March 22, 2011 CBCC Conference Call
Community-Based Collaborative Care Working Group Meeting
- Mary Ann Juurlink
- Richard Thoreson CBCC Co-chair
- [mailto: Ken Salyards
- Ioana Singureanu
- Serafina Versaggi
- Tony Weida
- Craig Winter
Agenda Items to cover)
- (05 min) Roll Call, Approve Minutes & Accept Agenda
- (15 min) SHIPPs Update
- (15 min) NQF measure comments
- NEW PROJECT Proposal HL7 Confidentiality Codes - review and update; Confidentiality codes - topics for discussion in May 2011 meeting; preliminary discussion needed to structure – Suzanne, Mary Ann, Jon Farmer
1. SHIPPs Update Richard – I wanted to take stock of where we are; we are still targeting the public comment response and we have the SHIPPS ballot, right? I want
Ioana – I have an update on that for Don – he will need that content before April 1st. I forwarded the functionality the supportive functionality to Lynell (?)—I may have to go back to the list serve to find the information of whether it will or not be a part of this ballot. Part of our priority is to provide a change request with a supportive function section in the EHR functional model (version 2) to include privacy consent management. So what we have is the information analysis for the NQF measures and the information maturity discussion which comes around HIT policy committee argue about—structure content, vs. coded content, local terminology etc. which we have in this ballot.
Richard – I want to describe the model architecture which we talk about when describing the NQF measures (attempting a PowerPoint). Let’s spend some time reviewing what’s going into the NQF and how we are building on that and using it in the ballot on what we are about here.
Ioana – as far the changing of the measures we will have to engage NQF directly; what we are doing here is to model for the content of the measures. The next thing that would be relevant is to try to draft that would be outcome based.-new eMeasure and propose those to the bodies that develop qualities measure based on clinical evidence and what is the process for NQf to sponsor that
2. NEW PROJECT Proposal HL7 Confidentiality Codes - review and update Confidentiality codes - add as a topic for discussion in May 2011 meeting; preliminary discussion needed to structure - Mary Ann, Jon Farmer, Suzanne
Overview: It may be helpful to break the value set into about four - so that implementers are more likely to implement all aspects or can at least meaningfully claim which ones they implement.
- “levels” (Low, normal, restricted, sensitive)
- “sensitivity content category” type or “rationale” (PSY, HIV, SDV, VIP, Taboo) …. Segments
- “role” (clinicians only) … RBAC roles
- “policy model” (individuals having treatment relationship to patient, as actor) – this is a “policy model”
Richard – looking at the agenda, what is the new project about confidentiality codes?
Suzanne – the new confidentiality codes project started as a conversation between Mike and Ioana at HIMSS. I feel it is appropriate that CBCC take the lead on it as we are more focused on the privacy potion and Security would be an interested party. Talk about it more, and have a project scope available for the upcoming may meeting. We will probably need to have a project scope statement prepared for the May meeting. Currently, there is not project team in place. Little to no information has been given about parameters; Mike has only touched on what he wanted to do with the value set codes.
This was originally a conversation between Mike and Ioana...Ioana made comment that the confidentiality codes were not being used the way they were intended.
Ioana - We’ve had some problems with the codes as they are now. Currently some of the codes ONLY apply to anonymized information--not identifiable information and there is combination of role-based access specific codes so it needs to be refactored. There were some issues in the HL7 confidentiality coding system it contains a combination—there are codes in there that specifically are ‘’only allowed for anonymized data’’ and people are using them for identified patient information which are not good—those codes include HIV-related, substance abuse-related, mental health related. If you read carefully the definitions of those codes, they are not intended to be associated with patient identifiable information (PII) it’s only for de-identified, anonymized information for which those are allowed. It’s a bizarre pre-coordination of codes, those are anonymized codes only—and that should become more evident. There are other issues having to do with the fact that some coding systems combine role-based access rules with sensitivity levels. We discussed a long time ago to tease out what is the sensitivity of the data—what are confidentially rules pointers, that would be part of the refactoring that coding system. Presumably we will also be doing something about these codes that tend to used a little too much for data if the data is identified; but maybe it would be reasonable for de-identified information
The current confidentiality code list/value set is substantive –the value sets that are currently in the coding system are explicitly only intended for de-identified information; the other ones are applicable for any other kind of information—sensitive or not-sensitive; so sensitivity can apply to any information, identified or de-identified. But things like condition related confidentiality codes--would be a horrible thing if they were assigned to de-identified data.
Richard - If data were disclosed they would be horrible.
Ioana – for example; it’s like writing on the outside of an envelope--- and says on the outside that ‘this is an HIV-related medical record, that wouldn’t be good idea. If you are putting de-identified information in the envelope and saying this is about people who have HIV, that wouldn’t be good.
But if you can’t tell it’s me inside that envelope, but if I know it’s about you and it’s about HIV then I don’t even need to open the envelope , I’ll deny you coverage. To the credit of those who authorized the value set—they did mention this value set is NOT to be used for patient identifiable information—but it’s being used anyway. So we either need to strike It and defer it; something needs to be done; in the US we hear from people from Microsoft that they love it—
Suzanne - We can take that statement ‘that its not to be used with PII...and say that the course of nature goes, that we need to expand this; with privacy/HIPAA rules and whatever else--the popularity of privacy and that this value set needs to be look at and revised, updated, modified.
Ioana This would be a harmonization proposal—that how the v2 terminology is maintained.
Mary Ann - Discussion from November 1 – we came up with an action item to be discussed everything but we held off until we had a face-to-face meeting (which would be in May)
Jon – it’s a refactoring analysis; it would be helpful to put some conditional constraints as to when to use these values
Serafina – we did submit this information when doing the Composite Security and Privacy DAM. At that time the proposal wasn’t considered—I don’t remember the exact thread that happened at the harmonization proposal. We certainly participated—this is still something that has to be written up as a harmonization proposal. In terms of a project—the deliverable would be the harmonization proposal.
Suzanne- I agree with serafina that this needs to be written up as a harmonization proposal.
Jon – I think that is fitting.
Ioana – this was submitted a long time ago and definitely dismissed us. We were supposed to do additional work and propose to do a more detailed analysis showing how this coding system would be refactored—we were supposed to do this work in the following cycle. This was actually an action item for the Security WG to take those codes on—and it’s coming to CBCC.
Richard – I don’t think that’s bad, it’s fair to say the Security WG doesn’t have (except for mike) a bunch of strong privacy advocates—and its better that it’s coming from our group rather than their group
Jon – I agree. If this comes from a group that is under-appreciating how disastrous it would be for an HIV-related on an envelope as a stamp—someone needs to advocate for a patient who may have a breach of privacy that way.
Suzanne - Security can say ‘yes’ you can’t see that…they can mark yes or no—but really when detailing privacy; they don’t really interested why or what the information is except that it needs to not be shared. Where from CBCC we have more interested privacy advocates that would take on those details.
Ioana – so who is going to write up the project scope statement and who is going to do the work
Suzanne – I’ll start it
Richard – there will be more resources when this contract mod is done. There is reason to do this at SAMHSA too—there are behavioral vendors who are putting together a privacy consent standard.
Suzanne – there was another project that had to do with SAMHSA that I needed to talk to you about off-line
Richard – a project appeared at HIMSS with 2.77 million dollars—the $$ was given to SAMHSA but some of the work wasn’t done and the money was taken back by ONC and routed to the existing NIEM contractors (Doug Fridsma contractors). The privacy coding value sets coming out of HL7 need to be robust.
3. NQF measure comments NQF and SHIPPS update – we’ve had our priorities, but I do want to cross our t’s and dot our I’s. I want to do more systemic, to put things in the right place to be useful for developers and want to make sure that our efforts will create the biggest bang is likely to come of these quality measure and for future quality measures—
Jon – we know there are high level limitations of the current crop of quality measure and one is population coherence. So the numbers of any measures that don’t deal with that are suspect and not useful for policy analysis or clinical decision support. Also, lots of them are for process measures for structural and very few are actual outcome measures—but they know that. Am I right on that last one? But phase iii is supposed to fix that to some degree
Richard – the narrative that we’ve recently uncovered, that they know that the focus of individual providers are misplaced.
Jon – that is a document is something that I haven’t seen yet. When you sample data from a fragmented population (fragment of a market place); say DC—there are 3 dominate provider networks, if you sample data from each of them without mapping those data points to individual real world person as patients, that’s called population incoherency. You’re drawing data separate independent measurements of the same population and they are not coordinated with each other. And our classic illustration in what I’m seeing in the current quality measures is this notion ofNewEpisode. All three providers can say they have a NewEpisode but really its only one even.
Serafina - doesn’t this in fact reflect that we don’t have any coordination of care?
Richard – I think you’re right, in the current system we cannot do it right. If we make it a big deal, there are more that 100 measures and (we’re only talking about 2-8 of the measures
Serafina – we know from their documentation that include structural measures, outcome measures, etc
Richard – but we’re only commenting on the behavior health measures—they’re process measures; in the narrative; you discovered some documentation (a guide). They weren’t able to model these measures as a network of care only as individual providers.
Serafina – let’s clarify how we’re using the term ‘individual providers’ its really individual … submitter of an individual’s data—are you talking about provider X or Kaiser Permanente? How many lone providers are out there? Not many. Those entire providers X’s are rolled up to a particular umbrella organization that is their provider network,
Richard – I don’t think we know that; I’ve seen 80-90 % of care is provided in smaller practices—it’s a very high number. I’m not an expert on that, doctors like to be on their own, but maybe that’s changing.
Serafina – I hear just the opposite--those statistics sound very rural to me—they do not sound like urban areas to me... can the construction of … go ahead of the delivery of care?
Richard- that becomes a question mark. Population incoherence we can point it out that if these measures are trying to describe the quality of a service provided by a provider/providers in an isolated way caring for the same person in a network. I though they already recognized that as a problem
Jon – the usefulness of the statistic of a small practice of 1-4 as a sample size. A larger network is a much better sample size; but it is still not a coherent population—there is incalculable overlap – no one has figured out if the same persons are crossed over in the sample. It’s not just for lack of EMPI an identifiable correlation technology—even with all the technology you cannot dictate that someone correlates of the entire player in a geographical area
Richard – we don’t have the networks to care for all those problems. We just had to work with what we had today. We can say we’re modeling for stage II or III; we understand that there are limits to what they can do in stage one. I’m not about making criticizing what they could have done in stage one; I’m more focused on what can be changed to make it better than stage II or II. --if we don’t’ do the modeling of how we compute the measure…what are the developer’s supposed to do?
Serafina – I’m not arguing that. Say we come up with the right model and the way to do it; if you don’t have the ability to go after data across the disparate organization—what’s the use if it’s not centralized in some manner
Richard- I understand that; we’re not just talking about developers of EHRs, we’re also talking about the developers of HIEs
Serafina – that’s what I’m talking about. Let’s go back a bit—I mentioned in a past meeting that there is a concept National federal claims database. That’s been stopped.
Richard – it’s being done in a number of states.
Jon – it’s mute in the free market place where you do not have a single-payer you cannot mandate 100% integrated population in your statistic.
Richard – if that kind of argument is made; what we’re really talking about here is the base, the base system. That’s where the payer is the single-public payer in the state
Serafina – Santa Barbara care data exchange. The concept was the organization of caretakers, housekeepers of the rich people that live there—the safety net population. It was an original HIE concept.
Richard – You’re absolutely right. They have to be solved. We’re just trying to anticipate the kinds of issues i.e. Medicaid is well under-funded. We’re talking about state insurance exchanges which are going to be fairly regulated—or else they’ll be meaningless. ‘’How do we know if the care is up to snuff? ‘’ so you have add these quality measure and eventually based on outcomes and we have to be able to model those people will know how to build these monitoring systems or the billing and quality measure systems
Serafina; we need to ensure that in our model that we factor in this ‘how are you going to create this population coherence’ across the networks?
Richard – I don’t’ think we need to add this in the project… we can assume that you can uniquely identify people –and in the safety net world they do. You will need to solve the coherency problem otherwise the information is meaningless or close to meaningless. It leads to this churning of services and in the end we’ve shot ourselves in the foot
Jon – I think that the key thing that NQF already knows that we cannot do best practice analysis from process measures but that by outcome measures. But no matter what analysis you’re doing, anyone using any kind of statistics has to be aware of the limitations if they don’t have 100% identity correlation.
Richard – not just identify correlation but you also have to have your selected population. (Not just selective sampling)
Jon – Yes, there’s completeness apart from identity correlation
Richard - I don’t like identity correlation because its sounds like….bad practice. If we can’t solve the issue of population coherence/selection bias kind of stuff; then the measures are going to be compromised. We need to say that. We need to say what they are useful or not useful for.
Serafina - So, where do we want to go with our comments?
Richard – why don’t we make up a summarized list and bring it forth to the group to take a look at—and say how we should be spending out limited resource and schedule. We have to coordinate our resources. We’ll need to put together a package of things; things that are a sense a part of the SHIPPS project and that work together. I want to know how we’ve set our priorities and how our next 3 weeks—where are we going and where will be at this time--To make sure we haven’t let time slip by. I also want to plug in work to possibly upcoming projects.
Serafina – with respect to the comments. There are approximately 40 that Mary Ann put together when we were doing the analysis for the information model. I added to that list, higher level comments to the re-tooling of the eMeasures.
Mary Ann – I wanted to say something about the first bunch that I did. They were detailed. We would have to pull out the documentation material and show you the inconsistencies. We have to push back and let their developers look at what they’ve done. What I’ve got form their pseudo code was very complex. It was very low detail showing inconsistency errors. It’s not high level; it’s also not comprehensive. Especially for the measure 004—which is extremely complicated. It’s enough material that to show them they need to do more work and iterations on what they’ve produced. If people want to sit back and go through the same kind of thinking I’ve done—it's very messy. The pseudo code is very difficult. There is enough ambiguity to force them to look at their work again.
www.qualityforum.org/projects/e-g/eMeasures/electornic_quality_measures.aspx?section=PublicandMemberComment2011-02-012011-04-01#t=1&s=&p= electronic quality measures (eMeasures) Public and Member comment
I do have some general comments that should go in another box that Serafina has presented. They are also looking for comments from a clinical perspective. We need to propose other new measures that might be more useful for behavior health.
Richard – I’m looking now at the higher/bigger picture. If you look at too small a spot and now you’ve taking into account if the system is going to react. That’s not NQF problem—it’s focused on ONC. Is that part of this process comment—is the measures on the table now; Does NQF feel they have the stage II measures on the table?
Mary Ann - From the behavior health perspective, you need to put the clinical comments forward. I think we have the overview and the detailed part, but we need more in the clinical part. The questions I would ask—do these measures address behavioral health? I would take the opportunity to put my two cents forward—it’s a small perspective and might work today but it wouldn’t tomorrow. I would take some of those concerns and put them forward.
Ioana – I don’t know if I agree with serafina if I need to review the comments. You have to spend a lot of time to with the measures to weigh in on the comments. It may be useful and faster for those interested in developing new measures and to propose outcome measure to look at the eMeasure guide that describes how the pseudo code works and the numerator and denominator constrains are defined.
Allow yourself 2 hours to go over the comments/documents; you’ll need to look at the guide first so you’re not baffled with the document itself. (The guide was published in December); it was not part of the comment package. It should’ve been required reading and it should be better presented on their website.
Mary Ann - This still won’t get to the point of the clinical aspect of the behavioral health perspective.
Ken – we’re working on measures to the extent that you guys are—they are at a conceptual level. For example they want an SBI measure and quality measure. And that’s about most of the thinking that’s going on. We’re also thinking about some composite measures, but if we’re going to entertain these measures we would have to coordinate with other people.
Ioana - Where we can add value is provide examples and have NQF think about their quality data model as being more of a model—that would be a good contribution we can make here in HL7, the next idea is to work with the community of interest that Ken has identified (clinical experts) to develop measures that effectively would measure and evaluate the outcomes of the real quality of care. That’s where the information that Richard has laid out would be useful—you know what you are looking for in the measure and that would be the community to raise the challenge to create those measures. NQF would need only to merely endorse them once they have been developed and the stakeholders have agreed on them.
PHQ9(?) have already been re-tooled. We have another six that still need to be worked on—we need to loop in Pat and Fran Carter and find out what all that is. We can do it at anytime. Fran is out on leave at the moment. We’ll work with them to find out the measure tooling to do. We don’t necessarily have to go through NQF and just submit it to them for endorsement.
Floyd was very generous with his time—and I gathered that you can submit a completed eMeasure and they would follow the process to endorse it and from that point on and then it would be part of the government about that measure. We do have our medical computations as part of stage I—and that’s a good thing. It’s a big step to go from no automation to something non-zero; that’s a huge step.
The comments that are provided here—do you want to do a report out for the May meeting agenda? It doesn’t have to be an exhausted list—10 minutes of depth.
- ’’ (15 min) Request for items for May 2011 Working Group Meeting in Orlando, Florida. Please contact Suzanne to add to agenda
Meeting adjourned: 12:03 PST