Data mapping in healthcare is the explicit specification of how data elements, fields, and codes in one system or format correspond to those in another, so that information can be exchanged or integrated without losing its intended clinical meaning. It covers both structural alignment—placing data into the correct positions and data types in standards such as HL7 v2, FHIR, and DICOM—and terminology alignment, mapping local or proprietary codes to standard vocabularies like SNOMED CT, LOINC, ICD, and RxNorm. Effective data mapping is a foundational step for interoperability, quality reporting, and the safe reuse of clinical documentation across systems and programs.
For clinicians and clinic administrators evaluating documentation tools, data mapping is usually invisible until something goes wrong: a result lands in the wrong field, a code is translated imprecisely, or a report cannot be reused. Below, we cover how mapping works in clinical interfaces and terminologies, how mapping projects are validated, and where it is most often misunderstood.
Definition: What Data Mapping Means in Healthcare
A mapping is a specification, not a data transfer: it states which field, code, or value in a source system lines up with which field, code, or value in a target system or standard message format.
In integration and interoperability projects, the mapping describes how fields and values from a source system are transformed into the structure and codes required by a target system or standard message format. Without explicit mapping, systems may technically connect but still misinterpret each other’s data, which undermines both clinical use and reporting.
How Data Mapping Shows Up in Clinical Interfaces
In traditional HL7 Version 2 interfaces, data mapping links local system fields to HL7 message segments and fields. Typical examples include mapping internal patient identifiers, location codes, or test codes to HL7 v2 segments such as PID for patient identification, PV1 for visit information, and OBX for observation results. Implementers must place local data elements into the specific segment–field–component structure defined in the HL7 v2.x standard, which prescribes the position, data type, and meaning of each field in a message.
In HL7 FHIR-based integrations, data mapping links source system fields to elements of FHIR resources. Patient demographics are mapped to elements of the Patient resource, and laboratory results are mapped to elements of the Observation resource. Because FHIR resources are defined with explicit elements and data types, implementers must align each source data element with the appropriate FHIR element and transform values into the formats and code systems expected by that resource.
Imaging workflows require similar work. The DICOM standard defines how imaging-related data elements are encoded in DICOM attributes. When imaging systems exchange data with non-DICOM systems, implementers must map DICOM attributes, including study, series, and instance identifiers, to corresponding fields in HL7 messages, FHIR resources, or other data structures so that imaging data stays linked to orders, encounters, and reports.
Terminology Mapping: From Local Codes to Standards
A large part of healthcare data mapping is terminology mapping. This is the work of mapping local or proprietary codes used in an EHR or departmental system to standard terminologies such as SNOMED CT for clinical concepts, LOINC for laboratory and clinical observations, and ICD for diagnoses. This is central to reusing data across systems, analytics, and reporting.
SNOMED CT is designed to support mapping from other code systems and local terminologies. Its documentation describes mapping as defining relationships between source codes and SNOMED CT concepts to enable data reuse and interoperability. LOINC provides guidance and tooling for mapping local test codes and names to LOINC codes, describing this as a key step for laboratories and clinical systems that want to exchange or aggregate test results consistently.
The U.S. National Library of Medicine’s UMLS Metathesaurus links similar names for the same concept from nearly 200 different vocabularies. NLM also publishes a map from SNOMED CT to ICD-10-CM, while SNOMED International publishes an extended map from SNOMED CT to ICD-10. For laboratory content, the LOINC Ontology, developed through a collaboration between the Regenstrief Institute and SNOMED International, links LOINC and SNOMED CT. For medications, RxNorm provides normalized names and unique identifiers for drugs and groups synonymous names from source vocabularies into single concepts, supporting semantic interoperation between drug terminologies.
Mapping between terminologies is rarely perfect, and the intended use matters. SNOMED International’s guidance notes that where patient safety or data accuracy is important, a map should be an exact match, while for other purposes a broader target may be acceptable. Map patterns also vary: one-to-one, many-to-one, one-to-many, or one-to-none when a source term has no available target. A funding use case, for example, would be expected to use many-to-one maps but not one-to-many maps.
Data Mapping in Documentation, Quality and Reporting
Data mapping directly affects clinical documentation tooling because many external programs expect data in specific formats and code systems. The examples in this section are U.S. programs; the same mapping principle applies to any program that specifies required code systems and formats. In the U.S., the Office of the National Coordinator for Health IT’s certification program requires the use of the United States Core Data for Interoperability (USCDI), a standardized set of health data classes and data elements, so implementers must map their internal data structures to those elements.
U.S. quality reporting programs such as those run by the Centers for Medicare & Medicaid Services require submission of data using specified code systems and formats. In practice, this requires providers and vendors to map internal clinical data to the codes and data elements defined by those programs. The U.S. National Library of Medicine’s Value Set Authority Center distributes value sets, lists of codes from standard vocabularies such as SNOMED CT, RxNorm, and LOINC, including the official value sets CMS specifies for electronic clinical quality measures, and implementers map local codes to those value sets.
Technical frameworks reinforce this. HL7 implementation guides often include detailed mapping tables that show how data elements from one standard or profile correspond to elements in another. Implementers use these tables as authoritative references when building data mappings in software. The Integrating the Healthcare Enterprise initiative publishes technical profiles that specify how systems should use standards such as HL7 and DICOM together. Identity profiles such as Patient Identifier Cross-Referencing and Patient Demographics Query, for example, address how local patient identifiers and demographic data are linked across systems for cross-enterprise patient identity management.
Formal Mapping Artifacts: ConceptMap and StructureMap
Modern standards include explicit resources for representing mappings. The HL7 FHIR specification includes a ConceptMap resource designed to represent mappings between code systems, value sets, or concepts, including the relationship between source and target codes. In the current FHIR release (R5), ConceptMap relationship codes include equivalent, source-is-narrower-than-target, source-is-broader-than-target, related-to, and not-related-to, which supports more precise documentation of how data is mapped between terminologies. ConceptMap resources can be used by terminology services to perform code translation operations at runtime, effectively applying predefined mappings when systems exchange coded data.
FHIR also defines a StructureMap resource to represent mappings between different data structures, including how elements in one structure are transformed into elements in another. This can be used to formalize data mapping between different message formats or resource profiles.
How Mapping Projects Are Designed and Validated
Organizations that maintain major terminologies treat mapping as a discipline with its own requirements. SNOMED International’s mapping guide says mapping rules should be established and documented from the beginning, allowable relationship types should be determined for each map, and the degree of equivalence between source and target is determined by the map’s use case. Its reference sets distinguish simple maps, which support one-to-one maps, from complex and extended maps, which support cases where additional information is required to determine the correct target in a statistical classification such as ICD-10.
Direction matters too. SNOMED International notes that a map built in one direction will serve poorly, if at all, in reverse unless all links are exact semantic matches, and FHIR StructureMap maps are likewise one-way, so a reverse transformation is not implied. Maps must also be maintained so that both source and target content remain current as code systems are updated.
Validation is a recurring theme. SNOMED International’s mapping guide describes content review of map rows, including full review, sampling review, and dual independent mapping, alongside design review and testing in the context in which the map will be used, and it recommends peer and dual review supported by clinical and domain expert validation. It also advises keeping detailed records of map decisions, including the rationale for inexact, broader, narrower, or “no map” choices, and recording who created and reviewed each map. The same guidance notes that structured laboratory content lends itself more readily to automated mapping and checks, while ICD-10 coding depends on expert judgement.
Where Clinicians Commonly Misunderstand Data Mapping
Several misunderstandings tend to surface when clinicians and clinic administrators encounter data mapping during documentation or integration projects. First, mapping is often assumed to be exact. As the terminology and validation sections above show, the acceptable degree of equivalence depends on the use case, so a map that works for funding or analytics may not be precise enough for decision support, and it may not work in reverse.
Second, mapping is sometimes treated as a purely technical exercise that IT or vendors can complete in isolation. However, guidance from terminology organizations stresses clinical and domain expert validation and testing in the context of intended use to confirm that mapped codes preserve intended clinical meaning. This implies that clinicians or clinical domain experts have a role in reviewing and approving mappings that affect clinical documentation, decision support, or quality measures.
Third, the structural side of mapping is easy to overlook. Clinicians may focus on the terminology codes, but HL7 v2, FHIR resources, and DICOM attributes have specific structures, positions, and data types that must be respected. For HL7 v2, local data must be mapped into the precise segment–field–component locations defined in the standard. For FHIR, each source element must be aligned with the correct resource element and transformed into the expected formats and code systems. For DICOM, identifiers for studies, series, and instances must be correctly mapped to non-DICOM identifiers to keep orders, encounters, and reports aligned.
Fourth, it is sometimes assumed that mapping is a one-time setup. In reality, major organizations publish and maintain mapping tables over time—for example, between SNOMED CT and ICD-10 variants, or between SNOMED CT and LOINC—because terminologies and classifications evolve. Mapping rules and artifacts such as FHIR ConceptMap resources may need to be updated as code systems change or new quality measures and value sets are introduced.
Practical Implications for Evaluating Documentation Tools
For clinicians and clinic administrators evaluating documentation tools, the useful questions are concrete. Which HL7 v2 segments or FHIR resources carry your identifiers, demographics, orders, and results, and which code systems are attached to them? In Canada, Canada Health Infoway maintains and distributes the SNOMED CT Canadian Edition, an extension of the international edition with Canadian requirements, and pCLOCD, a Canadian constraint of LOINC, so it is worth asking vendors whether their mappings reference these national versions.
Ask, too, whether code mappings are held in an explicit form such as FHIR ConceptMap, and what relationship types are allowed when there is no exact equivalent.
Finally, ask who reviews the mappings, whether clinicians are involved, how map decisions are recorded, and how maps are updated when terminologies release new versions.
Key Takeaways
- Data mapping specifies how fields, codes, and values in one system correspond to another.
- Structural mapping places data into HL7 v2 segments, FHIR resource elements, and DICOM attributes.
- Terminology mapping links local codes to SNOMED CT, LOINC, ICD, and RxNorm.
- U.S. certification and quality programs show how reporting rules drive mapping requirements.
- FHIR ConceptMap and StructureMap make mappings explicit and machine-readable.
- Good mapping projects document rules, use dual or full review, and test in context.
- Maps are not one-time work; they need maintenance as code systems change.
Our Take
In our view, the hardest part of mapping is the case with no exact equivalent. A one-to-one map is easy to check; the inexact, broader, narrower, and “no map” decisions are where clinical meaning is most likely to shift. Those are the decisions clinicians are best placed to review, and the ones most worth asking a vendor to show you.