Stream: implementers
Topic: Negation in resources
Shlomy Reinstein (Feb 07 2017 at 11:55):
I work in on a FHIR application in the Oncology domain. For many pieces of clinical data (conditions, procedures etc), there may be 3 options:
1. The patient has/had this <condition/procedure/...>.
2. The patient does not/did not have this <condition/procedure/...>
3. We do not know whether or not the patient has this <condition/procedure/...>.
The difference between (2) and (3) is important to the clinicians. Thus, we need to model (in FHIR) the negation of the data item (e.g. the patient does not suffer from diabetes), so that we can tell the difference between "unknown" and "no".
Is there a standard way to do this in FHIR?
This applies to comorbidities, procedures (incl. surgeries, chemotherapies and radiation therapies) and maybe some other types of data.
Thanks!
Grahame Grieve (Feb 07 2017 at 12:00):
http://build.fhir.org/allergyintolerance.html#9.1.3.3
Grahame Grieve (Feb 07 2017 at 12:00):
http://build.fhir.org/condition.html#9.3.3.4
Grahame Grieve (Feb 07 2017 at 12:00):
http://build.fhir.org/observation.html#10.1.4.5
Shlomy Reinstein (Feb 07 2017 at 12:02):
Thanks! But I forgot to say we're using DSTU2. Are those new in STU3?
Vadim Peretokin (Feb 07 2017 at 12:04):
Having skimmed over that - List.emptyReason seems to exist in DSTU2 too: http://hl7.org/fhir/list-definitions.html#List.emptyReason
Grahame Grieve (Feb 07 2017 at 12:08):
yes, you'll have to back port those approaches to DSTU2 with extensions - we put a lot of effort into this stuff during the STU3 preparation process
Shlomy Reinstein (Feb 07 2017 at 13:31):
What about Procedure? How do we say that a patient never had any radiation therapies (vs. "we don't know whether or not the patient had any radiation therapies")?
Michelle (Moseman) Miller (Feb 07 2017 at 14:24):
Seems more like QuestionnaireResponse or Observation to me. Same rationale as http://build.fhir.org/condition.html#9.3.3.5 says: It is common as part of checklists prior to admission, surgery, enrollment in trials, etc. to ask questions such as "are you pregnant", "do you have a history of hypertension", etc. This information should NOT be captured using the Condition resource but should instead be captured using QuestionnaireResponse or Observation. In this case, the combination of the question and answer would convey that a particular condition was not present
That said, we do have a Procedure.notDone element (formally known as Procedure.notPerformed in DSTU2). However, its use is intended for a slightly different use case, such as 1) a procedure was scheduled, but not done as originally planned, or 2) when there is an order NOT to do something, so documentation follows that the procedure was not done as ordered
Lloyd McKenzie (Feb 08 2017 at 04:42):
Also, Procedure probably won't be the resource for representing radiation therapy - there's a bunch of unique information for radiation that likely makes it appropriate to have its own resource. (We're looking for implementers who want to push that forward as we don't have a lot of expertise "in the room" to do the design.)
Shlomy Reinstein (Feb 08 2017 at 09:49):
You're probably right, but for now we're "stuck" with DSTU2 and we're using Procedure to model some of the data about the radiation therapy. We also don't have much expertise in that, we're learning as we go.
Regardless of radiation therapy, there are a few things we model as Procedure, and for them, we also need to be able to say whether the patient had this procedure, did not have this procedure, or if we don't know the answer.
Lloyd McKenzie (Feb 08 2017 at 14:29):
Do you actually want a full procedure record? Typically questions like "Has the patient had an appendectomy" would be captured using Observation. Procedure would only be used if you have details - like a date, location, etc.
Last updated: Apr 12 2022 at 19:14 UTC
