 Good afternoon and welcome to the Green Mountain Care Board. My name is Kevin Mullen, Chair of the Board, and we're about to call this meeting to order. The first item on the agenda will be the Executive Director's Report, Susan Barrett. Thank you, Chair Mullen. Good afternoon. I have several announcements and scheduling, including scheduling, public comment, and some job openings at the board. First, I wanna start with our upcoming schedule. Next week, we will be hearing from MVP and Blue Cross Blue Shield on their qualified health plan requests. As a reminder, on Monday, July 19th, starting at 8 a.m., we will hear from MVP. And then on Wednesday, July 21st, again, starting at 8 a.m., we will hear from Blue Cross Blue Shield of Vermont. All of these meetings are on Teams. On Thursday, we do have time held if we need additional time for the rate review hearings. That, again, starts at 8 a.m. as a TBD and a tentative rate review hearing. Then on Thursday evening, from 4 to 6 p.m., we have the Rate Review Public Comment Forum. And that is where members of the public can share their comments with the board on the rate requests, again, through Teams, and again, from 4 to 6 p.m. The other announcement is that we did receive the hospital budget submissions. They were due July 1st. Those will be, they are on the website, the narratives, and we'll be continuing. Our team will be reviewing those and putting additional information on the hospital budget submissions on our website. We will be opening up an official public comment period at the end of this month. But as many of you know and as a reminder that we take public comment on any of our work at any time. So those are the scheduling announcements. Shifting to public comment, we have several ongoing public comment periods right now. We have until July 22nd at 11.59 p.m., public comment period on the proposed rates from Blue Cross Blue Shield of Vermont and MVP Healthcare. And then we also have an ongoing public comment period that will open on June 30th and will end on August 10th regarding our data submission rules and data release rule. And that information on those rules can be found on our website and a way to provide your public comment. And then last, as I've announced on previous meetings, we have an ongoing public comment period on a potential next agreement with CMMI and the all-payer model that is ongoing, as I said, and any of the public comments that are submitted to us, we share with our partners at AHS and the governor's office as they are leading those negotiations. And then my last announcement is regarding some job openings. We have two staff positions available at the board right now. And these positions are posted on our website. One is a senior health policy analyst and one is a healthcare project director. And then I'm also pleased to announce that yesterday the position for the vacancy that we'll have in September for a new board member has been posted. That hiring and review process takes place through the Green Mountain Care Board nominating committee. And we do put that posting on our website and there are links to find out more about that posting on our website. So that is all I have to announce, Mr. Chair. I'll turn it back to you. Thank you, Susan. The next item on the agenda are the minutes of Wednesday, June 30th. Is there a motion? So moved. Second. It's been moved by Maureen and seconded by Tom to approve the minutes of Wednesday, June 30th without any additions, deletions or corrections. Is there any discussion? Hearing none, all those in favor of the motion signify by saying aye. Aye. Those opposed signify by saying nay. Let the record show the minutes were approved unanimously. So now the focus of today's meeting is to get an update from our data team and I'm gonna turn the meeting over to Jeff Batista and Lindsey Kill. And I'm not sure which one of you will start off but take it away. And we do see your report but we don't hear either one of you. You give me, Mr. Chair. I'm just figuring out how to present after I share the screen. Any hot tips will be welcome. Let's see. Four to five. There you go. Wonderful. So thank you, Mr. Chair and good afternoon everyone. My name is Jeff Batista and Lindsey Kill is joining us today. We will be discussing our approach as we enter year two of our hospital report work. We've done presentations on this in August 2020 and bid on January 21 as well. The agenda for today is the hospital report. The hospital report as a whole complements the GMCB's projects such as sustainability, atrium, hospital budgets and broader regulatory integration. Really it's just the eight teams efforts as we delve into two data sets simultaneously. The claims database speakers and the hospital discharge database vuts as well as other data sources to yield richer insights for these processes. So the research questions of this report in general are where do Vermont residents seek care at hospitals? Where do Vermont hospitals patients come from? A more provider oriented view and how do these trends impact residents and providers? Now when we speak to the impacts we're not just talking about the cost and utilization but also as we delve deeper into the data, healthcare outcomes, perhaps some gaps in where providers may need to be placed. All this theoretical work and it's a work in progress as we move forward with it. So today's agenda, we're gonna walk through the data we have at our disposal. This may be old hat to some of the regulars here at the meetings but perhaps a refresher for others. How we have used these data to date, how we're improving the data connections as well as what can we do with the data moving forward? So dealing into the first data set we have vuts. This is the Vermont uniform hospital discharge database. It began in the early 80s as I understand we have data going back to 2006 including inpatient and outpatient clay episodes and episode is essentially everything that happens prior to a discharge at a hospital. The database includes the charge amounts that's the charge master not necessarily what's paid by the insurance payer diagnoses, procedures as well as revenue codes and demographic details for the patients. It includes all payers not just the ones that pay claims this includes self pay and free care though the numbers aren't too big on that to show at a public level quite yet. It does exclude Vermont residents at non Vermont hospitals actual paid amount and out of pocket amounts for care. Patient identifiers which is pretty typical for a large data set you don't want SSNs and birth dates on that sort of stuff but we can't point out the race data gender data to varying degrees of certainty. Also it excludes clinical data from EMR and clinical notes as well as all professional payments made for these hospital encounters. I'll pass it on to Lindsay. Thank you, Jeff. Hi everyone. This is Lindsay Kill. I'm gonna talk about the vCures database and then I'll also talk about the comparison between the two which is really relevant to later slides. So vCures is Vermont's all payer claims database. What we have in terms of what's included in vCures data for 2007 and on with six months of run out. We have medical claims and retail pharmacy claims as well as the insurance payment information for those claims. So what did insurance pay the portion of that? And then also we have what we call the expected member shares. So these are values like the co-insurance, co-pay and deductible. We have insurance eligibility information and we have that information for Vermont Medicaid, Medicare, commercial insurances, the QHP and large group but a caveat here. We have only about half of the commercial self-funded market is included in vCures. And then what's excluded from vCures, what doesn't often get talked about. We don't actually have any personal health information. So we don't have name, birth date, social security number, those individual identifiers. We do not have any clinical data from the EMR or clinical notes. We don't have health costs or utilization for uninsured persons. And similarly we don't have health costs or utilization for that other half of the commercial self-funded market. Workman's compensation, tri-care, VA plans, the federal employee plans, self-pay or payers who insure less than 200 Vermont residents. Then next slide please. So I'm on slide five of this presentation. Here we're looking at a very high level of the summary of differences between these two data sets. And we've broken out some of the variables that you would find in one versus the other and the variables you would find in both. So I'm just going to go over individually the variables that we have in both because this directly impacts the research we're gonna talk about on future slides. We have in both data sets the DRG that stands for the diagnostic related grouper which is something you will find on your inpatient hospitalizations. We have revenue codes in both data sets. We have a level of aggregation for inpatient episode in both data sets. Although it's worth noting that in vCures we build this from scratch and in Buds it comes pre-built. We have the hospital charge amount which as Jeff pointed out is reflective of the like the charge master information and is not the same as what insurance actually ultimately pays. We have the payer that pays for the episode of care. So your Medicaid or Medicare so on and so forth. And we have dates of service specifically discharge dates. Slide six please. Thank you. So before we move into currently what we're working on we just wanted to give a refresher of some of the similar projects leveraging hospital and HSA information that we've done to date. And so the project that we presented before was the patient migration project. The objective of this project was to follow the residents movement in state between these HSAs and out of state or in other regions. I think specifically we called out the Albany Medical Area region around that hospital and then there's the area around Dartmouth-Hitchcock. Important to note that in the patient migration project we aggregate expenditures as claim payments. The total medical and pharmacy payments paid by insurance. And we defined a patient in that project as anyone who had medical coverage at any point in the years that are covered in that project 2014 to 2019. We have a list of the variables that are in that project and values. And just important to note that with the latest work which we'll talk about in a future slide some of this project have been expanded on because they are related. And so the advances that we're making to patient migration are the addition of MSDR gene which really helps explain a little bit of the why people travel. So the types of care that they're getting when they travel. We have as I mentioned we've added this episode logic which needed to be built from scratch for vCures. So we've added that and then we are adding care type flags to flag ambulatory care sensitive conditions and tertiary care. And on slide seven is just a screenshot of part of the presentation that already exists. If you follow the link that's on slide six you'll land on the report page for the patient migration report. And you can see and play around with all of this. And of course let us know if you have any questions. Jeff. Thanks Lindsay. So moving on to the related patient origin report. It's one might call it a doppelganger of patient migration except using vCures. So it's very much from the provider perspective. The objective is to track changes to hospitals, patients and the care they receive. We define hospitals as all hospitals subject to budget review. Patients are everyone who has received care. Inpatient care in this case with version two as we're delving deeper into the data so we're limiting the scope of what we're working with at hospital or its practices. You can note the variables to the right as well as some new variables that advance our understanding of how we're moving through this. This includes MDC so major diagnostic categories. The MSDRG is that Lindsay has explained DRG weights which indicate roughly the severity of a given DRG and the length of stay. Moving on to a visual of where we're going with this. So here I've broken down the inpatient episodes, present invets 2014 to 2019. When we add the musculoskeletal paid by Medicare so payer and major diagnostic category we see the charges and the episodes change and the relationship among the hospitals change. And then filtering that down further to the type of hospital you can see sort of it's like a map of the solar system. You have a bunch of small planets and some large ones delved into those small planets get a higher resolution view of how people are performing at the musculoskeletal paid by Medicare among small hospitals. So you get deeper and deeper detail as you move forward and we're visualizing in different ways. The past presentations have shown maps. There are of course tables and files that go along with it. And as we think about the final published product for this version two of patient origin we can factor in opinions on that moving forward. So moving on to the other sorts of data at our disposal. So we take a broad view of what's available to us here at the GMCB analytics team. There is of course firm data available that's come up quite a bit in the hospital budget process that's in our adaptive use by the finance team and the sustainability work as well. Service lines, quality, other things specifically related to the hospitals as firms and their contacts. Also we have community data. This includes the US Census data 2020 is coming out in a more and more detail every week. So waiting on that. But before that we have the American Community Survey which estimates census information in the US where they don't conduct the census. We also have other federal and state resources whether that's CDC's SVI, social vulnerability index, other metrics that you can define census tracts or patients, we turn to every source we can and figure out what's best for the question at hand. We also have spatial data. That's something we've brought up over the past year including where people live at different levels whether it's a zip code, HSA tract really depending on how we wouldn't wanna link that for any given study and how people access care. We can model drive times to primary care for example. We have shapefiles representing the transportation system of Vermont at different levels and we can factor that all in as well. There is also human services data. Right now this mostly takes the form of reports we read on say the efficacy of SASH or blueprint for health, really existing resources we can use to figure out coefficients if we're running something down the line as well as a public insights, not just other studies in the realm of work that we do but the frame that they're portrayed so we can really deliver results that suit the public's needs. This is just a brief run through of the sorts of data we have on hand. So you'll note the distribution of people with health insurance, this is from one of our visuals on the data visual page on the GMCB website. Here's some of the drive time modeling and here we see some of the demographics taken from the social vulnerability index from the census or rather the ACS breaking down data both by tract and by HSA which requires a crosswalk to calculate that that we have to design ourselves. So we've thrown a lot of data at you and it's great to rattle dazzle people but how does this all come together? And by bringing in not just the claims and the discharges we can make us and bring in other sorts of data we can create a whole that's greater than the sum of all these parts. Now these are a bit messy sketches here. I don't have an overhead in front of me but I want to imagine this as if say you're Coco Chanel and you wanna make a hope, you're working in Oak Couture and you wanna sketch a dress and it's really the first stage of where you're going with making the dress you have to consider the types of fabric you have you have to consider the sizes, the costs really what the market wants and perhaps what your predecessors have done and how your dress or fashion advances beyond what's already out there. So when we're looking at how to analyze all these data together we wanna get somewhere where we can isolate different factors of whether it's the community or the patient sort of statistically separate those or control for them and then find out how say the regulatory processes shape the relationship between charges and prices or how hospital service lines are changing or how that associates with demographic changes on the ground, payer changes, stuff like that parsing together this 6.5 billion dollar healthcare industry we have with as much precision as we can bust it. Now breaking that down into the particular databases we've started with, Buds and V-Cures I wanna focus on this impacts arrow between charges and pay boxes in the book, the two visuals. It's quite, it's a good place to bring in the next study. So how about these databases complement each other to bring us a greater whole? Here we have the study, the effect of safety net closures in conversion on patient travel distances to hospital services from health services research by Patsoli Adele. And the main question of that study is how does physical access, they use it measuring drive time, how does the closure of safety net hospitals impact physical access and the behavior of people getting to care among different parts of society? Pardon there are some school kids outside walking by. So the data consists of several states discharge data in the 90s and 2000s. So I wouldn't generalize this stuff to Vermont necessarily about the states are California, Arizona, Florida and Wisconsin. US census data from the same periods at which the discharge data was taken from as well as some AHA annual survey data on hospital characteristics. And with this, they made drive time the Y of the equation for healthcare access. And they treat it as a function of patient, hospital, community, so the census tract level stuff and healthcare system characteristics. And they ran this model for five separate diagnostic categories, including births, substance use disorder and behavioral health and three others. And what they generally find is that the uninsured and Medicaid patients experienced greater disruptions to care when the safety net hospital's nearby closed. These groups had worse access particularly for behavioral health and substance use disorder and birthing centers after the hospital closures than their commercially insured counterparts. So we're seeing some socio demographic inequalities here. And it was done with discharge data but if we add claims, we can add some value there. For example, claims data have actual paid amounts which could factor into either the dependent or independent variable as a function of access. This also includes the expected member share including the out of pocket, OPAs, co-deductibles that shape affordability. The claims data also include nearby non-hospital services for, I mean, the hospital database will not include your ambulatory care centers if they're not affiliated with a hospital. So you have your non-hospital services as long as they're registered by claims. And we can measure how the broader healthcare system sort of interacts with these processes. In our context, isolating these effects in greater detail, we can answer some questions like how do changes in physical access to a type of care impact, utilization, actual spending and healthcare outcomes? And we can draw the outcome stuff either from 30 day readmissions in VODs or some sort of V-Cures version. And related, how did these charges in physical access to care impact competition on charges price in charge to price differences? If you take away one hospital that shapes the ecosystem around it, perhaps we can identify effects there for certain line items. Example two, you have measuring telemedicine impacts with claims. And the question here, it's from someone's dissertation at the University of Minnesota. I believe she teaches at Cornell now. Her surname is Yu. And the question she asks is throughout her entire thesis is how do telemedicine visits associate with healthcare utilization, quality and spending relative to in-person visits? So it's more recent data. It's from an APCD, not quite ready to generalize it to Vermont, but it's a source of a methodological inspiration here. So what she did was take commercially insured women under 65 in the claims database with a certain healthcare issue. And she measured the difference in their spending and consumption habits, splitting into two groups, those who had commercial insurance that began to allow direct to consumer telemedicine and those that did not cover it. And this difference in difference approach allows you to see the change in effects due to the change in what insurance will pay for subject to all the other things in the model. The results, fewer services used, fewer antibiotic prescriptions and lower total spending during the 30 day episode of care. She concludes that direct to consumer telemedicine could reduce utilization in spending while maintaining a comparable quality of care to in-person services. Again, not generalizable to Vermont as far as I see it. Adding value with discharges. Now the discharge records, they include non-claimed patients such as free care, workers comp and self-pimp, but exclude telemedicine if it's not from a Vermont hospital. So there's a limitation and a benefit there. Discharged records also break down outpatient episodes much more cleanly than a claims database and by a measurable magnitude. Claims databases are quite complicated, especially isolating outpatient claims or outpatient episodes. So that is an advantage bringing in the discharge database directly from the hospital there. What would bringing in discharges to this study yield in isolating the effects in greater detail? Some questions that we could answer with it could include does telemedicine spur price competition for onsite substitutes, onsite as in the same care, but not by telemedicine. Does telemedicine adoption associate with changing charges for hospital services? What about changes in price or changes in price to charge ratio? And do these patterns very based on broadband access or socioeconomic factors that equity component, which can measure at the patient's claim or tract level, ideally tract because that's more generalizable for this sort of analysis. And what might these patterns mean for health equity in some sort of normative way? I'm going to pass it off to Lindsay. Thank you, Jeff. I'm on slide 19 now and what I'm gonna switch to talking about after we've gone through patient migration and patient origin to previous and ongoing projects and some other data sources we have and then those two papers that Jeff just reviewed for methodological inspiration. We're gonna focus now on what are we currently working on and where are we with that project? So this project, we call it generally hospital markets but it's really in two pieces. We are trying to work on database alignment which we had a meeting the other day and someone said it was a very noble goal and we're also working on migration analysis to talk more accurately about patient in out migration. So in general, we're trying to create these two summary data sets. One from vCures, the claims data and the other from VUDS, the discharge data. And we're trying to get them both on a level of where those variables are really equivalent and meaningful to talk to each other and then combine the two and eventually use this new combined data set to do a more in-depth analysis of patient in and out migration. So this combination of patient migration and patient origin and we would summarize those by types of care, the payer and of course other patient demographics. We think that if we could be successful with this project that it would be potentially applicable to hospital budget review, to CON and who knows maybe other areas of our regulatory duties here at the GMCB. There are a lot of directions we could take this project in I'm just mentioning two here. One would be flagging and analyzing expensive or complex types of care that Vermonters are traveling for. And that could be within the state or Vermonters to traveling out of state. And then flagging care that Vermont offers for those coming in-state to try to understand why do people come and use our hospitals. So just a quick high level overview of the methodology that we are using to try to combine these two data sets. The first thing that we wanna note is that for this initial phase at least we are focusing on in-patient episodes. We would eventually like to add outpatient episodes but as Jeff already highlighted, they're quite complicated on the claims side and so we thought that starting with the in-patient episode which is a little bit more clean between the admission and the discharge was a good place to start. We are using the hospital where the care takes place. We are using the MSDRG which is grouped by the MDC codes. The type of care, the weight, Jeff mentioned DRG weight which is an indicator of severity of sorts and length of stay. We are also using the insurance provider. So is it commercial? Is it Medicare? Is it Medicaid? We are using the patient origin. Where did that person come from? We do have it down to zip code but at this point we are aggregating to the VDH version four hospital service area. We have charge amounts that is available in both data sets. Recall that VUDS does not have the actual insurance paid amount. So although charge amounts themselves are not indicative of real expenditures, it is one area we can start to try to find alignment. We also are counting episodes as an indicator of volume and of course other patient demographics all as inputs to this analysis. And what we're doing is we're trying to aggregate to relevant levels of detail to display the important and significant changes over time and over time for us in this project specifically is 2014 to 2019. Next slide please, Jeff. Thank you. So now I'm on slide 20. What we have completed to date is that we've identified our ambulatory care sensitive conditions, these are conditions of interest to us and we've flagged our tertiary care episodes. We have a chosen methodology for aggregating both data sets for comparison and we've created those summary files from both data sets from VUDS and from VE cures. Next steps, some of which are already currently in progress. We have a methodology review with VOS NSO for the VUDS side. VOS NSO provides us with VUDS. So that's why we're consulting with them and then on point is the organization that provides us with V cures. So we're gonna talk with them as well about making sure that we've extracted these data correctly and that we're leveraging the correct variables. We would like to continue to test data comparisons on smaller and smaller scales to assess where there may be differences. So for example, looking at how do episodes and charge amounts differ between VUDS in 2014 at UVMMC and V cures in 2014 at UVMMC and then breaking that down smaller and smaller. Is there more difference depending on payer? Is there more difference across DRGs and looking at it through that lens? And then of course, applying this comparison to other analyses such as the patient in and out migration and potentially even price variation. If we could eventually get these two data sources to a level of agreement to quantify the difference with some certainty, then we could really talk about the differences that we're seeing in utilization from the provider lens. That's the VUDS data and financial compensation from the payers, which is through V cures. All right, on to you, Jeff. Awesome. So keeping this going with our progress, I'd like to speak to our data linkages and how we're advancing both the level of detail and the efficiency of it. So one major facet of it is that we're using census tracks more often as a standard level of community measurement. HSAs are a bit old school. They were made in 2006. The patterns that they represent generally hold true, but if we bring in census tracks, we could really bring things to a community-based level, look at those effects much more robustly. And there's a lot more data available at the track level than the HSA level, straight out of the box. So this includes demographics, the boundaries, and even populations, centroids, which means you could calculate weighted access at a pretty good level at the state level. Also, we are able to convert VUDS and VQZIP codes to tracks now. It's just standard practice and we can throw it in and line that up with columns from the census at any time or other data at the census track level. In addition, we're streamlining our data collection and analysis. This includes dropping, not going into computer programs and clicking stuff a lot, just starting to write code that could just be replicated and easily rolled out into other procedures or diagnoses. What's the reason why, Jeff? GUI, it's a graphical user interface. Okay. So your old school-like word rather than a black box where you're typing in a career new font, yeah. So- I didn't want to get lost again. You lost me at Hope Couture. Yeah, I've got to speak to a large audience here, Kevin. So harnessing existing code. We certainly harnessing existing code. It's like Lego blocks. You build a Lego block, you could add another Lego block that you just build and soon you have a big structure. And this includes the Tableau workbooks as well where data can be swapped in and out as it comes in. And we are starting to disaggregate trends by service line in other dimensions to greater detail, which you'll see in future products. In addition, we're exploring the use of APIs, application, I don't recall the acronym. It's essentially a way to grab data directly from a database online and spin it out without having to download the entire table and load it up and all that. It's very efficient, especially when it comes to uploading things at a regular basis. We could just ping the census and the new data will come in and populate the data set that we're working with. And on the horizon, these are sort of stretch goals or medium term goals. I guess you can call them. Integrate clinical data with vCures. We're not too involved with that at the moment, but it's something on our mind, or at least me and Lindsay. Getting more people involved that includes democratizing the use of public use files, breaking more of those out for other groups, whether you're at a college or you're an advocate or whatever, to do some analysis by themselves and share that into the broader public sphere of what healthcare is. And connecting this effort with other ongoing data quality work, particularly the broad data validation project which is going on in different dimensions at this time. So I'll pass it on to Lindsay for limitations. You there, Lindsay? Yeah, sorry, I got booted off at the wrong time. So I'm here now. So on slide 22 to talk about limitations. So as we mentioned before with these two main data sources, Buds and vCures, one limitation is identifying a source of truth. There is value in both hospital discharge data and financial data from claims. We know that there have been efforts and we've been part of efforts on the claims side to reconcile claims and some of our reports from claims like total cost of care with what the payers have. And that's always looked good. And we also know that on the VUD side of things, they have VOS NSO has done work to reconcile what they're receiving from the hospitals with the hospital's records. And that's always looked good. So which one is right when we go to quantify this difference? So that's something that we think about a lot. And also finding agreement across these two different data sets that are already cleaned and curated. We mentioned that VOS NSO provides us with VUDs and they have their own aggregating and cleaning steps that they do. And then on point provides us with vCures. They have their own cleaning and curating steps that they do. And so understanding the correct levels of detail that we need to find agreement between these two data sets has been something that we've worked through iteratively but was worth mentioning. And of course weighing the ethics of connecting all of these data sources, not just vCures and VUDs but bringing in census tract information and adding any clinical indicators we can. And all of this kind of combines to make a great picture at the population level but we just need to remember that it's important stewards of the data to keep it at the population level and leverage it in a responsible way. And then lastly COVID effects. So right now our data comparison is 2014 to 2019. Obviously the next year to add on there is 2020 and we are expecting some very big changes in the trends in 2020. And so it will complicate our comparisons if and when we add on that year but both Jeff and I are hoping that it's an opportunity to measure the same event in the two different databases the impact of the same event. And so potentially it could be an opportunity to align the data in that way. So yeah, that's all I have on the questions. Thanks Lindsay. So I'm just gonna wrap up with this slide here. To summarize, we continue to advance our analytic work. We're leveraging multiple data sources to tease out all these trends across space and society fully acknowledging all the limitations involved for every particular way we look at a problem. Just so we can give an honest and transparent and replicable view of what's going on. And also we want to generate products with increasing efficiency through the use of coding APIs and understanding the connection between Vickers and Vutz ever more quickly so that we can begin doing stuff with greater depth. This is not a simple project, but it's an ongoing what is and how can we know the ontology and the epistemology of hospital activity and how it relates to the healthcare system. And it will shape how multiple projects down the line move forward. So I will wrap up with that and pass it on. Jeff, do you have an estimated timeline for your deliverables? If we're talking Tableau visuals or particular datasets? The completion of the different pieces of these work. Do you have any type of timeline when this will be done, when that will be done, that type of thing? Yeah, breaking it down, Lindsay may have her own perspectives. The alignment will continue. I think you're going to see certain diagnostic categories, certain types of care, certain payers being figured out more quickly than others. So it's really an iterative process there, at least in the short term. In terms of the census data and spatial data, a lot of that's already available. Perhaps some of the code needs to be written to draw that more automatically or at a regular interval, but it's relatively, it's not too difficult to do. What may be a bit more open-ended is some of these more specific analyses in looking at, so we have all this, how can we discern X effect from Y? That involves some special project chatting and figuring out the scale and scope of the project moving forward. Thanks, okay, questions from the board? I see blue around Robin. Yeah, thank you. Thank you, Jeff and Lindsay for the update. It's good to understand where you are and how things are going. It seems like there's a lot of promise, I think, for how this will fit in with HRAP, a health resource allocation plan in particular, because I love that you're thinking about the automatic, like how to make it more automatic and less staff-intensive, because that is something as we've been working on HRAP, but we have minimal staff able to focus on that, and so that kind of flows at the available time that people have, so that will be a great thing. For the hospital market stuff, I think sooner than later, my suggestion would be to really sit down with the hospital budget team and talk now about how that gets integrated into the regulatory process, because for us, having the information and understanding the trends as background is helpful, but it's infinitely more helpful if it's integrated into the regulatory process so that we can actually use it in our decision-making. So that would be my suggestion is that there may be things that would be higher priority when thought about from a regulatory perspective than from simply the data perspective, so that would be my thought. Thank you, though. It's great to see how the progress is going. Thank you. Thanks, maybe I can just on that, is that all right? Go ahead. Okay, yeah, so thank you, Lindsay and Jeff, that's really helpful, lots of possible projects, I think emanating from your work, and I just wanted to build on what Robin was saying. I think it's really important for us to think about how we use this data to improve our decision-making, and so if you're prioritizing, I would say certainly in the, I see three areas where this particular analysis could be useful. One is hospital budget process. Two is ATRAP, both that Robin mentioned, and third is in our Rural Hospital Sustainability work that we're doing with the report due to the legislature in the fall. So prioritizing, I'm not sure you're, I think Kevin's question around your timeline was really helpful if it could be kind of more concrete than that. The hospital budget process, this is really, the patient inflows and outflows within a hospital service area or a census track will be really helpful in guiding some of the board's decision-making around reasonable NPR growth rates. For example, if there's a lot of inflow of patients into a hospital, for example, NPR growth might be higher than expected, but if there's outflows and those trends are downward, then NPR growth would be expected to be slower. And that would be really helpful to us as we're making hospital budget decisions. So I agree completely with Robin, if there's a way to integrate this more quickly, even in the next, I don't know, few weeks, but I'm not sure. So that the, working side-by-side with the hospital budget team, so that the hospital budget team can have this data as part of their analysis on the requests made by the hospitals around NPR growth would be really helpful. With respect to the Rural Sustainability Report that we have upcoming this fall, we know that Patient Bypass is a driver of financial vulnerability of rural hospitals. This is where patients living in an HSA bypass their local hospital and go to a different hospital. We know it varies by payer type. And so to the degree that there's a way to take this analysis and quantify the amount of patient bypass that's happening within each hospital service area or census track, if that's the geographic track you're gonna use will be really helpful for that report. So really coming up with a way to quantify is there patient bypass happening here and what's the degree of it? What's the intensity of it? And is it, how do we break it down by payer or by service type? I think would be super helpful. And the ATRAP, as Robin mentioned, perhaps maybe that's the last of a priority just given the timeline on that, but still a priority in helping us. And I would love to hear how you think that we can use this kind of analysis and the merging of these data sets or the alignment of these data sets to think about improving our understanding on met need in hospital service areas by merging in data, census track data and really thinking about how do we measure on met need using patient inflows, patient outflows, kind of utilization and basic demographic characteristics of the community. So I guess the last one is a question and the other two are maybe asking you to think about how we might achieve some of those goals in our regulatory process, hospital budgets and the sustainability report in the timeline that we have upcoming. Long questions. And insightful ones. I'll answer the question of the questions. So measuring the, a good approach to measure gaps and it's something I saw in the literature reviewing which studies to put in this presentation would be measuring the frequency of ambulatory care sensitive conditions that occur in hospital because if it occurs in a hospital it's sort of an indication they can't access primary care. That's one example. I'm sure we can spin out, look at the literature to see what other health outcomes might associate with healthcare gaps and whether those healthcare gaps are due to perhaps demographic concerns, physical access concerns or hospital and healthcare market concerns among other things. And I'll just add to that answer, Jeff, specifically for the question or request for data around the patient bypass problem. We do have that data available. And so we can on that shorter timeline work on packaging that for you all. It would be by HSA and whether or not over time people are seeking care within that HSA or they're going to another HSA and which one they're going to. We have it by expenditures and we have it by claims volume. Great. And is that something that you're working with the hospital budget team now for them to be able to incorporate that in their analysis of the hospital budget submissions? We can, we will work with them. Perfect. Thank you. Thank you all. As a follow-up on that question, Lindsey, what's the time lag? So for example, we already have seen one hospital talk about the fact that they're seeing significant in migration. Would we be able to see recent in migration or are we still talking a two-year lag? So I may be wrong about this but your hospital budget review is for 2022. Is that correct? Correct. Yeah, so the most recent data that we have, unfortunately, is 2020. So, and we all know that that looks really different. So unfortunately, yeah, it is, there is a little bit of a lag there which I think is part of the rub with using some of this data for the imminently but it can certainly help. And to see the trends over 10, I think can be really powerful. So, if you'd like it, we can definitely provide the data through 2020. Thank you. Other members of the board? Maureen? Yeah, just adding on to the, I mean, I think the patient migration piece and being able to track that is huge. And I would add to what the areas just said and also say the ACL, because we know risk and everything was being associated with HCA's and where people moved to and out of, certainly impacts with payments received. Another thing I think will be interesting to look at in light of COVID is were there, was there more or less remonters getting care outside of the state? Did people stay here or did they go back to Florida? What happened there? And how is that returning as well? Because I think that's one of the pieces we're always missing is all of the care that remonters get outside the state. And then should they return to the state or should that change, that obviously impacts the hospitals and could create increases at hospitals or vice versa, if more people are moving out. So, especially with a lot more people that came in and stayed in Vermont during COVID, it would be interesting to see also, I guess we're not sure if you don't really track the out-of-state piece in that, but I think the migration is gonna be really important though and a lot of the work we do. So I just wanted to add the ACL piece too. But thank you, very interesting. Okay, other members of the board? So I can't, I have one quick question and then just a couple of observations. One is, I think a while back within the last year, so we made an effort to get the more of the self-insured commercial into our system. And I remember reading some letters that were sent out to some of these folks. Did that yield any result? Did we grow our self-funded commercial participants by any significant measure? That effort was picking up right before the pandemic hit. And so I think priorities shifted during the pandemic, but I know that it's something that we're still working on. Well, that's definitely a fair observation. And otherwise, I can only echo Jess and Maureen and Robin on this, I think I'm struggling as I listened to say what's kind of interesting from a hypothetical point of view and what can be used here in terms of boots on the ground, in terms of our hospital budget process and our rate review process. And it just seems there's a lot of opportunity there. And I think not only talking to our rate review team and hospital budget team, but if you folks could be talking to us about how to ask the right questions because it's a very, usually complex set of data that you have, there's a lot of variations that could be scrubbed. And I mean, do we ask a question that is as simplistic as we have this rate review filing from XYZ commercial payer, go find us $3 million of savings in their proposal. So, I mean, that's kind of a simple but hardball kind of question, but I think it reflects the kind of practicality that would be helpful in terms of enhancing affordability and being in a database environment, kind of going after savings rather than just throwing a dart at the wall and hoping that it will yield a result. So congratulations on this. It sounds like it is a kind of a wonderful integration of a lot of information out there that I think can serve Vermont as well. Okay, anything else from the board? So hearing none, I'm gonna open it up for public comment. Does any member of the public wish to offer comment? And I'm going to recognize Mort Wasserman first, Mort. Good afternoon. Good afternoon. I echo all the board members' respect and admiration for the work that you two have done here already it's quite a daunting task and it's amazing what you've accomplished so far. I did seem to hear in passing that you hadn't yet had a chance to focus on primary care and given that hospital visits and admissions are often a reflection of a failure of primary care. I was wondering how the work was going to be done on primary care and the changes that are happening in primary care which are certainly affected by the same geographic demographic factors that are affecting hospital use. I'd be happy to jump in, Lindsay could pop in later if she'd like. So the primary care, ATRAP has covered quite a bit of that or it gave us quite a good sense of primary care outside the hospitals themselves. We're advancing from those insights to, I mean, the hospital database, the products can be discerned by DRG way into primary care versus tertiary care at a very basic level or high level care, low level care. We can also break down specific MSDRGs and it's really just a matter of taking the variables and deciding how we define primary care and then popping it out there. In terms of the specific deliverables we could have say a worksheet in the next tab, low visual that shows primary care at the hospital level or at the claim level. But it's an open book. We can go in many different directions, especially after talking with the board and any other stakeholders. Great, thanks, yeah. Thank you more. Next I'm gonna call on Eric Schulteis. Eric? Hi, so obviously Jeff and Lindsay, wonderful job. It's great to hear about your move towards scripting and coding in terms of access and also replicability and also perhaps looking to update information in a slightly more timely fashion using application programming interfaces. I think looking at VUDs and vCures, I mean, I think it's important to remember that the consumer experience or the lived experience is missing from these datasets. And I think we have to be careful. I'm gonna speak from my own background, but I think if you look at urban planning in the, say, 1950s and this kind of technocratic brain, I mean, maybe perhaps best put out by Bob Moses. And then you look at someone like Jane Jacobs that's really focused on the lived experience that that's an element that's missing from these data sources. And it's important to acknowledge that behind these numbers, there are actual people and also that there are a phenomenon of why these people behave in certain ways that aren't captured. So for instance, going back to board member Holmes's question about, how do we measure potential access service gaps? So yes, you can kind of roughly estimate that population if they're going to the hospital, but that's always gonna lag a bit. It's also gonna leave out the people who are sick but couldn't get to the hospital or were sick and were afraid to go to the hospital because of a bill. So I just, I think as you're moving forward and the data team has always been outstanding in my opinion in this before is to try to frame what the data can and can't tell you and where potential future expansions could go. Because I think there's the issue of the connecting this to actual regulatory processes, but there's also as we implement regulatory processes we wanna have them be responsive to the actual experience of our monitors and not just what we're measuring in these two databases. Thanks. Thank you, Eric. Other members of the public? Other members of the public? Hearing none, I wanna thank Lindsey and Jeff for an excellent presentation. Jeff, I'll do a little studying and figure out what Ho Kutur is and get back to you on that. And with that, we'll move on. Thank you so much. Next item on the agenda is old business. Is there any old business to come before the board? Seeing none, is there any new business to come before the board? Seeing none, is there a motion to adjourn? So moved. Second. It's been moved by Robin and seconded by Maureen to adjourn. All those in favor signify by saying aye. Aye. Those opposed signify by saying nay. Thank you, everyone.