LOINC — Naming Observations and Lab Tests

What LOINC names

LOINC — Logical Observation Identifiers Names and Codes — is the universal standard for identifying observations: the questions a clinical system asks and records an answer to. Its most familiar use is laboratory testing, but its scope is broader. LOINC also identifies vital signs, anthropometric and other clinical measurements, survey instruments, and even whole document types, so that a “discharge summary” or a “blood pressure panel” has a single agreed code wherever it travels ¹. Where one code system is needed for what was observed, regardless of who performed it or which instrument was used, that system is LOINC.

Each LOINC term has a numeric code and a Long Common Name. The code 2345-7, for example, denotes “Glucose [Mass/volume] in Serum or Plasma” ², and 718-7 denotes “Hemoglobin [Mass/volume] in Blood” ². The code is the stable identifier two systems exchange; the Long Common Name is the human-readable label rendered alongside it.

The six-part name model

The reason a LOINC code can be trusted to mean exactly one thing is that every term is defined along a fixed set of axes. This six-part name fully specifies an observation, so that two terms that differ on any axis are genuinely different observations ¹:

1. Component / Analyte — what is measured. For a glucose test the component is Glucose; for the hemoglobin term it is Hemoglobin.
2. Property — the kind of quantity being reported, such as a mass concentration (mass per volume), a substance concentration, a count, or a ratio. “How much glucose by mass per unit volume” is a different property from “how much glucose by moles per unit volume,” and LOINC distinguishes them.
3. Time Aspect — whether the result is a single point in time or is measured over an interval (for instance, a 24-hour collection).
4. System / Specimen — where the observation was made: the specimen or body system, such as Serum or Plasma, or Blood.
5. Scale — the type of value the result takes: Quantitative (a number), Ordinal (ranked categories), Nominal (named categories), or Narrative (free text).
6. Method — how it was measured. This axis is optional, and is left unspecified when the term is intended to apply regardless of method.

Mapping the two verified codes onto these axes makes the model concrete:

Axis            2345-7                         718-7
-----------------------------------------------------------------------
Component       Glucose                        Hemoglobin
Property        Mass concentration             Mass concentration
                (mass per volume)              (mass per volume)
Time Aspect     Point in time                  Point in time
System          Serum or Plasma                Blood
Scale           Quantitative                   Quantitative
Method          (unspecified)                  (unspecified)
-----------------------------------------------------------------------
Long Common Name:
  2345-7 = "Glucose [Mass/volume] in Serum or Plasma"
  718-7  = "Hemoglobin [Mass/volume] in Blood"

Notice that the bracketed fragment “[Mass/volume]” in each Long Common Name is simply the Property axis surfaced into the readable label. Read the name and the axes together and you can see how the formal model and the common name describe the same observation.

What LOINC does not carry

The single most important boundary to internalize is this: LOINC names the question, never the answer ¹. The code 2345-7 says “a glucose mass concentration in serum or plasma was measured.” It does not carry:

• the result value — the number 5.3 or 95 is supplied separately;
• the units — whether the glucose is reported in mg/dL or mmol/L is the job of a units standard such as UCUM, not of LOINC;
• the interpretation — flags like “high,” or a coded finding such as “abnormal result,” come from elsewhere, typically coded answers drawn from SNOMED CT.

This division of labor is deliberate. By keeping the identity of the observation in LOINC and pushing the value, units, and meaning of the answer into other standards, each system stays small and reusable. A glucose code does not have to enumerate every possible value, and a units vocabulary does not have to know what glucose is.

Where the code lands: OBX-3

In the HL7 v2 course, the OBX (Observation/Result) segment carried a single result, and its third field, OBX-3 (Observation Identifier), said which observation the segment was reporting. LOINC is exactly what populates OBX-3: the glucose code 2345-7 ² goes there to declare the question, tagged with its coding system so the receiver knows it is LOINC and not some local code. The answer then lands in the surrounding fields — the value in OBX-5, the units in OBX-6 — keeping LOINC’s “question, not answer” boundary intact even inside a single segment.

OBX | 1 | NM | 2345-7^Glucose [Mass/volume] in Serum or Plasma^LN | ... | 95 | mg/dL | ...
          ^^^^   ^-- OBX-3: LOINC identifies WHAT (the question)      ^-- value   ^-- units

That is the through-line of this module: a code system answers a precise question — what observation is this? — and answers it the same way everywhere. LOINC owns that question for laboratory tests and clinical observations, and its six-part name is what makes the answer unambiguous.