Units, Reference Ranges, and Abnormal Flags in OBX

Why a bare number is dangerous

A previous lesson covered the shape of the OBX segment: the value type in OBX-2, the observation identifier in OBX-3, the value itself in OBX-5, and the result status in OBX-11. Those fields tell you what was measured and what number came back. They do not, on their own, tell you whether that number is good news or a medical emergency.

Consider a result of 13. Is it high, low, or unremarkable? You cannot say, because a bare number carries no meaning. Thirteen grams per deciliter of hemoglobin is a healthy adult value; thirteen millimoles per litre of potassium would be incompatible with life. The number only becomes a clinical fact once you know its units and the reference interval it should be judged against, and once you know how confident the lab is in it. Three more OBX fields supply exactly this context: OBX-6 (units), OBX-7 (reference range), and OBX-8 (abnormal flags) ¹.

OBX-6: units, ideally UCUM

OBX-6 carries the units of measure for the value in OBX-5. It is a coded field, so units should be expressed as codes from a defined system rather than as free text. The strongly preferred coding system is UCUM (the Unified Code for Units of Measure), which gives every unit one unambiguous, machine-comparable representation — mmol/L, g/dL, 10*9/L — instead of the dozens of local spellings that “mg per cent” or “K/uL” tend to accumulate ².

Coded units matter because the receiving system may need to convert, trend, or compare values across sources. It can only do that safely if it can read the units mechanically. Free-text or missing units force humans back into the loop and invite silent unit mismatches — one of the most dangerous classes of lab error.

OBX-7: the reference range

OBX-7 holds the reference range the result should be interpreted against, typically as a low–high interval such as 3.5-5.1. It can also express one-sided ranges (<200), sex- or age-specific intervals, or descriptive ranges for non-numeric tests. Crucially, the range is supplied by the performing lab, for its method and population. It is part of the result, not a universal constant, because different analyzers and calibrations produce different normal intervals ¹.

OBX-8: the abnormal flag is the lab’s interpretation

OBX-8 is the lab’s own verdict on the value: an interpretation code that summarizes whether the result is normal, abnormal, or critical. This is not a field a receiver computes for itself — it is the source’s judgment, and downstream systems lean on it to drive alerts, highlighting, and clinical decision support. A critically high flag is frequently what fires a “call the provider now” notification.

A worked example

OBX|1|NM|2823-3^Potassium^LN||5.9|mmol/L|3.5-5.1|H|||F

Reading the fields in order: OBX-2 is NM (numeric); OBX-3 identifies the test as serum potassium (LOINC 2823-3); OBX-5 is the value 5.9; OBX-6 gives the units mmol/L; OBX-7 states the reference range 3.5-5.1; OBX-8 carries the flag H. Only with all of these present can a reader conclude that 5.9 is above the normal ceiling of 5.1 and therefore high. Strip out the units or the range and the same 5.9 becomes uninterpretable.

The final field, OBX-11, is F — final. This status governs how much weight to place on the result. A P (preliminary) value may still change and should be treated cautiously; an F value is verified and releasable; a C (corrected) value supersedes an earlier result that was wrong, and systems must overwrite, not append, when they receive one. The units, range, and flag are only as trustworthy as the status that accompanies them ¹.

Pitfalls

• Missing or wrong units. A value with no OBX-6, or with the wrong unit code, can be misread by orders of magnitude. Always treat unit-less numbers as suspect.
• Mismatched reference ranges across labs. Because each lab sets its own intervals, the same value can be flagged high by one lab and normal by another. Never carry a range from one source onto another lab’s result.
• Flags computed against a different population. An abnormal flag is only meaningful for the population and method that produced it. Trusting an inherited or recomputed flag — especially across age, sex, or pediatric versus adult reference sets — can trigger false alerts or, worse, suppress real ones ².

Together, OBX-6, OBX-7, and OBX-8 turn a raw measurement into an interpretable clinical observation, and OBX-11 tells you how far to trust it. A robust results interface treats all four as essential, not optional.