#280: Medication ontology orgainzed by ingredient
-------------------------+----------------------------
Reporter: preeder | Owner: jay.pedersen
Type: enhancement | Status: assigned
Priority: major | Milestone: snow-shrine-2
Component: data-stds | Resolution:
Keywords: | Blocked By: 568, 592
Blocking: |
-------------------------+----------------------------
Comment (by campbell):
Keith
If you check the notes from the ?third call you hosted, discussion on the
PROCEDURES table concluded that medication administration events could be
stored there…a surprise to me. At any rate, GPC is simply trying to
achieve consensus on where to store medication events. For right now, and
based upon your response, I am concluding:
ORDERS(Inpatient, clinic and prescriptions) ==> CDMV31.PRESCRIBING
DISPENSE EVENTS(Inpatient and community pharmacy) ==> CDMV31.DISPENSING
ADMINISTRATION EVENTS(Inpatient, clinic) ==> CDMV31.PROCEDURES
You further caution us to record events indexed to the most specific level
of clinical detail: NDC codes for dispense and administration events and
RXNORM SBD and SCD (or I assume BPCK or GPCK) for orders.
Have I interpreted you correctly?
At Nebraska and for SCILHS, we have our metadata organized so that we can
aggregate by ingredient for searches, but do not store events at that
level.
Jim
>>>
Jim,
You can end up looking at this from a couple of different angles. If you
are an investigator, and want to research on chemotherapy treatments, you
can look at the available tables that may contain medications and decide
which to investigate. For chemotherapy, that’d probably be PROCEDURES and
PRESCRIBING. It’s likely that an institution would have data in one of
those tables, or both, depending on their source system.
If you’re approaching the problem as an ETL programmer (which is what I
think you’re asking), then it’s a different question. Given chemotherapy
data, where do I put it in the CDM? Assuming that the chemotherapy
treatments look something like medication administration data, I think
we’d want the orders to be put into the PRESCRIBING table. Right now, we
don’t really have a place to store the administration event itself, but
that’s something we hope to address in the next version of the CDM.
In the antibiotic study, the main challenge was getting network partners
to map their data to the right level of RxNorm. The preference is to use
a more inclusive term level, like Semantic Branded Drug (SBD) or Semantic
Clinical Drug (SCD). I understand that storing by ingredient can make the
ontology more straightforward, but it greatly complicates any downstream
analysis. So my recommendation would be to store the SBD or SCD RxCUI as
part of the observation (perhaps as a modifier) even if you use ingredient
as a concept code. That way there isn’t any loss of analytic utility when
it comes time to populate the CDM.
I’m copying Laura in case she’d like to add anything.
Keith
>>>On Apr 4, 2017, at 1:08 PM, Campbell, James R <campb...@unmc.edu>
wrote:
Keith
From the Data characterization calls with DRNOC a couple weeks back, I
took away to a confusing discussion within GPC about how we are deploying
medications in CDMV3+. In particular, you noted that medication
administration events were to be expected in PROCEDURES in NDC codes but
orders are in PRESCRIBING (at least ambulatory) in RXNORM and dispense
events in DISPENSING with NDC. Discussing this with our Cancer workgroup,
they have a requirement for tracking chemotherapy treatments and are
asking for advice on standardization of deployment.
We are organizing our i2b2 medication metadata from NLM by ingredient
(RXNORM) and subsuming all clinical formulations (RXNORM) and packaging
(NDC) in a tree structure with top level organization by VA drug classes.
I understand you are involved in a national antibiotic study and ask if
you can provide some advice or direct me to guidance from PCORI/Sentinel
about how we should be planning our i2b2 deployment of medication events.
Jim
--
Ticket URL:
<http://informatics.gpcnetwork.org/trac/Project/ticket/280#comment:23>
gpc-informatics <http://informatics.gpcnetwork.org/>
Greater Plains Network - Informatics
_______________________________________________
Gpc-dev mailing list
Gpc-dev@listserv.kumc.edu
http://listserv.kumc.edu/mailman/listinfo/gpc-dev
