0

Devices sometimes send NaN (Not a Number) or +/-Inf (plus/minus) infinity when these special cases arise. I do not see a means to encode them into the FHIR Observation resource reliability or to place them as codes into the value element itself. There is no way to handle this in HAPI FHIR either.

Does anyone know how to report such special values in an Observation FHIR Resource?

Brian Reinhold
  • 2,313
  • 3
  • 27
  • 46
  • Does the difference between NaN, +infinity, -infinity matter? – Grahame Grieve Jul 06 '15 at 13:40
  • Yes. NaNs typically mean the measurement sensor failed in some manner (often in periodic streams), +/- Inf means, in the case of Glucometers, that the concentration was above/below the sensor's capability. IEEE 20601 also defines a reserved value and a not-at-this-resolution value. All of these are special values since they are encoded as MDER FLOAT/SFLOATs with 0 exponent and special mantissas. – Brian Reinhold Jul 06 '15 at 17:23
  • there's rather an obvious distinction between >(measure limit) and infinity. That's why we don't have a value for infinity. There are appropriate codes for this in SampledData, but not for an ObservationDirectly. See http://gforge.hl7.org/gf/project/fhir/tracker/?action=TrackerItemEdit&tracker_item_id=8389 – Grahame Grieve Jul 06 '15 at 20:53

1 Answers1

0

These concepts would need to be conveyed using CodeableConcept rather than Quantity. There are codes for PINF and NINF in the v3 null flavor code system (http://hl7-fhir.github.io/v3/NullFlavor/index.html). Though if what you really mean is "below detectable limits" and "above detectable limits", it would likely be better to have codes for that. I don't believe a code exists in HL7-maintained code systems for NaN. Would suggest raising change requests for Observation to recommend common codes for all of these concepts.

Lloyd McKenzie
  • 6,345
  • 1
  • 13
  • 10