September 24, 2003
The Honorable Tommy Thompson
Secretary
U.S. Department of Health and Human Services
200 Independence Avenue, SW
Washington, D.C. 20201
Dear Secretary Thompson:
The National Committee on Vital and Health Statistics (NCVHS) commends you for your commitment toward government wide adoption of clinical data standards that you first announced on March 21, 2003. NCVHS recognizes and appreciates that there is new momentum to adopt clinical data standards that is driven by you and the Consolidated Healthcare Informatics Initiative (CHI). Consequently, NCVHS is now working closely with CHI to study, select and recommend domain specific patient medical record information (PMRI) terminology standards. We have mutually developed a process that allows NCVHS to discuss in open, interactive sessions CHI recommendations as part of the CHI Council acceptance process.
The NCVHS has the following comments on the attached set of CHI domain area recommendations:
- The NCVHS concurs with the CHI recommendations for Unit Domain.
- The NCVHS concurs with the CHI recommendations for Lab Result Content as modified in the attached document.
- The NCVHS concurs with the CHI recommendations for Demographics Domain as modified in the attached document.
We understand that the next stage is formal government adoption which the NCVHS supports. We are excited about the value of this continuing process.
Sincerely,
/s/
John R. Lumpkin, M.D., M.P.H.
Chair, National Committee on Vital and Health Statistics
cc: HHS Data Council Co-Chairs
Enclosures
Consolidated Healthcare Informatics Initiative
Final Recommendation Information Sheet[1]
Domain Title(s) and Team Lead
Lab Results Contents and Units: Steve Steindel, CDC
Scope
This domain is defined as a terminology that is used to exchange results of laboratory tests between facilities. These results are contained within a laboratory report that includes additional items such as patient demographics, laboratory test name (expressed as a LOINC code as approved by CHI Council and adopted as a federal standard), specimen type and other items as may be required by business needs or messaging structures.
The workgroup defined what constitutes a laboratory result to include four basic parts: 1) the result itself; 2) results units if applicable; 3) normal ranges or indicator flags; 4) any comments that may be associated with the result.
The result itself has many forms:
- Numerical results, including titers and ranges (#1 – #2), representing quantities found and do not need a terminology.
- Ordinal results such as 1+ that infer relative quantities but are not standardized. These also do not need a terminology but it was felt that in order to achieve the goal of CHI, interchange understanding between agencies, a proposal should be made to introduce standardize reporting of these results.
- Results versus a cut-off value, including >< numerical results where the cut-off value is generally not known and requires interpretative information in the form of normal ranges or indicator flag. These also do not need a terminology
- Alphanumeric results that now generally take the form of free-text but could be expressed as coded values. These include results from anatomical pathology, microbiology, hematology, molecular pathology and immunohematology. These also do need a terminology
The group felt that we should emphasize in the report the need to transmit normal ranges and any indicator flags in an appropriate place in messages for all applicable laboratory results, but the form was test dependant and outside the scope of the workgroup. Indicator flags are of particular importance for microbiology and serology results and must be included. Applicable laboratory results were considered to be those non-descriptive in nature.
Though not part of the terminology recommendation, we note the need to modify the laboratory result message to include the name of the methodology used to produce the result or pointer to unique resource identifier (URI) used to find the unique protocol used. Knowing the method is essential for report interpretation. Also the actual method lends significance to non-human test results when the test was performed using standard methods such as those accepted by AOAC International or non-clinical specimens such as those used for forensic testing.
The group noted that comments were a part of many laboratory results. Standardization of the form of these comments was required before we could achieve interchange understanding. A terminology should be selected that has the ability to code comments. Associated with comments were abbreviations and the group felt a common set should be developed and placed within the terminology. Jules Berman has been compiling a list of those found at NCI.
The group felt that abbreviations form a nomenclature in laboratory medicine and should be considered part of this domain.
For non-human samples the identification of specimen type is viewed as part of this domain.
| Domain/Sub-domain | In-Scope (Y/N) |
| Numerical results including titers | N |
| Normal result and other flag indicators | N |
| Out of range results | N |
| Ordinal results | Y |
| Anatomical Pathology report codes | Y |
| Living Organism codes | Y |
| Hematology result codes | Y |
| Immunohematology (Blood Bank) result codes | Y |
| Units | Y |
| Other descriptive laboratory test result codes | Y |
| Standard Comments | Y |
| Abbreviations | Y |
| Non-human Specimen Type | Y |
Alternatives Identified
- MESH (Medical Subject Headings)
- The National Cancer Institute (NCBI) Taxonomy
- SNOMED CT
- LOINC
- Other UMLS Category 0 Terminologies
- HL 7 Version 2 and 3 Unit Coding
Lab Results Contents: Final Recommendation
SNOMED CT codes, when available as Category 4 codes in National Library of Medicine (NLM) Unified Medical Language System (UMLS) Metathesaurus (estimated date 1/04), were found to adequately cover the domain of laboratory result content coding. Where codes do not exist, an adequate mechanism exists to add new codes in a timely fashion. Codes for living organism are augmented in Category 0 of the UMLS by the National Center for Biotechnology Information (NCBI) Taxonomy codes, which are more complete than the SNOMED codes (Note: Information on the UMLS is found at www.nlm.nih.gov/research/umls and on the NCBI taxonomy at www.ncbi.nlm.nih.gov/Taxonomy/.)
The specific locations in the SNOMED CT hierarchy that form the basis of our recommendation are:
Quantifier Value:Unit
Qualifier Value:Modifier:Linkage Term and/or Qualifier
Organism:Living Organism
Body Structure:Morphologically Altered Structure:Morphologically Abnormal Structure
Disease:Disease by Body Site:Disorder of Body System:Disorder of Hematopoietic System
Findings:Clinical History and Observation Findings:General Finding of Observation of Patient:Genetic:Molecular and/or Cellular Finding:Genetic Finding:Phenotype:Blood Group Phenotype
Findings:Finding by Method:Test Finding:Laboratory Test Finding.
Substance:Dietary Substance
It is our recommendation for laboratory result content coding that we use US effective Category 0 UMLS codes in this terminology domain where they exist. This will effectively encompass all of SNOMED CT for this domain. For microorganism, the UMLS Concept Unique Identifier (CUI) will refer to the SNOMED code or the equivalent NCBI code. We recommend the sending of both SNOMED and UMLS codes when available, but the sending of a SNOMED code alone is essentially equivalent to sending a UMLS CUI because of the UMLS mapping. As this represents an expanded use of the CUI, we have contacted NLM and they support this use. We anticipate that this approach will be sufficient for anatomical pathology, microbiology, hematology, immunohematology, and units. We did not explore, but anticipate that the approach will also be adequate for other general laboratory results. Gaps found in the area of comments and ordinal results are discussed below.
Note: Category 0 UMLS Codes have no license restrictions on their use beyond the minimal restrictions provided by the National Library of Medicine on UMLS. They can be used and distributed without further license fees. While SNOMED CT is classified as Category 4 in the UMLS, the effective nature of that category makes it equivalent to Category 0 within the US.
The investigation by this workgroup has also revealed existing coverage for the Unit domain described below. These recommendations are sufficient for defining that domain as well.
Conditions
No conditions apply to the above recommendation but gaps are noted with respect to standardization of comments and abbreviations that may for part of a laboratory result. Also, the CHI messaging recommendation for laboratory results needs to include the specific method used.
Units: Final Recommendation
HL7 is being recommended for the Units domain. The HL7 Units vocabulary is based on ISO/ANSI standards and further work is underway to enhance the standard to reflect the Unified Code for Units Measure (UCUM). This standard has already been adopted by DICOM and is heavily based on ISO and ANSI standards. The UCUM has more than 3X the amount of symbols as in ANSI X3.50.
Content Coverage
Units:
UMLS CUIs were found for most common laboratory units. They exist in the section “Quantitative Concepts” of which approximately 3190 exist, most not laboratory related. Most of the UMLS codes derive from the Read code system that was subsumed by SNOMED and hence exist, even more extensively, in SNOMED CT. The SNOMED codes are found under Quantifier Value:Unit.
Additional unit descriptions, but not codes are found in various standards used by HL7 derived from ISO 2955 83 (withdrawn by ISO in 2001) and ANSI X3.50. Inclusion in the HL7 documentation without restriction indicates free use for interchange messaging. HL7 defines a number of common relative units used in medicine such as mg/dL and the ISO/ANSI standards define ways to create others from core units. These combined and core units include mixed case forms where appropriate. While a complete analysis of the HL table was not conducted against all possible units, investigation of the laboratory domain and a quick review of other areas of medicine indicates it is sufficient to recommend.
Living Organism codes:
CDC distributes for use the SNOMED hierarchy Organism:Living Organism, containing approximately 17,000 concepts. A cross-mapping of those of public health interest to concepts found in MESH reveals adequate coverage for reporting but not case-investigation purposes of about 2500 organism. No count is readily available for the NCBI taxonomy, but the UMLS lists approximately 107,570 concepts. No check was made to see if all of the SNOMED organism codes correspond to NCBI codes but it is assumed they do. If that is the case, it is safe to assume that MESH forms an adequate subset for reporting purposes, SNOMED for clinical use and NCBI for research purposes. Organisms in each should map to the same UMLS CUI, making interchange of codesets easy.
Anatomical Pathology report codes:
Anatomical pathology concepts were found in previous versions of SNOMED. The last version of SNOMED International (V 3.5) had approximately 5,900 morphology terms. In the present version of SNOMED CT these are found primarily in the following hierarchy: Body Structure:Morphologically Altered Structure:Morphologically Abnormal Structure. It is difficult to compare the completeness of the SNOMED Terms with other terminologies due to classification differences. For example, Hodgkin lymphoma and Morbilliform rash are both found in this hierarchy of SNOMED CT. In UMLS, Morbilliform rash is classified as a finding and appears in six other terminologies. Hodgkin lymphoma is classified as a neoplastic process and appears in 25 other terminologies.
Anatomy forms an important part of the anatomical pathology report but the workgroup offers no recommendation as to a terminology. We note that another Workgroup, History and Physical, is charged to look into this domain. We note that one of the difficulties in an anatomy terminology is that the concepts and hierarchy are use dependant. For instance a dermatologist assigns a topology of “skin” to a demrofibroma of the foot. The podiatrist would assign a topology of “foot” or something more detailed, such as “planter foot” for the same lesion. A pathologist may assign a topology of “soft tissue” to the same lesion. The Workgroup requests that others looking into anatomy take these comments into consideration.
Hematology result codes:
Codes for hematological disorders appear focused in SNOMED CT under the Disease:Disease by Body Site:Disorder of Body System:Disorder of Hematopoietic System hierarchy, as well as other areas. UMLS places the hematological diseasaes in a similar hierarchy. A count of these codes is not available. Codes for various hematological cells are found primarily in the Body Structure hierarchy, Cell Structure, Blood Cell. Again, a count is not possible. UMLS places the hematological cells in a similar hierarchy.
Immunohematology (Blood Bank) result codes:
Blood groups appear to be classified in SNOMED CT under Findings:Clinical History and Observation Findings:Clinical History and Observation Findings:General Finding of Observation of Patient:Genetic:Molecular and/or Cellular Finding:Genetic Finding:Phenotype:Blood Group Phenotype. In UMLS, these appear to also be findings, but are noted as laboratory results. A count is not possible.
Other descriptive laboratory test result codes:
SNOMED CT includes many general laboratory result codes, including some appropriate to the above specific sections, under Findings:Finding by Method:Test Finding:Laboratory Test Finding. UMLS Groups similar terms under Laboratory Test Results. A count is not possible.
Standard Comments:
SNOMED CT includes many general laboratory result comments also under Findings:Finding by Method:Test Finding:Laboratory Test Finding. UMLS Groups similar terms under Laboratory Test Results. In addition, other general comment codes are found in Qualifier Value:Modifier:Linkage Term and/or Qualifier. A count is not possible.
Abbreviations:
In general, neither SNOMED nor UMLS contain abbreviations. UMLS contains an Abbreviation Table that is a list of acronyms used within the system but does not represent common clinically used abbreviations. The US Army has a standard set of abbreviations used for their internal purposes and the Workgroup recommends this be reviewed and considered as a basis for a National Standard Abbreviation set to be maintained by a designated organization, not necessarily the DoD.
Non-human codes:
Under Organism:Living Organism, SNOMED has concepts for many plants and animals. Basic food products are listed under Substance:Dietary Substance but have no commercial codes.
Acquisition
Lab Results Contents: UMLS is available at no charge to anyone who agrees to the license terms. UMLS license terms allow use for all patient record uses and messaging. An agreement has been reached that provides, in the US, SNOMED CT, a Category 4 codesets essentially allowing free distribution and use in the US. The agreement also allows for perpetual use of these codes.
Units: Hard and computer readable forms of the standards are available from HL7 and cost $50-$500 depending on the specific standard and member status.
No cost is associated with using the standard.
Gaps
While the implementation of the coding of the actual laboratory results should precede expeditiously, the following areas are noted where future development is needed.
Content:
Except for the ever present need to add new concepts as the area expands both SNOMED and UMLS appear complete in the coverage of coding for laboratory results.
Comments:
No universal agreement exists for the use and contents of comments associated with laboratory results. In many cases, especially if the comment involves the condition of the sample or the inability to collect a sample, the comment has direct implications with regard to the result. In other cases, the comment provides understanding with regard to the interpretation of the result. For example the comment “Some gonorrhea seen” is used by one laboratory as a culture result code, but the meaning is not clear. Similar ambiguous language exists in other laboratory disciplines. A professional organization needs to step forward and undertake the standardization of comments so that we can achieve better understanding when these are received.
Abbreviations:
As with all professions, abbreviations and acronyms are widely used in laboratory medicine and are not standardized. When used in a result they could result in ambiguous understanding. A professional organization needs to step forward and undertake the standardization of comments so that we can achieve better understanding when these are received. One of the work group participants has compiled a list of approximately 12,100 laboratory associated abbreviations, many having multiple meaning and one having 40. Clearly some interchange standard is needed to use these successfully. The US Army has a standard set of abbreviations used for their internal purposes and the Workgroup recommends this be reviewed and considered as a basis for a National Standard Abbreviation set to be maintained by a designated organization, not necessarily the DoD.
Ordinal Results (Method Codes):
While ordinal results such as 1+ or Positive are beyond the scope of this report, the workgroup wanted it noted that some conformance is needed before these can be understood between sites.
One set of tests reports results with respect to a sample containing a fixed amount of the material tested. If more material is noted in the patient sample it is reported as positive, less is negative and many tests define a borderline zone. Generally these are tests for infectious material as determined by the presence of either the antibody developed in response to the presence of the organism or direct measurement of an antigen found on the organism. Without knowledge of the cutoff value comparison between positive and negative results obtained at different institutions is difficult.
Similarly, values reported on an ordinal scale, such 1+, depend on the amount of material present. Also, within a test, the difference between 1+ and 2+ varies in a non-standard fashion. Again, without knowledge of how these scales are derived interpretation between institutions is difficult.
Ideally, standards should be developed for the tests and universally applied by all test developers, perhaps enforced by the FDA as part of their medical device oversight. Realistically, this will not happen, in part because technology changes and also the sensitivity and specificity of test, which helps determine the cutoff value and ordinal scales, varies with clinical need. Hence, as part of the CHI recommendations we should advise that testing method be sent as part of a laboratory test result so these values can be understood.
It is recommended that the FDA develop a standard and informatically sound means to identify existing approved methods to the device level and publish that information as part of their approval process. This classification system should be also be applied to legacy devices. It is our understanding that the FDA is currently exploring the use of the Global Device Nomenclature for similar purposes. The current list maintained for the Clinical Laboratory Improvement Act (CLIA) should be reviewed as a source for these codes.
Non-Human Content:
Coverage is weak and will need through review and expansion with topic experts. It is recommended that the government partner with the American Veterinary Association to more fully support the animal portion of SNOMED and find suitable partners for the plant and food portions. Actual food product codes are outside the scope of a terminology and will require agreement between FDA and USDA.
Additional non-human content that was not explored by the workgroup but it is important include environmental samples such as air, soil and water, storage containers, manufacturing sites, etc. While many of these types of sample are not generally involved in health information, it is to the government’s benefit for common terminology to exist so that when potential health impacts occur semantic understanding is achieved. It is recommended that and interagency taskforce be formed from DoD, HHS, and EPA at a minimum to make better recommendations in this important area.
[1] Information Sheet designed specifically to facilitate communication between CHI and NCVHS Subcommittee on Standards and Security resulting from May 20, 2003 testimony. CHI may seek assistance to help further define scope, alternatives to be considered and/or issues to be included in evaluation process.
Consolidated Health Informatics Initiative
Final Recommendation Information Sheet[2]
Domain Title(s) and Team Lead
Demographics: Beth Franchi, VA
Scope
The standard, as identified in the following section of this document, will be used to set the requirements for collecting and storing specific patient demographic data, to be used for various purposes, primarily that of correctly identifying a patient. Through later research, demographic information can serve as an important component in the identification of health trends within various segments of the population and can focus on these demographic groups to recognize reactions to specific health care regimens and clinical trials. The collected information can be used to track the elimination of health disparities, a high priority initiative for many government departments and agencies, such as the Department of Health and Human Services.
| Domain/Sub-domain | In-Scope (Y/N) |
| Patient Care (Patient and Clinical Management) | Y |
| Financial / Billing | Y |
| Legal and Regulatory Concerns (including Records Management) | Y |
| Research | Y |
| Communication | Y |
Alternatives Identified
- HL7 (Health Level 7)
- SNOMED CT (Systematized Nomenclature of Medicine Clinical Terms)
- ASTM E1384 (American Society for Testing and Materials)
- X12
- MedDRA (Medical Dictionary for Regulatory Activities)
Final Recommendation
Health Level Seven (HL7), Version 2.4 and higher, with conditions. This complies with the OMB’s Race and Ethnicity standards for reporting.
The workgroup performed a comprehensive review of standards and early elimination of any that did not properly meet the immediate demographic requirements for the data elements or concepts identified as patient demographics. The group developed a checklist of data elements needed for proper demographics collection and unique patient identification, performed a comparison of each standard against this checklist, and selected the standard that best meets the overall requirements identified along with recommendations for filling any gaps. Refer to the attachment which demonstrates the elements and robustness of the different standards.
Content Coverage
See attachment.
Acquisition
Standards are available from HL7. HL7 asserts and retains copyright in all works contributed by members and non-members relating to all versions of the Health Level Seven standards and related materials, unless other arrangements are specifically agreed upon in writing. No use restrictions are applied.
Hard and computer readable forms of the standards are available from HL7 and cost $50-$500 depending on the specific standard and member status.
No cost is associated with using the standard.
Conditions
This is a conditional recommendation based on the following changes:
- Marital status: The available values should be re-examined due to the many overlaps currently in the standard. One recommendation is to make it similar to the ASTM standard, which has better choices.
- Gender: The selections within this data element contain the necessary selections (male, female, unknown) for demographic purposes. There are additional values in the code set that may be needed by others outside the federal enterprise or for other purposes. However, since a subset of the values meets federal needs for demographic purposes, the workgroup recommends adoption with the understanding that enhancements may be re-evaluated for future applicability in other areas. Specifically, the workgroup did not expand the gender category into gender identification and chromosomal sex. That level of specificity is not yet available in the HL7 tables. In addition HL7 has a separate field for gender status. Upon satisfactory listing of these distinct vocabulary tables the gender section could be considered for adoption.
- Insurance Status: This data element and its associated values should be added.
- Living Status: This data element should be considered for those situations when information related to the presence of another individual in a residence is needed to determine the supervisory care needed. Such information is also becoming more necessary where homelessness needs to be considered or tracked. ASTM has this data element for tracking whether a patient lives alone or with someone.
- It is also recommended that a separate group maintain a mapping to X12 and where appropriate initiate a harmonization effort.
- Any other required federal standards that should emerge will require a harmonization between HL7 and the new required standard.
[2] Information Sheet designed specifically to facilitate communication between CHI and NCVHS Subcommittee on Standards and Security resulting from May 20, 2003 testimony. CHI may seek assistance to help further define scope, alternatives to be considered and/or issues to be included in evaluation process.