2017-09-19 IA Small Group Call

Date

Attendees

Scribe

Goals

  • Discuss with members of the CDS team the status of the CDS KNARTs and concerns that priority has been given to order sets vs. phenomenon in terms of terminology modeling.

Discussion items

TimeItemWhoNotes
60minOverview of Status of CDS KNARTs and TSRsGroup
  • Keith:  Let's review where we are with the various documents.  Do we have specific action items we need to discuss?  Claude:  We are getting to the point now where we will submit a TSR to Terminology.  The question is what should it look like for Deloitte?  Keith:  For Phenomenon Measurement, it's pretty much LEGOs and we are set.  Do you have any idea how many will be of Phenomenon type vs. request or goal?  Claude:  The majority will be action requests.  Keith:  But for a KNART, don't you need to have a trigger for the ordering of the order set?  Claude: There are some.  We are instructed by a clinical team (did not capture what else he said).  Keith:  He will push back immediately on that and if we need to bring Diane in, we will.  He had a call earlier with her, Steve, Mike Lincoln, and others.  By selecting action request vs. phenomenon, you are tilting what the Terminology team needs to deliver.  Claude:  It depends on the KNART, if you look at a documentation template or Event Condition Action (ECA) rule, then phenomenon definitely is applicable.  (Missed part of the conversation as I had to contact Diane to see if she could join the call.)
  • Keith:  We need to have work for the Terminology team to do.  While we are still working out the details for action request and action performance, we need progress on the KNARTs.  There is pressure on us from many sources.  (Missed part of what he said as audio was lost temporarily.)  There might be complex text that we can work to refine.  As the models improve, we would move away from putting in placeholders.  Also, we would go back and redo the things that were added as primitive.  You would get credit for modeling them a second time than what was modeled in the first place.  Claude:  (Much of what he said was inaudible due to an echo present during this portion of the call.)  The action requests are very simple.
  • Keith:  Informed Steve on what has been discussed thus far, particularly that Cognitive is focusing on action requests and not phenomenon clinical statements.  He had said the clinical SMEs asked them to focus not on phenomena but on orderables.  That is not delivering the value that we need for this effort.  While waiting for Steve to join this call, he asked that we insert a primitive string to be able to work in an iterative way.  this really impacts Kirsten's group, and he would like to understand the amount of effort from Terminology.  Steve:  He understands there is one clinical white paper either close to launching or is ready to launch.  Claude:  It is the chest pain white paper for it.  There is a documentation template and an order set.  The documentation template is very complex.  Keith:  They are doing what is easiest for them but what is harder for the other group.  He is concerned you are choosing not to tackle the difficult work. Claude:  We want to get the process started and define and start with the simpler models.  Keith:  He thinks we need to do some refocusing here and reallocate resources as needed.  Steve:  It sounds like you are doing partial deliverables with the easier part of the work being done first, with more coming later.  Keith:  His concern isn't the white paper itself, rather focusing only on the order aspect of it.  It's not a comprehensive slice, and we should pick our highest risk areas first.  If we don't have alignment between the two groups, that is a problem.  He doesn't want to cherry pick but focus on them all equally and you are being selective.  Selectivity causes other problems for the project.  Steve:  We need to try to balance out the work amount for Terminology.  We have a better understanding of observations vs. orders.  Terminology has sort of bought ahead on models.  A fair number of what would be easy models for them are pre-paid but are going on the shelf.  If it's not this particular form, if we can start using some of the pre-paid modeling effort that is sitting on the shelf going stale, that would be good.  Starting to give Terminology something to get them moving would be good.
  • (Diane joined the call at this time time).  Keith:  He felt he needed for her to attend the call since this work falls under her authority for the CDS contract.  Keith recapped what transpired during the call thus far.  To him, it seems the whole purpose of a KNART order set is that you need to be able to tie it to a phenomenon.  Thus, it was the opposite of what he was expecting. And, the SMEs don't want to deal with the observation/phenomenon aspects of it and only the order sets.  He believes the SMEs need to be redirected to spend more effort to define the conditions that are the indicating/non-indicating aspects of the order sets.  Has there been a directive from CDS to do this or was this something the SMEs came up with on their own and should they be redirected to focus on the indication for the order set?  An example would be an order set for someone coming into the ER for chest pain.  What are the factors you might choose to do something that resulted in the order? Are they indicated every time someone comes in with chest pain or are there physical characteristics that come into play?  Yes, we can model an order set for TPA but what are the factors that contribute as to why it should be administered?  What are the contraindications, what if they had TPA in the past, etc.  If all we focus on is the order sets, we have lost the beginning.
  • Diane:  There is a fair amount of this discussion occurring with the SME group, most densely for those KNARTs being combined as a composite consult request because that workflow demands that type of information.  However, at the stage of KNART development we're in, we're treating many of the order sets as standalone KNARTs.  You evaluate the patient with a documentation template.  For the order sets we're working on this year, we had that assumption in mind.  The use of the order set would be triggered by clinical decision making.  Keith:  He proposes for each element that comes in the order set, it would also come in on a documentation template.  The work that comes out needs to be balanced.  Claude:  We are working with a composite KNART that has an order set and a documentation template.  We are starting with the structuring of the order set, but when the KNART is delivered, it will contain both the order set and a documentation template.  We only started with the order set.  Keith:  His choice to do the order sets first is exactly the opposite of the skill sets we have set aside on the Terminology side.  He recommends switching the order of the KNARTs.  It is a short-term, logistical issue.  Steve:  The Terminology team is waiting to have widgets to model, and those that they would model would be from documentation templates vs. order sets.  It would be nice to get them in the pipeline and we pre-paid for those.  The sooner Terminology can begin working on documentation templates, the better.  Diane:  The fundamental problem is we have people waiting to make widgets, like LEGO modeling.  Steve:  Agreed.  Diane:  If that's the fundamental problem, what is the relationship to Keith's concern?  Keith:  Part of the conversation from our call this morning is that we don't know how to model the requests, only the old-style LEGOs.  If everything that is requested falls into what is disputed, there is a problem.  Claude:  Documentation templates will have the same types of challenges as order sets.  It doesn't make the problem easier.  You can partially model the documentation template but not fully model it yet.  For those things we can't model yet, such as procedure request and procedure performed, we'll use placeholders. 
  • Keith:  We may need to revisit this in the small vs. the large.  Gather additional facts, and maybe he, Diane, and Steve should have a discussion.  Diane:  She still isn't convinced that she fully understands the problem and needs more information.  Steve:  We have to ensure understanding and coordination between both teams.  He will manage the constraints and integration between those teams.  Diane:  What is Terminology doing?  Keith:  He thinks they are doing work but the order of the work doesn't align well with what we wanted.  Diane:  We need to come to agreement on what we think the fundamental problem is.  There is a 3-step process to create the KNARTs and the contractor must fulfill those steps.  There is a term that is meeting every other week and this is the type of conversation they should be having.  Claude:  We can't just do a KNART for phenomenon measurement.  Keith:  He agrees, and what he is advocating for is a balance where 50% of TSRs are related to a phenomenon and 50% to orders and it should be hard for the SME groups. 
  • Catherine:  There are definitely KNARTs that are only documentation templates as well as ECA rules.  The problem is getting these back to B3 after we have reviewed them.  Diane:  There is a delay in the feedback to B3 from the Program Office regarding the content.  It's not due to performance of her team but the amount of revisions they are having to make.  We are on a learning curve.  She agrees with Catherine.  Keith:  He is talking about something that isn't as time critical as what you are describing.  On a TSR, we need an early balance between order set and documentation templates.  Catherine:  What is the source of the TSR you have received thus far?  Keith:  We need to say that both of those KNARTs need to be prioritized and not just order sets.  Claude:  If we want to focus on phenomenon, if the Women's  Health white papers can be made a priority, then they can work on them.  The problem is we have the Chest Pain white paper that we started with.  Keith:  This same small group is scheduled to meet on Friday.  Let's end this call and discuss on Friday.