Semantics


1. Introduction

The following presents the basic considerations for dealing with terminologies and classifications used to represent microbiological diagnostics.

An overview of all terminologies and classifications used can be found in the chapter Basics and Overview/Terminologies, and the value sets can be found in the chapter Resources/Terminologies/ValueSets.


2. Basic Considerations for Mapping Microbiological Diagnostics

The fundamental principle for the profiling of FHIR resources and the handling of semantics presented here is that the overarching goal of interoperability of health data in Germany and Europe is also being pursued for antibiotic resistance surveillance. For the mapping of microbiological diagnostics, we follow rules that are intended to form the basis for the use and interaction of LOINC and SNOMED-CT in particular. These rules are intended to enable the mapping of diagnostics as comprehensively and coherently as possible using LOINC and SNOMED-CT.

The mapping of microbiological diagnostics should be implemented in a form that is suitable for various use cases in Germany and Europe. Even if a level of detail on procedures and results of microbiological diagnostics relevant for other use cases is not required for ARS, the profiling and handling of semantics presented here have been repeatedly checked to see whether they are also suitable for use cases with a greater level of detail.

The created value sets are subject to continuous development and refinement based on the needs of the participants in accordance with the established rules for dealing with semantics. Currently, the value sets containing LOINC codes also include codes that do not always conform to the established rules, but represent widely used or currently available codes. New code applications are submitted and reviewed on an ongoing basis. A detailed discussion of this can be found in section "4.1. LOINC" and in the subsection Semantics Examples.


3. Rules for Handling Semantics

To ensure that microbiological diagnostics can be represented coherently across pathogens, specimen, and methods, and that the use of coding systems for data senders is as effective as possible, we have formulated rules for handling semantics. These rules describe the goal for handling semantics for ARS. For full implementation of the rules presented here, both new LOINC codes and codes for SNOMED-CT must be requested. In addition, adjustments to laboratory information systems will likely be necessary to maintain coded content in accordance with the rules presented here.

Compromises are necessary at various points and times and are also part of the discussion presented in the examples. The rules for dealing with semantics presented here are intended to provide the concept of what to aim for when developing and expanded mapping of microbiological data on the part of the data senders and are continuously subject to coordination processes with other use cases in Germany, such as the MIO laboratory report and the microbiology extension module of the Medical Informatics Initiative, and on European level.


The rules for using the central terminologies LOINC and SNOMED-CT for ARS are:

General:

  • The steps of microbiological diagnostics should be mapped as a process.
  • A unique mapping of pathogen, specimen, and method in the FHIR resources should be possible.
  • The LOINC code reflects the question behind the diagnostic step. Defined value sets are created, especially for LOINC codes.
    • Value sets for ARS are not binding, and codes other than those recommended here may be used.
  • In principle, established value sets, such as the German SNOMED-CT Germany Edition and value sets from HL7 Germany and HL7 Europe, are used wherever possible.

For diagnostic procedures for pathogen detection and identification, usually based on culture techniques:

  • For procedures that involve broad pathogen detection and identification, the nonspecific LOINC code 41852-5 Microorganism or agent identified in specimen is used.
  • Pathogen, specimen, and method are imaged using SNOMED-CT.

For diagnostic procedures that address a more specific question, e.g., the targeted detection of a pathogen, antibodies, or genes:

  • If a diagnostic procedure is used to address pathogen-specific questions, more specific LOINC codes are used. The LOINC code should be specific to the analyte (component axis) but, if possible, nonspecific to the method (method) and specimen (system).
  • The more specific LOINC code reflects the question that the microbiological diagnostic step can answer.
  • The answer using SNOMED-CT is often nominal (e.g., positive, negative) or contains characteristics or specifications (e.g. "103480009 | Neisseria meningitidis serogroup B (organism)").
  • The method and specimen are mapped using SNOMED-CT.

4. Discussion on value set referencing

The key terminologies for mapping microbiological diagnostics are LOINC, SNOMED-CT, and UCUM.


4.1. LOINC

For direct and indirect pathogen detection in bacteriology, virology, mycology, and parasitology, phenotypic and genotypic resistance testing, as well as resistance mechanisms and classifications, value sets with LOINC codes are offered for ARS. The LOINC code should reflect the question underlying the diagnostic step, in accordance with the semantic rules established for ARS. The following value sets are defined:

TestBakLOINC; TestVirLOINC; TestMykLOINC; TestParLOINC

  • These value sets contain a code for non-specific pathogen detection, which corresponds to the question "Is a pathogen detectable in the submitted material, and if so, which pathogen is it?" and is usually based on culture methods, as well as codes for more specific, non-culture methods for pathogen detection.
  • Methods for non-specific pathogen detection and identification
    • For pathogen detection based on culture methods, which determine whether a pathogen is detectable and, if so, which pathogen it is, the value sets contain the LOINC code "41852-5 Microorganism or agent identified in specimen." Particularly for bacterial pathogen detection and the detection of fungi in culture materials, this LOINC code corresponds to the non-specific question posed by the submitter and the test procedures used to identify the pathogen.
    • Currently, the LOINC code "41852-5 Microorganism or agent identified in specimen" is used for all steps of non-molecular diagnostics for pathogen identification, e.g., culture cultivation, microscopy, MALDI-TOF. See also Example 1, Example 2, Example 3.
  • Methods for Specific Pathogen Detection and Identification
    • Methods for direct and indirect pathogen detection are also used for both bacteriological detection and the detection of viruses, fungi, and parasites, which address a more specific question. According to the rules for dealing with semantics, this more specific question should be more specific in the "component" axes of the LOINC code, depending on what the test procedure used detects, and, if possible, nonspecific for the "system" and "method" axes, e.g., detection of Clostridioides difficile using organism-specific culture (34712-0 Clostridioides difficile [Presence] in Stool). See also Example 4 in subchapter Semantic Examples).
    • Here, the value sets contain LOINC codes, particularly for serological diagnostics for the detection of antigens and antibodies, which contain for example serum or cerebrospinal fluid for the LOINC axis "system". The possibility of requesting LOINC codes with a non-specific "system" in accordance with the established rules for handling semantics, in order to also be able to represent tests with other starting materials such as plasma, urine, or other body fluids, is under discussion.
    • For molecular methods for pathogen detection, e.g. "35492-8 Methicillin-resistant Staphylococcus aureus (MRSA) DNA [Presence] in Specimen by NAA with probe detection", LOINC currently predominantly contains codes that specify "by NAA with probe detection" or "by NAA with non-probe detection" as the method. It is becoming apparent that these codes are increasingly being provided without this methodological addition. The value sets will be adjusted accordingly in the future.

TestResistancePhenotypicLOINC

  • This value set contains the existing LOINC codes for active substances approved in Germany and tested in laboratories. They are available in the following format: component: active substance name, property: [Susc], time:Pt, system: Isolate, scale: OrdQn, method: (unspecified).
  • The value set must be supplemented, in particular, with LOINC codes for active substances for which the interpretation of the result according to EUCAST requires information on the application of the active substance (oral, IV) or the underlying clinical picture (e.g., uncomplicated urinary tract infection, meningitis). LOINC proposes to map the underlying disease pattern in the "system" axis, e.g., system: isolate.UTI in "103672-2 Amoxicillin+Clavulanate [Susceptibility] for UTI," and the route of administration in the "component" axis, e.g., "35783-0 Cefuroxime Oral [Susceptibility]." Missing LOINC codes for mapping differential breakpoints in EUCAST are requested.

TestResistanceMolecularLOINC

  • Carbapenemases:
    • Rapid immunochromatographic tests and PCRs allow for determination of whether a carbapenemase has been detected and, if so, to which carbapenemase group this carbapenemase belongs. The value set contains LOINC codes for the individual carbapenemase groups on the "component" axis.
      • Example code for the carbapenemase group using immunochromatography: "101673-2 KPC carbapenemase [Presence] in Isolate by Rapid immunoassay".
      • Example code for the carbapenemase group using PCR: "49617-4 Carbapenem resistance blaKPC gene [Presence] by Molecular method" LOINC codes for 25 carbapenemases are available for determining carbapenemase by sequencing and are included in this value set.
    • This value set contains currently available LOINC codes that enable the mapping of molecular resistance detection in routine microbiological diagnostics.
  • The value set contains additional codes for the molecular detection of methicillin resistance genes (mecA, mecC), colistin resistance genes (mcr-1, mcr-2), vancomycin resistance genes (VanA, VanB), and beta-lactamase resistance genes.
    • A code for PBP2a testing does not currently exist and is requested.
  • For the molecular detection of resistance genes, we pursue the goal of being able to explicitly state the diagnostic result and thus also the lack of detection in the interest of interoperability. This requires mapping the individual analytes of a test kit, such as a rapid carbapenemase test, using specific LOINC codes and transmitting the result for each analyte in individual "observations" (see also Example 1).

TestResistanceMechanismLOINC

  • This value set contains the non-specific LOINC code "41852-5 Microorganism or agent identified in specimen", currently used as a placeholder. LOINC codes are being requested that enable the detection of an ESBL, an AmpC beta-lactamase, cefoxitin-resistant bacteria, and bacteria with inducible clindamycin resistance. This addresses the need to be able to map results from automated phenotypic resistance testing procedures, e.g., by VITEK, according to the output result.

4.2. SNOMED-CT

Value sets with SNOMED-CT are offered for pathogens, including typing, specimen, sampling site, methods, genes, nominal responses, and diagnostic requests:

ResultPathogenBakSNOMED; ResultPathogenVirSNOMED; ResultPathogenMykSNOMED; ResultPathogenParSNOMED; ResultPathogenDescriptionMicroscopySNOMED

- The German SNOMED-CT Microorganism Reference Set serves as a reference for the value sets for pathogens and their typing. The value sets contain common pathogens already included in ARS. Additional pathogens can be added as child concepts, particularly of "409822003 Domain Bacteria", "414561005 Fungi", "417396000 Kingdom Protozoa", and "49872002 Virus".

  • The ResponsePathogenDescriptionMicroscopy value set contains codes from the "organism" domain, which include microscopy results after Gram staining and acid-fast bacilli staining. The description of the microscopic result is usually of secondary importance for ARS. The value set was included for interoperability reasons and can be expanded.

MaterialSNOMED; CollectionLocationBodySiteSNOMED; CollectionLocationMorphologicAbnormalitySNOMED;

  • This value set contains a selection of codes from the "Specimen" value set from the German SNOMED-CT Reference Set. The selection includes codes that are relevant for current specimen mapping for ARS and microbiology in general. Codes are selected that describe the specimen or sample arriving at the laboratory. For example, an aerobic and anaerobic blood culture bottle are considered two different specimen.
  • Some of the codes describe the specimen in great detail, e.g., "258574006 Mid-stream urine specimen (specimen)." For ARS, detailed sample specimen will be summarized for the body system after transmission. In the example mentioned, midstream urine is considered "urine." The detailed codes are permitted here to ensure interoperability with other use cases.
  • Furthermore, some of the codes describe the collection location in addition to the specimen. For example, a throat swab can be described as a specimenl under "Specimen.type" using the code "472881004 Swab from pharynx". However, it is also possible to specify an unspecified swab as the specimen under "Specimen.type" as "257261003 swab (specimen)" and further specify the collection location in the "Specimen" resource under "Specimen.collection.bodySite" or via the "BodyStructure" resource under "BodyStructure.location". The corresponding code for a complete description of the collection location would be "54066008 Pharyngeal structure (body structure)". See also Example 5.
Option 1 Option 2 Option 3
Specimen.type:
- 472881004 Swab from pharynx
Specimen.type:
- 257261003 swab (specimen)

Specimen.collection.bodySite:
- 54066008 Pharyngeal structure (body structure)
Specimen.type:
- 257261003 swab (specimen)

BodyStructure.location:
- 54066008 Pharyngeal structure (body structure)
  • The collection location of a sample can be a body site (value set with codes from the "body structure" domain) or a morphological abnormality, such as a wound. For this reason, a second value set is provided that contains codes from the "morphologic abnormality" domain (e.g., 13924000 Wound (morphologic abnormality)). The value sets can be supplemented with codes corresponding to the concepts 442083009 (Anatomical or acquired body structure) or 442083009 (Morphologically abnormal structure).

MethodSNOMED

  • The "Laboratory Techniques" value set, which is currently being developed and published as part of the commentary on the HL7 Europe Laboratory Report, will be used as a reference for the method value set. In creating the "Laboratory Techniques" value set for the HL7 Europe Laboratory Report, the change in the description of the methods from "procedure" to "qualifier value" will be implemented, which we will follow for ARS. The value set is supplemented by methods that were not included in the "Laboratory Techniques" value set for the HL7 Europe Laboratory Report but are important for microbiological diagnostics in Germany.
  • Codes are being requested for methods for which no SNOMED codes are currently available as "qualifier values."
  • In the future, the Reference Set for Methods for Germany will be used here.

ResultNominalSNOMED

  • The value set based on SNOMED-CT is defined as a reference for the value sets for nominal responses.
  • Regarding the representation of phenotypic susceptibility testing (RIS) results and the standard used for assessment, the representation and use of SNOMED codes is currently being discussed at the national and European levels. The value set currently contains the approved and available codes for describing RIS.

ResultResistanceClassificationSNOMED

  • This value set currently contains a reference to SNOMED-CT as a placeholder code. SNOMED codes for 2MRGN, 3MRGN, and 4MRGN are being requested.

InterpretationSNOMED

  • For the interpretation of laboratory results, coding is possible both according to the basic profile of HL7 and with SNOMED-CT.

DiagnosticRequirementsSNOMED

  • This value set contains two codes that, for ARS, distinguish between a screening sample and a sample suspected of infection. This information should be provided for each sample in ARS.

DiagnosticRequirementsMeaslesSentinelSNOMED

DiagnosticRequirementsChlamydiaSentinelSNOMED


5. FHIR Profiles and Use of Terminology

The following FHIR resources are available to illustrate the steps of microbiological diagnostics:

  • Sample (Specimen)
  • Laboratory Testing (Observation)

5.1. Sample (Specimen)

The process of microbiological diagnostics starts with the laboratory receiving the sample. The sample is represented in the "Specimen" resource. Only SNOMED-CT codes are used, drawing on the German SNOMED-CT Reference Set for specimen.

In the "Specimen" resource, in addition to the specimen as "Specimen.type", the collection location "Specimen.collection.bodySite" can be described which allows for a clear description of both the specimen and the collection location. Alternatively, for reasons of interoperability, a description is also possible using the "BodyStructure" resource, specifically "BodyStructure.location" and "BodyStructure.morphology." When defining the specimen and collection location, we follow the SNOMED-CT Germany Edition for materials.

For blood cultures, we aim to treat the aerobic and anaerobic blood culture bottles for ARS as one specimen each ("866033003 Blood specimen in aerobic blood culture bottle (specimen)" and "866032008 Blood specimen in anaerobic blood culture bottle (specimen)") (see Example 1).

For the purpose of clear mapping, the description of the specimen should, if possible, be post-coordinated via the "Specimen" or "BodyStructure" resource with SNOMED-CT.


5.2. Laboratory Testing (Observation)

The "Observation" is the central resource for mapping microbiological diagnostics. Here, the question behind a diagnostic step is mapped under "Observation.code," the result of the test is mapped under "Observation.value," and the interpretation of the test result is mapped under "Observation.interpretation." The methods used are described under "Observation.method."

The goal is to map the differences in diagnostic procedures, such as bacteriological culture methods, serological methods, phenotypic and genotypic resistance testing methods, etc., solely by using the terminologies LOINC, SNOMED-CT, and UCUM in the "Observation" profiled for ARS. The presented value sets are designed to comply with the principle of a unique description of the diagnostics performed and their results, as far as possible with the existing codes. Limitations exist where LOINC codes are currently not available, e.g., without specifying the specimen. The application examples are intended to illustrate the principle pursued in dealing with semantics for mapping the steps of microbiological diagnostics.

For ARS, a single "observation" is profiled to map all possible tests. It can therefore map both quantitative and qualitative results and can be applied to bacteriological, virological, mycological, and parasitological diagnostics (see also "Map of Qualitative and Quantitative Results").

When using the "observation" to map the results of phenotypic resistance tests, the mapping of the norm used to interpret the results (EUCAST, CLSI, others) should be used under "Observation.referenceRange.extension:norm." Profiling a second "observation" solely for mapping phenotypic resistance testing has been omitted. Accordingly, the extension for mapping the norm is stored with cardinality 0...1. However, we ask that you always include the norm when mapping phenotypic resistance tests.


5.2.1. Observation.code und Observation.value

Pathogen detection and identification

For "Observation.code," a LOINC code is required that reflects the question underlying a diagnostic step according to the rules described above.

One pillar of bacteriological diagnostics for pathogen identification is culture methods, which ask the nonspecific question of whether a pathogen is detectable in culture. Following enrichment culture, other methods for pathogen identification are applied.

For bacteriological pathogen detection and pathogen identification, and thus for all diagnostic steps where the question is whether a pathogen is detectable in a specimen and, if so, which pathogen, the nonspecific LOINC code "41852-5 Microorganism or agent identified in Specimen" can be used. Example 1, Example 2, Example 3, and Example 8 demonstrate that this code can be used for methods for pathogen detection and identification, such as enrichment culture, microscopy, and mass spectrometric methods (for the latter, see "Molecular Methods for Pathogen Identification"), etc.

The answer to the question of whether a pathogen is detectable can be coded under "Observation.valueCodeableConcept" by describing the pathogen or the lack of detection. The corresponding answer-value sets are provided as "ResponsePathogenBakSNOMED", "ResponsePathogenVirSNOMED", "ResponsePathogenMycSNOMED", and "ResponsePathogenParSNOMED". If no pathogen has grown in the culture enrichment, the result can be represented using the code "264868006 No growth (qualifier value)" from the value set "ResponseNominalSNOMED".

If molecular methods are used for pathogen search and identification or for direct and indirect pathogen detection, LOINC codes are available in the value sets "TestBakLOINC", "TestMycLOINC", "TestVirLOINC", and "TestParLOINC". These codes usually have the property [PrThr]. The answer can be given using SNOMED codes contained in the value set "AntwortNominalSNOMED" (see also Example 4, Example 5, Example 6 and Example 7).


Phenotypic Resistance Testing

For phenotypic resistance testing, all resistance tests performed must be mapped to create a complete antibiogram. The value set "TestResistancePhaenotypicLOINC" contains LOINC codes that map the active ingredient on the "component" axis and have [Susceptibility] as the "property," e.g., "18864-9 Ampicillin [Susceptibility]." LOINC codes that map methods or the unit of measurement results, such as the MIC, have been omitted in accordance with the aforementioned rules for using terminology in ARS.

An exception are LOINC codes that represent susceptibility testing of active substances for which differential breakpoints are defined according to EUCAST depending on the underlying disease or the type of application, e.g. "50633-7 Ceftriaxone [Susceptibility] for meningitis". The FHIR resource "Observation" allows, as an extension under "Observation.referenceRange", the mapping of the standard used (EUCAST, CLSI) according to which the result of the phenotypic susceptibility testing was interpreted.

The codes contained in the value set have the "scale" [OrdQn] and thus allow both qualitative and quantitative answers; see also Example 1 and Example 2.

In principle, the various "Observations" describing the results of phenotypic resistance testing for a drug are understood as a panel. All "Observations" of this panel are summarized with a bracketed "Observation". The LOINC code "29576-6 Bacterial susceptibility panel" should be specified as "Observation.code" in the bracketed "Observation". See Example 1 and Example 2. The bracketed "Observation" references the "Observations" describing the pathogen detection via "triggeredBy". The "Observations" within the panel are linked to each other via "hasMember".


Molecular Methods

Molecular methods are used for pathogen identification and typing, for genotypic resistance testing, pathogen-specific antibody and antigen detection, and for the detection of virulence factors, especially exotoxins.

In principle, according to the established rules, a LOINC specific to the analyte (the "component" axis) should be used for the specific detection of, for example, a gene. Currently, the value sets are structured so that all analytes of a test kit are explicitly mapped into individual "observations." The mapping of a multiplex PCR for blood cultures can therefore contain >40 "observations". Whether this mapping represents the optimal approach or whether mapping approaches that, for example, only allow positive detection will be chosen will be part of the discussion during this pilot phase.


Mapping qualitative and quantitative results

Qualitative results can be mapped according to the "property" of the LOINC code by describing the detection or resistance for LOINC codes with the "property" [Presence], of the pathogen or resistance gene for LOINC codes with the "property" [Identified], as described in the previous sections.

Quantitative results are generated by:

  • phenotypic resistance testing in the form of the minimum inhibitory concentration (MIC) or the inhibition zone diameter
  • molecular methods in the form of the cycle threshold (CT value)
  • in serology in the form of ratios

The breakpoints used to evaluate the results in phenotypic resistance testing do not need to be specified in "Observation.referenceRange" if the standard used is specified instead.

For serological tests with manufacturer-dependent reference values, these should be specified as "Observation.referenceRange".

If it is important for molecular methods to represent the cycle threshold (CT value) as a quantitative result, the quantitative result cannot be represented as "Observation.valueQuantity" in conjunction with a LOINC code with the property [Presence]. To be able to represent CT values, specific LOINC codes that request the CT value for the analyte should be selected according to the established procedure and submitted in a separate observation in addition to the pathogen detection, e.g., "94643-4 SARS-CoV-2 (COVID-19) S gene [Cycle Threshold #] in Specimen by NAA with probe detection".


5.2.2. Observation.interpretation

The "Observation.interpretation" describes the interpretation of the microbiological diagnostic result. A value set with codes from SNOMED-CT is provided.

For qualitative results, the "Observation.interpretation" is sometimes a duplication of the result as "Observation.valueCodeableConcept." The "Observation.interpretation" is particularly important for quantitative results, e.g., in the context of phenotypic resistance testing. When specifying a quantitative value with a unit, the description of the interpretation of the result (e.g., RIS) using the specified standard is important.


5.2.3. Oberservation.method

Codes from SNOMED-CT are available for the method used. The corresponding value set from the HL7 Europe Laboratory Report is used. Some methods have been added to the value set.

If the method used is not yet available at the time the ARS interface is first implemented in the laboratory information system, the general SNOMED code "272394005 Technique (qualifier value)" can be used as a placeholder under "Observation.method," or the information can be omitted (cardinality of "Observation.method" 0...1).


6. Content to be submitted for ARS

In accordance with the objectives of ARS, final results for pathogen identification as well as phenotypic and genotypic resistance testing should be submitted. Although the structure of the profiles and the handling of semantics make it possible to map every step of the microbiological diagnostics performed, not all of these intermediate results need to be submitted to ARS.

In principle, the results of the following diagnostic steps or methods should be submitted:

  1. diagnostics that clearly identify the pathogen phenotypically and genotypically

    • Example of a diagnostic chain from:
      • aerobic blood culture: positive
        • microscopy: gram-positive rods
        • MALDI-TOF: Klebsiella pneumoniae
        • multiplex PCR: Klebsiella pneumoniae
      • anaerobic blood culture: negative
    • For aerobic blood culture as the specimen, the MALDI-TOF result and the multiplex PCR result should be submitted;
    • For anaerobic blood culture as the specimen, the negative result with no cultural enrichment should be submitted (see also Example 1 and Example 2).
  2. diagnostics that are used for phenotypic susceptibility testing;

    • The results of all tested active substances, i.e., the complete resistogram, must be shown, regardless of the result;
  3. diagnostics used for genotypic resistance testing;

    • all procedures and results should be submitted to ARS.

If in some cases it is easier to submit further diagnostic steps to ARS due to reasons of interoperability, the entire data can also be submitted to ARS. The data will then be cleaned up after submission to the RKI.