Stream: implementers
Topic: link observation to condition
Jose Costa Teixeira (Sep 13 2021 at 13:31):
We need to assert "this observation is (probably) due to that underlying Condition". e.g. a breathing difficulty due to COPD or Asthma
Is there a preferred way? I see these extensions, but they don't cover this only http://build.fhir.org/extension-condition-related.html http://build.fhir.org/extension-condition-dueto.html
Daniel Venton (Sep 13 2021 at 13:37):
Wouldn't every observation be influenced by every known condition, every unknown condition and every absence of conditions in some way?
Jose Costa Teixeira (Sep 13 2021 at 14:16):
probably. What we we want to support is the explicit assertion between one Obs and one condition
John Moehrke (Sep 13 2021 at 14:31):
wow, I fully expected to find that in Observation. I guess you would need to do the link thru an encounter?
Eric Haas (Sep 13 2021 at 14:42):
Condition.evidence
John Moehrke (Sep 13 2021 at 14:44):
so the expectation is that a Condition will exist after some Observations... rather EHR centric.
John Moehrke (Sep 13 2021 at 14:44):
then the Condition would need to continue to be updated each observation? Each blood-sugar measurment will also need to update the Condition?
Lloyd McKenzie (Sep 13 2021 at 16:52):
I think there's a difference between "This observation supports the assertion of the Condition" and "I believe the cause of this Observation is this known Condition".
Lloyd McKenzie (Sep 13 2021 at 16:52):
That said, the difference is subtle, and the likelihood of people messing it up is non-negligible.
Jose Costa Teixeira (Sep 13 2021 at 17:36):
I mean the latter
Jose Costa Teixeira (Sep 13 2021 at 17:36):
And in our meeting today we encountered both - sometimes the condition exists and you want to point at it, sometimes the condition will later be created based on the observation
John Moehrke (Sep 13 2021 at 19:07):
agree, different meanings might be needed. but also agree that the difference is not obvious. I think the meaning that exists today is a medical-decision statement. This condition has been declared because of this evidence.
John Moehrke (Sep 13 2021 at 19:16):
Is the other use-case better served by the indirection thru encounter or careplan?
The use-case I have that seems to fall somewhere inbetween all of these are
a) Patient asserts a Condition, and a set of their own Observations that they think justify that Condition.
b) Patient provides updated observations (self entered health data) in support for a Condition that is not yet formally declared by a clinician and/or part of an encounter or CarePlan.
Eric Haas (Sep 13 2021 at 19:39):
http://build.fhir.org/extension-workflow-reasonreference.html is your other option, but there is issue of cross referencing
Robert McClure (Sep 15 2021 at 23:44):
Wow, I'm kinda stunned that is there no simple way to link conditions and observations directly without having to create something else too. This most definitely should not require an encounter. And Eric's condition.evidence is not what was requested here.
Jose Costa Teixeira (Sep 16 2021 at 15:21):
@Robert McClure you are stunned because we can't
a) link an observation to an existing condition
or
b) link a condition to an previously entered observation?
Jose Costa Teixeira (Sep 16 2021 at 15:22):
(just to figure out which one is more urgent. I'm looking for the former)
Jay Lyle (Sep 16 2021 at 15:39):
I think (b) is covered by Condition.evidence. The assertion of cause (rather than evidence) seems to belong on the Observation. So (a) looks like a gap. It's touched on in the Concern DAM. reasonReference looks close but not right.
Eric Haas (Sep 16 2021 at 16:58):
Jay Lyle said:
I think (b) is covered by Condition.evidence. The assertion of cause (rather than evidence) seems to belong on the Observation. So (a) looks like a gap. It's touched on in the Concern DAM. reasonReference looks close but not right.
I agree that (b) is covered, why do you think (a) is not? can you explain?
Jay Lyle (Sep 16 2021 at 17:03):
Because it's for the final cause, not the efficient cause. The reason (justification) something was done, not the reason (explanation) something happened.
"Indicates another resource whose existence justifies this event."
Eric Haas (Sep 16 2021 at 17:06):
The reason (justification) something was done, not the reason (explanation) something happened.
This sounds like passive vs active case to me. how are these different in the real world?
Lloyd McKenzie (Sep 16 2021 at 17:51):
'reason' says "I did X because of Y". 'cause' says "I believe this happened because of this pre-existing factor". They are very different statements. One is a concrete assertion of justification for a behavior, the other is an assertion of a belief in a hypothesis as to causation.
Eric Haas (Sep 16 2021 at 18:50):
now we are in ClinicalImpression Land.
Eric Haas (Sep 16 2021 at 18:51):
but nevermind create a new extension see if it sticks and call it a day.
Jose Costa Teixeira (Sep 16 2021 at 19:34):
Eric Haas said:
but nevermind create a new extension see if it sticks and call it a day.
If I would do just that, I woudn't get the 2 bitcoins that HL7 give for every FHIR JIRA tracker submitted
Mark Kramer (Sep 19 2021 at 01:40):
Observation.focus
Lloyd McKenzie (Sep 19 2021 at 02:10):
Focus is "what" you're observing, not "why you believe the thing observed is happening"
Hank Lenzi (Dec 21 2021 at 16:15):
Jose Costa Teixeira said:
We need to assert "this observation is (probably) due to that underlying Condition". e.g. a breathing difficulty due to COPD or Asthma
Is there a preferred way? I see these extensions, but they don't cover this only http://build.fhir.org/extension-condition-related.html http://build.fhir.org/extension-condition-dueto.html
Wouldn't it be better to link the condition directly, via a code system binding (LOINC, SNOMED, ICD), restricting it to, e.g., respiratory illness?
Now, if you're talking some kind of inference, I suspect Turtle might be what you're looking for (but I am not sure, this is just a hunch).
Last updated: Apr 12 2022 at 19:14 UTC
