Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Biomed Inform. Author manuscript; available in PMC 2010 June 1.
Published in final edited form as:
PMCID: PMC2693286

The impact of SNOMED CT revisions on a mapped interface terminology: Terminology development and implementation issues


Large-scale mapping efforts have been done in attempts to migrate systems that use proprietary concepts to ones that use terminological standards such as SNOMED CT. As efforts move towards implementation, the target maps should retain a predictable structure including those targets requiring post-coordination of SNOMED CT concepts. In this paper, we compared the editorial guidelines of two versions of SNOMED CT (January 2005 and July 2006 respectively) and noted how the revisions affected a single, comprehensive set of mapped concepts (n = 2002) from a legacy system. Changes made to the categories and guidelines for approved attributes were noted and then evaluated against the post-coordinated maps (n = 1570) from the original mapping effort. 71% (n = 1118) of the concepts were affected due to changes made in either SNOMED CT categories or to the revision of approved attributes. While the efforts of each subsequent SNOMED CT version aim for continual improvement, changes made to its core structure and post-coordination guidelines make it more difficult to migrate proprietary data to this reference standard. Attention must be paid to auditing the processes used in terminology development to include the impact that their revisions may have on real-world clinical implementation.

Keywords: terminology mapping, terminology audit, SNOMED CT, interface terminology, terminology versioning, SNOMED CT implementation


In 2005, Vanderbilt University Medical Center (VUMC) in Nashville, TN, USA engaged in a large-scale mapping effort to evaluate the correlation between interface terms used in one of their proprietary structured documentation systems [1] and SNOMED CT. A total of 2002 interface concepts were mapped to SNOMED CT [2]. This work was done in an effort to migrate towards leveraging a standard clinical terminology to help make EHR systems be interoperable, to drive decision support, to enable data aggregation, etc [3-6]. This interface terminology evolved from clinical finding frames originally developed by Miller and Masserie as part of an early NLM UMLS project to support the Internist/Quick Medical Reference (QMR) diagnostic expert and decision support system [7-8]. Many of the interface terms included phrases. The mapping of these concept phrases required the use of post-coordinated concepts in SNOMED CT for the resulting maps. In the original work [2], concepts were categorized into clinical topic areas and mapping rules agreed upon in advance including how similarly grouped concepts would be post-coordinated so there would be consistency in the resulting maps. At the conclusion of this mapping effort, 1570 out of 2002 interface source concepts (78 %) were represented as post-coordinated concepts in SNOMED CT following the guidelines set forth in the SNOMED Clinical terms® User's Guide (January 2005 release) [9].

Since this effort was completed, SNOMED CT continued its ongoing editorial work [10-11] with biannual revisions of its terminology through established editorial processes [12-13]. Changes were made to the structure of SNOMED CT (i.e. category changes) as well as the post-coordination guidelines (i.e. rules for approved attributes). In this paper we compared the changes made to SNOMED CT from the January 2005 release to their July 2006 release in terms of structure and rules and examined how these two versions impacted the original maps.


Comparison of the changes made to SNOMED CT was done using the SNOMED Clinical terms® User's Guides of the January 2005 release (Version 1) [9] and the July 2006 release (Version 2) [14]. This included an examination of the top-level categories as well as the guidelines (rules) for the approved attributes used for post-coordinated expressions. The same terminologists (GW and STR) who did the original mapping based on Version 1 also did the review and comparison with Version 2.

First, the changes made to SNOMED CT's top-level categories (Table 1) were examined. These changes included the renaming of Context-dependent categories to Situation with explicit context and the subsumption of Attribute under the Linkage concept category. Next, the changes made to SNOMED CT's approved attributes were noted. The attributes are a category of concepts that can be applied to one or more SNOMED CT hierarchies referred to as the “domain” of the attribute and they are used to create post-coordinated expressions. Each post-coordinated expression consists of several concepts related by attributes. These attributes were also revised in Version 2. Table 2 depicts the attribute changes noted between the two versions that were examined.

Table 1
Changes made to top-level categories in SNOMED CT
Table 2
Attribute changes

After the changes made to SNOMED CT were documented, we examined how each of the changes affected the original set of mapped concepts. A separate mapping of the legacy concepts to version 2 was not done. An Access database with the recorded resulting maps was used for searching and analysis. This database was a direct extract of the flat file of final maps done in the original mapping work and included a total of 2030 rows. (2002 concepts with resulting maps; 28 concepts had duplicate maps); analysis was done on all rows that included post-coordinated maps (n =1570) and on pre-coordinated maps using the Context-dependent categories (n = 31).

First, all concepts that used the Context-dependent categories were re-evaluated. Those concepts included “Family history of”, “History of”, “History/symptoms”, “Past medical history of”. The number of affected maps was recorded as well as how they would be revised under Version 2. Next, the attribute changes as recorded in Table 2 were examined against the resulting maps. The addition of “FINDING METHOD” as an approved attribute in Version 2 was applied to the maps where the resulting maps included a clinical finding associated with a procedure. “FINDING METHOD” provided specificity for use with Clinical Findings as opposed to “ASSOCIATED WITH” that allowed more general associations with several SNOMED CT categories. For example, an original map from the interface concept “ABDOMINAL TENDERNESS PALPATED” was mapped to:

118242002 | Finding by palpation (finding) |:
 47429007 | ASSOCIATED WITH (attribute) | =
  43478001 | Abdominal tenderness (finding) |

With the addition of “FINDING METHOD” as an approved attribute, this concept expression would be revised as:

43478001 | Abdominal tenderness (finding) |:
 418775008 | FINDING METHOD (attribute) | =
  113011001 | Palpation (procedure) |

This was applied to several categories where the legacy concepts were clinical findings that included a procedure. Examples included interface concepts with the words Auscultated, Observed, Palpated or Percussed.


Of the total original post-coordinated maps (n = 1570), 71% (n = 1118) were affected by the SNOMED CT changes and required updating. Table 3 depicts the changes between Version 1 (January 2005) and Version 2 (July 2006). Most of the concepts from the Context-dependent categories in Version 1 were reassigned under the Situation with explicit context category in Version 2 except for the History/symptoms concepts. History/symptoms was changed to a Navigational (Special) concept that did not allow for post coordination. This affected 342 of the original maps. Other maps that involved clinical findings associated with a procedure were revised because the newer version added the attribute “FINDING METHOD” with a procedure value. This attribute-value combination provided specificity for use with Clinical Findings as compared with the more general use of “ASSOCIATED WITH” that applied to several SNOMED CT categories. It also allowed the source clinical finding to be mapped primarily to a Clinical Finding in SNOMED CT with a procedure assuming a more secondary role. For example, the source concept ABDOMINAL SHIFTING DULLNESS PERCUSSED was mapped in Version 1 to:

Table 3
Examples of SNOMED CT changes and the affected maps.
118244001 | Finding by percussion (finding) |:
 47429007 | ASSOCIATED WITH (attribute) | =
  366463006 | Shifting abdominal dullness finding (finding) |

With Version 2, the concept was revised to:

366463006 | Shifting abdominal dullness finding (finding) |:
 418775008 | FINDING METHOD (attribute) | =
  75180006 | Percussion (procedure) |

This extended to all of the original maps where the source concepts included the words “Auscultated”, “Observed”, “Palpated” and “Percussed”. A procedure that allowed for the attribute COMPONENT to be used in Version 1 was restricted in Version 2 to only those procedures that were measurement procedures. For example, the interface concept ACTIVITIES OF DAILY LIVING ASSESSMENT OF PUTTING ON SHOES was mapped to:

304492001 | Activities of daily living assessment (regime/therapy) |:
 246093002 | COMPONENT (attribute) | =
  284978003 | Ability to put on footwear (observable entity) |

(Note: regime/therapy is a procedure.) Since COMPONENT was no longer approved under Version 2, the source concept was revised as:

365224008 | Finding related to ability to put on footwear (finding) |:
 418775008 | FINDING METHOD (attribute) | =
  304492001 | Activities of daily living assessment (regime/therapy) |

Several additional examples of the affected maps are shown in Table 4. Some of the original maps involving measurements were retained as preferred maps to permissible (approved attribute) alternatives. For example, the map for the source concept BODY MASS INDEX QUANTITATIVE MEASURED was retained as:

Table 4
Examples of original maps and revised maps based on changes made to SNOMED CT
118245000 | Finding by measurement (finding) |:
 363714003 | INTERPRETS (attribute) | =
  60621009 | Body mass index (observable entity) |

An alternative map was not used:

122869004 | Measurement procedure (procedure) |:
 246093002 | COMPONENT (attribute) | =
  60621009 | Body mass index (observable entity) |

The pre-coordinated maps that were affected by the revisions (n = 31) involved name changes but, interestingly, no changes to the concept ID. For example, the legacy concept “PREVIOUS PHYSICAL ABUSE” mapped to:

313215004 | History of physical abuse (context-dependent category) | (Version 1)
313215004 | History of physical abuse (situation) | (Version 2)


In this paper, we examined how a mapped interface terminology would be impacted by SNOMED CT revisions made between versions (January 2005 and July 2006, respectively). Specifically, changes made to categories and attributes affected 71% of the mapped legacy concepts. While we anticipated that the ongoing maintenance of the SNOMED CT terminology would include the addition and retiring of concepts, our observations showed unexpected findings in terms of the structural changes made to the SNOMED CT corpus as well as rules changes that affected most of the post-coordinated expressions. Our analysis provides empirical evidence that the editorial changes made between these two versions of SNOMED CT would adversely impact mapped legacy concepts that have been implemented as part of an interoperable enterprise system.

While it is understood that the revisions made to SNOMED CT with each release represent efforts at continually improving the terminology, it is important for editors to consider issues of implementation during the auditing process. The SNOMED CT editorial staff now under the auspices of the International Health Terminology Standards Development Organisation (IHTSDO)[15] has processes [12-13] in place that carefully address how concepts are added, edited and retired while maintaining history files since the inception of SNOMED CT in 2002. During its evolution, other revisions to the terminology such as those highlighted in this paper involving category changes and permissible (i.e. approved) attribute-value pairs for post-coordination have not been as readily understood and documentation referencing these editorial decisions is not publicly available.

Auditing of the terminology development processes should include measures of the impact of changes made to its core structure and rules or other methods must be considered for the real-world implementation of SNOMED CT when attempting to migrate concepts from clinical applications [16] that have been in long-standing use. One such effort currently underway at the IHTSDO is the development of a Machine Readable Concept Model (MRCM) [17], a prototype that seeks to assist content editing by providing a consistent representation of SNOMED CT concepts and a means of validating post-coordinated expressions. If successful, this may help to overcome some of the obstacles in transforming legacy concepts to SNOMED CT and in maintaining them in an implementable system. Mapping legacy concepts to SNOMED CT is labor-intensive especially if it is large-scale but once this is completed, there needs to be a predictable way of updating and maintaining these mapped concepts within an enterprise system. Based on our observations here using two versions of SNOMED CT, we conclude that there were no editorial measures in place that considered the implementation issues related to post-coordinated expressions. We do agree that improvements are necessary and overall beneficial from the perspective of improving concept model expressiveness but the impact on applications in use should also be considered if we want to support the use of a standard reference terminology in the applied clinical setting.

Other strategies under consideration for implementation include data model changes to focus on the representation of pre-coordinated concepts, use of extensions [12] or implementing the most frequently used concepts in the clinical setting that may have more clinical relevance for decision support, aggregate analysis, etc.

We agree that SNOMED CT is quite unique among terminologies in terms of clinical data representation and in allowing the creation of post-coordinated expressions to capture the correct semantics of clinically complex concepts. These unique features limit our ability to contrast the editorial changes made to SNOMED CT with other terminologies and code sets. Our experience as described here is limited to SNOMED CT and to a single set of revisions (from Version 1 to Version 2). It is likely that future revisions will not have major category changes and rule changes that would show a similar impact on mapped legacy concepts. Additionally, our results may not be typical for mapped legacy concepts used in other clinical environments in terms of scale and content. The legacy concepts that we used for the mapping were complex (included many phrases) and could have been more challenging. Also, we were not able to contrast our findings with those of others, as we were unable to find any published work of similar scale at the time of the analysis.

In conclusion, while other investigators have demonstrated different methods of representing concepts [18-19] and implementing SNOMED CT [20-21], our observations highlight the challenges of maintaining already migrated legacy interface concepts through the ongoing terminology revision process. Additionally, while methods have been proposed for transforming ontologies [22] and for auditing terminologies [23-25], there is also a need for ongoing analysis of the impact that terminology changes have on clinical applications already in use or those undergoing implementation. Until a more predictable means of following the changes made to SNOMED CT can be made, alternative methods must be sought as a more realistic way [26-27] of implementing SNOMED CT in the clinical setting


The project was supported by a Grant from the United States National Library of Medicine (K22 LM008576-02).


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


1. Shultz E, Rosenbloom ST. Quill: A novel approach to structured reporting. AMIA Annu Symp Proc. 2003;2003:1074. [PMC free article] [PubMed]
2. Wade G, Rosenbloom ST. Experiences Mapping a Legacy Interface Terminology to SNOMED CT. BMC Med Inform Decision Making. 2008 Oct 27;8(Suppl 1):S3. [PMC free article] [PubMed]
3. Rosenbloom ST, Miller RA, Johnson KB, Elkin PL, Brown SH. Interface terminologies: facilitating direct entry of clinical data into electronic health record systems. J Am Med Inform Assoc. 2006;13(3):277–288. [PMC free article] [PubMed]
4. Chute CG, Elkin PL, Sherertz DD, Tuttle MS. Desiderata for a clinical terminology server. Proc AMIA Symp. 1999:42–46. [PMC free article] [PubMed]
5. Elkin PL, Brown SH, Carter J, Bauer BA, Wahner-Roedler D, Bergstrom L, et al. Guideline and quality indicators for development, purchase and use of controlled health vocabularies. Int J Med Inform. 2002;68(13):175–186. [PubMed]
6. Spackman KA, Campbell KE, Cote RA. SNOMED RT: a reference terminology for health care. Proc AMIA Annu Fall Symp. 1997:640–4. [PMC free article] [PubMed]
7. Miller RA, Pople HE, Jr, Myers JD. Internist-1, an experimental computer-based diagnostic consultant for general internal medicine. N Engl J Med. 1982;307(8):468–476. [PubMed]
8. Miller RA, Masarie FE, Myers JD. Quick Medical Reference (QMR): A Microcomputer-based Adaptation of the INTERNIST-1 Diagnostic System for General Internal Medicine. In: Salamon R, Blum B, Jorgenson M, editors. MEDINFO 86. Amsterdam: North Holland Publishing Co; 1986. p. 1143.
9. College of American Pathologists. Attributes used in SNOMED CT SNOMED Clinical terms® User's Guide-January 2005 release. Northfield, IL: 2005.
10. Wang AY, Sable JH, Spackman KA. The SNOMED clinical terms development process: refinement and analysis of content. Proc AMIA Symp. 2002:845–9. [PMC free article] [PubMed]
11. Spackman KA. Rates of change in a large clinical terminology: three years experience with SNOMED Clinical Terms. AMIA Annu Symp Proc. 2005:714–8. [PMC free article] [PubMed]
12. International Health Terminology Standards Development Organisation. SNOMED Clinical Terms Editorial Guidelines, Version 1.08. May 1, 2008.
13. International Health Terminology Standards Development Organisation. SNOMED CT Style Guide: Introduction and Overview, Version 0.01. Apr 10, 2008.
14. College of American Pathologists. Attributes used in SNOMED CT. SNOMED Clinical terms® User's Guide-July 2006 release. Northfield, IL: 2006.
15. International Health Terminology Standards Development Organisation (IHTSDO).
16. Rosenbloom ST, Miller RA, Johnson KB, Elkin PL, Brown SH. A model for evaluating interface terminologies. J Am Med Inform Assoc. 2008 Jan-Feb;15(1):65–76. [PMC free article] [PubMed]
17. International Health Terminology Standards Development Organisation. Machine Readable Concept Model – Abstract Logical Model, Version 1.01. Nov 27, 2008.
18. Andrews JE, Patrick TB, Richesson RL, Brown H, Krischer JP. Comparing heterogeneous SNOMED CT coding by examining normalized expressions. J Biomed Inform. 2008 Feb 5; Epub ahead of print. [PMC free article] [PubMed]
19. College of American Pathologists. SNOMED Clinical Terms®Guide: Transforming Expressions to Normal Forms, Version 5. Aug, 2006. Revision.
20. Wasserman H, Wang J. An applied evaluation of SNOMED CT as a clinical vocabulary for the computerized diagnosis and problem list. AMIA Annu Symp Proc. 2003:699–703. [PMC free article] [PubMed]
21. Green JM, Wilcke JR, Abbott J, Rees LP. Development and evaluation of methods for structured recording of heart murmur findings using SNOMED-CT® post-coordination. J Am Med Inform Assoc. 2006;13(3):321–333. [PMC free article] [PubMed]
22. Ceusters W, Smith B. A realism-based approach to the evolution of biomedical ontologies. AMIA Annu Symp Proc. 2006:121–5. [PMC free article] [PubMed]
23. Wang Y, Halper M, Min H, Perl Y, Chen Y, Spackman KA. Structural methodologies for auditing SNOMED. J Biomed Inform. 2007 Oct;40(5):561–81. [PubMed]
24. Min H, Perl Y, Chen Y, Halper M, Geller J, Wang Y. Auditing as part of the terminology design life cycle. J Am Med Inform Assoc. 2006 Nov-Dec;13(6):676–90. [PMC free article] [PubMed]
25. Cornet R, Abu-Hanna A. Auditing description-logic-based medical terminological systems by detecting equivalent concept definitons. Int J Med Inform. 2008 May;77(5):336–45. [PubMed]
26. Schulz S, Suntisrivaraporn B, Baader F. SNOMED CT's problem list: ontologists' and logicians' therapy suggestions. Medinfo. 2007;12(Pt 1):802–6. [PubMed]
27. Nachimuthu SK, Lau LM. Practical issues in using SNOMED CT as a reference terminology. Medinfo. 2007;12(Pt 1):640–4. [PubMed]