Standardizing the MDS with LOINC and Vocabulary Matches


U.S. Department of Health and Human Services

Standardizing the MDS with LOINC® and Vocabulary Matches

Mark S. Tuttle, FACMI, Vice President StrategyTony Weida, PhD Vice President, Consulting ServicesApelon

Thomas White, MD, MS, MA, Director, Bureau of Mental Health InformaticsNY State Office of Mental Health

Jennie Harvell, MEd Senior Policy AnalystUS Department of Health and Human Services

April 23, 2007

PDF Version (36 PDF pages)

This report was prepared for the U.S. Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation, Office of Disability, Aging and Long-Term Care Policy. For additional information about this subject, you can visit the DALTCP home page at or contact the ASPE Project Officer, Jennie Harvell, at HHS/ASPE/DALTCP, Room 424E, H.H. Humphrey Building, 200 Independence Avenue, S.W., Washington, D.C. 20201. Her e-mail address is:

The Consolidated Health Informatics (CHI) Initiative unanimously endorsed selected content and messaging standards for use with federally-required patient assessments that include functional and disability content. The National Committee for Health and Vital Statistics (NCVHS) has recommended that the Secretary approve and adopt these CHI standards into federal programs.

The choice of CHI standards was based, in large part, on the results of an U.S. Department of Health and Human Services (HHS), Office of the Assistant Secretary for Planning and Evaluation (ASPE) contract with Apelon and a subcontract with Thomas White, MD. This effort resulted in: (i) a representation and coding of the federally-required nursing home Minimum Data Set (MDS) questions and answers in Logical Observation Identifiers names and Codes (LOINC®1); and (ii) an identification of standardized vocabulary matching terms (i.e., semantic matches) for the LOINC® represented and coded MDS questions and answers. The standards are intended to support standardized information exchange and re-use, such as may occur with the newly adopted Continuity of Care Document (CCD) standard.

The following briefly describes the MDS standardization process and the attached spreadsheet2 identifies the current vocabulary and LOINC® matches that were identified through the ASPE contract. For additional information, the following resources are available:


HHS/ASPE contracted with Apelon to identify "usefully related matches" between elements of MDS and the CHI endorsed terminologies (such as SNOMED-CT), and to create a representation for transmitting these matches using messaging standards (such as HL7). Thomas White, MD, State of New York collaborated, focusing on the use of LOINC® as a standard representation for MDS elements and associated vocabulary matches.

MDS Deconstruction, LOINC®-ification, Semantic Matching, Messaging and the UMLS

MDS Deconstruction

Several iterations of the MDS standardization effort produced the "deconstruction" of the MDS into the questions and answers that appears in columns D, E, and F in the attached spreadsheet. The purpose of the deconstruction was to maximize the linking of MDS question-and-answer pairs with CHI-endorsed standard vocabulary matches, while at the same time retaining the Centers for Medicare and Medicaid Services (CMS) intent (meaning) of each question-and-answer option, and to enable the exchange of assessment content using health information technology (HIT) messaging standards. Specifically, the deconstruction enables the representation and coding of the MDS questions and answers in LOINC® and supports the linking of these LOINC® coded items with any available semantic-matching terms from the CHI-endorsed standard vocabularies.


Given a deconstruction of an assessment instrument, such as the MDS, "LOINC®-ification" is the process of representing and coding the elements resulting from the deconstruction. In this case, the elements are question-and-answer pairs.

Given a LOINC®-ified assessment instrument, elements of that instrument -- the question-and-answer pairs -- can be matched -- by meaning -- to terms in standard vocabularies, as further explained and illustrated in the example below. The spreadsheet attached identifies the LOINC®-coded MDS item and any available matching vocabulary concepts (such as those that were found in SNOMED).

The LONIC representation and coding of assessment question-and-answer pairs, and associated vocabulary matches is specifically supportive of data exchange and re-use because: (i) exact semantic matches to assessments concepts are not always available; and (ii) often identified semantic matches provide "usefully related" (cut frequently not exact) matches to an assessment item. Thus, coding in LOINC® ensures the availability of codes that can be exchanged in an HL7 message even when there is not an exact semantic match. Currently (and in the accompanying spreadsheet) LOINC® codes are provided for only the MDS "questions." LOINC® is working to assign "answer codes" (i.e., LOINC®-Answer IDs), to each answer option on the MDS forms (and other LOINC®-coded patient assessments). These codes will further support health information exchange and re-use, particularly when exact semantic matches are not available.

Semantic Matching

Most MDS elements, whether they are questions, answers, or combinations of the two, do not have exact equivalents in SNOMED (or any other CHI standard terminology). However, and electronic health record (EHR) or personal health record (PHR) for a given patient that makes use of CHI standard terminologies may include terms that are "usefully-related" to an assessment question and/or answer. The presence of terms in an EHR and/or PHR that are usefully related to (or are an exact match of) an MDS question or answer option, might be of interest to a provider charged with completing an MDS assessment, and they may facilitate the completion of the assessment or other clinical applications (e.g., care planning). Simple examples include standard terms describing the patient's visual acuity or incontinence; such terms rarely translate directly into the related MDS assessment responses, but they are certainly "usefully related" to such responses.


Because of prior work on the problem, assessment instruments represented in LOINC® can leverage existing HL7 standard messages. To exchange assessment results in a standard way, an HL7 message will include:

  • The LOINC® code for the native assessment question-code from the assessment form.
  • The response for the specified patient including, at minimum, the LOINC® answer ID code and also including, if available, the semantic matching term.


A future release of the National Library of Medicine/Unified Medical Language System (NLM/UMLS) will represent "LOINC®-ified" assessment instruments. This representation will display for each standardized assessment the:

  • LOINC® code for each assessment question.
  • LOINC® answer IDs for each answer option.
  • All identified semantic matching codes and terms for each assessment question-and-answer option.

The UMLS representation will allow the re-use of standard terms, for example, by HIT vendors as they develop software that supports the completion of and/or use of standard assessments.

The deconstructed, LOINC®-ified, semantically matched MDS will be the first assessment instrument to appear in the UMLS. As additional semantic matches are identified (or retired), these changes will be updated in the UMLS.

The Spreadsheet

Accompanying this document is a spreadsheet representing the current state of an effort both to link MDS question-and-answer pairs to: (i) concepts and codes in SNOMED-CT (and other CHI-endorsed vocabularies); and (ii) LOINC® codes.

The spreadsheet columns D ("Concept Name"), E ("MDS Code"), and F ("QA Pair for Matching") contain, respectively, MDS elements (e.g., "H1.b.4. Incontinent"), MDS Question Numbers (e.g., "H1.b.4."), and the phrases used for matching (e.g., "Is incontinent"). The latter phrases were "synthesized" by domain experts as part of the matching effort. To improve readability, duplicates entries have been removed from these columns. That is, if 20 different SNOMED "matches" are linked to a given MDS element, the spreadsheet has been "uncluttered" by removing the 19 repeats of the MDS element name, code and match phrase from columns D, E, and F. Columns A, B, and C (initially hidden3), have the duplicates still in place, in case users wish to sort or extract selected rows.

Some MDS elements -- such as "SECTION G: PHYSICAL FUNCTION" -- provide the MDS hierarchical context for nearby question-and-answer pairs. These elements are not assigned a LOINC® code, although sometimes their presence is required to interpret the meaning of the nearby coded elements.

Column G contains the name of the standard terminology the MDS element is linked to (either SNOMED or LOINC®); column H has the associated SNOMED or LOINC® code, and column I has the SNOMED or LOINC® concept name. Thus, in spreadsheet row 1036, MDS element "H1.b.4. Incontinent," represented in columns D, E, and F as shown above, is linked to LOINC® code "45619-4" and the LOINC® concept named "BLADDER CONTINENCE:FIND:PT:^PATIENT:NAM:MDS." In spreadsheet rows 1037-1049, are codes and concept names from SNOMED that are "usefully related" to the MDS element "H1.b.4. Incontinent." That is, for example, if a patient's EHR or PHR contained the SNOMED term "Abnormal bladder continence (finding)," or its SNOMED code "150003" (or both), the provider is alerted to a potentially relevant finding. In most cases, the "usefully related" SNOMED terms ignore any MDS-specified "look-back" period or other "count-related" notions, but the presence of the related terms in an EHR or PHR may help the assessing provider find the portions of the record needed to complete MDS's quantitative requirements.

Apelon-MDS Database

For a short time, interested parties may apply for Web browser access to the Apelon database which can be used to search for the semantic matches included in the spreadsheet and review them in context. Access will be free, but on a first-come first-served basis. Those interested in such access should send an email message:


A brief telephonic training session may be required.


John Carter led the MDS deconstruction and matching effort, and introduced the "synthetic" terms created to aid MDS matching. Several persons from LOINC® spent many hours working on refining the LOINC® data model, extracting the question data from RAVEN, building the hierarchies and answer lists, as well as developing new capabilities in RELMA to display such information. Notably, we are very appreciative of the efforts by Clement McDonald, Daniel Vreeman, and Kathy Mercer from LOINC® for their assistance with these efforts. Betsy Humphreys, Clement McDonald, and Kin Wah Fung of the NLM also provided key leadership in this project, including working to include the LOINC® represented MDS and linked semantic matches in the UMLS.


  1. LOINC® is a registered United States trademark of Regenstrief Institute, Inc.

  2. The Excel spreadsheet, described in this paper, is available at

  3. To view these columns, "drag" the left-hand column boundary for column D to the right until the desired portions of the next column to the left is visible, and then do the same for column C and then column B.

The spreadsheet is also available at:Excel file: file (at the end of the PDF of this paper): standardizing-mds-loinc-and-vocabulary-matches-pdf