LOINC and SNOMED CT on the public health test card

A public health result that leaves the laboratory as free text is a result the next system has to guess. Electronic laboratory reporting, case notification, and a receiving epidemiologist all need the test and the value to mean the same thing in every system that touches them. LOINC is how this site names the observation. SNOMED CT is how it names coded results and organisms. The hero of the marketing site lists both vocabularies because they are part of the product, not a future integration slide.

The pathogen and test library is the public view of that choice. Cards span respiratory, arboviral, enteric, bloodborne, mycobacteria, biothreat, antimicrobial resistance, pathogen-agnostic, fungal, water, and vaccine-preventable domains, among the domains the library defines. A card shows the method, the specimen, the destination laboratory, reportability notes, ELR notes, and the analytes. Each analyte has a LOINC. Each coded pick has a SNOMED CT term. The library page says the card carries the code, the laboratory, and the reason it matters. Narrative comes second.

This article does not list a number of codes, a percent of the national reportable menu, or a count of states live on the map. The site does not publish those figures. Coverage is whatever is on the card you are reading and whatever your laboratory has built or installed. If a pathogen you test is not in the public catalog, the absence is not a claim that the assay is out of scope for the LIMS. It means the public card is not there yet, and the Test Builder is how a laboratory defines the test itself.

In the laboratory management module, a laboratory creates a test with its acceptable result set and assigns LOINC to the observation and SNOMED CT to coded results and organisms. That assignment is the default for the test. The same definition can be scoped to one laboratory or made available across the deployment. When a specimen is split into picks, each pick keeps its place on the parent accession, so the code travels with the work rather than with a loose row in a send-out file.

The DMI page states the wire shape. Observation identifiers are LOINC on OBX-3. Qualitative and organism values are SNOMED CT on OBX-5. Panel LOINC stays in OBR-4. The catalog ships those codes with the test so the map is configuration, not a professional-services project that starts from a blank spreadsheet every time a panel changes. Conrad can validate coding before a human transmits. He does not send the message. A clean map reduces rejects. It does not let the site advertise a zero rejection rate. Rejection is a destination decision. The DMI page says that in those words.

Community registry bundles use the same fields: domain, pathogen, method, LOINC order, LOINC observation, coded results, SNOMED, ELR notes, reportable flag, owning laboratory, and last verified. A laboratory that installs a bundle should land it in the department that runs the method, which the library describes as Virology, Molecular, Enterics, LRN, or Water rather than a junk drawer. Install is an adoption of a version. It is not an automatic clinical go-live. A person still completes validation.

Comparable codes are what let a state see influenza A, a carbapenemase, or a select-agent rule-out as the same kind of event across submitters. They are also what a genomics result needs if the sequence is going to stay on the specimen instead of on a side laptop. The genomics pages say organism calls get SNOMED when the organism is identified, and that pipeline version is part of the method. LOINC and SNOMED do not replace the quality metrics of a sequencing run. Depth, Q30, and the pipeline identity still have to be on the record a supervisor can read.

Codes are not an interpretation. A detected or not-detected pick can be coded and still be the wrong clinical story if the specimen was expired, the wrong matrix, or held past its limit. Holding time, referral verification, and the director's sign-off sit beside the code. The code makes the message computable. The quality system makes the message trustworthy.

Codes are not a diagnosis. Nothing on the card authorizes the software to tell a patient what disease they have. The platform's own boundary list says Salus does not issue a patient-level diagnosis or interpretive result. A coded organism on an isolate is a laboratory observation. What a clinician or an epidemiologist does with it is outside the software's authority.

Open the card. Read the method, the specimen types, the quality-control note, the reflex, and the reportability line. Check the reviewed date. Follow the sources the card cites. Then compare the LOINC and the SNOMED picks to the package insert or the laboratory-developed design record you will actually run. The workbench outline repeats those codes into a draft validation skeleton so the packet and the message start from the same terms. The outline is a draft. The medical director signs the validation summary. The quality manager files it and assigns an SOP version.

If a trading partner requires a different code, put the override in the interface map and keep a record of who approved it. Do not 'fix' a reject by editing the outbound value in a text editor the LIMS cannot see. The audit trail the integrity page expects includes accession, result, verify, and transmit. A silent edit is how a surveyor loses the plot, and how the next person on call sends the old code again.

Vocabulary changes. LOINC and SNOMED CT publish new terms. APHL toolkits and notification lists change. The living catalog and the regulatory change log are how this site expects a laboratory to notice. They are not a promise that every code on every card was refreshed this morning. Read the reviewed date. If you need a term the card does not carry, build it in the Test Builder and say who checked it.