Microbiology molecular testing is the term family where post-coordination has been worked through in the most detail, and it illustrates both the problem and the approach.
The pre-coordinated pattern
Today, LOINC contains many separate terms for what is essentially the same microbiology genetic test, because the specimen type and the testing technique are built directly into the term. Detection of Neisseria gonorrhoeae DNA, for example, requires a different LOINC term depending on whether the specimen is urine, a genital specimen, a throat specimen, or another source — and again depending on the molecular technique used. The result is a large family of highly specific terms that are difficult to create, manage, and maintain over time, and that grows with every new specimen type or technique.
The post-coordinated pattern
Instead of a separate term for every combination of organism, specimen, and technique, a single term represents the organism target being tested. The specimen type and the technique are reported separately, as structured data elements, at the time of result reporting.
Terms in this family use curly-brace placeholders in the two axes that are supplied at the point of use:
| Axis | Value |
|---|---|
| System | {Specimen} |
| Method | {Molgen} |
A DNA/RNA detection term matching Property = PrThr, Time = Pt, Scale = Ord in Class MICRO therefore reads:
[Pathogen] DNA:PrThr:Pt:{Specimen}:Ord:{Molgen}The valid substitutes for each placeholder are not left to local interpretation: {Specimen} and {Molgen} each resolve through a defining LOINC and its Answer List, as described in Section 14.2. For {Molgen}, one such valid resolution is the Method Part LP6464-4 (Probe.amp.tar).
This model reduces the number of microbiology molecular terms that must be maintained, while improving flexibility: as new techniques and specimen types emerge, they can be represented without creating additional LOINC terms. It also aligns with the direction of modern FHIR genomics reporting, which emphasizes capturing these details in structured data rather than embedding them within the code itself, and with the way the corresponding attributes are represented in SNOMED CT.
For the naming convention itself, including when it applies to a submitted term, see Section 2.1.8.