Applying nullFlavor and negationInd to Problem Observations for "generic problem"
Brian Weiss
In the course of working on our “No Known Problems” sample, we have gotten into some questions re. variations on the “No Known Problems” theme…
On page 450, the IG gives two examples of “non-problem” (for lack of a proper phrase) observations. In Figure 213 a nullFlavor=”NI” (on an observation of Asthma) is used to indicate “specific problem not observed” for an observation on a specific problem (Asthma, in this example). In Figure 214 a neagtionInd=”true” is used on an observation (an HL7ActCode ASSERTION of a SNOMED generic "Problem") to indicate "No Known Problems".
What would be the correct combination of nullFlavor and negationInd on an Observation template (possibly the same for meds, allergies, and other conceptually similar templates in this context) to say the following:
1) Reflect variations on “No Known Problems”:
1.1) I have no information one way or the other about Problems (e.g. I am generating a CCD from my EMR and nothing was entered for Problems). (along the lines of nullFlavor="NASK")
1.2) I asked the patient if s/he has any problems, and s/he said "no". (along the lines of negationInd="true")
1.3) I asked the patient if s/he has any problems, and s/he didn't know. (along the lines of nullFlavor="ASKU")
1.4) I asked the patient if s/he has any problems, and s/he said "yes, but I'm not sure what". (along the lines of nullFlavor="UNK")
1.5) I asked the patient if s/he has any problems, and s/he said "I think so, my dad will fill you in on the details tomorrow". (along the lines of nullFlavor="NAV")
1.6) I asked the patient if s/he has any problems, and s/he said "yes, but I'm too embarrassed to tell you ". (along the lines of nullFlavor="MSK")
THE HIGH LEVEL QUESTION HERE: Can you apply nullFlavor to an HL7ActCode ASSERTION of a SNOMED generic “Problem” the same way Figure 214 applies negationInd=”true” to it?
Also: What do we really mean by “No Known Problems”. To me, Figure 214 is 1.2 above (“I have reason to believe there is no problem”) rather than 1.1 above (I have no idea one way or the other). But 1.1 is probably the more common case where “No Known Problems” is needed practically (as per the EMR example called out at the end of 1.1 above).
2) Reflect combinations of negation of a specific problem, with the variations above on “No Other Known Problems”:
2.1) asked the patient if s/he has Asthma and looked into it and s/he indeed does not have Asthma, I have no information one way or the other about any other Problems
2.2) I asked the patient if s/he has Asthma and looked into it and s/he indeed does not have Asthma, I asked the patient if s/he has any other problems, and s/he said "no"
2.3) …
2.7) I asked the patient if s/he has Asthma and s/he said she didn’t know, I have no information one way or the other about any other Problems
2.8) …
THE HIGH LEVEL QUESTION HERE: Can you combine a nullFlavor and/or negationInd on an HL7ActCode ASSERTION of a SNOMED generic “Problem” alongside the documentation of information about other specific problems? Can/should you combine them in one Concern Act?
Happy to get a pointer to where I can learn more about this if it is already documented. Please note that this is not a question about the different kinds of nullFlavor – so please don’t just point me at that. This is a question about how nullFlavor and negationInd are applied to “generic problems” (not a specific problem) and how that co-exists with specific observations (e.g. to communicate about “other problems”).
Brian Weiss
Lisa Nelson
Brian,
You are opening a very large can of worms here.
I have brought up the issue of the use of ASSERTION in the code element of an observation several times before. I was instructed to take the issue to the Vocab group where I learned that the only time we should be using ASSERTION in the code element in a observation is when the full semantics of the observation is represented in the code used in the value element. This is described in TermInfo. I came away from that experience in Vocab (and then in taking the Terminfo class at the last HL7 meeting) realizing that we in SDWG may not have considered fully the vocabulary requirements for templates where we are designed ASSERTION to be the code element. We would need to look at these and make sure that an appropriate value set of codes with full expressivity of the needed concepts, had in fact been specified for the value element of these templates.
In the case of the one example you note where a Problem Observation template has been shown with the observation/@negationInd=”true” and then a code element of “ASSERTION”. I have already submitted a comment about this example in the DSTU comments. (For some reason, the link to the DSTU Comments for CCDA was not working for me this morning, so I can’t supply the number.) According to the C-CDA spec, the code of the observation SHOULD come from this value set:
SHALL contain exactly one [1..1] ValueSet Problem Type 2.16.840.1.113883.3.88.12.3221.7.2 STATIC 2012-06-01 (CONF:9045). code, which SHOULD be selected from
Table 142: Problem type value set Value Set: Problem Type 2.16.840.1.113883.3.88.12.3221.7.2 STATIC 2012-06-01 | |||
Code System(s): | SNOMED CT 2.16.840.1.113883.6.96 | ||
Description: | This value set indicates the level of medical judgment used to determine the existence of a problem. | ||
Code | Code System | Print Name | |
404684003 | SNOMED CT | Finding | |
409586006 | SNOMED CT | Complaint | |
282291009 | SNOMED CT | Diagnosis | |
64572001 | SNOMED CT | Condition | |
248536006 | SNOMED CT | Finding of functional performance and activity | |
418799008 | SNOMED CT | Symptom | |
55607006 | SNOMED CT | Problem | |
373930000 | SNOMED CT | Cognitive function finding | |
There is no reason why ASSERTION should be shown in that example, in my opinion.
Based on what you are specifically recording, you could use Problem or even some of these other codes to make a slightly different semantic twist on the basic theme. For “no known problems” I think 55607006 seems the best choice.
The use of ASSERTION in this example of the Problem Observation is, in my view, a mistake. The use of ASSERTION in other templates needs to be reviewed with greater scrutiny to determine if the intensions set forth in TermInfo have really been met in the design of these other “ASSERTION-using” templates.
Lisa
Lisa R. Nelson, MSc, MBA | Consultant | Life Over Time Solutions | cell: 401.219.1165 | Westerly, RI | LisaR...@cox.net
--
You received this message because you are subscribed to the Google Groups "C-CDA Samples" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ccda_samples...@googlegroups.com.
To post to this group, send email to ccda_s...@googlegroups.com.
Visit this group at http://groups.google.com/group/ccda_samples?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
Brian Weiss
OK – let’s park the question of ASSERTION as that was not where I wanted to focus here….
Let’s assume we do it without an ASSERTION and use the SNOMED codes both for the code element and the value element.
We still have all the same questions I asked below re. the use of nullFlavor and negativeInd to express the various flavors of “No Known Problems” as per what I outlined below.
So, ignoring/setting-aside the ASSERTION issue – now what?
Brian
Robert Hausam
I would essentially echo Lisa's comments. And even if we take ASSERTION out of the picture for now, as you are doing, it's well recognized that the use of negation in its various forms in CDA (and elsewhere in HL7) is often quite problematic and confusing. The TermInfo project in Vocab is intending to address the negation issue specifically, and hopefully help everyone in bringing some clarity and reproducible guidance to this topic. I won't try to specifically answer your questions now, but I think you have a good set of questions/examples that collectively should be worked through. I'm assuming that you are interested in getting answers in the relatively near term. Let's see if we can work out a plan between TermInfo, Structured Docs, you and all of the other interested parties to get that done. Then we can include that guidance as part of the TermInfo IG on negation that we are intending to produce - and hopefully help reduce the confusion on this for everyone.
Rob
OK – let’s park the question of ASSERTION as that was not where I wanted to focus here….
Let’s assume we do it without an ASSERTION and use the SNOMED codes both for the code element and the value element.
We still have all the same questions I asked below re. the use of nullFlavor and negativeInd to express the various flavors of “No Known Problems” as per what I outlined below.
So, ignoring/setting-aside the ASSERTION issue – now what?
Brian
From: Lisa Nelson [mailto:LisaR...@cox.net]
Sent: Thursday, June 06, 2013 14:46
***********************************************************************************
Manage your subscriptions | View the archives | Unsubscribe | Terms of use
--
Boone, Keith W (GE Healthcare)
Sent from my iPhone
On Jun 6, 2013, at 7:20 AM, "Brian Weiss" <brian...@cdapro.com<mailto:brian...@cdapro.com>> wrote:
In the course of working on our “No Known Problems” sample, we have gotten into some questions re. variations on the “No Known Problems” theme…
On page 450, the IG gives two examples of “non-problem” (for lack of a proper phrase) observations. In Figure 213 a nullFlavor=”NI” (on an observation of Asthma) is used to indicate “specific problem not observed” for an observation on a specific problem (Asthma, in this example). In Figure 214 a neagtionInd=”true” is used on an observation (an HL7ActCode ASSERTION of a SNOMED generic "Problem") to indicate "No Known Problems".
What would be the correct combination of nullFlavor and negationInd on an Observation template (possibly the same for meds, allergies, and other conceptually similar templates in this context) to say the following:
1) Reflect variations on “No Known Problems”:
1.1) I have no information one way or the other about Problems (e.g. I am generating a CCD from my EMR and nothing was entered for Problems). (along the lines of nullFlavor="NASK")
1.2) I asked the patient if s/he has any problems, and s/he said "no". (along the lines of negationInd="true")
1.3) I asked the patient if s/he has any problems, and s/he didn't know. (along the lines of nullFlavor="ASKU")
1.4) I asked the patient if s/he has any problems, and s/he said "yes, but I'm not sure what". (along the lines of nullFlavor="UNK")
1.5) I asked the patient if s/he has any problems, and s/he said "I think so, my dad will fill you in on the details tomorrow". (along the lines of nullFlavor="NAV")
1.6) I asked the patient if s/he has any problems, and s/he said "yes, but I'm too embarrassed to tell you ". (along the lines of nullFlavor="MSK")
THE HIGH LEVEL QUESTION HERE: Can you apply nullFlavor to an HL7ActCode ASSERTION of a SNOMED generic “Problem” the same way Figure 214 applies negationInd=”true” to it?
Also: What do we really mean by “No Known Problems”. To me, Figure 214 is 1.2 above (“I have reason to believe there is no problem”) rather than 1.1 above (I have no idea one way or the other). But 1.1 is probably the more common case where “No Known Problems” is needed practically (as per the EMR example called out at the end of 1.1 above).
2) Reflect combinations of negation of a specific problem, with the variations above on “No Other Known Problems”:
2.1) asked the patient if s/he has Asthma and looked into it and s/he indeed does not have Asthma, I have no information one way or the other about any other Problems
2.2) I asked the patient if s/he has Asthma and looked into it and s/he indeed does not have Asthma, I asked the patient if s/he has any other problems, and s/he said "no"
2.3) …
2.7) I asked the patient if s/he has Asthma and s/he said she didn’t know, I have no information one way or the other about any other Problems
2.8) …
THE HIGH LEVEL QUESTION HERE: Can you combine a nullFlavor and/or negationInd on an HL7ActCode ASSERTION of a SNOMED generic “Problem” alongside the documentation of information about other specific problems? Can/should you combine them in one Concern Act?
Happy to get a pointer to where I can learn more about this if it is already documented. Please note that this is not a question about the different kinds of nullFlavor – so please don’t just point me at that. This is a question about how nullFlavor and negationInd are applied to “generic problems” (not a specific problem) and how that co-exists with specific observations (e.g. to communicate about “other problems”).
Brian Weiss
***********************************************************************************
Manage your subscriptions<http://www.HL7.org/listservice> | View the archives<http://lists.HL7.org/read/?forum=strucdoc> | Unsubscribe<http://www.HL7.org/tools/unsubscribe.cfm?email=keith...@ge.com&list=strucdoc> | Terms of use<http://www.HL7.org/myhl7/managelistservs.cfm?ref=nav#listrules>
Brian Weiss
Thanks, Keith. Indeed a very helpful post on this that addresses my 1.x questions. Essentially you are saying that the two IG samples can be combined and it's OK to apply a nullFlavor to a generic problem/allergy/med SNOMED code and not just to a specific one.
Note that you, like the IG sample, have that questionable use of HL7ActCode ASSERTION rather than the SNOMED Problem Type as guided to via SHOULD language in CONF:9045 - but as per my exchange with Lisa, we said we would set that aside for now...
Now maybe you can help me out with how to say "I know the patient is not allergic to anything ELSE" or "I don't know whether the patient is allergic to anything ELSE". At the core you have an observation of an allergy the patient DOES have. You can't set negationInd on that and you can't nullFlavor that (they do have it and you do know it). Would you just add a second observation that is the same as your samples for the simple case when ALL allergies are known to not be there or not known? In some ways the two observations contradict each other?
I think what typically happens is that the C-CDA has the known allergies/meds/problems and doesn't say anything about the others (implying what? Not known? Known/believed not to exist?).
The main use-case for "No Problems (/meds/allergies/etc.) Known" is to meet the requirement that a section is present (for certification and/or for conformance) when you have nothing to say about it. So, usually what is really desired for "Not Known" is the nullFlavor form (which says "I don't know"), not the negationInd form (which says "I know they don't"). Perhaps not surprisingly, the IG does the opposite - in Figure 214 to demonstrate what it terms "No Known Problems" it uses the negationInd which actually means "Known that there are no problems"...
Clearly we are going to have TWO samples here - one for what the IG calls "No Known Problems" (which really is "Known that there are no problems") and one for what the implementation community primarily needs here ("filler XML" for when you're required to populate a section that you would otherwise leave blank, if it wasn't for those darn validator error messages...)
Brian
Brian Weiss
Thanks, Rob.
Here’s what I plan to do:
1) With Lisa’s continued help, complete our two samples of “Known/Believed to Be No Problems” and “No Idea if There Are Problems” (note that I am now avoiding the problematic/ambiguous wording, at least for me, of “No Known Problems” – which apparently means “Known/Believe to Be No Problems” even though it sounded to me more like “No Idea If There Are Problems”). Hope to do that by the start of next week.
2) Post a pointer to the results on the SDWG HL7 listserv.
3) Await comments/feedback and, in parallel, move on to the next samples…
Meanwhile, to the extent your TermInfo/SDWG/Other-Interested-Parties work moves the ball forward in a different direction, happy to go back and change the samples to match…
Brian
From: ccda_s...@googlegroups.com [mailto:ccda_s...@googlegroups.com] On Behalf Of Brian Weiss
--
You received this message because you are subscribed to the Google Groups "C-CDA Samples" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ccda_samples...@googlegroups.com.
To post to this group, send email to ccda_s...@googlegroups.com.
Visit this group at http://groups.google.com/group/ccda_samples?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
***********************************************************************************
Manage your subscriptions | View the archives | Unsubscribe | Terms of use
Reed D. Gelzer
Kibitzing here, you’re welcome to ignore.
From an HL7 Records Management and Evidentiary Support WG point of view, I find this discussion very reassuring.
Our WG has periodically identified substantial problems with non-definitive statements of “unknown”. Among the most worrisome are defaults to No Known (XYZ) rather to some form of null state for such as allergies, medications, prior conditions, and etc. when in fact no information was gathered, along with insufficient means to capture gradations or variants of “know”.
It is great to know that SD is so attentive to the records management and evidentiary implications of “unknown” vs. “no info available to the EHR-S”.
Keep up the great work.
RDGelzer, MD, MPH
Co-Chair EHR-S Records Management and Evidentiary Support WG
In the course of working on our “No Known Problems” sample, we have gotten into some questions re. variations on the “No Known Problems” theme…
--
You received this message because you are subscribed to the Google Groups "C-CDA Samples" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ccda_samples...@googlegroups.com.
To post to this group, send email to ccda_s...@googlegroups.com.
Visit this group at http://groups.google.com/group/ccda_samples?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.