NLM Health Data Standards Executive Summary for 2016
As the coordinating body for clinical terminology standards within HHS, the National Library of Medicine has a diverse and ever-growing portfolio of products and services that supports interoperability and the unambiguous exchange of health data. Below is a summary of the major activities and accomplishments in this domain in calendar year 2016.
On June 23, 2016, VSAC published downloadable files of C-CDA r1.1 value sets available, authored by the HL7 Terminology group. For access, select the C-CDA Value Sets section in the VSAC Downloads tab.
On September 14, 2016, VSAC published the CDC Race and Ethnicity Code Set Roll-up Code Supplement File. In support of the 2015 Edition Common Clinical Data Set, this value set supplement file allows systems to discover the appropriate roll-up race code, given a detailed race code from the CDCREC Race & Ethnicity Code System. For access, select the CDCREC Roll-up Codes section in the VSAC Downloads tab.
On January 6, 2017, CMS used VSAC to publish nearly 2,000 updated value sets to be used by eCQMs in electronic health records for the 2017 eCQM reporting period. In addition, VSAC continues to provide value set authoring and retrieval services for many medical specialty societies, clinical data registries, academic institutions and other federal health IT initiatives.
Throughout 2016, the Value Set Authority Center (VSAC) Collaboration Tool saw an increased usage of its services by the Centers for Medicare & Medicaid Services (CMS) electronic Clinical Quality Measures (eCQMs) program; the Council of State and Territorial Epidemiologists Reportable Condition Knowledge Management System; and the Federal Health Interoperability Modeling and Standards program. The VSAC Collaboration Tool, open to all users with a UMLS License, provides communication, knowledge management and document management tools for value set authors and stewards with the goal of supporting the highest quality value sets for use in electronic health records. VSAC Collaboration provides tooling that helps un-silo the value set authoring process, by allowing value set authors to get feedback on their content in development and draft stages, resulting in the publication of higher quality value sets.
In 2016 the NLM continued to support SNOMED CT with the March and September SNOMED CT US Edition releases. These releases resulted in the addition of nearly 1,300 new concepts specific to the needs of U.S. healthcare practitioners and systems. Content from IHTSDO and NLM partner Kaiser Permanente has continued to be included at the International core and US-specific SNOMED releases. Areas of musculoskeletal, radiology, cardiology, ophthalmology, and mental health among other areas, were included. Support for US users’ needs was provided through the NLM's U.S. SNOMED CT Content Request System (USCRS).
The NLM continued to maintain the SNOMED CT to ICD-10-CM map (84,238 SNOMED CT concepts from the September 2016 US Edition of SNOMED CT mapped to the 2016 version of ICD-10-CM) to facilitate the integration of clinical and administrative data. This map continues to be a highly used resource for the support of reimbursement from clinical data encoded with SNOMED CT. The SNOMED CT CORE Problem List Subset was also updated and now contains over 6,159 of the most commonly used clinical problems and diagnoses. Additionally, with over 30 SNOMED CT releases (International, refsets, subsets and harmonization projects) in 2016, users have been able to find and utilize SNOMED CT data releases for specific use case interoperability needs.
As previously announced by IHTSDO, the July 2016 release of SNOMED CT was the last time the files will be distributed in Release Format 1 (RF1). Going forward Release Format 2 (RF2) is the authoritative distribution format for SNOMED CT International as well as the US Edition of SNOMED CT. To support this transition, NLM hosted webinars to inform and assist users in resources available for the conversion to RF2 format. Antecedent versions of SNOMED CT (e.g. SNOMED International, SNOMED RT, etc.) will no longer be available for use other than for historical or research purposes. The NLM has hosted webinars on both topics and will continue providing additional documentation to assist users in both areas.
In addition to the maintenance of terminology data resources, NLM continued supporting SNOMED CT as a U.S. clinical standard via several avenues. We continued working with the IHTSDO to ensure US priorities are represented. Additionally, work continues with SI and SI Members to meet the needs of now 30 Members around the world working towards clinical data standards and harmonization.
Lastly, the NLM continues to update our SNOMED CT web resources including newly redesigned, mobile responsive SNOMED CT webpages, new nursing documentation and many other resources introduced in 2016. In 2017, NLM will continue to provide SNOMED CT data resources and services to our customers while upgrading and migrating all internal production tooling behind the scenes to help support the growing need for SNOMED CT products and services for US clinical health information data exchange.
Several improvements to the repository were added, including the ability to export forms in all available formats (JSON, XML, XML/ODM, XML/SDC, XML/SDC with XSL Transform, and REDCap), support to variable level mapping to datasets (PhenX/dbGaP mapping added), add a form inside another form, display a synonym in the permissible values field if the code is in the UMLS (via the API), the addition of Google Analytics and schema.org elements, and a new printable rendering of a form. NLM collaborated with ONC to improve the functionality and documentation of the repository in support of the ONC Structured Data Capture Initiative.
The NIH CDE Repository also made several enhancements to support collaboration and CDE/Form collection management. Users are now able to participate in live discussions on CDEs/Forms and Boards. Those developing CDE collections on behalf of an organization can create and share a board with team members and now add a reviewer before promoting the collection. Administrators of a CDE/Form collection can now annotate their collection topically by MeSH terms, have customizable validation rules at the organization level, and can embed their CDE collection in a CDE Repository search widget on their own website.
MedlinePlus Connect is a free service of the National Library of Medicine linking health IT systems, patient portals and electronic health record (EHR) systems to targeted, authoritative patient health information from the MedlinePlus.gov website. Both a Web application and a Web service, it uses standard clinical vocabularies for diagnoses (problem codes), medications, and lab tests to facilitate the retrieval of patient education materials. In 2016, MedlinePlus Connect provided patient education materials to more than 11 million EHR users. MedlinePlus Connect supports the HL7 Context-Aware Knowledge Retrieval (“Infobutton”) Knowledge Request Standard, and has since release in 2010.
LOINC is a set of universal codes and names to identify laboratory and other clinical observations. It facilitates the exchange and pooling of clinical results for clinical care, public health and research. The Regenstrief Institute, Inc., an internationally respected healthcare and informatics research organization, maintains the LOINC database and supporting documentation, and the RELMA mapping program. In 2016, through many collaborations, LOINC extended its coverage in many domains relevant to clinical care providers, researchers and public health:
- In a collaboration with the Radiological Society of North America (RSNA), LOINC finished converting its terms for magnetic resonance imaging, nuclear medicine, ultrasound, and x-ray reports into the LOINC/RSNA unified model, which is based on the useful aspects of LOINC Radiology and the RSNA RadLex Playbook, and added new radiology LOINC terms when needed, for a total of approximately 5,000 terms, which are being adopted by the VA, DoD, and other large hospital systems. Funding came from the National Institute of Biomedical Imaging and Bioengineering (NIBIB).
- In collaboration with the Institute of Electrical and Electronic Engineering (IEEE), LOINC mapped about 70% of IEEE’s active 11073, 10101 and 10101a codes for observations generated by instruments, such as ventilators and ECGs, to appropriate LOINC codes. This will permit easy integration of the output of such instruments into Electronic Health Records.
- LOINC is part of the Centers for Medicare and Medicaid Services (CMS) effort to standardize its major assessment instruments. including their Post-Acute Care (PAC), Nursing Home Minimum Data Set (MDS), Long-term Care Hospital CARE Data Set (LCDS), In-Patient Rehabilitation Facility Patient Assessment Instrument (IRF-PAI), and the Home Health Outcome and Assessment Information Set (OASIS). LOINC is providing LOINC codes that will be linked to the individual items in the CMS Data Element Library (DEL) for easier integration and use by clinical systems.
- LOINC and NLM participated with the FDA, CDC, IICC and the IVD industry in a number of public meetings to produce a draft electronic format in which IVD venders could represent vetted mappings from their instrument test codes to the corresponding LOINC codes. The IVD manufacturers could then deliver these mappings in an industry standard format to their laboratory customers and relieve them of the mapping work needed to get LOINC codes into their system
Provided new codes and health IT technical assistance for nationwide newborn screening public health efforts
The Newborn Screening and Coding Health IT Standards Guidance project created and mapped codes for Newborn screening conditions (Mucopolysaccharidosis type I (MPS I) and X-linked adrenoleukodystrophy (X-ALD)) and the analytes used to screen for them. These codes were added because the conditions were added to the Recommended Uniform Screening Panel promulgated by the federal advisory committee (Advisory Committee for Heritable Disorders in Newborns and Children), and to some state public health screening protocols. Also, NLM helped created, mapped, and revised additional codes related to newborn screening interpretations, feeding types, and infant NICU factors that affect newborn screening interpretation. These additions and changes to the guidance were based on requests and input from 23* state newborn screening programs that are mapping to NLM-supported standard codes including LOINC and SNOMED, as well as the Association for Public Health Laboratories and the Newborn Screening Technical assistance and Evaluation Program (NewSTEPs) Health IT Workgroup. NLM assisted with a process led by NewSTEPs to begin gathering updated HIT-readiness information from the state public health departments. Additionally, NLM served on a subcommittee of the HL7 Laboratory US Realm Working Group to begin to develop a Newborn Screening Dried Blood Spot profile for the Laboratory Results Interface Implementation Guide.
*Colorado, Delaware, Florida, Georgia, Iowa, Kentucky, Michigan, Minnesota, Nebraska, New York, Ohio, Pennsylvania, Texas, Utah, Virginia, Washington, Wisconsin with Maine, Massachusetts, New Hampshire, Rhode Island, Vermont served by the New England Newborn Screening Program
The HL7 v2 Newborn Dried Blood Spot Screening implementation guide, based on the original LHC HL7 message structure, has been balloted as part of the HL7 v2.5.1 Implementation Guide: Lab Results Interface (LRI).
The Lister Hill Center for Biomedical Communications (LHC), the R&D component of NLM, was the principal contributor to the Clinical Genomics section of the Health Level 7 (HL7) Version 2.5.1 Implementation Guide: Lab Results Interface (LRI), as well as tools for creating example HL7 V2 clinical genomic messages via the LHC-Forms Widget. LHC has led the development of this guide to encourage and make easier for laboratories adding structured results to the pure narrative genomic reports that are common today. LHC has also created a clinical table search service that provides look-up tables for much of the genetic fields required for the HL7 genetic messages, as well as look-tables for ICD9, RxTerms, and others, which many clinical applications often require.
Additionally, NCBI created MedGen as the phenotype backbone of the Genetic Testing Registry (GTR) test descriptions and ClinVar variant interpretations, and to support interactive and computational needs of the genetic community. MedGen does not create terms - it draws from authoritative sources such as OMIM, HPO, Orphanet, NCI Thesaurus, MeSH, and SNOMED CT - as well as terms from submitters. MedGen organizes the attributes of phenotypes - such as preferred names, synonyms, definitions, associated genes, modes of inheritance and related conditions - and assigns to each phenotype a concept unique identifier (CUI). The CUI from Unified Medical Language System (UMLS) are used when available; otherwise MedGen creates CUI that are distinguished because they begin with 'CN'. Concepts come in the semantic flavors of conditions and clinical features, and HPO – OMIM mappings make it possible to see all features of a disorder or find all disorders with (a) particular feature(s) (e.g. clinodactyly).
MedGen builds on UMLS to resolve the plethora of synonymous terms for conditions - reflecting the enzyme deficiency, affected gene, constellation of clinical features, abbreviation, person who first reported the condition, etc. - by revealing all condition names for a given concept and prioritizes a preferred term based on vocabulary source. Synonyms and other record features such as definitions and mode of inheritance provide confidence to users about having found their record of interest. High-value literature augments MedGen e.g., pre-calculated PubMed queries and connections to 521 practice guidelines, recommendations, position statements and evidence-based reviews for 878 conditions. Medical images enhance MedGen e.g., from Elements of Morphology and links to A.D.A.M. Medical Encyclopedia and PubMed Health.
Laboratories, manufacturers, hospital systems and databases can no longer operate in isolation, thus the benefit of sharing computationally accessible information globally and across the continuum of care. We will demonstrate how MedGen contributes to the precision medicine initiative by harmonizing genetic nomenclatures that facilitate integration of electronic health records (EHR), streamline the billing process, enable accurate interpretation of the clinical significance of sequence variation as well as deep phenotyping of sequenced individuals to inform variant interpretation. Some ways to accomplish this is the use of SNOMED CT for most preferred terms to support EHR data mining and the freely available mapping of MedGen, HPO and OMIM concepts on the ftp site.
In 2016, NLM added DrugBank as the newest source in RxNorm. DrugBank is a bioinformatics and cheminformatics resource that combines detailed drug (i.e. chemical, pharmacological and pharmaceutical) data with comprehensive drug target (i.e. sequence, structure, and pathway) information. The addition of DrugBank to RxNorm provide a direct link between drug information in the clinical and research settings.
Following discussions with stakeholders and feedback received while exhibiting at the 2016 National Council for Prescription Drug Programs (NCPDP) annual meeting, NLM added a new attribute to RxNorm that allows users to more easily differentiate between specific drug products, such as sugar-free diabetes medications, products that aid with smoking cessation, and opioids with abuse-deterrent properties. NLM also refined its process for editing National Drug Code (NDC) content to better identify active NDCs on the market in the United States. NLM worked with many of its data providers to suppress more than 7,000 inactive NDCs in RxNorm.
RxNav gets a new design
NLM released a mobile-responsive, web-based version of RxNav, which allows users to search and browse RxNorm drug information without downloading any data or software. RxNav 2.0 provides new features and functionality through an easy-to-use interface. In 2016, users queried the NLM Drug APIs, which power RxNav 2.0, over 800 million times. The number of unique users has continued to increase sharply, from 24,000 in 2015 to 40,000 in 2016 (monthly average). The top query is for resolving NDCs into RxCUIs. Mapping drug names and identifiers accounts for over 50% of all queries. MedlinePlus Connect was the single largest customer, accounting for 20% of traffic.
NLM hosts 4th annual DailyMed Jamboree
On September 27, 2016, NLM hosted 98 in-person attendees and 127 online attendees in the Lister Hill Auditorium for the 2016 DailyMed/RxNorm Jamboree. The Jamboree featured presenters from Geisenger, the United States Pharmacopeia, the FDA, ONC, and more, with topics ranging from drug allergies to lab culture content to risk evaluation and mitigation.
In 2016, NLM added five new and updated sources to the Unified Medical Language System (UMLS) Metathesaurus to support interoperability among health systems:
- Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) - a reference used by clinicians and researchers to diagnose and classify mental disorders
- DrugBank - a bioinformatics and cheminformatics resource that combines detailed drug (i.e. chemical, pharmacological and pharmaceutical) data with comprehensive drug target (i.e. sequence, structure, and pathway) information
- Geopolitical Entities, Names, and Codes (GENC) - an information model for representing names and codes of geopolitical entities and administrative subdivisions, with supporting information
- Healthcare Cost and Utilization Project Clinical Classifications Software (HCUP CCS) - a diagnosis and procedure categorization scheme based upon ICD-10
- National Uniform Claim Committee - Health Care Provider Taxonomy (NUCCPT) - a data code set designed for use in an electronic environment, specifically within the ASC X12N Health Care transactions, including the transactions mandated under HIPAA
Users queried the UMLS APIs over 29 million times in 2016. The number of queries increased 16% from 2015. The new REST API accounted for 52% of all queries in 2016. The number of unique users increased from 558 in 2015 to 632 in 2016, an increase of 13%. NLM added new functionality to the REST API to allow retrieving UMLS content view information, retrieving subset information, and crosswalking source-asserted identifiers from one terminology to another.