Stream: implementers
Topic: Observation.status lifecycle & guidance/change feedback
Travis Stenerson (Nov 09 2017 at 17:01):
First, the OO group's new guidance on code-value pairing is excellent. I don't know when that was added but it is very helpful to see that laid out, I've profiled 40 or 50 observations and I will likely add exactly which pattern each of them is to the SD. Thanks!
Regarding the change to Observation.related, I have a use case in which I made heavy use of the 'replaces' related-type, but I believe it may be easier to accomplish with observation.status. The explanation of it doesn't quite answer all my questions. The situation is a sequence of the same observation from a series of procedures meant to confirm N Category of a cancer. (X is any of 5/6 procedures). I would like to know what the status is of each of these four observations. It may be that they are all 'final', but I am hoping that the first three are either preliminary (or amended?), I would just like confirmation.
N1 at diagnosis (preliminary) -> X | N2 (status: preliminary) -> X | N3(status: preliminary) -> X | confirmed N3( status: final)
Is this correct? Or are those all technically final observations? I use the 'replaces' field to differentiate the next three, replaced by may also worked.
Eric Haas (Nov 09 2017 at 17:10):
that is typo should be 'replaces'
Travis Stenerson (Nov 09 2017 at 17:15):
By typo you mean 'replacedBy' is actually going to be 'replaces'?
Any guidance on status for this example? @Eric Haas or @Michelle (Moseman) Miller
Eric Haas (Nov 09 2017 at 17:18):
The statuses would be about a single observation's state over time. The 'replaces' would a way to trace within observation other observations that the current one over-rides. I am concerned that we are overloading observation if this is the primary way to monitor a progression of a lesion, etc. There may be some overarching resource that provides a better way to follow a case over time.? @Lloyd McKenzie ?
Travis Stenerson (Nov 09 2017 at 17:23):
I would love further explanation of those statuses with examples. I'm not quite sure I understand preliminary vs amended vs final. When does one set an observation's status to amended vs preliminary. Are all observations that require further confirmation set to preliminary?
Lloyd McKenzie (Nov 09 2017 at 17:28):
In FHIR, a single Observation instance (same 'id') may be updated many times over its lifespan as it evolves towards completion. You would never use "replaces" to link two versions of a single Observation instance. That linkage is implicit in that both are versions of the same instance (and have a timestamp that says when each version was created). The idea of "replaces" would be for something like "we had a baseline for Observations of type x, but that baseline is now replaced with this new Observation instance". So both Observations would typically have a status of completed and they'd have different "id" values (and different identifier values). You're establishing that one Observation result should now be used in preference for the other.
Lloyd McKenzie (Nov 09 2017 at 17:29):
Setting the status to "preliminary" for results that require confirmation would certainly be typical - but there's one thing we've learned is that "typical" does not mean "always"...
Travis Stenerson (Nov 09 2017 at 18:37):
Excellent, thank you. I'm going to have these observations set to preliminary and communicate that in the implementation guide.
Last updated: Apr 12 2022 at 19:14 UTC
