NR 512 Week 3 Terminology Comparison for One Finding Example

Terminology Comparison: One Finding, a Saturation of 88% on Room Air That Rose to 94% on 2 Liters, Written in LOINC, SNOMED CT, ICD-10-CM and a Nursing Diagnosis, and What Each One Keeps and Drops

Student Name

College of Nursing, Chamberlain University

NR 512: Fundamentals of Nursing Informatics

Instructor Name

Month Day, Year

Terminology Comparison: One Finding, a Saturation of 88% on Room Air That Rose to 94% on 2 Liters, Written in LOINC, SNOMED CT, ICD-10-CM and a Nursing Diagnosis, and What Each One Keeps and Drops

The Finding

The finding comes from an admission on a medical unit. A 71-year-old man with community-acquired pneumonia had a pulse oximetry reading of 88% on room air at 1510. Oxygen was started at 2 L/min by nasal cannula, and at 1530 his reading was 94%. A nurse, a physician, a coder and a quality analyst will each need to say something about this, and each will reach for a different vocabulary. This paper compares four of them against that one finding, not against pneumonia in general and not against the patient's whole stay.

The Four Vocabularies Side by Side

Table 1 sets the four vocabularies against the same finding. The prose after it does what the table cannot.

Table 1

Four Vocabularies Applied to One Finding

Note. The two LOINC codes shown are the ones the US Core profile for oximetry specifies (HL7 International, n.d.). The finding is a composite built for teaching.

Granularity: Who Keeps the Numbers

Only LOINC keeps the numbers. It does not say whether 88% is bad; it names the measurement so that 88 and 94, with their units and times, can be sent between systems and understood as the same kind of observation. McDonald et al. (2003) describe LOINC as a set of universal names and codes for observations, created because each laboratory and system had its own local names. The HL7 pulse oximetry profile then pairs the saturation code with the flow code, which is how the rise from 88 to 94 on oxygen stays interpretable after it leaves the unit (HL7 International, n.d.).

SNOMED CT keeps the meaning but not the numbers. A hypoxemia concept says what the clinician concluded, and it can be linked to other concepts to say more. Lee et al. (2014) reviewed 488 papers on SNOMED CT published from 2001 to 2012 and found most were theoretical or about design, with only 34 describing implementation in the later period; they concluded that most real implementations are not reported in the scientific literature. That is relevant here: the concept is precise, but a system still needs the LOINC observation to know how low the saturation was.

ICD-10-CM keeps a category. R09.02 tells a payer and a statistician that hypoxemia was part of the encounter. It cannot hold 88% or 2 L/min, and it was never meant to. The choice between R09.02 and J96.01 depends on what the physician documents, not on the numbers, which is why the same oximetry values can end up in two different categories.

The nursing diagnosis keeps the nurse's judgment and the evidence for it. Impaired gas exchange, with defining characteristics such as low saturation, says what the nurse will act on and evaluate. It keeps something none of the others do, the link to a care plan, but most hospital systems cannot exchange it with anyone outside the building.

Precoordination and Postcoordination

ICD-10-CM is precoordinated: a single code carries a fixed combination of meaning, and anything the code does not include is simply absent. J96.01 carries acute, respiratory failure and hypoxia together; it cannot be split to say chronic with an acute worsening unless a separate code exists for that combination. SNOMED CT allows postcoordination, in which a base concept is combined with others to express a meaning no single concept holds. That power costs something: two systems can express the same meaning in different combinations, and matching them later requires rules. LOINC sits apart from both, because it names observations rather than conclusions, and the value is stored beside the code instead of inside it.

What Happens When the Record Crosses a Boundary

Suppose the unit's system stores the finding in SNOMED CT and the hospital's billing system needs ICD-10-CM. Maps exist, and the National Library of Medicine's Unified Medical Language System integrated, in its 2004 description, some 900,000 concepts from more than 60 families of vocabularies, so that a term in one can be found in another (Bodenreider, 2004). But a map moves a concept, not the evidence behind it. Mapping the hypoxemia concept to R09.02 is straightforward. Nothing in the map carries the 88%, the 2 L/min or the fact that the patient improved on oxygen. If the quality analyst later wants to count patients who were hypoxemic on room air, the ICD-10-CM data cannot answer, and the SNOMED CT data can answer only if someone recorded the context. The LOINC observations can answer it directly, provided the flow value was recorded with each reading.

Which Vocabulary for Which Use

None of the four is better; each is better at its job. For deciding whether this patient is deteriorating, the nurse needs the LOINC-coded observations, since only they carry the trend. For summarizing the clinical picture in the problem list, SNOMED CT's hypoxemia concept is right. For billing and national statistics, ICD-10-CM is required and sufficient. For the nursing care plan, the nursing diagnosis carries what the others cannot: the nurse's reasoning and the interventions it leads to.

The practical lesson for a nurse is that the vocabulary is chosen by the question that will be asked of the data later. A nurse who records a saturation with its oxygen flow has made the finding answerable by all four. A nurse who writes only "hypoxic, placed on O2" in a free-text note has made it answerable by none, since none of the four can read the note until someone codes it by hand.

The comparison also shows why the terminology question cannot be left to coders at the end of a stay. By the time a coder sees the chart, the moment when the saturation was 88% on room air has already been recorded in one form or another. If it was recorded only as a sentence, the coder can choose R09.02 but cannot recover the numbers, and the quality analyst can count the code but cannot tell how severe the hypoxemia was or whether it responded to oxygen. The decision that matters most for all four vocabularies is made at the bedside, in the seconds it takes to key a value and its flow.

References

Bodenreider, O. (2004). The Unified Medical Language System (UMLS): Integrating biomedical terminology. Nucleic Acids Research, 32(Suppl. 1), D267-D270. https://doi.org/10.1093/nar/gkh061

HL7 International. (n.d.). US Core pulse oximetry profile (US Core Implementation Guide, Version 9.0.0, STU 9). Retrieved September 30, 2026, from https://hl7.org/fhir/us/core/StructureDefinition-us-core-pulse-oximetry.html

Lee, D., de Keizer, N., Lau, F., & Cornet, R. (2014). Literature review of SNOMED CT use. Journal of the American Medical Informatics Association, 21(e1), e11-e19. https://doi.org/10.1136/amiajnl-2013-001636

McDonald, C. J., Huff, S. M., Suico, J. G., Hill, G., Leavelle, D., Aller, R., Forrey, A., Mercer, K., DeMoor, G., Hook, J., Williams, W., Case, J., & Maloney, P. (2003). LOINC, a universal standard for identifying laboratory observations: A 5-year update. Clinical Chemistry, 49(4), 624-633. https://doi.org/10.1373/49.4.624