OBX: Observation result

A blood test comes back as a sheet with one line per measurement: what was measured, the number, the units, the normal range, and a flag if it is out of range. OBX is one of those lines. A panel of ten results is ten OBX segments.

OBX carries exactly one observation: what was measured, what the answer was, what units it is in, and how confident the sender is that the answer is final.

The shape of one line

Where OBX appears

OBX appears in 68 of the 202 message structures the standard defines.

A whole message

MSH|^~\&|LABSYS|STJOSEPH|EPIC|STJOSEPH|20240314141207||ORU^R01^ORU_R01|LR20240314-88412|P|2.5.1
PID|1||1000345678^^^STJOSEPH^MR||CARTWRIGHT^ELEANOR^MAY^^^^L||19510822|F
OBR|1|ORD8842^EPIC|LAB77213^LABSYS|58410-2^CBC PANEL^LN|||20240314103000|||||||||1447312^OKONKWO^ADAEZE^^^^MD||||||20240314141000|||F
OBX|1|NM|718-7^HEMOGLOBIN^LN||11.8|g/dL|12.0-15.5|L|||F|||20240314141000
OBX|2|NM|6690-2^LEUKOCYTES^LN||14.2|g/dL|4.0-11.0|H|||F|||20240314141000
OBX|3|ST|33914-3^INTERPRETATION^LN||Mild anaemia with leukocytosis|g/dL|||||F|||20240314141000
NTE|1|L|Specimen slightly haemolysed. Result confirmed on repeat.

Every field

Open this OBX message in HL7 Studio

Field notes

• OBX-1Set ID – OBX

  Set ID, numbering the OBX segments within their parent OBR. It restarts at 1 for each OBR, so it is unique within a result group and not within the message. Using it as a key across the whole message collides as soon as a message carries two orders.

  Open OBX-1 in HL7 Studio

• OBX-2Value Type

  Value type, and the field that tells you how to read OBX-5. NM is numeric, ST a short string, TX text, CE or CWE a coded value, DT a date, TS a timestamp, SN a structured numeric, ED encapsulated data, RP a reference pointer. Parse OBX-5 according to this field and nothing else. Code that assumes numeric because the field usually is numeric will break on the day a result comes back as `<0.01`, which arrives as an SN and not an NM.

  Open OBX-2 in HL7 Studio

• OBX-3Observation Identifier

  Observation identifier, the CE or CWE saying what was measured. LOINC in components 1 to 3 and a local code in components 4 to 6 is the pattern to expect, and both matter: LOINC for interoperability and the local code for reconciling against the sending system. A sender that populates only the local triplet is common and forces a mapping table at the receiver.

  Open OBX-3 in HL7 Studio

• OBX-4Observation Sub-ID

  Observation sub-ID, which groups OBX segments that belong together under one identifier. Its main use is the multi-part result: a susceptibility panel where each organism is a sub-group, or an imaging report split across several segments. Ignoring it merges results that were meant to stay separate.

  Open OBX-4 in HL7 Studio

• OBX-5Observation Value

  The observation value, and it repeats. For a TX result, one OBX with several repetitions is how a long narrative arrives, and each repetition is a line. Joining them without a separator runs the report together into one paragraph. For an ED value, the content is base64 in component 5 and the MIME type in components 2 and 3.

  Open OBX-5 in HL7 Studio

• OBX-6Units

  Units, ideally in UCUM. This field is not optional in any real sense: a haemoglobin of 11.8 means different things in g/dL and g/L, and a receiver that discards units and stores the number has stored a number that cannot be interpreted. Where the sender uses local unit codes, map them before storing.

  Open OBX-6 in HL7 Studio

• OBX-7References Range

  References range, as free text and not as structured bounds. `12.0-15.5` is the common form but `<5.0`, `>=60` and `Negative` all occur, and the field is a string in the standard precisely because ranges are not always numeric. Parsing it into a low and high pair works for most rows and has to fail gracefully for the rest.

  Open OBX-7 in HL7 Studio

• OBX-8Abnormal Flags

  Abnormal flags, bound to table 0078, and repeating. L and H are low and high, LL and HH are critically low and high, A is abnormal, N is normal, and S, R and I carry susceptibility meanings. The critical values are the ones that drive alerting, so treating LL as merely a variant of L removes the distinction that matters clinically.

  Open OBX-8 in HL7 Studio

• OBX-11Observation Result Status

  Observation result status, and the field that governs whether the result can be acted on. P is preliminary, F is final, C is a correction to a previously final result, X means the result cannot be obtained, D means delete the result. A C is not a new result. It replaces one already sent, and a receiver that appends it will show a patient two conflicting values.

  Open OBX-11 in HL7 Studio

• OBX-14Date/Time of the Observation

  Date and time of the observation, which is when the specimen was collected or the measurement taken. This is the clinically meaningful time. OBR-22 is when the result was reported and MSH-7 is when the message was built, and the three can differ by days for a culture.

  Open OBX-14 in HL7 Studio

Vendor behaviour

What follows describes patterns common to these systems' standard interfaces. Every installation is configured, so confirm against the interface specification for the site you are integrating with.

Mapping to FHIR

One OBX becomes one FHIR Observation, and OBX-2 decides which Observation.value[x] to use: NM becomes valueQuantity, ST and TX become valueString, CE and CWE become valueCodeableConcept, SN becomes valueQuantity with a comparator. The parent OBR becomes either a DiagnosticReport that references these Observations or a member Observation with hasMember entries, and which of the two is right depends on whether the OBR represents a report or a panel.

Parsing pitfalls

• OBX-5 parsed without reading OBX-2

  The value type field exists because the value field has no fixed shape. Assuming numeric, or assuming a single line of text, produces a parser that works on the sample data and fails on the first structured numeric or the first encapsulated document.

• A corrected result appended instead of replacing

  OBX-11 set to C means this value supersedes one already sent for the same observation. Storing it as an additional result leaves the superseded value visible, and a clinician looking at the chart sees two answers with no indication which is current.

• Multi-line text joined without separators

  A TX result arrives as repetitions of OBX-5 or as several OBX segments, and each represents a line. Concatenating without a line break turns a structured report into a wall of text, and concatenating with the wrong separator can join a heading to the sentence below it.

• Units discarded

  A number without its unit is not a result. Where the receiving system has a fixed unit per test, convert; discarding the unit is not an option. If the sender uses local unit codes the conversion needs a mapping table, not string matching.

• OBX-4 ignored on panelled results

  Susceptibility results, differential counts and multi-organism cultures use the observation sub-ID to keep groups apart. Flattening them attributes an antibiotic sensitivity to the wrong organism.

Related

• OBRthe order or report this observation belongs to
• NTEnotes and comments attached to an observation
• ORCthe common order control information above OBR
• SPMthe specimen the observation was made on

Message types carrying OBX

OBX appears in 68 message structures, belonging to 44 message types.

• ADR
• ADT12
• BAR2
• CSU
• DFT2
• DOC
• MDM
• OMB
• OMD
• OMG
• OMI
• OML3
• OMN
• OMP
• OMS
• ORF
• ORL2
• ORM
• ORU2
• OUL4
• PEX
• PGL
• PPG
• PPP
• PPR
• PPT
• PPV
• PRR
• PTR
• RAS
• RCI
• RDE
• RDS
• REF
• RGV
• RPA
• RQA
• RRI
• RSP5
• SIU
• SRM
• SSU
• VXR
• VXU