Early Lessons Learned: Data Sharing Transcript

July 25, 2018

[Ryan] Alright folks, we´re going to get started. This is Ryan Ashe, Director of Medicaid Payment Reform at the Department of Health, and I´m joined by Rachel Hajos and others from the Department of Health. I want to welcome and thank Dr. Vasquez and Dr. Brill for presenting today. This is the second presentation in the Early Lessons Learned webinars that are rolling out, as part of the VBP Pilot Program. I touched on this a little bit in the first presentation, but a tremendous amount of work and effort has gone into the early adopters of Value Based Payment arrangements within the New York State VBP program both internal effort but also effort on behalf of the providers and payers that have moved forward in VBP. Moving forward to the next slide to show what I think most folks already know is the timeline, and I think, as I said, everyone really knows this and is working to this timeline, but we just passed April 2018 which was our first key milestone in VBP which is to move at least 10% of managed care expenditure to Value Based Payment arrangements. Now we´re working towards April 1 of next year where again we will work towards the next key milestone to transition at least 50% of managed care expenditure to Value Based Payment. These webinars are extremely valuable. I think the education and outreach that the department has pushed out based on feedback we´ve heard has, has been helpful, but I think this really rounds it out and, and offers up the opportunity to hear directly from the payers and the providers who are the stakeholders implementing VBP and making it a reality. The focus of today will be on data, on data sharing, leveraging data for, or to support the VBP arrangement. There´s a lot of interest in that area so I think this´ll be extremely helpful, and again thank you Dr. Vasquez and Dr. Brill for presenting. I´m going to stop there so we don´t take up too much more time and turn it over to Rachel Hajos who will touch on a bit more about the pilot program.

[Rachel] Thank you, Ryan. Some of you were on the phone with us last month, but there are a fair amount of you who weren´t, so for the newcomers we just want to, review or introduce and make you aware that the VBP Pilot Program is one of the initiatives to help the state reach the April 2020 goal that Ryan just discussed and described. This program launched in the Fall of 2016, so coming up on two years, and it was created with the purpose of generating energy in the movement from the Fee For Service system to the VBP model. The VBP Pilot Program is comprised of six different providers and eight unique MCOs who are piloting three different arrangement types, that of HARP, IPC, and TCGP. The series of Lessons Learned webinars that we kicked off last month and will continue to roll out over the remainder of the year capture the experiences of those involved in the program. The early lessons learned from each of them are providing intelligence to the state and all of you on the call today who are likely at varying stages in the VBP movement. You´ll notice on this slide that the pilots are required to get an approved Level 2 contract and report on quality measures. It may seem easily achievable to the bleacher seats, but to the professionals on both ends of the field the research, analysis, security compliance and negotiation alone that goes into the end result can be more complex than a complicated surgery. Last month we focused on contracting, but before the parties can get to contract there´s a long list of data readiness must–dos. GBUACO, the Greater Buffalo United ACO, and YourCare is one of the pilots who have blazed the trail of data exchange and joins us today to share with the audience what they did along the way to get where they are today. These webinars are meant to be a learning opportunity, and as such we encourage you to ask questions. Throughout the webinar, attendees are invited to ask questions using the Q&A function on your screen. We will respond to questions at the end of the presentation. Now at this time I am delighted to turn the webinar over to Dr. Raul Vasquez, Chief Executive Officer of GBUACO, and Howard Brill, Senior Vice President for Population Health Management and Quality at YourCare Health Plan. Dr. Vasquez, we are transferring control of the slides to you to be able to advance, as necessary, and we are also unmuting your and Dr. Brill´s phone lines so that you may both speak simultaneously.

[Raul] Thanks for having us, okay. I think we´ve all had a really nice time here in this interaction. We´ve never had, I´ve never had a close relationship with a company like I have with YourCare, and I think both of us have learned quite a bit. By working together, we actually can achieve a lot more than separate. So, one of the things that we wanted to kind of show, is the quality measurements, because when we started this whole process one of the things I´ll explain with the next slide is that it was just so much stuff that we were trying to put together. We found that if the quality measures were not integrated in any way, the shared savings really wouldn´t be something that we would be able to achieve. So, we´re trying to merge the clinical data in an intelligent way and really trying to use that information to help the practices transform, in terms of their workflows. So, when you´re looking at this particular slide which is what we designed, on the left side you´re seeing the provider groups. Now, the interesting thing about our value base is we were trying to bring in ten thousand because YourCare was new to the area, so in order to have the threshold for the value base we brought in different groups. So, on the top we actually have an IPA, group of eight providers, we had a big medical group, another medical group, and then we also have NF2AC which is Jericho. On the bottom part where you see the two separations, we have two hospital clinics, so we have Kaleida clinic and an ECMC clinic, each one on a totally different electronic medical record. As a result of that when you´re trying to figure out well how do I get all these guys, how do we get all this data in one place to make it work, and do it fast enough so that, you know, this thing becomes practical. Now, the nice thing about the IPA and Jericho is that they that would help us kind of really do a lot of things. We work with a company called Clinogens as a data aggregator and they already have an interface that allows for there to be daily data pools coming in, and we could pretty much work with the data to filter it and make sure we were getting the right things within the system. We were able to track and map some of the documents in there too. Now, the two areas that were difficult were Kaleida and ECMC because they were on a different server. There is All Scripts and medical, all these different electronic records and in terms of creating the interfaces, it was going to take us quite a bit, and some of those venders are not always sort of user friendly to do this. So, what we thought was, well you know what, if I can´t get the data from the clinics maybe at least I can scrape some of the data out of the hospital systems that are already porting information in. So, we actually created a connection through the reel with the right agreements in order to scrape. Because what happens a lot of times is the hospitals are billing global, that´s got to be built, there´s certain things that are just not being built, but if we´re able to get that information from the reel we were able to then import it into the Clinogens ,you´re looking at Clinogens on the top, YourCare was able to dump all their unfiltered data in there so we were able to kind of play with it and then use the Clinogen data aggregator and use Clinogen´s population to, Qualimetrics is on the right side. So that´s how we kind of play with it, but what was nice about the Clinogen´s aggregator tool was we were also able to kind of bridge when gaps occurred. So, let´s say a mammogram was done in a hospital, in an office setting, that wasn´t tagged, and I think that´s the biggest problem is filing is taking place within the electronic records but people aren´t tagging those documents, so not making it a colonoscopy or mammogram, which then you would know right then. If they were able to find mammogram, Clinogens was the source of us testing that that test was done and that information was transferred over to YourCare, and I think that that was a very valuable piece, and then the core metrics as well. We´ll talk a little bit more on the next slide, but it gave us a way to really analytically look at different aspects of the whole systems, and Howard do you want to give them an idea of how the Clinogens poured and the real data coming in actually made a difference with some of the things we were doing.

[Howard] Yes. So, I want to emphasize something that Dr. Vasquez had mentioned and really kind of call that out. The data flow that we have here is integrating both the health plans administrative data and clinical data coming from the EMRs. In addition, there´s other data sources, state data, and, uh, lab data, um, that the MCO, YourCare, is also pulling in. so that, one of the key pieces here is the integration of that clinical data and administrative data sources. The challenge in the absence of that is up from a practice perspective is that the EMR data, while extremely detailed and important, doesn´t include all the information about what´s occurring with a patient or a member. At the same time on the administrative side, there´s claims lag. We know that there´s issues with getting particular pieces of information appropriate ways so sometimes you miss activity that´s actually going on at the practices. So, this data flow allows us monthly to integrate administrative and clinical data. Now, in some of the details of this, you know we have a lot of arrows on this diagram. There´s probably actually a few more that would make it even more accurate but the key features of it is that the EMR data is being sent to the aggregator, Clinogens, as well as data flows going to the healthy light, and those data flows get combined with administrative data that YourCare is also sending to Clinogens. In addition, YourCare is sending eligibility data to the QE so the QE can pull the correct data from their data sources to be able to send to Clinogens. What happens then is that Clinogens sends information both back to, the ACO, GBUACO, and to, uh, YourCare. Uh, GBUACO then uses that information to provide a variety of different reports and analyses to practices, some very high level as you´ll see but some very detailed also. At the same time YourCare also provides high level information to the ACO, about the overall progress of the ACO against targets, and again, is combining information from a variety of different sources. Of course, ultimately, this information needs to get processed according to HEDIS specs and ultimately submitted to NCQA in the state, for compliance with quality measures that the state has established, and so we take this combined data and feed it through, our HEDIS software. The software then calculates the HEDIS measures according to NCQA specifications and those measures are enhanced by the supplemental data that we´re able to pull from the EMR, the EMRs and healthy link, um, data. As you´ll see in the upcoming slides, a very important part of this is a legal framework allowing all this data to be exchanged, and of course as we go on we´ll talk about some of the challenges as well. Dr. Vasquez, please go on.

[Raul] Sure, okay. Yes, so again, like we talked about, there are multiple EMRs, the reel, just figuring out the right agreements and using the right aggregator tool. The legal framework was tricky in the beginning, so trying to get the BAAs in from here to there where we went to the to the healthy link, the reel, you know we had BAA between GBUACO, between YourCare, between the provider groups, so that´s a lot of BAAs, but once that piece was out of the way everything worked out pretty well. The next stage was using Clinogen as a third–party aggregator tool it gave us. Like we mentioned before just to kind of expand on it, it gave us where the care coordinators or the PHMs could actually have a place if they found something to kind of attest for it so that it would get back to, to basically the MCO or YourCare, in terms of having that information. In time an area that we found was the gaps, and we´ll show you later on in the slide, was blood pressure, I mean not having blood pressure transfer at all even though it´s being done at all levels, it´s a challenge for us, and we, and Howard will expand a little bit later on that. The core metrics tool is a really cool tool because it took all this aggregated data and allowed us to look at a lot of different things, and I think this is what the tool that we use to really use our team to work around. What I´m saying about the tool was we were able to get not only pharmacy data but adherence data from YourCare, and I think that helped us with the quality metrics because you knew with certain medications that we were giving maybe 30 tablets and 90 tablets made more sense to make the core metrics, we were able to talk to the insurance and really work on ways to kind of ship that out. On the other side too, that was very helpful with the data coming in is we were also looking at pharmacy cost. We were able to change eight drugs and really save close to 2.3 million in the first year and it was minimal substitution, so weren´t even big substitutions. There, again, looking at the data was really important. In the hospital where you would know what the frequent diseases were, where the hospital leakage was, and it was interesting because we were part of one system that had a lot of leakage to another system and they didn´t even know that. So, we were showing this information to the hospital and they were quite impressed. I think for us too is the predictive cost of using this system was really good, because the case management through John Hopkins was already built into it, so we could actually look at the numbers that were costlier, the pharmacy, the hospital areas, and use our teams to really work around that area. The avoidable ER, and I´ll show you some slides on it, they were using NYU criteria and that gave us a really good perspective as to, you know, who should be in that emergency room and not, and talking to the hospitals we started to create a diversion type of plan to move us forward. You´ll even see within the system, which gave us an idea of what specialty should be ordering what at what time, and that´s a feature that sometimes as providers we don´t have. The unique thing about this is that this system was put together by a provider, me. You know, I´m the user, and I think a lot of times you have administrators for hospitals that are doing that but not the guys that are touching the field. That was important to me. The ER utilization, the super utilizers, really creating the groups on the diagnosis really helped their team really do a lot of the things that was important. This is what I´ll say, I´ve got to tell you that on this is a particular slide on the QMX Tool that kind of gave us an idea of where to go so on the whole team it said you had about three million dollars in opportunity and these are your categories where you see that you should basically go after it. We talked about the ER visits, that´s using NYU criteria. We can actually click on the 121 million and it would go to the providers, it would go to the patients, to the diagnosis, to every aspect of the cost of that group. The super utilizers, what we ended up doing was actually putting two personal health navigators that did a lot of the care coordination to address that particular issue, and again knowing what you have in terms of data allowed us to really use it when we met with our teams. Now, the nice thing about this is we were able to go by a specialty and figure out what risk scores that individual had, what was the average patient risk, and throughout this whole thing, because we were teaching the providers about coding, now ICD10 coding not TPT coding, because it´s you know, physicians are always concerned about TPT coding not the ICD component of it, so we were able to kind of help them kind of change the way that they manage these patients. Then you expect them to kind of follow through the different comparison, because we had a point where we were comparing efficiencies of care with low performing providers and that was, that´s one of the features we totally used. This one again looked at avoidable admissions, but here you´re able to break down the facilities where a lot of the activity´s taking place. You´re also looking at the providers and which PCPs had, you know, higher entity so we can really focus attention, but if you look on the lower part of this thing you´re actually looking at the way that the members were there, and here you´re actually seeing what each member that you clicked was giving you in terms of the avoidable ER. So, the green means avoidable. If it was a laceration, viral syndrome, gastric, all those were avoidable. The actual hospital tries to create diversions within the ER, because the interesting thing that we found was that these individuals were actually going to the emergency room at the time of clinic visits which wasn´t a thing that we wanted to see. So here on another slide of the QMX Tool, there´s looking about KPIs but this is what we were talking about in helping the specialist´s kind of get better. So, our cardiologist was here compared to the general within that specialty, and a lot of the misdiagnosis for ICD10 are really from when they don´t check hypertension, heart disease without heart failure, because a lot of these individuals don´t just have hypertension, they already have LDH and they have to be treated a little more aggressively including, and, a lot of times they´re managing these patients but this is the coding they´re using which is not helpful. Again, the procedures help because we learn from the different teams what was being done, what made sense to do and what didn´t make any sense to do. Putting together a value base for us was really important, because what that allowed us to do was to really use different formats. So, we had dashboards, we could run reports, our teams were helping the practices on a daily basis. We have the next slide on Tableau, we had a Tableau server, and with Howard´s team he was able to filter it and give us a lot of information on the Tableau, and that was used more for a lot of the meetings that we had. When it came down to more population cost models at the ACO level, we were using the QMX Tool, and I think that became quite important, but generating the reports, guiding people through that process so they weren´t intimidated by their data, because data does intimidate people, and too much data isn´t good. So, this is an example of the tableau that´s in our servers, and, Howard basically would generate these things for us, and we would just filter them through and kind of work out, so these are our quality metrics, you know, for that time period. The blue actually is where we need to be in terms of the metric, and the green is where we are right now. Again, the \meta hearings helped with a lot of the features that we have. Stem was just one that was changed over, so there´s been a lot of focus, but you can tell in the blood pressure and Howard, you want to tell them a little bit of what we came together in terms of the CPT codes? Because we were able to generate it from the electronic record, when the vitals were being done and then export it out as a bill with the Tax Payer ID code to YourCare. Would you like to tell them more?

[Howard] Yeah, yeah. Yeah, so, a few things here. So, one of the things, as Dr. Vasquez says, is that the quality metrics tools are extremely helpful for them and it´s organized in a way that works well with them in understanding the key cost drivers for the ACO. We provide a tableau work package that we can modify very, very quickly so we can address specific issues that, come up with the ACO in our, in our meetings that occur every two weeks. What you´re seeing here is a particular set of views that we have which compare the actual scores for different measures to the contract thresholds for different periods of time. So, the graphic can of course be drilled down to different practice groups and to different providers and show how the ACO and the groups within the ACO are doing against contractual thresholds. Some of the other panels in the view indicate information about how things are changing over time and it also provides essential gaps of care. You can see all the way down to the member level which particular members are not compliant or missing services. One of the things that Dr. Vasquez pointed out which is very important is that the administrative data doesn´t have some information that´s critical for some of the important measures, such as blood pressure, in particular, as it´s an area that in the HEDIS specs requires chart review. This again is where the information coming through from the clinical systems, the EMR systems, is extremely important. As you´re also aware all the pilots are involved in a VBP pilot, and one of the things that we´re doing right now together is looking at several different modalities for collecting blood pressure information. One of the ways is through CPT codes and in addition to that through the EMR we´re also able to get information through those too so again, in terms of the pilot and the experiment that´s currently going on with VBP, we´re going to be able to evaluate how much information comes through CPT coding versus how much information comes over on the EMR, and again, what´s nice about having extremely flexible analytical tools is that we´re able to address very detailed specific issues, as they come up.

[Raul] Hey Howard, do you want to expand a little bit of when on the information we were able to kind of use to bridge the gap, like you talked about with the HEDIS measures…

[Howard] So none of the areas that that´s had, I think the most impact is on each of the A1C values. We were able to get that information both from the EMR data and from the data coming from healthy link. So, the two hospital systems that are not using the same systems, they are sharing CCDs with healthy link and healthy link then provides those CCDs to Clinogens. We then are able to pull off information from that and use that as supplemental data in our calculation of the HEDIS scores, particularly important in the ones that are not well covered by administrative data.

[Raul] And so, you know, the challenges that we encounter, again, was the fact that there was a lot of EMRs, the fact that, you know, when they got documents in there wasn´t a lot of tagging or structuring, putting the data inside, so later on a lot of the EMRs allow you to create formulas but you got to have something to draw from, and that was one of the difficulties. The other thing was the organizational chaos, and we have a lot of local state things going on, and so a lot of these hospital structure are absorbing practices, so that was something that was difficult to kind of keep people on task. But our value–based team did a pretty good job. What helped there too was because we were able to drill down what we needed, for example mammogram, instead of having everybody run through the mammograms we signed an agreement that we can basically have that list sent to an imaging place and let them go do the searching. For the colorectal, the same thing, so we were trying to narrow down the scope because we´re talking about 100 to 120 thousand people. This was an easier approach, and then we would meet as a group and really go over the data and I think the last piece was with Howard commented on in terms of the auditing. Did you want to add anything else?

[Howard] Yes so I´ll go into that, because I think that´s, that´ll be helpful information for people in the audience. Also I do want to state that there´s not much of this data exchange with a straightforward set of standards, and so there was, there was significant work involved in being able to consume the data. In particular for the clinicians´ data, the data´s formatted in XML files so we did have a staff person on our team who has expertise in XML, and he was able to work through parsing that information and then reformatting it so it could be accepted in the various other software systems including the HEDIS software for use in that system. Based on a variety of comments that we´ve had from other people who tried going down this route, we knew that we were going to face significant challenges in the NCQA audit process around getting supplemental data approved as standard data. So, we understood very early on that that was going to be a challenge, so we worked from the start with our NCQA auditor to understand what her requirements would be and what kind of materials that she would need to see in order to support approval of the process which involved very detailed documentation of the data flows, of the data formats, and the process which the data was transformed in various steps. So, we provided this information very early on in the process. In addition, the auditor went through a detailed review of records also to validate that the information was being received and processed correctly. So again, one of our recommendations is as you go down this path it´s extremely important to work with the NCQA auditor very early in the process so you can get the appropriate documentation and testing information available for them to accept the data. I would say that, I don´t want to underestimate the challenges and being able to consume the data but it is a doable process.

[Raul] So that´s, that´s our presentation. I guess we´re open for questions.

[Rachel] Okay, so just checking that both of you can see the questions to the presenters that are located on what would be the right–hand corner of your screen.

[Howard] Yeah, so, I´m looking and it looks like the first question here though is directed at the Department of Health. It says; "When does New York State expect to implement VBP for consumer directed personal assistance program?" Uh, and the second part is; "Are managed care programs currently allowed to partner with CDPA providers to implement VBP, or are both parties need to await DOH guidance?"

[Ryan] Sure. We´ll take that one back and follow up on that question and provide some more specific guidance to that question. So, we´ll follow up with that.

[Howard] Yeah. So, the next question is appropriate for us. The question is; "Do you access DOH MAPP data at all for this pilot? If so, what data and how? Thank you." So, there is actually a little bit of a MAPP data that we are using, and that´s about health home engagement, and so we do include health home engagement in the Tableau data that we send to GBUACO based on what we´re seeing for upcoming initiatives, our use of that data will increase. Right now, it´s limited to identifying health home engagement. On psyches right now. The next question is; "Is psyches data incorporated into–"

[Raul] Yeah, Howard, let me just interject, too. What we were trying to do too is we were trying to tie into the server data, and that just became unbearable because we were trying to see if, if we could get PHI data just based on the clients we were watching to be able to draw that in and really find gaps, or areas that we didn´t have any information on, so we were trying to do that. The Department of Social Services, we would love to get some of that data, especially on the Social Determinants of Health, to use within some of the things that we did, and we weren´t able to kind of make those two connections. I just wanted to say that, Howard. I´m sorry.

[Howard] No problem. So, the next question: "Is psyches data incorporated into the data aggregator?" Currently no, it is not incorporated into the data aggregation. That´s something I´m looking at now, but that´s definitely not something that is in the current data flow. The next question is for DOH; "In September 2016, DOH indicated that it would be rolling out a more robust MAPP VBP data warehouse. Was this released yet? If so, when? How can we get more detail on this new MAPP delivery? Thank you."

[Ryan] This is Ryan Ashe from the Department of Health. So, there have been releases of the MAPP Tool over time that are specific to DSRIP, and then we are continuing to build out and design what MAPP will look like for Value Based Payment. We´re starting with the pilots to prioritize the Value Based Payment pilots in the initial rollout and we hope to have access to MAPP for pilots completed this year and then there´ll be rollouts thereafter that will encompass a broader spectrum of VBP contractors and MCOs. So, the short answer is that this year will have an initial rollout and then there´ll be more rollout of MAPP thereafter.

[Howard] So, the next question is; "Could you please discuss how the pilot addresses substance use treatment data was integrated while remaining compliant with 42CFR Part 2?" So that´s a very good question. It´s a question we will be exploring more deeply as we tackle some program initiatives around substance abuse that we´re developing. Within the current flow the data we´re exchanging is very limited. It´s essentially administrative data indicating payment for services, and the exchange of that data is covered, in the agreements between the ACO and the health plan, but I do agree that it becomes more complex as we move deeper into actual treatment data.

[Raul] Yeah, and Howard, one of the things that we were able to see on the core metrics based on information we got, especially on the avoidable ER visits we saw patients going in and when you drilled down to what they were there for it was opiates. They were there for opiates or they were being managed opiates we saw. They gave them two pills, so guess what happened on day three? They came right back. So, you know, that became an interesting model, and that´s why we´re working with YourCare and proposing in our piece of the state to really manage this in a different way to address that, but that was something that popped up, and it was quite costly to the system. In light of what we saw too was one of our cost drivers within the system, number one was schizophrenia.

[Howard] So it´s very clear that behavioral health, both serious mental illness and substance abuse are very crucial areas. As we´re moving forward we´re working closely with legal counsel and, and trying to address the 42CFR Part 2 issues, and again and an area that I think the state can also, help provide greater clarity for too. The next question is; "Can you share a sample of the BAA?" We certainly can do that & I´ll just say that we follow, very close to the model BAA that is in the regulations. Again, another question on 42CFR Part 2; "It would not seem that only a BAA is sufficient. Is it true that there are issues around 42CFR Part 2?" Again, within the context of the health plan and the ACO and the limited information that we´re providing in terms of what, what´s being paid for, um, we believe that, that´s covered, but I agree that you get into more issues as you provide more detailed information. The next question is; "Since the reel is a QE and not a covered entity is a BAA or DUA most appropriate?" So, the reel actually, the QE in this particular case had participation agreements, and the participation agreements actually have there. I would say they´re broader and more complex than BAAs, and they basically have a set of participation agreements both with us, with the ACO, and all the different participating providers. So again, on that the QE healthy link established a structure participating agreement that they use around that data use exchange. We´re at another question on 42CFR Part 2 data. Again, in the exchange between the plan and the ACO, the focus really is on, administrative data and cost data. I agree that it´s a major issue in how to exchange that data. Next question is; "Is the data presented just Medicaid VBP pilot members or all payers? It´s just Medicaid VBP pilot members. Have you started to think about how data exchange or sharing will happen with CBOs that will be incorporated into VBP contracts? Have you started to think about how data exchange or sharing will happen with CBOs that will be incorporated into VBP contracts?" You know, that´s a great question. We have not started thinking that through.

[Raul] It´s a hard, one of the things that we did, was talk about really using the Clinogen´s aggregator tool, because in the care coordination which becomes sort of a health home, this could be a source where the CBOs can actually enter data. You know, they get permissions to do a certain amount of things, and that data can actually be pushed to us, so whether it´s the blood pressure or whatever, and remember we´re trying to also look at patient reported data from the electronic side and being able to pull that in there, uh, whether it comes down to other screening tools that we´re using and they may be able to use this as a form of data coming back to the MCO. That´s what we´ve been exploring with Clinogens.

[Howard] Yeah. So, the next question is; "Can you speak more about the NYU criteria?" I can talk a little bit about that, Dr. Vasquez might want to talk a little bit more about it, but what NYU did is they did a study where they looked at a variety of different codes for ED visits and looked at the diagnostic codes, and then compared that to what was in people´s charts, and they came up with a probabilistic set of formulas for what the underlying cause of the ED visit was. Which then gets applied or projected into the particular data that you have. So, it´s probabilistic, it´s not going to be perfect, but it does have a value of showing places where you´re going to have hotspots where there´s issues, where there´s particular, members or providers that seem to have a higher proportion of ED visits related to diagnoses that are potentially treatable in a practice setting. So, it´s not a perfect system but it´s an attempt to understand hot spots.

[Raul] I agree with Howard.

[Howard] The next question is in terms of unstructured data, and I´m probably not going to be able to answer this one. "In terms of unstructured data, what did you do about this? Did you set up work groups to determine how you´re going to measure and what you´re going to measure? I ask this because BH organizations, so you can imagine there is a lot of unstructured data." I would say that for what we presented today there really wasn´t use of unstructured data. What Dr. Vasquez was talking about is that in the EMR, and the EMR data can be more effectively used if there´s appropriate tags for information set up. I mean, maybe you can talk a little more about that process, Dr. Vasquez.

[Raul] Yeah, sure. So, I mean, it´s how information comes in, because remember things are coming in through a fax, tag or an image tag, and when you get the documents in there, if you have a mammogram and you just file it in not really tagging it into… so we create documents that say mammogram so then later on if we have to create a logic formula in the background we can actually say, you know, who within the state does not have a mammogram, this is the actual document code, and it´s easier to track. We do it for colonoscopies, but there could be things that you don´t think about. When you´re running reports in the background you actually know that things are being done, and so at the chart level with the electronic records. You have a DMHM report and you´ll know within that patient all the logic stuff whether it´s chronic disease based or preventative services that needs to be done. But again, it has to be put in the right way, because if you just, you know, garbage in garbage out. It´s got to be in a way that the system can digest, and later you can use it to really run some intensive formulas that could be run on a weekly basis creating triages, sending patient, patient portals, and doing a lot of stuff in the background, machine learning. It really requires you to kind of set up the right way, and our teams are doing that because it became easier with one system. The one in the hospital is really difficult to do because, you know, that was a test beyond what we wanted to take.

[Howard] So the next question; "What was the timeline for this project? How long did it take to create what you´ve shown us today?" So, I would say that we actually started doing work in early 2016 for the project. I would say a year, year and a half, timeline.

[Raul] I agree.

[Howard] The next question is; "Once you have the data, how do you make it actionable within the system?" You know, Dr. Vasquez probably can talk a little bit more about that but as he was describing his teams are presenting this information, to practice groups, and the team, the data is actually being presented in multiple different formats, because what works for some people doesn´t work for other people. In addition to that, there´s some higher–level things that have been done both at the ACO and plan levels. One of the things that we´ve seen for instance is that there is an impact of health home participation so there´s been work done on both the plan and ACO to increase health home participation. Dr. Vasquez, maybe you want to talk a little bit more about that question?

[Raul] Yeah, so I mean, the nice thing about is that because we can go in and kind of create the documents that we want them to kind of track especially for the quality metrics, and the formula to somebody else. We have the formulas from one place and we drop them another and then from there we can actually say okay, who do you want to assign to take care of this? John will do it? So, we would have that report run on a weekly basis in the background creating a triage for John and getting a notification or having it used to reach the patient. So, we were trying to do as much as we could in the background, because I want doctors to focus on what they are supposed to do: patient interaction. Then have all the stuff taking place in the background, and I think that worked out well for them, in terms of how we designed that model.

[Howard] Okay, and so the next question is; "It looks like you have one main payer and one main ACO in your area." Well, not in our area but you´re right in terms of the relationship of its ACO and payer, YourCare, in this relationship. The follow up question was; "For those of us in a more crowded region, how can we convince payers to give us more actionable data and better analytics?" I mean, frankly, that is part of the reason of a pilot, right? And part of the focus in having this kind of relationship is that it´s a lot easier to work out the kinks and problems in this more simplified set of relationships.

[Raul] And I´ll mention to them, because what we did was we actually combined committees, we had a set of committees. Because I think that helped to really get us all on the same page. So, we had players from your side, players from outside–our side, kind of deciding different things. You want to describe the committees?

[Howard] Yeah, there´s, there´s actually a very interesting committee structure. So there´s a joint set of committees between the plan and the ACO, and, you know, depending on the committee they are meeting every two weeks or, once a month but the committee structure involves both data and analytics enrollment membership medical management and quality, financial analysis and review of financial progress, and in addition there´s a steering committee also. So, there is a, there is a rich set of committees that are involved in developing this relationship. Dr. Vasquez, anything you wanted to add to that?

[Raul] No, I think it allows the dialog to take place, so you know, one of the areas that we´re looking, to get into is tele–health. I think it´s really practical using an aggregated model, but this is something that, you know, hasn´t really been done, so it allows us to kind of say listen, we´re actually looking at the resources that we can create through savings for using these types of vehicles, and the insurers are listening. I mean, it´s not like no, absolutely not. They´re listening, especially when we went with the hearings and said listen, we want to move from 30 tablets to 90 tablets. There was no resistance. It was like okay, let´s do this, and I think that´s, that´s a dialog that really needs to take place in order for us to achieve any value in a value–based pilot.

[Howard] The last question that´s kind of a proprietary question is: "What were the shared savings?" You know, I think what I can kind of share about shared savings is what we´ve seen is, as we start our second year, is us getting closer to achieving the objectives that we had. Obviously, it takes time for this to get up and for processes to get into place.

[Raul] One thing, one last thing, for the first year of one of the nice things was is because of the working in terms of an arrangement with the insurance company, we were able to actually get 571 thousand dollars in terms of dollars back to the physicians or the PCPs that were providing the services, so depending on the acuity, the lives that they had attributed to, when you actually got to dollars in the first year, and that was good, I mean, it was a way to kind of reward the PCPs, because they´re putting a lot of work into this, but this year I´m hoping that we´re going to really do quite well, so we´ve got to cross our fingers, but everything´s looking pretty good.

[Rachel] And gentlemen, that does, wrap up the Q&A that had been submitted from the participants. I´d like to thank you all for your engagement and your interest, and the thoughtful conversation that has just taken place, and to Dr. Vasquez and to Dr. Brill, thank you, to both of you, for the wealth of information that you shared, and especially the considerations people on the phone want to give extensive thought to, and the speed bumps that people might expect to face during their VBP journey. To everyone on the phone, we thank you for your attendance today. The Early Lessons Learned series will pick up again in September. At that time, we will focus on quality measure testing, stakeholder engagement, and CBO collaboration. So, at this time, again, thank you to our wonderful presenters, our hundreds of participants on the call, and I wish you all a great rest of the day, and enjoy the rest of everything that summer has to offer.

[Howard] Thank you very much.

[Raul] Thank you very much.