Stream: implementers
Topic: Interventional lab resource
Roy van Pelt (May 18 2016 at 15:50):
Hi all. I'm new to the list and have some questions right away. We're discussing which resource is most suitable to model the time a patient resides in the interventional (cath)lab, for instance for a PCI intervention. Thanks!
David Hay (May 18 2016 at 18:00):
sounds like an Encounter...
Roy van Pelt (May 20 2016 at 11:21):
Thanks David. That seems intuitive to me too. I'm a bit puzzled about the use of medication. If I would need to know all medication used during an encounter (say in the cathlab), how could I obtain that information? Would you follow the route of MedicationAdministrations for that encounter, or would you divide the period of time in the cathlab into procedures, which in turn reference the Medication resource?
Grahame Grieve (May 20 2016 at 11:27):
MedicationAdministration is the intent. But you'd have to check that all of them were actually recorded
Roy van Pelt (May 20 2016 at 11:44):
Great. Of course it is up to the user to check completeness. So where do you see the relation between Procedure and Medication stepping in? Is it the intend to divide an encounter into procedures, e.g., diagnostic and therapeutic?
Roy van Pelt (May 23 2016 at 06:13):
Can anyone comment on the intended use of Procedure in this context, also in relation with Medication?
Yunwei Wang (May 23 2016 at 21:01):
I think you can link Procedure and Medication using Encounter.
Roy van Pelt (May 24 2016 at 11:33):
Indeed there is a link between Procedure and Medication - that is useful. Just wondering if I can query the administered medication per procedure, assuming there can be multiple procedures per Encounter. The same holds for observations by the way. How can I know which Observations are performed during a certain Procedure?
Erich Schulz (May 24 2016 at 12:18):
timestamps tell all...
Erich Schulz (May 24 2016 at 12:19):
often with procedures there is anaesthetist giving medication independent of the main procedure...
Erich Schulz (May 24 2016 at 12:20):
and then nurse will step up and slip a suppository up the nethers...
Erich Schulz (May 24 2016 at 12:31):
but on reflection I can see a need to tie a medication administration directly to a procedure...
Roy van Pelt (May 24 2016 at 12:32):
We were at some point thinking in that direction as well. I see your point that some tasks, e.g., anaesthesia, can be performed independent of the main procedure. These should be related to the Encounter. However, other tasks (e.g., MedicationAdministration or Observation) will be part of the Procedure, and there is no way to distinguish from the other tasks purely on timestamps.
Erich Schulz (May 24 2016 at 12:32):
looking here for example https://www.hl7.org/fhir/valueset-route-codes.html
Roy van Pelt (May 24 2016 at 12:32):
Indeed, that would require the possibility to tie MedicationAdministration and Observation instances to Procedure. Or am I missing something?
Erich Schulz (May 24 2016 at 12:33):
i agree @Roy van Pelt
Erich Schulz (May 24 2016 at 12:34):
there are many nerve blocks being a clear example
Erich Schulz (May 24 2016 at 12:35):
sometimes (especially children) an entire anaesthetic is simply to give a medication (eg chemo into the csf via a lumbar puncture)
Erich Schulz (May 24 2016 at 12:36):
and its not too unusual for surgeons to apply/inject a medication directly into wherever they are
Erich Schulz (May 24 2016 at 12:37):
eg the opthamologists inject antibiotics below the conjunctiva
Erich Schulz (May 24 2016 at 12:37):
when the obstetricians are unhappy they inject things directly into the uterus
Erich Schulz (May 24 2016 at 12:38):
(ie during a Caesarian)
Erich Schulz (May 24 2016 at 12:39):
theoretically a combination of expanded route codes and the timestamp would capture all this (together with who administered)
Roy van Pelt (May 24 2016 at 12:45):
Thanks, these are interesting examples. These seem to relate mainly to access routes. It seem Procedures as such can be used in a broader scope as well, e.g., chemo treatment in your example.
Erich Schulz (May 24 2016 at 12:46):
another example is use of contrast medium injected by surgeons
Erich Schulz (May 24 2016 at 12:47):
local anaestheics can be injected by both surgeons and anaethetists... often v close to the maximum dose...
Erich Schulz (May 24 2016 at 12:50):
vascular surgeons commonly inject contrast, as do general surgeons doing cholecystectomies, and urologists doing RPGs...
Erich Schulz (May 24 2016 at 12:50):
gynaes use blue die to look at fallopian tubes...
Erich Schulz (May 24 2016 at 12:51):
and breast surgeons inject blue dye below the nipple to find the "sentinel node"
Erich Schulz (May 24 2016 at 12:54):
it is interesting to ponder if the "administrator + timestamp" link is enough...
Roy van Pelt (May 24 2016 at 13:04):
Indeed @Erich Schulz. There are many different situation that can occur, and it seems not all information can be resolved with the current references between the Resources. Perhaps the intended use is to query for all medication administered or observations during an Encounter, and if you want this divided per Procedure, you need to introduce your own Extension?
Erich Schulz (May 24 2016 at 13:06):
I actually thinking that if you have "who" and "when" administered an agent then that is probably sufficent
Erich Schulz (May 24 2016 at 13:06):
the fact the same person was performing a procedure at the same time is the link
Erich Schulz (May 24 2016 at 13:07):
specifying the precise route of administration is a bit tricky
Roy van Pelt (May 24 2016 at 13:22):
Perhaps "Who" and "When" is clinically sufficient indeed - I'm curious if this is the way it was intended in the specification? Using timestamps to find out the associated Procedure is a kind of reverse engineering for a lack of structure. Going via the Practitioner is an interesting thought, but seems a roundabout way of doing things.
Yunwei Wang (May 24 2016 at 15:14):
I think you need to use workflow in such complicated scenario. So you can have multiple ProcedureRequest paired with MeidcationOrder.
Erich Schulz (May 24 2016 at 22:28):
well its not that complicated a scenario really (at least it isn't a rare scenario by any means)
Erich Schulz (May 24 2016 at 22:29):
it's a recurring theme in records - much information is just a tranformation of some records into others
Erich Schulz (May 24 2016 at 22:29):
ie many "conditions" are simply a restating and summary of certain observations
Erich Schulz (May 24 2016 at 22:30):
the issue is how much redundant implied linkage to make -and how to achieve that reliably
Yunwei Wang (May 25 2016 at 01:02):
I know this is a simple case. The problem is that there is no direct link between these two resources.
Yunwei Wang (May 25 2016 at 14:14):
@Roy van Pelt I have raised this question to Pharmacy WG. We will discuss this topic in next WG conference call on June 6th. I will get you updated.
Lloyd McKenzie (May 26 2016 at 23:20):
As part of a general pattern for "event" resources, the Workflow project is recommending that most/all of them have "parent" relationships, so you can indicate that a particular MedicationAdministration is "part of" a Procedure (or vice versa). Note that this is a recommendation - it'll be up to the committees to decide whether to do so and whether to do so as a core element or an extension, based on what they see common practice to be.
Roy van Pelt (Jun 09 2016 at 12:46):
Thank you all for you kind responses. @Yunwei Wang : I'm looking forward to hear the result from the conference call. @Lloyd McKenzie : my first impression is that this suggestion is a fair generalization. It would be great if each "event" resource has a "part of" relationship. For MedicationAdministration we're bound to the Encounter for now, but it would be convenient if that will become more flexible, and a relation to Procedures becomes possible as well.
Lloyd McKenzie (Jun 09 2016 at 13:09):
@Roy van Pelt The linkage to Encounter/EpisodeOfCare will remain as that's a different sort of assertion. I expect the ability to have the "partOf" relationship will end up everywhere, though it may be extension in some places.
Yunwei Wang (Jun 09 2016 at 21:01):
@Lloyd McKenzie Do you suggest that Order/Request resource could have a parent/child relationship just like what Task resource has now?
Lloyd McKenzie (Jun 09 2016 at 23:21):
Not for "request" resources, no. They have an "in fulfillment of" relationship, but we decided against allowing parent/child.
Yunwei Wang (Jun 10 2016 at 00:05):
What are the "event" resources you referred to?
Lloyd McKenzie (Jun 10 2016 at 00:25):
Things like Procedure, Observation, Encounter, etc. Things that reflect actions rather than requests for actions.
Yunwei Wang (Jun 10 2016 at 00:25):
Now I understand. Thanks.
Roy van Pelt (Jun 14 2016 at 05:25):
That was very helpful, thanks. Continuing on the topic of 'event-type' resources, we see that for Medication there is a MedicationAdministration resource. In an interventional setting, a DeviceAdministration could be convenient resource as well (e.g., sheath, catheters). Has this been considered?
Yunwei Wang (Jun 15 2016 at 13:46):
I believe that's also event-type resource.
Roy van Pelt (Jun 16 2016 at 05:41):
Yes, indeed MedicationAdministration seems to be an event-type resource. However, DeviceAdministration does not currently exist, and I was wondering if such a resource was considered? It would be convenient in the interventional setting.
Last updated: Apr 12 2022 at 19:14 UTC
