Stream: argonaut
Topic: Smoking status
Jay Lyle (Sep 01 2018 at 13:22):
Just has a conversation about the meaning of Argonaut labels for Smoking Status, here.
Michele Mottini (Sep 04 2020 at 13:52):
The smoking status observation profile appears not to require a category - am I right? That would make it the only observation without a category - that seems wrong?
Josh Mandel (Sep 04 2020 at 13:53):
Indeed. This has come up in our discussion about granular scopes too (which I know will not bother you :-))
Brett Marquard (Sep 04 2020 at 14:38):
Fun time. Thanks @Michele Mottini please log a tracker
Josh Mandel (Sep 04 2020 at 14:39):
Ideally we'd want to use a specific category like smoking history rather than a general category like social history, based on experience from several Argo vendors trying to keep very different kinds of data under distinct categories.
Michele Mottini (Sep 04 2020 at 15:31):
Ah not ideally for us...
Michele Mottini (Sep 04 2020 at 15:36):
https://jira.hl7.org/browse/FHIR-28436
Brett Marquard (Sep 04 2020 at 15:44):
Thank you sir
Eric Haas (Sep 04 2020 at 15:53):
(deleted)
Eric Haas (Sep 04 2020 at 15:55):
if up to us ( and might as well be) would be category = social-history
Eric Haas (Sep 04 2020 at 15:56):
although the lack of category makes it easier to implement and anything we propose will meet opposition somewhere.
Josh Mandel (Sep 04 2020 at 21:39):
if up to us ( and might as well be) would be category = social-history
Can you explain why @Eric Haas ? We had very strong negative reaction to lumping this level of 'harmless' (smoking) data along with 'sensitive' (e.g., alcohol abuse, family life) data. @Pascal Pfiffner, @Jenni Syed , @Isaac Vetter expressed concern about this on our Argonaut calls if I recall correctly.
Josh Mandel (Sep 04 2020 at 21:40):
This was one of the few concrete suggestions we heard about how to improve policy-based access control.
Pascal Pfiffner (Sep 04 2020 at 21:50):
Yes, lumping "smoking status" and e.g. "sexual history" together is problematic for us when we think about how users would share this data with 3rd parties. Currently this is based on category. I can see appeal in pulling these together under "social history" but this is something that creates an issue for us specifically. In fact we have the same problem with labs, "harmless" and sensitive labs are still just "laboratory" after all, so maybe creating a different category to separate items of different sensitivity is not the right approach?
Josh Mandel (Sep 04 2020 at 21:53):
The "right" approach would be to use security labels here -- the challenge as always is how to apply them consistently. keeping tobacco history in its own rum given that it is a distinct category in the US core seems like the low-hanging fruit approach.
Eric Haas (Sep 05 2020 at 01:51):
Well what should it be then? We are at loggerheads with categories which is why there is nothing now.
Btw. I think one can make the same argument for all the categories.
Sent from my iPhone
Josh Mandel (Sep 05 2020 at 01:53):
I'd just categorize it under "smoking history" or "tobacco history". I agree this is basically an adhoc answer, but I still think it solves a practical problem.
Eric Haas (Sep 05 2020 at 19:29):
I think the same argument could made other categories as well. The base categories terminology was not designed to discriminate what is sensitive, There are security and privacy tags that do that. What other solutions were proposed? This lack of consensus is why is not specified in US Core btw.
Eric Haas (Sep 05 2020 at 19:31):
Josh Mandel said:
I'd just categorize it under "smoking history" or "tobacco history". I agree this is basically an adhoc answer, but I still think it solves a practical problem.
IMO just use the code instead, that is essentially what super fine grained categories are...
Eric Haas (Sep 05 2020 at 19:34):
How is this security issue solved in CCDA, since that is one of the primary inspirations for these categories?
Josh Mandel (Sep 05 2020 at 19:56):
I've only ever seen CDA files shared wholesale, never broken into parts and used to fuel an API (except for a couple of demos I built in like 2012...)
Grahame Grieve (Sep 06 2020 at 00:13):
it happens inside products
Josh Mandel (Sep 06 2020 at 01:12):
I've seen meds, problems, allergies extracted (that's required in the US) but often with a pretty manual importing step.
Josh Mandel (Sep 06 2020 at 01:15):
I haven't come across different security/sharing rules being applied to different parts of a CCDA
John Moehrke (Sep 08 2020 at 16:06):
I would agree. although DS4P theorizes it, the implementation of these fine-grain controls always tend to be too difficult. Further the recipient must be authorized to receive the most sensitive of the content, thus the marking of some stuff at less sensitive than that seems to be of limited benefit. We did make this easier in FHIR, but the effort is still going to be a big effort, and the fact that the recipient must be authorized to receive the most sensitive tends to leave us in the same place. That is, you either are authorized to receive THIS data, or you are not. I need to know the distinction within my walls, but all I need to do is filter out the stuff you are not authorized and send that subset. Hence why most use of DS4P today is at the Document level (bundle). There are Normal and there are Restricted; there is just not much benefit in mixed. so, like Grahame said, this happens inside the product.
John Moehrke (Sep 08 2020 at 16:11):
category vocabulary should be big enough to group reasonable things together; but not so big as to group things that are different. What is not clear is if the category should try to group similar sensitivity things (like social-determinants-of-health would include smoking status and drinking status and sexual health indicators); or if that distinction is enough to make a more fine grain social health categories. I suspect that these are two different vectors, and any attempt to have category be broken down into likely-sensitivity fine grain; will fail.
Josh Mandel (Sep 08 2020 at 17:54):
and any attempt to have category be broken down into likely-sensitivity fine grain; will fail.
Agreed, re: general-purpose strategies. But today smoking status is the only kind of social history data required in ONC's core data set, and it sounds like Argonaut participants would be happy enough with a "spot fix" for this issue.
Eric Haas (Oct 28 2020 at 20:54):
J#28436 proposed disposition
Josh Mandel (Oct 28 2020 at 21:02):
Gah, that's too bad in my view (it lumps smoking history in a large-and-generally-quite-sensitive bucket).
Josh Mandel (Oct 28 2020 at 21:03):
(It means that even with the addition of SMART's v2 "category-based scopes," we still lump together a lot under social history.)
Eric Haas (Oct 28 2020 at 21:03):
I would like to get others to chime in too - it can be pulled for discussion.
Last updated: Apr 12 2022 at 19:14 UTC
