Definitions
- Concepts - Concepts are known mappings stored in MR which include a code, label, and Concept Class, and description.
- Concept Classes - A concept class is a specific group of codes or a Code System that has been loaded into Mirth Results. (Can be SNOMED, LOINC, NDC, HL7, …)
- Code System (aka Code Set) - Is a set of known codes as determined by a governing body (ex: SNOMED, LOINC, NDC)
- Concept Loader - This is a feature in Mirth Results that allows administrators to load a code system into Mirth Results. Some code systems come out of the box with MR. MR currently has data and meta-data files for: HL7, ICD-9, LOINC, NDC.
- Coded Elements - Coded Elements are a mapping between a code, label, and/or description to a known Concept. They are unique per facility. A coded element's "scope" is defined by its Code System. Coded Elements in MR are ALWAYS the code we got from the source, coded elements may then be linked to a concept. See Coded Element Usages.
- Concept Mapper - The Concept Mapper is a feature within MR that tries to map unknown codes to known Concepts. You can search by usage and facility to see unmapped codes. This is used to automatically match incoming codes to a known set of codes, like LOINC.
- Concept Alias - Concept Aliases are setup on a per Source basis and are used to map data/codes from incoming data to existing/known Concepts. This is often used when a system is sending custom or non-standard codes which need to be linked to standard codes. When data sent to MR matches a Concept Alias, a new Coded Element will be created and cached. Once cached, the coded element will no longer be mapped to a concept alias.
Configuration
OR
From the UI, go to Administration --> System Settings and search for com.mirth.results.autoConceptLinking
If the System Setting for "com.mirth.results.autoConceptLinking" is enabled (i.e. Value equals "1"), when Mirth Results receives a coded element, such as a SNOMED code in a medication message or a LOINC code from a lab. It will attempt to match that code to a known concept and normalize the data.
MR comes with several concept classes pre-loaded. If you need to load your own or a newer version see Concept Loader
Relevant Application Screens
- Coded Elements - Located under the Setup Menu in the MR UI.
- Concept Classes - Located under the Setup Menu in the MR UI.
- Concepts - Located under the Setup Menu in the MR UI.
- Concept Mapper - Located under the Setup Menu in the MR UI.
- Concept Alias - Located under Setup -> Sources -> (click name of source to go to Edit Source page) -> then look at the bottom of the screen.
- Concept Loader - Located under the Setup Menu in the MR UI. Part of the Concepts feature.
What happens at runtime? What does this feature really do?
There are two major scenarios where code normalization applies. The first is where a source is sending codes that are known to not be normalized and there is an explicit mapping from those codes to concepts. The second scenario is where MR is receiving generally good data and does some simple matching to link those codes to known concepts.
Examples of each scenario:
1) Patient Gender is being sent to MR incorrectly. Example: (see PID.8.1 - Sex)
This should be "Male" or "M", but the sending facilicty (Source) is sending incorrect data. We can use Concept Aliases to map this data to correct data.
- From the MR UI, Under Setup -> Sources -> (your source, i.e. "LAB1") -> Click "Add Concept Alias".
- Alias is "blah", Coded Element Usage is 1 - Subject Gender.
- Concept Class is "HL7 v2.6"
- Search for concept "Male" and select the proper Concept from the search results.
- Click Submit and you should now have a Concept Alias created (at the bottom of your Edit Source page)
The next time a message comes to MR with the Patient Gender "blah", it will be linked to "Male". The Patient Gender may still show up in the UI as "blah". Under the hood when code data comes to MR, and matches the Concept Alias, a new Coded Element will be created and will be linked to the correct Concept based on the Concept Alias.
The Patient Gender will continue to show as "blah" in the MR UI, but when you download the CCD of the patient, you will see:
Note the translation code.
2) Map an incomplete code to a known and complete code (ex: map an incomplete code to a complete LOINC code)
OBX|2|NM|14682-9^Creatine^LOINC| <-- complete code, label, and code system
OBX|2|NM|^Creatine^LOINC| <-- missing code
Concept Mapping can be used to determine what the missing code should be. MR will try the following to match that coded element to a concept.
1. MR will determine the concept if there is a defined Concept Alias for that source that matches (See scenario 1).
2. If the concept class/code system is given, MR will look up the concept by code system. In this case, LOINC. MR will attempt this by OID (as seen in CCD) or by label (like the string “LOINC”)
3. MR will attempt to find the code in the default concept class for the usage.
4. Finally, MR will try to match on label and code system (if given). This step is really a very last ditch effort.
Limitations of Concept Mapping
MR does not map known and good codes to other known and good codes of another system. (ex: Mapping a SNOMED code to an NDC code).
MR is very dependent on having the correct configuration and aliases. Be sure to test and review each code that you're concerned about.
MR is making a best effort to identify the correct code, but it is not a full featured code normalization system. Review your needs with Mirth and decide if this is adequate for your needs.