This is a very important question-- I can't imagine doing a meaningful
retrospective study without addressing it.
I noticed that whenever an encounter has a diagnosis with MODIFIER_CD =
'DiagObs:PROBLEM_LIST' that encounter will also contain the same
diagnosis but with MODIFIER_CD = 'PROBLEM_STATUS_C:X' where X is 1,2, or
3. And in these cases I've been filtering so that there are only
'PROBLEM_STATUS_C:1' if I want the active diagnoses.
I also filter out anything that has MODIFIER_CD = 'DiagObs:MEDICAL_HX'.
There is a second pattern for what I suspect are also active diagnoses.
These have MODIFIER_CD = 'DiagObs:PAT_ENC_DX'. Some of these have a
second copy of the diagnosis in the same encounter with MODIFIER_CD =
'DiagObs:PRIMARY_DX_YN' and some do not. I have been assuming that the
'PAT_ENC_DX' ones are current if they exist at all and have been
ignoring 'PRIMARY_DX_YN' because I'm interested in active diagnoses
regardless of whether or not they are primary for their respective visits.
So, together this means that through the I2B2 interface, every single
diagnosis term in my query I would include twice in the same pane: once
with the 'DiagObs:PAT_ENC_DX' modifier and once with the
'PROBLEM_STATUS_C:1' modifier (and never just a bare diagnosis).
BUT, that's just my current best guess. I'm just as uncertain as you,
Dan. How do the rest of y'all do?
Something I've been meaning to do is actually pull consecutive visits
for patients and see if 'PAT_ENC_DX' for something you would expect to
go away (fracture, laceration) actually goes away in subsequent visits.
But the opinion of any Epic experts on the list will save us all a lot
of trial and error.
PS: KUMC also has diagnoses where MODIFIER_CD is one of
'DiagObs:UHC_DIAGNOSIS', 'DiagObs:Clinic', or 'DiagObs:Primary'. My site
doesn't have those, so I have no idea how to distinguish active
diagnoses from among the ones with those modifiers.
On 09/24/2014 01:10 PM, Dan Connolly wrote:
Russ,
I have a note from the six month evaluation: "4.c currently have a
diagnosis." i.e. distinguishing between historical/resolved diagnoses
and current diagnoses.
I can imagine some somewhat manual approaches to this, but I don't
know of a crisp, generalizable technique for it. Is it clear to you
(or anybody else in gpc-dev) how to do it?
In any case, is this something I/we can leave in the someday pile?
--
Dan
_______________________________________________
Gpc-dev mailing list
[email protected]
http://listserv.kumc.edu/mailman/listinfo/gpc-dev
_______________________________________________
Gpc-dev mailing list
[email protected]
http://listserv.kumc.edu/mailman/listinfo/gpc-dev
