative p roc ed ures, etc .
It ha s a solid mapping w ith ICD-10 that a llow s using ICPC a s an acc ess me thodolog y
to ICD-10.
It is designed to be used op tiona lly within a c are e p isod es da ta model. Ep isod es
are made up of one or more co nsulta tions; ea c h consulta tion is c od ified acc ording
to the rea son for consulta tion, the d iagnosis and any resulting p lan. The ep isod es
are c ha ined , d isp laying the evolutionary proc ess of the d iagnosis.
For examp le, a c onsulta tion d ue to a d ry cough may be e ntered w ith a non-spec ific
"c oug h" d iagnosis and a thorac ic x-ray ma y be req uested . A sec ond c onsulta tion is
scheduled for interpreting the x-ray and a tumor is detected, resulting in a new
d iagnosis of lung c anc er . This ep isod e ha s two c onsultat ions, and it d em onstrates
the e vo lution of the symp tom in the final diagnosis.
-
8/4/2019 Module i Unit 02 en v01.3
20/32
20
DRG (Diagnosis Rela ted Groups)DRG (Diagnosis Related Groups) are a different kind of classification, designed for
b illing in the U.S. The DRG are c ombinat ions of a ll ICD-9 CM c odes assigned to a
hospitaliza tion ep isod e rep resented as a single cod e.
A DRG includes hospitalization episodes with different diagnoses, which might begrouped together with resource consumption criteria in one instance and with
c linica l c ha rac teristics in a sec ond insta nc e.
DRG is an isocost grouper. Each DRG has an economic value that determines
payment in the United Sta tes hea lth system.
Voc ab ularies for Drugs
ATC (WORLD HEALTH ORGANIZATION)
ATC is c onsidered the mo st imp ortant c ont rolled voc abulary for drugs. It is part o fthe WHO (World Hea lth Organiza tion) drug d ict iona ry.
It c lassifies d rugs ac c ording to Ana tomica l-Therap eut ic-Chemica l criteria , thus its
abbrevia tion ATC.
This voc abula ry has internationa l c harac teristics, co mb ining the c linic a l expe rienc e
of more tha n 34 c ount ries.
Every yea r, ATC is up dated with som e 2,000 new d rugs.
NATIONAL DRUG CODES (NDC)The Nationa l Drug Co des (NDC) is a d rug c lassifica tion o f the United Sta tes FDA.
Each drug is identified by an 11-digit code, made up of three parts. NDC has flaws,
such as the inability to group certain similar drugs, which has led to the
development of a new, more functional vocabulary called RX-Norm, which has a
muc h mo re solid sem antic structure.
Controlled Vocabulary for Laboratory
LOGICAL OBSERVATION IDENTIFIERS, NAMES AND CODES (LOINC)
Logical Observation Identifiers, Names and Codes (LOINC-http :/ / www.reg enstrief.org/ loinc / ) is a c lassifica tion for c linic a l observat ions. It is
primarily used for lab results, but can also be applicable to aspects of a physical
examina tion or any o ther c linic a l observation.
LOINC was deve lope d by the Reg enstrief Institute a t the University o f Ind iana .
For ea c h ob servation the follow ing is spec ified:
Prop erties type of m ea sure, e.g., concentration, numeric frac tion, etc .
Time po int in time
-
8/4/2019 Module i Unit 02 en v01.3
21/32
21
Samp le e.g ., b lood , c ereb rospinal fluid
Method e.g., qua lita tive, qua ntita tive, and it som etimes inc lude whe ther it is
automa tic or manual, etc.
These aspec ts a re rep resented within a text system with p red efined sep ara tors andabbrevia tions to spec ify ea c h d imension (see Figure 6).
Figure 6: LOINC
This is an exam ple of LOINC, showing first the d esc ript ion of the observa tion to b e
rep resented , in this c ase lab ob servat ions, and then the enc od ing in LOINC. Even if it
looks c onfusing a t first g lanc e, LOINC is very useful for automated informa tion system
processing.
Controlled Voc ab ularies for Procedures
Current Proc ed ural Terminolog y (CPT-4)
CPT (Current Proc ed ura l Terminology) w as c rea ted by the AMA (Americ an Me d ic a l
Association) to represent procedures and services provided exclusively by
physicians. It is used for reporting and reimbursement for medical practices, and its
use is compulsory in the outpatient environment for government reimbursement in
the United Sta tes.
HCFA COMMON PROCEDURE CODING SYSTEM (HCPCS)
HCPCS (HCFA Com mon Proc ed ure Cod ing System) is another p roc ed ure
c lassifica tion c omplementa ry with CPT. It is mainta ined by the U.S. government. The
first leve l c onsists of C PT-4 c od es; the sec ond leve l inc lude s the c odes
corresponding to procedures performed by personnel other than doctors and for
b illab le sup p lies suc h a s p rostheses.
This c lassifica tion is used for hospita l inpa tient scena rios.
-
8/4/2019 Module i Unit 02 en v01.3
22/32
22
Controlled Voc ab ularies for NursingTrad itiona lly c ontrolled voc abularies have foc used on pa tholog ies and symp tom s,
but have failed to contemplate the needs of nurses, who must address concepts
from different func tiona l evaluations such as ac tivity intolerance .
This ha s led to severa l initia tives to introduc e c ontrolled vocabularies in the nursing
dom ain, resulting in the c rea tion o f seve ra l non-trad itiona l classific a tions:
North Am erica n Nursing Diag nosis Assoc iat ion (NANDA) Taxonomy II
Nursing Intervention/ Outc om es Classific a tion (NIC/ NOC)
Omaha System
-
8/4/2019 Module i Unit 02 en v01.3
23/32
23
OTHER CONSIDERATIONSControlled voc abularies p resent seve ra l limitat ions when using the m, nam ely:
The terms inc lude d in the voc abula ries often d o not relate to the terms used
naturally by physicians, ma king their adop tion and use d iffic ult.
In the na tural expression o f d iagnoses or c linica l cond itions, mo difiers a re used
which o ften c annot b e expressed in the c ontrolled voc abula ry (mild , seve re, ac ute,
rec urrent, etc .)
Another freq uent p rob lem is tha t b illing needs d istort the c a teg ories of a
c lassifica tion, unifying d ifferent c linic a l co nc ep ts tha t ha ve the same signific ance for
billing within the same c a teg ory.
Unified Medical Language System (UMLS)
As the number of c ont rolled voc abula ries inc rea sed , there a rose seve ra l initiatives toprovide a unifica tion o f suc h voc abularies. The most imp ortant o f these is the UMLS
http://www.nlm.nih.gov/research/umls/, an NLM p rojec t (United Sta tes Nat iona l
Library of Med icine)
UMLS is made up of three parts:
1. Metathesaurus: A combined repository of all the vocabularies, with
interconnec tions. Here the m ost w idely know n voc abula ries c an be found , inc luding
their Spanish translations.
2. UMLS semantic network: Relationships tha t p rovide informat ion on the meaning ofconcepts.
3. Spec ialist lexic on: An application targeted at facilitating the association of
natural language terms with the words included in the Metathesaurus (only
ava ilab le in its Eng lish version).
There a re some limita tions to the use o f UMLS.
It o nly includes one -on-one relat ionships.
It d oes not p rovide for the add ition o f terms, but only for the use o f terms inc lude din the source vo c abula ries.
It d oes not have a hierarchy tha t unifies a ll the c onc ep ts, only those hiera rc hies
(where they exist) that a pp ly to ea c h source voc abula ry.
It is not extensible (a s op posed to SNOMED or LOINC).
These c ha rac te ristics have limited the utility of UMLS, which c hiefly func tions on ly as
a rep ository of controlled med ic a l voc abula ries.
-
8/4/2019 Module i Unit 02 en v01.3
24/32
24
UMLS ac cess is free of c ha rge fo r academ ic a nd resea rc h purposes, but to use a ny
vocabulary included in the metathesaurus in clinical practice one must seek a
license from the a uthor of the voc ab ulary and a fee o ften a pp lies.
-
8/4/2019 Module i Unit 02 en v01.3
25/32
25
PART 2 - HL7 VOCABULARY DEFINITIONSThis part o f the unit is devo ted exc lusively to HL7 definitions reg ard ing voc abulary.
HL7 VOCABULARY WORKGROUP
Mission
To ident ify, orga nize a nd ma intain cod ed vocabulary terms used in HL7 sta ndards.
Strategy
HL7 Voc abula ry d eve lop me nt stra teg y includes:
Referenc e existing voc abularies: SNOM ED CT, LOINC, RxNorm, FDA id entifiers Collaborate with other SDOs: NCPDP, DICOM, ASTM, X12, CEN, ISO Co llaborate with g overnment-sponsored effo rts: US: NCVHS Patient Medica l
Rec ord Informa tion (PMRI) sta ndards and Consolida ted Hea lth Informatics (CHI)
stand ards, Ca nad a Infowa y
Ad d va lue b y c rea ting linkage b etw een HL7 message s and existing voc abula ries Only c rea te new items tha t do no t a lrea dy exist Collab orate with voca bulary develop ers to a dd need ed c ontent to existing
vocabularies
Co de systems used in HL7 sta ndards need to b e reg istered for co nforma nc e tothe standa rd
Must be registered by a c o-cha ir or ad min
HL7 VOCABULARY MODEL
Concep t Doma in
An HL7 Conc ep t domain is a na med c ate go ry of conce pts that w ill be bo und to
one o r mo re coded elements.
Conc ep t dom ains exist to c onstra in the intent of the c od ed element while d eferring
the a ssoc ia tion o f the element to a spec ific c od ed terminology until la ter.
Conc ep t d omains are indepe ndent of a ny spe c ific voc ab ulary or cod e system.
They p rov ide a high leve l grouping for all things possible in a g iven dom ain from
which va lue sets will be c onstructed .
A c onc ep t dom ain rep resents an ab strac t conc ep tua l spac e such as "countries of
the w orld ," "the g end er of a p erson (fo r administrative p urposes)," languages of the
world, etc .
Every coded a ttribute ha s a voc abula ry constra int: this is the Conc ep t d om ain of
that attribute.
-
8/4/2019 Module i Unit 02 en v01.3
26/32
26
You ma y ask: But hey, wa it! What is a coded attribute or element?
We w ill loo k at tha t in more d eta il when reviewing V2.x messaging , HL7 V3 and CDA,
but think about c od ed a ttributes as the informa tion items or fields or co lumns in
da ta bases (or attributes in c lasses) de fined as c od ed : gend er, d iagno stic,
langua ge, etc.
Code System
A c od e system is a set of unique c od es tha t represent a c orresponding set of c lasses
in the rea l world , at va rious time s referred to as an onto log y, c lassific a tion,
terminolog y, or c od e set.
Within the HL7 c ontext, concep t c od es within a c od e set must not c hange
meaning.
Cod es ma y be a dd ed or retired Definitions ma y be c la rified New rela tionships ma y b e esta b lished Codes ma y not be reused Changing the mea ning o f conc ep t c od e(s) results in c rea tion of a new c od e
system
Code system s ma y co nsist of anything from a simp le c od e/ va lue ta b le to a c om plex
referenc e te rminolog y like SNOMED CT.
CODE VALUE
M MALE
F FEMALE
U UNDIFFERENTIATED
Code system exam ples: LOINC, ISO 3166-2 count ry codes, ICD 9-CM, SNOM ED CT,
ISO 4217 currenc y c odes.
Concept
A concept defines a unitary mental representation of a real or abstract thing, an
ato mic unit of thought.
It should b e unique in a given te rminolog y The mnemonic o r term referenc ing the c onc ep t ma y have synonyms,sometimes referred to as surface forms or interface vocabulary (with interface
me aning related to user interfac e o r the final user . These synonyms may have their
ow n ident ifier Desc ription ID
Externa l Code System s
HL7s policy is to use existing c od e systems whenever possible. HL7 will not d evelopits own c od e system unless a ll exte rnal possibilities have p roven unw orkab le.
-
8/4/2019 Module i Unit 02 en v01.3
27/32
27
Externa lly ma inta ined : HL7 refe renc es the c ontents of the c od e system b ut does not
ma intain or distribute content (e.g . , LOINC)
Internally ma inta ined : HL7 ma inta ins an imag e o f the c ontents for the c onvenienc e
of its mem bers (e.g., ISO 3166 c ountry codes)
Internal Code Systems
Interna l Co de Systems are c od e systems develop ed and ma intained within the HL7
organization.
We c an c lassify inte rna l co de systems as:
Structura l Codes: Significa nt p ortions of HL7s mod els a re rep resented as c onc ep t
codes.
Short Code Lists: Short tab les of c od es tha t a re tightly linked to HL7s models and
have not wa rranted external references (e.g., administra tive, gender, etc .)
Version 2 Tab les: Ma inta ined by HL7 (we ll go through these cod e systems when
studying V2.x)
Coded Concep t
A coded concept is a concept associated with an explicit machine-rec og nizab le ident ifier. It s unique within the Cod e system tha t defines it.
A Coded concept has the follow ing a ttributes: code- an identifier that uniquely names the class or "concept" withinthe c ontext o f the d efining Cod e system .
status- represents the current status of the Coded concept within theCod e system
A single conc ep t c an m ap a broad numbe r of terms. Exam ple: myoc ardial
infarction, cardiac infarction, heart attack, myocardial infarct, MI - Myocardial
infarc tion, infarc tion o f hea rt a ll map to c onc ep t ID 22298006 in SNOMED CT.
-
8/4/2019 Module i Unit 02 en v01.3
28/32
28
Value Sets
A Value Set represents a uniquely identifiable set of valid conceptrepresentations, where any concept representation can be tested to determine
whe ther or not it is a me mb er of the va lue set.
Complexity may range from a simple flat list of concept codes drawnfrom a single code system to an unbounded hierarchical set of possible expressions
drawn from m ultiple cod e systems.
Exist to constrain the content for a coded element in an HL7 staticmod el or da ta type property.
Cannot have null content, and must contain at least one conceptrepresentation where any given concept is generally (but not necessarily)
rep resented by only a single cod e w ithin the Value Set .
Is used to identify the subset of coded concepts important in a particularbusiness c ase:
The SNOMED CT Organisms tha t c orrespond to Nationa lly Rep orta b leDisea se infec tious agents
All LOINC tests for brucellosis The p ermissible op erat ions on a n appointment
-
8/4/2019 Module i Unit 02 en v01.3
29/32
29
A va lue set is de fined by a ma c hine-proce ssab le set o f me tad ata that ma y be
resolved at a point in time to a uniquely identifiable set of specific valid coded
c onc ep ts from one or more c od e system s.
Binding Rea lms
International differenc es in voc abula ry must b e ena b led This voc abulary is known as Rea lm-spec ializab le This is a pa ramete riza tion p ermitting internat iona liza tion o f vocabulary binding different value sets to the same message design being used in a different
country
Eac h HL7 Inte rnat iona l Affilia te has bee n assigned a Binding Rea lm a p riori Every Concep t Doma in used in a mo del must be b ound to a va lue set in thec ontext of a Binding Rea lm in order for the m od el to b e instantia ted
COMMON TERMINOLOGY SERVICES
It s an HL7 ANSI sta nd ard defining the minimum set o f req uirem ents for
interoperability across disparate health-care applications.
It s a spec ific a tion fo r acc essing te rminolog y content: The C TS identifies the
minimum set o f functiona lcha rac te ristics a te rminolog y resource must p ossess for
use in HL7.
It s a funct iona l mod el: Defining the funct ional c harac teristic s of voc abula ry servic e
as a set of Ap p lica tion Prog ramming Inte rfac es (APIs)
It s not a software pac kage, a lthoug h c ertain software exists tha t imp lem ents thespe c ification, and it s not a common voc ab ula ry da ta structure, although the
stand ard c an be imp lemented to interfac e with varying mod els (da ta a nd
terminology.)
Its purpose is to spec ify a com mon Ap p lica tion Prog ramm ing Interfac e (API) to
ac c ess terminolog ica l c ontent: Client software d oesn t have to know about spec ific
terminology d a ta struc tures and / or how to ac c ess them , and server softw are c an
plug and play with many c lients
-
8/4/2019 Module i Unit 02 en v01.3
30/32
30
CTS ARCHITECTURE
Application
Service
Interface
Data
CTSCTS
. . .
-
8/4/2019 Module i Unit 02 en v01.3
31/32
31
Som e CTS Runtim e Message API Examples
validateCode Determine whe ther the supp lied c od ed a ttribute is
valid in this voc abula ry doma in and c onte xt.
va lida teTranslation Determine whe ther the t ransla tion portion of the
c od ed a ttribute is va lid in this dom ain and conte xt.
translateCode Transla te the input c od e into a form tha t is va lid in
the ta rget c ontexts.
fillInDetails Fill in the d eta ils for the c od ed a ttribute , inc luding a ll
c od e system name s, versions and d isp lay na me s.
lookupValueSetExpansion Return a hierarchica l list of selec ta b le c onc ep ts for
the voc ab ulary dom ain and co ntext
lookupCode SystemInfo Return de tailed informa tion a bo ut the name d c od e
system
isConceptIdValid Determine w hether the c onc ep t c od e is valid in the
c od e system .
CTS Implem entations and rec ommended (non-mandatory) readings
NCI caBIG (aka LexBIG)
USE: Voc abulary Service for NCI c aBIG
Platform: Java
Backend: Relational Database
Please see: http s:/ / xmd r.org/ p resenta tions/ Thomas%20Johnson%20-%20LexBIG%20-
%20Mayo%20-%20January%202007.ppt
Mayo Clinic
USE: Open Source Refe renc e Imp lem enta tion
Platform: Java
Bac kend : SOAP, LDAP, Rela tiona l Data base, Protg
Please see: http://informatics.mayo.edu/LexGrid/index.php?page=ctsimpl
VHA Enterprise Term inolog y Servic e
Use: Internal voc abula ry ma nage ment
Platform: Java
Midd le-tier: Web log ic
Bac kend : Relationa l Database (Orac le)
VHA Enterprise Term inolog y Servic e
http:/ / library.ahima.org/ xpe dio/ groups/ pub lic / do c uments/ ahima/ bo k1_028716.pdf
-
8/4/2019 Module i Unit 02 en v01.3
32/32
Summary:
We have examined general concepts and characteristics of real-world healthcare
information systems, in order to understand the need for standards. We have alsosee n the essent ia l elements of the imp lem enta tion of an HIS and we spec ific a lly
exam ined interoperab ility, bec ause it is the axis of this c ourse.
Without Terminology Standards...
- Hea lth da ta is non-com parab le
- Aggregation is difficult, if not impossible
- Hea lth system s c annot intercha nge da ta
- Second ary uses (resea rc h, effic ienc y) are no t possible
- Linkage to dec ision sup port resource s is not p ossible
Interoperability cannot accomplish its purpose without dealing effectively with
terminology!
We hope you now understand that useful information interchange depends upon
the existence of an approved set of semantic and syntactic rules and that
controlled vocabularies, especially terminologies, are a key component for
ac hieving interop erab ility o f hea lth information systems.