Stream: IPS
Topic: Discharge based on IPS
John Moehrke (Oct 27 2021 at 16:24):
Kind of working on a Discharge Summary derived off of IPS, adding in the Encounter... has anyone else done this?
John Moehrke (Oct 27 2021 at 16:41):
Hmmm, it seems the IPS is forcing the .type to be 60591-5.... so it is impossible to derive my Discharge Summary off of the Medical Summary.... right?
Elliot Silver (Oct 27 2021 at 16:43):
Yup. However, you could create another composition with a different type, and reuse the section content from IPS. There have been discussions about developing an IPS document library which would probably follow that approach.
John Moehrke (Oct 27 2021 at 16:44):
I would thus replicate EVERYTHING that is IPS
John Moehrke (Oct 27 2021 at 16:44):
except for the .type element
John Moehrke (Oct 27 2021 at 16:45):
so, it was not designed like C-CDA to be refined ... drat
Elliot Silver (Oct 27 2021 at 16:46):
No, you wouldn't need to re-profile AllergyIntolerance, you wouldn't need to re-profile MedStatement, ... you'd only need to create your own composition.
John Moehrke (Oct 27 2021 at 16:48):
ah, that is indeed different
John Moehrke (Oct 27 2021 at 16:50):
I still think I will copy the IPS Composition XML, rename it, remove the .type, and then further profile that in FSH.... unless someone has already converted that to FSH?
Elliot Silver (Oct 27 2021 at 16:52):
I think @Rob Hausam has converted IPS to FSH in the connectathon branch.
Elliot Silver (Oct 27 2021 at 16:54):
https://github.com/HL7/fhir-ips/blob/connectathon/input/fsh/CompositionUvIps.fsh
Rob Hausam (Oct 27 2021 at 17:01):
Yes, that's correct - 'connectathon' is is FSH.
John Moehrke (Oct 27 2021 at 18:27):
so, as far as I can tell... I will need to copy that Composition FSH and adjust it to my needs. I will still depend on IPS so as to benefit from the other stuff... right?
Elliot Silver (Oct 27 2021 at 18:47):
Yes.
Peter Jordan (Oct 27 2021 at 19:32):
This certainly fits in with the 'building block' approach to IPS and deriving from the IPS for local use. However, whether a Discharge Summary - a legal transfer of care document with (in many/most countries) a pre-defined structure based on an encompassing encounter - is a good fit is open to debate. It is a very different use case to the provision of a patient summary, e.g. ALL events that occurred in the facility discharging the patient must be included.
Rob Hausam (Oct 27 2021 at 19:55):
I think that's partly right, @John Moehrke. But if your profile is depending on the IPS profile (if I'm understanding that correctly) then you will only need to differential in your profile, so your "copy" of the IPS Composition FSH likely will be (or at least can be) stripped down pretty significantly. And I agree with Peter's questions about the use case and whether it is a good fit - so that probably needs to be explored further.
John Moehrke (Oct 27 2021 at 20:00):
Rob, I initially was having my profile profile based on the IPS.. but ran into the problem right away that your IPS profile requires the .type to be "Medical Summary"... well a discharge (by any definition) is not that...
Rob Hausam (Oct 27 2021 at 20:16):
@John Moehrke I do think that logically a "discharge summary" is a type of "medical summary". The document type codes might not be organized in that way (I'm not sure if they are or not), but if that's the logical understanding then I don't think that actually should be a roadblock.
John Moehrke (Oct 27 2021 at 20:22):
so, you think I can just have the .type.coding hold both Medical Summary (60591-5) and add to .type.coding Discharge Summary (18842-5)? That felt like I would get yelled at for that./
Elliot Silver (Oct 27 2021 at 21:02):
John, for clarity, what I was suggesting was that your IG derive from the IPS IG, not that your composition derive from the IPS composition.
John Moehrke (Oct 27 2021 at 21:09):
what?
John Moehrke (Oct 27 2021 at 21:11):
@Rob Hausam is there an editable version of your pretty IPS graphic?
Rob Hausam (Oct 27 2021 at 22:03):
@John Moehrke I think we have a few pretty graphics :) And nearly all of them (maybe actually all) are attributable to @Giorgio Cangioli, and he probably has editable copies which may (I don't know for sure) be available?
Rob Hausam (Oct 27 2021 at 22:04):
And @John Moehrke, I wasn't necessarily suggesting to "have the .type.coding hold both Medical Summary (60591-5) and add to .type.coding Discharge Summary (18842-5)". I think that could be a possible option, but it may not be the only one. We could look at whether there are other options.
Elliot Silver (Oct 27 2021 at 22:35):
John Moehrke said:
what?
Is that to me? For the approach I was envisioning, your IG is going to need to have a dependency on the IPS IG. Your composition however does not reprofile the IPS composition, but instead is a new composition profile (with many similarities to the IPS composition) that indicates which sections you need for your purposes.
John Moehrke (Oct 28 2021 at 12:27):
@Rob Hausam and @Giorgio Cangioli it would be nice to have an editable form. I often use PowerPoint (not my favorite, but sometimes it just works). I put that PPTX file into an images-source folder, and export the PNG as you have. This enables future people to refine graphics, while having github doing binary change tracking. Not as nice as plantuml, but plantuml is not as pretty as these.
John Moehrke (Oct 28 2021 at 12:30):
@Elliot Silver I think what you describe is where I am at right now. I am dependent on IPS. I "copied" the FSH version of IPS, renamed the profile, and eliminated the .type constraint.
I could do all my profiling in this "copied" body, but by leaving it as close to unchanged, I give myself a chance at benefitting from future improvements to IPS.
Thus I named this "copied" version "MostlyCompositionUvIps", from which I have my purpose specific Composition that builds upon that with the constraints I have. Making some of the IPS sections mandatory, adding a discharge section.
This is all ugly, and I expect we all will figure out a more friendly way to do this. But it seems to work for now.
John Moehrke (Oct 28 2021 at 12:31):
https://github.com/IHE/QRPH.QORE
John Moehrke (Oct 28 2021 at 12:32):
actually I am just the muscle, @ANDREA FOURQUET is the brains.
Giorgio Cangioli (Oct 28 2021 at 12:38):
Reusing IPS for Discharge Letter
As mentioned by other answers the best option I see is to reuse IPS profiles and section (as they are now or profiles ) to build other kind of documents. This is - for example - how we are going to propose a EU Cross Border Hospital Discharge Report by reusing the IPS CDA (https://art-decor.org/art-decor/decor-templates--eehrxf-?section=templates&id=2.16.840.1.113883.2.51.10.12) [in progress] and FHIR IPS (https://build.fhir.org/ig/hl7-eu/x-ehealth/StructureDefinition-Composition-ips-xeh.html) [less than a draft ;-) ]
Giorgio Cangioli (Oct 28 2021 at 12:39):
John Moehrke said:
Rob Hausam and Giorgio Cangioli it would be nice to have an editable form. I often use PowerPoint (not my favorite, but sometimes it just works). I put that PPTX file into an images-source folder, and export the PNG as you have. This enables future people to refine graphics, while having github doing binary change tracking. Not as nice as plantuml, but plantuml is not as pretty as these.
I'll search and send it to you as ppt
Elliot Silver (Oct 28 2021 at 15:24):
John Moehrke said:
Elliot Silver I think what you describe is where I am at right now. I am dependent on IPS. I "copied" the FSH version of IPS, renamed the profile, and eliminated the .type constraint.
I could do all my profiling in this "copied" body, but by leaving it as close to unchanged, I give myself a chance at benefitting from future improvements to IPS.
Thus I named this "copied" version "MostlyCompositionUvIps", from which I have my purpose specific Composition that builds upon that with the constraints I have. Making some of the IPS sections mandatory, adding a discharge section.
This is all ugly, and I expect we all will figure out a more friendly way to do this. But it seems to work for now.
Ideally, the "MostlyCompositionUvIps" would move into IPS, and the current IPS composition would be a specialization of that, just like your document is a specialization, and others could do similar specializations.
Rob Hausam (Oct 28 2021 at 15:47):
Possibly, but I'm not yet quite sure if this actually is needed (I probably need to read the details more thoroughly). If we may have over-constrained this in the current IPS profile, we can look at that.
Elliot Silver (Oct 28 2021 at 18:15):
I don't think you've overly constrained the current IPS document, if what you are wanting to generate is a true IPS document. However, John's case brings up that people want to generate other kinds of documents with similar sections. We saw with vaccine certificates that people want to generate IPS-like documents with less content. And I'm dealing with a situation where there is a desire for "almost IPS" documents. Having some flexible layer of profile between Composition and true IPS seems useful.
John Moehrke (Oct 28 2021 at 18:45):
I think the current IPS is okay today... but I do think that we should be open to these other use-cases that would drive for a more modular approach.. I worry that starting with a modular approach gets overly modularized into something that no one finds helpful. So for those of us that are trying to expand beyond IPS, it is up to us to clearly identify the constraints that should be relaxed in a reusable not-IPS profile.
Derek Ritz (Nov 03 2021 at 19:09):
This idea... of creating IPS-ish documents out of the IPS building blocks... is one that various groups are exploring (including Canada Health Infoway and its jurisdictional collaborators). Having a clear and consistent way of approaching this would be very beneficial. :+1:
Derek Ritz (Nov 03 2021 at 19:10):
@Attila Farkas
Rob Hausam (Nov 03 2021 at 19:52):
We definitely agree with that regarding using the "building blocks" and doing that consistently. We will need to work on further guidance.
Last updated: Apr 12 2022 at 19:14 UTC
