 Welcome to Green Mountain Care Board. The first item on the agenda is the Executive Director's Report, Susan Barrett. Thank you, Mr. Chair, I have a few announcements. First, I wanna remind those that next Wednesday, October 2nd, in this room, we're hosting a panel discussion on an update on healthcare workforce solutions. And then another scheduling reminder that on Wednesday, October 16th, the board will be going on the road going to St. Albans, Vermont, where we will be conducting a public board meeting but also in the morning meeting with healthcare community members. So looking forward to that, more details to follow on our website. And in terms of hospital legends, those will be released by October 1st. And then I wanted to just mention that on our agenda today, under our old business, that we'll be discussing and clarifying a matter under old business. And just to mask that to the group. Anything else that you have in mind? Just that that matter is really to the hospital that the board had your week. Great. Okay. Hey, the next item on the agenda are the minutes of Wednesday, September 18th. Is there a motion? No. I move to second. Second. Been moved and seconded to approve the minutes of Wednesday, September 18th, without any additions or additions or corrections. Is there any discussion? Seeing none, all those in favor, stand by by saying aye. Aye. Any opposed? Okay, with that, we're going to turn it over to King O'Neill to kind of get us started on the afternoon activities. Okay. Good afternoon, thank you. I'm Kate O'Neill and I'm staff here at the Green Mountain Care Board and I work on the data and analytics team. So today we are going to be showcasing some examples of how our all-payer claims database is used in research. So the Vermont Healthcare Uniform Reporting and Evaluation System, we call it VCURES. It's Vermont's All-Payer Claims Database and it includes eligibility information and medical and pharmacy claims provided by commercial health insurers as well as Medicaid and Medicare and we've been collecting such data since 2007. Through data use agreements, VCURES data is used by Vermont state agencies and non-state entities both within Vermont and outside of Vermont, such as contractors, academic researchers and the like to support analysis of healthcare access, spending, utilization, and quality. And today I'm pleased to introduce you to three entities who under data use agreements have conducted research. The first presenter who you'll hear from is Tammy Mark and she's the senior director in behavioral health financing and quality management at RTI International and she's going to be talking about a study that they conducted and published that they titled Opioid Medication Discontinuation and Risk of Adverse Opioid Related Healthcare Events. The way we structured this is that each researcher will present their work and then the board and the public will have an opportunity to ask some questions and then we'll move to the next researcher. But I just want to give you a preview. After RTI International, the Vermont Department of Health, Lauren Cosshigan and Laurel Omland, their Laurel is an epidemiologist and CDC assignee. Laurel is the director of the child adolescent and family unit and they will be presenting to you an analysis that they did on mental health related emergency department claims for Vermont children. And then finally, Valerie Harder, associate professor of pediatrics and psychiatry and the director of the health services research team at the UVM Child Health Improvement Program, V-CHIP. She will be sharing examples of how that team uses V-Cures for pediatric health services research. So really pleased to be able to share these researchers, their work with you today and I'm going to sit back down and invite Tammy. Come on up. Thanks so much for the opportunity to come here and talk about our research. I was in Vermont two summers ago in July and we were having a heat wave so I'm very pleased to come back under these nice weather conditions. So RTI International is a little bit of background on the study. RTI International is a nonprofit research organization. Our mission is to improve the human condition. My background is as a health economist. I spent most of my career focused on trying to improve the mental health and substance disorder treatment system. I've worked with a lot of different data sets and I was very personally interested in trying to do some work with these growing availability of APCDs and RTI was also interested in APCDs which were becoming more and more of a resource for researchers. And so RTI gave us a little bit of funding and some time to explore different APCDs and then to conduct a pilot study with one of them. So we looked at what different states had and how accessible they were, how complete they were. For example, then they have substance use disorder data and then we decided we'd like to work with Vermont data. So that's how we came to request to get access to the future's data. The focus of the research I'm gonna talk about is opioid use disorder medication. RTI has a long history and a lot of 200 folks actually who just focus on substance use disorders and addiction treatment. And with the emergence of the opioid epidemic, the Institute really wanted to make sure that those resources were being put to addressing the issue of the opioid epidemic. So this topic is really one that resonated with our Institute and we're doing a lot of work around opioids in general. And as was mentioned earlier, the study's been published in the Journal of Substance Abuse Treatment and there's copies of it and the entrance. So, there we go. Okay, so I'm sure you're aware that opioid overdose deaths have been arising at alarming rate. Preliminary data from 2018 suggests that we may see, we may be seeing some moderation, but up until then we've lost possibly 300,000 people to opioid deaths. And initially the epidemic started as a prescription opioid pain medication epidemic. If you look at the trajectory of the growing use of opioid pain medications and the trajectory of deaths from opioid overdoses, that you can see they're pretty parallel. Then around 2009 or so we saw a shift in the epidemic where deaths increased from heroin and use of heroin started to go up. And then more recently we saw another shifting of the epidemic where we saw increased deaths from synthetic opioids such as fentanyl. So of course a key response of federal, state, local governments, health plans, providers is to try to reduce the amount of opioids being prescribed. And we've actually been successful at that if you look at this chart, opioid prescribing rates peaked at about 81 per hundred in 2012 and they've been falling pretty rapidly since then. The little line underneath, this is all opioids and this is high dose opioids and high dose opioids prescribing rates have also been falling. As part, when you look at this data it's hard to think about two different kinds of prescribing, there's prescribing for acute pain. So for example, if you go to the ED with a strang ankle or you have a toothache and the doctor gives you a supply of opioid medications, one approach that we've been trying to take that the US has been trying to take is to reduce the amount of opioids prescribed for acute pain. So rather than getting a 30 day supply if you've twisted your ankle, now we're moving to regulations and guidelines and say just maybe hand out three or seven at most for those acute instances. There's also been efforts to address and reduce use among folks who are using at high doses for long periods of time. So for example, doctors have been encouraged to look into prescription drug wiring systems to see who might be using a lot of opioids. They've been getting alerts through Medicaid claims data through electronic health records to indicate which patients are using at high rates and for long periods of time which may indicate misuse or inappropriate use. And that's really the focus of our study. We wanna understand what's happening to that population of folks who are identified as high dose users. And so we began by looking at the literature what does the literature say about how do you reduce use when you identify somebody or how do you taper folks off of opioid pain medications? In their guidelines that the CDC put out in 2016, they recommended that people be tapered about 10% of the dose per week. So as you'll see going forward, we're focused on people who are using at 120 what's called morphine milligram equivalents. And so if you were to taper those at 10% a week, that's going to translate into about 10 weeks. So I'll come back to that statistic, just keep that in the back of your mind. The CDC also said that sometimes rapid tapering, which they defined as two to three weeks is appropriate. For example, if someone is suspected of substance use disorder or has a medical condition such as unstable cardiac condition that it might be appropriate to taper them more rapidly than that. They did note that patients who are diagnosed with an opioid use disorder should be given medications to treat that opioid use disorder or medication assisted treatment it's sometimes called. Also what's important to note is that they are pretty frank in the guideline in saying that there's really little evidence to support these tapering recommendations. The research is very thin. So again, that was part of the motivation for doing this study. So what we wanted to look at in this research is what's the typical time at which it takes someone who is using opioid pain medications at a high dose to discontinue the medication? R is the tapering following the clinical guidelines. And what's the association between the tapering time and the risk of negative healthcare consequences? So we are fortunate enough to be able to access and use the V-Cures data for this. We had hope to use all of the payers because that of course is the beautiful thing about APC data, but we ran out of resources so we were able to look at this in the Medicaid population and hope to replicate it again in the private insurance and Medicare population. The study sample were individuals who were using at high doses which was defined as 120 MMEs per day, morphine milligram equivalents, and folks who had used that dosage for at least 90 consecutive days. This is a definition that's part of a National Quality Forum endorsed quality measure of opioid misuse or potential misuse or inappropriate use. So it's a measure that health plans have been implementing and using. So we wanted to see, well, if it's being used, what's the action that the clinician is taking or the patient is taking following identification with this measure? So the key independent variable was taper time, the number of days between starting to reduce the daily dosage and discontinuation. So from the time we first saw them lower their dose from 120 days, how long did it take them to where we saw no additional fills? And our dependent variable was time to adverse opioid related healthcare event which were events where someone had an emergency department visit or inpatient admission for an opioid poisoning or any substance use related condition. The statistical method that we used was Cox proportional hazard models that we controlled for age, sex, opioid prescription fill patterns and physical and behavioral health comorbidities. So here is what's called a consort diagram. It shows you how we got our sample. We started with data from 2013 to 2017. There are 752,000 Medicaid members in that sample. And then we limited it to adults age 18 to 64. We limited it to folks who had full Medicaid benefits. We took out the dual eligible since they get their prescription drug coverage from Medicare Part D. We limited it to people who were continuously enrolled so we didn't have dropout, we didn't have censoring. Took out cancer diagnoses, applied that measure of using at 120 MMEs for 90 or more consecutive days. And we ended up with a sample of about 500 Medicaid members. So this picture shows you graphically how we structured the study. So again, we were identified people who were potentially misusing or at least not using appropriately or who perhaps should be tapered. Those were folks who were using at 120 MMEs for 90 days. And then we tracked how long was it until they discontinued completely where we saw no more prescription fills for any opioid medication. And that's time is what we call the time to discontinue. And then we looked at the correlation between that time and the time to an opioid related adverse event. So here's some of the key descriptive findings. Almost half the population had an opioid related adverse event. The median time to an adverse event was 77 days meaning that half the sample had one of these events within about two months. The typical time to discontinue was one day. So what that means is that in half the sample there really was no tapering. We just saw them using at 120 days and then we didn't see any more fills for any type of opioid pain medication. So it just suddenly stopped. 86% of the sample were discontinued rapidly which the CDC defines as less than three weeks. And these folks were taking these opioid pain medicines for a fairly long period of time. The median time in which they had taken it was 510 days. In terms of fills from different providers, 24% had filled from one or more providers. Sometimes that's used as an indicator of potential doctor shopping or potential misuse. And that's, there's a typo there that instead of saying I'm at it, should say Matt, medication assisted therapy. We only saw that 0.6% of people who were tapered were transitioned on to medications to treat their opioid use disorder. So less than 1% after they were tapered were transitioned on to any kind of treatment for their opioid use disorder or medications to treat their opioid use disorder. And this was a fairly medically complex population, a third of them had diseases of the nervous system, that could be migraines, that could be a degenerative neuropathies. About a third had a disease of musculoskeletal system. A third had endocrine problems, for example, diabetes might be an example, and 60% had a substance use disorder diagnosis on their insurance claim. So this shows you graphically the relationship between the percent of the population that had an adverse event and the time to discontinue. So if we look at 365 days, the folks who discontinued rapidly in less than 21 days, about 50% of them had an adverse opioid related event within that year. If we look at the population that discontinued between 21 days and 90 days, it drops to about 30%. And then if we look at the population that discontinued in more than 90 days, it drops to about 20%. And when we put that in a regression, where we control for all these different factors, we see that there's a statistically significant relationship between the time to discontinue and the risk of an adverse opioid related event. The way you can interpret that is that for each day, additional day you spend tapering, you reduce the probability of an adverse event by 1%. So for each additional week, you spend tapering, you reduce the probability of an adverse event by 7%. So this summarizes the findings again, 50% of the patients were suddenly discontinued, there was no tapering, 86% of the patients were rapidly tapered, 49% of the members had an opioid related hospitalization or emergency department visit. Again, each additional week of discontinuation time is associated with a 7% reduction in the probability of having an opioid related adverse event. And 60% of members had a substance use disorder diagnosis on their insurance claims prior to tapering, but less than 1% were transitioned on to an opioid medication to treat that opioid utilization condition. So since we published this paper, this issue has gotten more attention. The CDC in April put out a statement saying that they believe their guidelines were being misinterpreted, that people were interpreting it as you should stop prescribing, which they did not mean. They also point out that it's really important to transition people on to medications for their opioid use disorder or not to just fire them from the practice. And so there's been a lot of more concern lately about the interpretation and the approach to tapering people. There's a lot of limitations to this study, a lot more we'd like to do with it. We're limited in our time and our resources. I'm sure a lot of things are probably going through your mind. But again, we were limited to Medicaid data. We'd like to look at this another insured populations. There are other important outcomes that we did not capture in the study, but we could with the insurance claims. We did look at suicide hospitalizations. We didn't look at all CODS, ED visits and hospitalizations. We didn't look at mortality. We didn't examine lower tapering doses. Now the recommendation is approved from 120 to 90. And we didn't look at what happened in people who are tapered from 90. We didn't look at whether folks were getting transitioned on to alternative pain treatments. And really our focus was on time to taper. It wasn't on whether tapering is a good idea at all. So that's just another really important question. And with that, I will open up for questions and comments. Thank you. Thank you. Questions from the board? When you go to your summary of findings, I think there's so much more to talk about because the median length of discontinuation to one day and then those were the patients and seeing that obviously had the higher issues potentially. Do we have any data on how many of those people maybe found their, got drugs illegally after that because it seems if they were on such a high dose and then they went from zero to nothing, that's gonna create some huge withdrawals and we're really talking prescription versus street. Yeah. I mean. That's really what worries me. And I don't, we couldn't look at this, but when you look at the correlation between the fact that as we started clamped down on prescription opioids, we saw a rise in heroin use. It begs the question, are we pushing people into the illicit drug market? And that brings the question beginning when you showed the chart with fentanyl and not really creating a lot of overdoses. And it appears when you read about it that there's a lot higher incidence of someone that might use that drug to overdose and potentially die than have they just used prescription medication? So that's obviously creating like it seems like a big jump there. And then I wondered how if anything, reversal drugs like oxalone and things like that, that's become much more prevalent. So that might be hopefully helping the number of deaths, but are we getting the right picture if there's a lot more overdoses out there? I did that factor in at all. Yeah, I mean it's tough to look back and this is a retrospective study. We're looking into, we stopped in 2017, we started in 2013. I mean, the oxalone's really been distributed much more aggressively over the time period, which is great. Next one I think is the sample size of 494, is that a thing enough sample size? And we started out with like 800,000 and then it went all the way down to 494. Yeah, I think that's, you know, again, that's partly why we'd like to look at this in the 90, you know, because 120 was pretty high. So I think if we look, we did this with the 90 enemy, we get a larger sample size. I mean, the results were statistically significant. So, and if you look at the pictures, it's pretty, it's pretty stocky then. So I think, you know, 500, you get spoiled when you work with claims data. You know, people doing a randomized trial or pulling charts are gonna think 500 is a lot. Those of us that work with claims data. Okay, good, thank you. Sure. Other questions from the board? Jess? Yeah, I just took a question. Since you talked about this as a part of a larger initiative to test drive, you know, all kinds of database. So I'm wondering if there's any lessons you could share with us about using Vermont's, why you chose Vermont's, strengthening weaknesses of Vermont's database? Yeah, I mean, one strain frankly was it wasn't as expensive as some of the other databases. Again, it had the substance use disorder claims included and that's since that was our focus, that's important. It had the three payers included. Still, the process wasn't, I mean, the DUA process was still a process. You know, it took time. We had to go answer questions and go through review and it'd be nice to, as a researcher, to just get the data more quickly. Okay. Thank you. I had a question on the numbers of people themselves. You got it down to the 494, but how did it start with a larger population in the entire state? Is it not identifying repeat performers between new year's or how is that? I'm trying to find that picture for you. The picture in the honor? Yeah, yeah, so it's four years. So in the final in the 494? Yeah. Could some of those have been the same person? No, I mean, no, we only limited it to the extent that you're, yeah, the member, to the extent the member ID is unique, it shouldn't be the same person. Other questions on the board? John? Your question, Kevin, was exactly the first one on my list and the second is it's part of the data collection had to do with folks that had four prescribers or more and is there a database that tracks that that you can get that information from to see that an individual is getting their information to them from an array of providers? Yeah, we were able to capture that in the each year's data because we could use the different MPI or provider IDs to indicate if they were getting prescriptions from different providers. And that's another measure that the pharmaceutical quality alliance has developed for health plans and Medicaid programs to use to try to identify prescription misuse. So that's why we use that measure. Okay, at this time we'll open it up to the public for any comment? Yes, yeah. Kindly capture off to the healthcare advocate and I apologize if you said this and I missed it, but I was curious in your analysis of this data if you were able to identify any aspects that caused people to be split into those two groups of a longer shorter time like that. Was it, did it seem to depend on the vision they were seeing or some aspects about their health history or their history? Yeah, that's a great question. We did redo the analysis just for people with the SUD diagnosis on the claim and those without we found the same relationship regardless of whether they had the SUD diagnosis or not. We don't know really, and we didn't look at whether this varies by the physician that there, any other characteristics of a provider, but that's a great question to look at further. Okay, other comments from the public? Seeing none, I wish to thank you very much. Thank you. Do we have our colleagues here from the Vermont Department of Health, Lauren and Laurel? I think we could switch. So in that case, we'll ask Valerie to come on up and share your work from the Beechit program. And thank you Abigail and that if we're able to share with you the Vermont Department of Health work, we'll do that after Valerie, thank you. I'm Dr. Valerie Harder, I'm an associate professor for pediatrics and psychiatry from the University of Vermont. I'm also the director of the Health Services Research Team at the Vermont Child Health Improvement Program or also known as Beechit. For those of you who aren't familiar with Beechit, we're a maternal and child health quality improvement and health services organization. We are often doing work with the maternal and child health division at the Vermont Department of Health. We are sometimes referred to as the implementation arm, working with the pediatric primary care practices across the state. Thanks for having me and thanks to the data analytics team for asking me to present today. I have no conflicts of interest to disclose. Also all the analyses, conclusions and recommendations from the data that I'll present today are my own or from my research team. I have funding from the state of Vermont, also from the Agency for Healthcare Research and Quality and the National Institutes of Health. Rather than go in depth on one of our many research projects, I thought I would give you a brief or overview of five projects that we're working on that utilize the V-Cures data to give you more of a breadth of our health services research that I'm conducting at Beechit. The first I'll talk about, we call attribution. This is where we place every child zero to 26 years at the primary care practice in the state for each year of V-Cures. The second project I'll tell you about is related to the quality improvement that Beechit does. And we wanna look at child health outcomes that are impacted by the quality improvement projects we do at the primary care practice level. Third project I'll tell you about is our work at looking at trends over time in V-Cures. Some of the data that I'll show you are related to developmental screening for children one to three years old. The fourth project I like to call Workforce, we took a look at where pediatric patients are seeking primary healthcare, pediatric practices, family medicine practices, natural path practices, all those different locations of practice site. And we published this work and it's now published very recently last month. And finally I'm gonna tell you about a policy related project that we're working on and we're similar to the project that was just presented. We looked at opioids in relation to the 2017 opioid prescription policy change that happened in July of 2017. And we wanna see the impact of that policy change on opioid overdoses. So I'll tell you a bit about that project. So our first project is attribution. It's quite an endeavor that our team does every single year. My health services research team, they work really hard to place pediatric patients ages zero to 26 at their primary medical home practice site each year. First thing we do is we place every child with a single primary care physician for the year. We use a hierarchical algorithm that prioritizes well care visit over acute visits for example. We also use the frequency of visits and the date if there's any ties between physicians seeing patients. Next we place every physician at a single practice location for each year. We assess several administrative data sources. We don't do this all by ourselves. I wanna thank my colleagues at the blueprint for providing some of the information that they collect. And other administrative data sources in the state and nationally. Surprisingly many of them don't agree where a physician works. And so we do our best to have the majority rule or talk to people who know these practices very well and the physicians that work there. So these two connections from the child to the physician and then the physician to the practice allow us to attribute each child to a practice. So you might be wondering why do we do all this work? Why is this important? Our goal is to measure the impact of each of quality improvement efforts at the practice level. Health outcomes in Vermont children. And so we have the data and v-cures at the child level and v-cures does not tell us which practice they're at. So we have to do this work behind the scenes in order to look at the child health outcomes as a result of our practice-based quality improvement. So we work closely with the maternal and child health division at the Department of Health as I mentioned. And we look at outcomes related to the quality improvement projects we do at each practice each year. They might focus on subsets of children. I'll give an example of one of the quality improvement projects around asthma that we've done. And that's not at every practice. That's just the practices that volunteer to do our quality improvement project. We also have looked at children with special healthcare needs and been able to attribute them to different practices. So it's therefore really important for us to measure the impact of v-chips QI efforts by attributing children in v-cures to a primary care practice. So here in the data that you see up here, you can see that in 2017, the majority of children in Vermont receiving care are at pediatric practices followed by family medicine. Interestingly, there are three times more family medicine practices in the state of Vermont than pediatric practices. Why do we see these differences? You can look across on the ages. This is something else we do in attribution. You see what age groupings are attributed where. You see that the majority of the younger newborns and infants are seen at pediatric practices and the family medicine practices are seeing older patients up to age 26. There are so many questions that we can ask using these data. And we've already started working with our maternal child health division to answer some of those questions. One is where are the newborns being seen in Vermont? Do they cluster by hospital service area? Are they in certain regions, certain practices? Or we just answered a question of how many children are seen at federally qualified health centers in Vermont? Later in the presentation, I'll show you the project around our workforce where we attributed children from 2009 all the way up to 2016. And we looked at their movement between different types of practices. For the second project, I'd like to highlight one of our quality improvement initiatives, which was around asthma control in the primary care practice. In October, 2015 to April, 2016, Bechip led a quality improvement learning collaborative at the practices to help improve asthma care. There were 20 participating practices improving asthma care for their asthmatics at their practice. Some of the things that they did was improve the asthma action plans given to these kids, helping making sure the kids had the medications they needed, also having timely recall of patients who needed to be seen every six months to monitor their asthma. So they improved all of these systems at their practices. And what we wanted to look at was before and after participation in this quality improvement program, what was the emergency department visit rate of the children with asthma at the practices who participated compared to the practices who did not participate? So our measure here is the rate of emergency department visits, which also includes hospitalizations for asthma among patients with asthma at the participating practices, which is an orange in this diagram, compared to 15 control practices who's in blue. The 15 control practices are practices that are part of our quality improvement network at Bechip. And so they're highly engaged in quality improvement each year. They don't all participate. They get to choose voluntarily what they participate in. And so these 15 chose not to work on asthma this year. Maybe they're working on a different project. We don't know the exact details. But the decreases that you see in the orange line is that prior to the quality improvement project, you saw that there was a little bit of difference between the non-participating and participating practices. But over time, the participating practices out to 2017, there was a significant decrease in the emergency department visit rate per 100 person years for the children with asthma at the participating practices compared to controls. The way you interpret this outcome is that in a year, if you have 100 children with asthma, this is the number of emergency visits they'll have for asthma. So this is really exciting news for us. And we're working on the publication right now to share these results. Some of our work is related to looking at trends over time. We look at health measures over time to inform us of the progress we've made on different quality improvement projects we're working on. We use these trends in response to questions from our colleagues at the Vermont Department of Health. For example, this past year, they wanted us to look at developmental screening over time and wondering how are we doing in the state in terms of developmental screening for our children who are one to three years old. So the objective of this project was to assess the percentage of children ages one to three years old receiving developmental screening from 2015 through 2017. The table that you see here shows the results to the question, does the proportion of children who have had at least one developmental screening within the first three years of life increase over time? So you can see in the red numbers on this table that indeed in 2015 we had a 47% of children one to three received at least one developmental screening in the year and that has significantly increased over time up to 59% in 2017 that children one to three were receiving at least one developmental screen in the past year at their primary care. So this is encouraging news. The Department of Health has been working to improve developmental screening and there have been a lot of efforts around this and this is one way for us to help support them so they can show developmental screening is going off in these age groups. It's not where they wanna be yet, but it is improving. You can see among two-year-olds which is the age group where the majority of developmental screenings are supposed to be happening. In 2017 we're at 74% if you follow that column over or the row. So there's still room for improvement but these data we shared directly back to the Division of Maternal Child Health and they make plans and move forward with their efforts as a result. I'd like to take a moment just to discuss something that you could see in these data as a result of the GO-BAY versus Liberty Mutual decision over time. This is something that we and I'm sure many other people using V-Cures is struggling with that we have a loss of data since 2015. So if you look at the columns above overall between 2015 to 2016 in black, you can see that the sample size, the overall sample of one to three-year-olds dramatically decreases and that's directly related to that decision that has limited the number of people who are giving claims to V-Cures. So we just estimate that there are approximately 5,500 newborns every year in the state. It's around that. If you multiply that by three you get 16,500 and in 2015 we're at 15,600 not so far off. It's about 95% of the one to three-year-olds in the state but by 2017 we're down to 12,400 and that's closer to 70% of the newborns in the state. So one thing that we're hoping for is more guidance on how these missing children impact our research questions specifically, how to really think about them, think about their characteristics, how's that gonna impact our work? So we are looking for more guidance on that. The fourth project I'd like to tell you about is the one that was recently published. So I've put the picture up here. It's called Change in Sites of Children's Primary Care, a longitudinal population-based analysis that was just published in the Annals of Family Medicine. We thought the family medicine physicians would be most interested in our findings for this. So this study used our attribution data that I showed you on that first slide. From 2009 to 2016 we attributed each child to a primary care practice and we desegrated that practice as a pediatric practice or family medicine practice or a nurse practitioner, et cetera. And we looked over time at the proportion of children at each of these different practice types. From 2009 to 2016 and we saw that children are receiving care at family medicine practices less frequently over time, since 2009 up to 2016. So these were important findings. We spoke with some family medicine physicians and we've been reviewing the literature and there is some reports of family medicine physicians not choosing to do obstetrical care as much as they used to and so they're delivering through our babies potentially and there are fewer infants coming into care at family medicine over time. So this might be one of the reasons. We did also look at these data broken out by a morality indicator in the state and found that some of the more striking differences were found out in the isolated rural areas of Vermont. So some things to think about related to the workforce for physicians in the state. Final project I'd like to highlight here is our work on the policy on opioid prescribing that went into effect in July of 2017. This work is ongoing. I'm not showing our data here because it's still preliminary results. We're still waiting on the last three months of V-Cure's data from 2018 to complete these analyses but we're close. So we looked at two different objectives and two different outcomes in these data and we wanted to see whether or not there's an impact of the opioid prescribing policy on opioid overdose rate in the state from before the policy to after the policy and also on what I'm calling opioid related adverse effects. The phrase adverse effects are related to things like sedation, slow respiration, and altered mental status. It's not overdosing but it is related to symptoms that are a direct result of taking too many of the opioid pills. So here I call, I say opioids but we are encapsulating obiates which is heroin and fentanyl underneath the terminology of opioids. But the adverse effects as we talk to the ED clinicians are really related to those effects they see among people who come in by using too much of the opioid pills like OxyContin pills. So in July 1st, I'm sure many of you are already aware of the opioid prescribing policy that went into effect. It limited the dosing and also number of pills that are able to be prescribed and requiring physicians to check the prescription monitoring programs as they're prescribing opioids pills. So we look at data from January 2016 up until June of 2016 for the pre-time period and then immediately starting when the opioid policy went into effect from July 2017 out to right now September 2018 but we wanna have a full 18 months afterwards and go to the end of December 2018 to publish our results. So some of the preliminary analysis that we found a really interesting effect of seasonality. We've seen this with opioid deaths that there are more deaths in the summer months but we're seeing this definitely in Vermont that there are more opioid overdoses in June, July, and August months and so that's a time-varying factor in our modeling. I'm using an interrupted time series model to look at the pre versus the post time series data and controlling for the seasonality or the summer months effect underlying these data but preliminarily we're seeing possible increase in opioid overdose rate after this policy change in our Vermont data. I didn't mention that this project is in collaboration with colleagues in Maine so they also have an all-payer claims database in Maine and we are working with their all-payer claims and seeing this, looking at the same outcomes in the Maine APCD so we'll be able to compare what's happening in Maine, what's happening in Vermont, similar, different and really think about that using our two APCDs. Another interesting outcome, as I mentioned those adverse effects, we are seeing a decrease in the opioid-related adverse effects after the policy change and again these were those slow respiration and altered mental status that are seen in the emergency departments, people come in using too many pills so this is a positive finding that we're seeing so far in the Vermont data after the change. So I don't do this work alone, I have a fantastic team of analysts and research assistants at VCHIP. Also I have many national collaborators and regional collaborators that help support these work and I'd also like to thank our community partners, some of their logos are up here. We try to have them involved in our work and help inform our work. So thanks. Thank you. So we'll open it up to questions. Well thank you, really interesting. One question from the last slide. Any hypotheses about the summertime? What is driving the summertime? Sure, so people, being outdoors, being more active, someone was telling me that the substance of choice in the winter is alcohol and in the summer is opioids. They're seeing that in primary care as well. So I mean I haven't done that work myself but those are some of the thoughts. Just having access, open area areas, meetings outside with people, access, maybe supply, getting into the state more recently and the roads aren't horrible. And that's not, I would suspect if it's access and roads and that it would be the same for all drugs, heroin and other types of drugs that over this would be the highest in the summertime as well, it was easier to get them in the open season. I would think that as well. Yeah, okay. I was wondering that maybe that, well I was wondering that it had to do with access to providers over the summertime and providers in the limited locations. I mean I was trying to, in my head, think through this and I don't have an answer to that. I haven't investigated that in the data but it's a question that comes up a lot. Can you get a slide three or the trends over time and develop a screening? Sorry, the next one, yep. So this is really intriguing and I'm wondering, obviously early intervention is such a key determinant of success in school and mental health and physical health and so I had a couple of questions about this. One was the drop off at three years old and just in terms of screening is it just that there's more screening that happens at two years old there's less standard screening that happens at three? So screening happens at nine, 18 and 30 months of age and so the three year olds would be hit at the 30 month visit to see if they had a screening and they, that visit isn't as well attended as some of the other earlier visits. That is something that we know about for well care visits so it just might be that fewer children are coming in for their well care visit and I think that also if people, children are screened early and they're screened that they need further evaluation then they are sent off to children's services to have further investigation into their development and then when they come back for their next visit maybe when the three they aren't screened again because they don't need to be screened again that's already in the system so that happens with the smaller percentage of children who do have developmental delays so both of those things I think are going into that where we don't see as high screening rates among the three year olds. And I was wondering if you did any variation in the screening rates by HSA or by site of service for example whether it was a family practice or a pediatrician or whether it was a independent practice or a hospital practice or by payer, Medicaid versus private insurance just to unpack some of that. Yeah so we have done those. I would like to have the data in front of me to make sure I have the right conclusions from those but those are exactly what we did. We looked at payer, we looked at family medicine and pediatrics, just sort of the trends and helping us know where we need to target our efforts moving forward. Often the quality improvement though that we do with the practices are system-wide it wouldn't be targeting a specific group of children at the practices based on insurance or anything like that but we might want to target different areas and reach out to practices where we might see an area that has lower screening rates. I think that is very important. Great, thank you. Thanks. I have a question on the next slide. I mean you talked about the movement from more people at the pediatricians versus family practices. Has there been a change in the number of practices and more pediatricians in the area as well or? Well actually recently there's been a decrease in the pediatric offices. I know that from 2015 to 2016 there's several practices that even closed up in Brinkland County. I don't think there has been any other significant increase in numbers of practices of pediatrics or family medicine. I know that the practice type that is increasing the most are the naturopath physicians and so there are many more naturopath practice locations today than there were even four years ago. And then on the next slide kind of an attack on some of the questions that Jess had but when you look at the overdoses and the adverse effects I mean is there a way to see how many of those people are actually prescribed medications versus illegal drug use and it seems like the fentanyl is what's creating a spike up even more so it's unfortunate that this new policy is implemented and it's possible that things are going higher but also when we dissect that is it fentanyl is it the illegal drugs and how does that correlate to when we're looking at the prescribed drugs? Sure, that's a great question. So one of our hopes and one of our objectives when we started this research project was to separate the overdoses due to opioid prescriptions from the overdoses due to heroin or fentanyl and while those distinctions are in the ICD-10 diagnosis you have very specific descriptions it's a heroin overdose, it's an opioid overdose, it's other types of drug overdoses. When people present to the ED the ED physicians don't take the time to be specific in diagnosing so much of the diagnoses are the broader categories of narcotic and opioid overdose and they make all they need to do is put one diagnosis in there to move along and move along to helping the patient. So we aren't able to determine if it was specifically, I mean there are a few that do say heroin was the reason for the overdose for prescription that the opioid pills were the reason for the overdose but the majority there's even a non-specified category of overdose that we need to look at and there are many other papers out there that talk about this restriction and having to have a broader look at what you're pulling in for opioids and opioid overdoses. So, while we would like to do that we weren't able to do that in the claims data but we are, to get at your other question, this project is also looking at another data source that is not V-Cures, it's looking at the electronic health record data that's available to us through the University of Vermont Medical Center and in those data we have looked at the prescriptions and we're categorizing people into chronic opioid users, intermittent opioid users and never users and looking at the overdose rates of those individuals and seeing if they had a history of chronic opioid use after the policy change what's happening with them with overdoses or adverse effects. We could also look at something like this in claims it might be more challenging but we could look at the prescription records and claims we just haven't gotten into that yet. But it seems like there's different things going on and I also wonder in summer months if it has anything to do with since a lot of the overdoses are males between 30 or something school and vacations and things like that, in the summer. Actually, we've done a breakdown by age categories and those who overdosed are 26 to 34 year olds. I mean is it a national trend too that it peaks more in the summer? For deaths, I haven't seen this reported for overdoses specifically yet. So that's something we found that we were surprised to see this sign wave through our data essentially but then as we were looking into it others have reported that for deaths that they get from the medical examiner records and things like that. Thanks. So I notice in your chart of the collaborators that one of them is the Vermont Center on Behavioral Health. And I have spent some time with students against the Dr. 80s just kind of exploring their clinical work and the benchmark plan for example, for the QHP population. And just asking them, I saw an article in a paper where they got these very large $30 million plus grant from the Center for Disease Control and just trying to make see if that connection between their work and the world that we live in which is approving hospital budgets and rate review from tourist companies whether or not there's a tight relationship there in terms of the data that's available to clinical data and the decisions that we make. And we found for example, that one area of pre-IVs there is very little connection between the data that says what works and how we spend our money. And I'm just wondering from your perspective, do you feel that the access that you have like this and maybe more formally in the decision making process to connect good data with the right decisions? Do you feel those avenues are pretty much across the board well paid or is there a lot of room for improvement? Well, I think that it is moving in the right direction and I have been pleasantly surprised with how many people are open to having me come and present even before we started analyzing the BQS data to tell them about what we were doing with the opioid project. There is a group in Burlington led by the mayor and the police chief called the statistics community statistics meeting and they bring together everyone who is working on opioids into county at least and they had me present there about what we were going to do, I got feedback from them and then I presented some of our preliminary results to them a couple months ago in August and there was great response from them. I've also gone to the community health centers around the area and shown them and talked to the primary care providers. This was fostered by Andrea Vellante who is a team member and a professor in the Vermont Center for Behavioral and Health. So that's our connection there. We also, I'm also working on the evaluation of their new rural center for addiction treatment. So I think through that rural center, hopefully there'll be more connections that can be made so we can help bring what we find to those people who are making more of the decisions. So I think that's wonderful. So whatever we're doing in our work, I'd be happy to come back here and make any connections so you could suggest for us to share it with. On the analysis, were you able to do any deeper dive into variations from health service areas? So as I mentioned for this one, we're looking at a rural urban commuting area. So that is more of a morality indicator. So we have more of an urban area up in Chittenden County. And then we have three other levels of morality. It's not specific to hospital service area. How do you break that down to those three levels? So that is a designation that's a federal designation called rural urban commuting areas. And so we use their designations based on our zip code of the population is put into these. There's four categories in Vermont. There's 10 that are usually, but we don't have the more urban areas in the state. So we're down in the bottom four categories. Similar in Maine as well. Was there significant variation? For this right now, we're not seeing that a specific area in Vermont had more or less of a response following the opioid policy change. So what we're seeing in one area is similar to these other rural areas. Hospital service area though. Let me think, we did not, we had also broke this down by your rural areas. For developmental screening, we could apply hospital service area to this. And I think that was one of the first questions of Jessica was that, yes, we have that information. We have all the primary care practices in their hospital service area and we can easily group them by hospital service area. We didn't do that yet, but it is definitely something we could do. And asthma control, this was very specifically looking at who participated and who did not. And it's across the whole state. So we didn't further separated them by hospital service area. But for our attribution, we do have all of them grouped by hospital service area and thinking about where children are being seen based on the area that they live in. Are there comments from the public? Thank you very much. Thank you. Do we have partners from the Health Department? Yes, we do have our partners from the Vermont Department of Health who will share with you the analysis that they have done related to mental health, the emergency department claims for Vermont children. So at this time, I'll invite Laurel and Lauren up to the mic and your slides are coming. Good afternoon. I'm Laurel Oland. I'm the director of the Child LSD family unit at the Department of Mental Health. I'm Lauren Kaseigin. I'm a Centers for Disease Control and Prevention epidemiologist assigned to the state of Vermont's Department of Health and Department of Mental Health. So we're here to present a project that actually was conducted by an epidemiology fellow under the supervision of Lauren and in partnership with our Children's Division at Department of Mental Health. And so to give a little bit of context for what it was that we were trying to learn through the V-CURS data. As you are probably well aware, there's a concern about children and youth who are in a mental health crisis, waiting in hospital emergency departments. And yet at the department, we lack information and data about those individual children and especially those on voluntary status because they are not, there's no requirement that information is reported to the department. And so we wanted to better understand kind of the picture of who those kids are, what's the, how many are going to emergency departments with a mental health need and what is kind of the clinical picture behind that. And hoping that that could help inform some policy or program development and perhaps even some workforce development needs. With the ultimate goal to really inform changes so that we can reduce the use of emergency departments for kids. While it will always be a need there, we do believe that that may not be the best setting for children who are experiencing a mental health crisis. So this project helped to gather more information for us. So for some background, this is some national information. This is not any to Vermont, unfortunately. So about one in six U.S. children age two to eight are estimated to have a diagnosed mental health or behavioral developmental disorder from 2016. Psychiatric visits accounted for about eight to 10% of all emergency department visits over the years of 2011 to 15, again, national data. And there has been an increase in the number of youth visiting emergency departments with psychiatric needs. And it is, of course, acknowledged, as I said, that emergency departments are a safety net, they're essential for our system. But we do believe there are some challenges in really being able to address the needs of children in those settings and so wanting to, again, understand who is showing up there, for what reasons, and how can we get, perhaps, upstream and divert and address needs outside of emergency departments. So I'm going to cover a lot of the data. There are a ton of slides. I'm going to give you a high-level picture of the slides. We wanted to acknowledge Anita's deep work on this. This is the first time that the Department of Mental Health has actually used V-Cures data. So this is a first foray for us, and now the Adult Services Division is very jealous. So what did we already know? So from the Department of Mental Health, we know that from our designated agencies or those community-based mental health services and incentives that about 11,000 children were served in fiscal year 2017. 1170 of those clients received emergency or crisis assessment, support, and referral services. About 283 received emergency or crisis bed services. 2,600 received emergency or crisis services that were actually delivered. And then about 3,300 days of emergency crisis bed services were delivered. So what did we want to know? So we asked Anita to kind of give us a landscape of children's mental health at a population level. So looking at all of the claims, regardless of whether or not they were a client of the Department of Mental Health. So at a population level, we wanted to know how frequently children used emergency departments for mental health-related conditions. We wanted to get a better idea of what types of diagnoses were reported on those claims. And from a broader scope, we wanted to figure out how to use those data to drive what services were needed within an ED and what could we do before a child enters the ED. So here's some highlights on methods. We're gonna cover very broadly how we chose our data source, what types of tables that we used, how we created our mental health categories, and then some of our analysis and then our findings. So as I said, this was a first use by the Department of Mental Health. We considered hospital discharge data, syndrome surveillance, and Medicaid claims as well. But for the wants and needs of this project, we really wanted to get a sense of the records for every Vermont child that visited an ED. We wanted the ability to identify unique children. And we weren't able to do that with some of these other data sources. We also wanted to follow children across multiple years. Again, we couldn't connect both child and year in some of these other data sources. And we wanted the ability to see multiple diagnoses associated with a singular visit. So you'll be curious with the winner. And we can go to the next slide. We also entertained both the pros and cons of using V-Cures. We felt that it presented an opportunity to look at the majority of children who visited an emergency department. We had the ability to identify unique children, to follow them longitudinally, and to look at all of those diagnosis codes. The cons for this was the policy change that Valerie mentioned about the GOBE decision. Some of the limitations that we have with claims data, they are only good as they are billed. So if we weren't seeing what was billed, we couldn't make that assessment. There's no time of admission to the ED on these claims. So if a child was there for less than a day or for more than one day, that's a little bit hard to assess. And about the time of day when the admission occurred. And there's a complicated layout where particularly compared to hospital discharge data. So our inclusion criteria were primary facility claims of Vermont residents under the age of 18 years of age. We visited an ED, excuse me, in Vermont or in New Hampshire. We also included New Hampshire primarily because Dartmouth Hitchcock can be a big draw from the Eastern side of the state. And parents may choose to use that emergency department system. And then you can see the various codes. One thing that we were looking at was how to characterize our data. So we needed to know both ICD-9 and ICD-10. And we looked at various categorization methods. So some people use something like just grouping the codes together. So for ICD-10, you might just look at quote unquote the F codes for all of the mental health diagnoses. And it's fairly similar to the DSM, but not exactly. So another way that we could look at this is something called the faces of Medicaid. It also didn't fit very well, I'm sorry. As well as something called the clinical classification software. And some of you may be aware of what that is, but for those of you who aren't, it's from the Agency for Health Care Research and Quality. And it's a way to cluster the patient diagnoses into a manageable number of clinically meaningful categories. So we decided to use this as our group. And these are the mental health categories. I'll just let you read them for yourselves. But it really ranges from adjustment disorders all the way to substance related disorders. There's also a category for screening and history of mental health and substance abuse codes. They can be pretty specific. And we were comfortable with these classifications. So some things to keep in mind. There were three big changes in 2015, 2016, and 2017. We had the ICD-9-CM to ICD-10-CM transition. We had the GO-Bay decision. And for the Vermont Department of Health, which is where we used their dataset, our data ends on September 13th, 2017. And we've had no updates to that since we got that first load of data. So we're looking very much forward to the new platform. This is kind of an effect of the ICD-9-10 transition. So in the top kind of beige color, if you're looking at suicide and intentional self-inflicted injury, there were a handful of codes that described this really well. When we moved to ICD-10, there was much, much specificity. And so in the bottom, the same things that were being coded for an ICD-9 went to this and ICD-10. So much more difficult to get your arms around. We also had the GO-Bay decision and that data, we just didn't have the quarter four data for 2017. So when we look at slides in the future that talk about data across the various years, you'll see a dotted line of one, two, or three. And that just designates that this was the limitation for that particular year. So I'll go over some high level results. So the slide here talks about the pediatric population. Where knowledge, the population under 18 Vermont from this time period was 123,000 plus children. On the right hand side, you can see the distribution by age group. So about one third are under six years of age, almost a third are six to 10, about 18% or 11 to 13 and 25% are 14 to 17. And about a 50-50 split on the distribution of sex. This slide highlights the percentage of claims that come from a mental health diagnosis in blue. And that's from the first six diagnosis fields. So it ranged from about 6.2 to a high of 7.7. And then the top numbers are the numbers of claims related to EDs for each year. And you can see the same effect that Valerie showed about the number of claims decreasing, particularly in 2016 and 2017. And that's just a highlight. So from this time period that we looked at, from 2009 to the third quarter of 2017, there were just over 105,000 unique individuals. 2,200 were unique individuals who had at least one claim of the primary mental health diagnosis and 4,800 with one claim from the fields of one through six. About 8% and 7% respectively were children who had a pediatric emergency department visit or a pediatric mental health related visit and went to New Hampshire. The remaining of the claims were from Vermont. The slide highlights the sex distribution for the Vermont pediatric population at the bottom and then without a mental health diagnosis in the top tier and in the middle tier with a mental health diagnosis, not much difference. But we did notice over time that there are more claims filed for females than for males. And that split occurred sometime after 2012 and has been maintained. We also looked at just how the distribution occurs in the pediatric population by the age categories. So without a mental health diagnosis, EDs are often attended most of the claims come from the population of under six. But for a mental health diagnosis, the bulk of the claims, 63% come from children who are 14 to 17. We also looked at the rate of claims. And again, you can see that the bulk of the claims are really from the age group of 14 to 17 over time. When we looked at age and sex together, we saw a different pattern for claims that were without a mental diagnosis compared to those with. Again, for both males and females, the bulk of the claims are for individuals 14 to 17. Then the next kind of difference, if you'll go to the next slide, for children six to 10. So we also looked at claims based on insurance type. And you can see for both mental health diagnosis and without mental health diagnosis, it exceeds 70% versus the population of children in Vermont, about 50% of those are on Medicaid. So what did we make of this? So when we looked at this, we had several interpretations if you go for that. That's okay. The larger proportion can be that there are more estates with expanded Medicaid. Vermont is one of those. So it could be that this is both a factor of Vermont having a greater allowance. So 312% coverage along with the reduction of claims from the self-funded or where the, we don't really know in that commercial sector that includes, that would include both a parent self-funded employee plan or a federal health insurance or self-paying or from small companies that are required to report. So we think that all of those things may be influencing in that maybe an artificial inflation of the publicly funded emergency department claims when you just look at it by proportion. Okay. So some of the takeaway points from the findings, really both primary diagnosis and in diagnosis fields of one through six, there were four order, four diagnosis kind of categories that were very similar, mood disorders, anxiety, attention deficit, suicide and attentional self-entry. The two that were unique between primary and the first six fields are adjustment disorders versus a screening in history of mental health and substance abuse codes. The next one. And this is just a different way to depict that, showing what's happened to these categories over time. That light green one that takes a deep dive from 2012 down to 2016 is the screening in history of mental health and substance abuse codes. All the remainder from around 2012 have increased just in terms of the rate for 1,000 claims. So when we focused on the more, on the, I guess on the mental health claims in the categories that were of great interest to us, we found that for the most part in three of the categories, females had more emergency department claims. The exception to that was in attention deficit, conduct and disruptive behavior disorders where young males led. So I'm gonna cover all of these on separate slides because it's too much to take in. I'll have the next one. Our takeaway for anxiety disorder diagnoses, there's a great discrepancy between the percentage of claims by males versus females for 14 to 17. On the right-hand side is males and on the left-hand side is females. And the same thing for six to 10-year-olds, there are more males compared to females. For attention deficit, we focused on the six to 10-year-olds where there are a number of claims and that's different between males and females by 10 percentage points. It's also a little bit different if you look at the older children from 14 to 17. For claims with suicide and self-intentional, self-afflicted injury, we'll note that it's from age 13 to 17. There's not a drastic difference but females account for 98% of those claims in an emergency department versus 91% of males. And for mood disorders, the big difference seemed to be with the six to 10-year-olds having the mood disorder and it's 9% versus 2% male-female. So our key highlights, there were over 22,000 claims related to a mental health diagnosis in fields one through six. 6.8% of all ED claims related to a mental health diagnosis and that represented over 4,000 unique children. Mental health claims were overwhelmingly from public-insured children, 51% female and 63% for 14 to 17-year-olds. Age and sex, distribution of age groups differed by sex of child and our top three mental health categories of mood disorders, anxiety disorders, attention deficit, conduct and disruptive behavior disorders. Some final thoughts. The general proportion of the claims related to mental health was consistent between 6.2 to 7.7 over this entire time period and the counts of mental health categories varied over time. Some of the limitation is its claims, it's only as good as what gets built. The caveats of the past three years that we discussed earlier and also the use of the beta version still for the clinical classification software, it has not yet emerged out of that beta software yet. So there could be some differences when the final software is engaged. Anita raised some issues rather that this analysis actually led to more questions than it solved anything. We want to know more about readmissions. We want to know more about geographic location of the patient, perhaps even their time to an emergency department for care, more about procedures and comparing with other Vermont data sources. We also want to know how to use this data better for both policy and practice implications. So an example of that that we have been talking about is a proposal that DMH is working with other departments across AHS on about mobile response and support services for children in Vermont as a way to think about getting more upstream and infusing our system in a way that we can have more of a going to where families want to receive supports in moments of distress for their child. So this is not just about doing an emergency crisis screening to determine whether a child needs inpatient or a crisis fed, but really when a family is in distress with their child's behavior, their mental health needs, that there can be a more rapid response to that and in the home for the setting of the family's desire. So this is a proposal that we've been working on with input from, as I said, departments across AHS as well as a representative from our family advocacy group and the designated agencies. One of the reasons why we're pursuing this as we've been talking with children's programs in other states, they also have been developing mobile response services in different ways and what we've learned is that many of them have done it in reaction to something, whether it be a lawsuit or a tragedy and so we're trying to get ahead of that here in Vermont. And part of our process of putting together this proposal has been helpful to access some of this data because as I said, we have information about children who are waiting in emergency departments on involuntary status but we don't have a bigger picture of it and so this has been really helpful to get a sense of what are some of those needs when it gets to the point where a family feels that they need to go to an emergency department. That's just one part of the picture. We're also looking at use of residential and other mental health services across our system of care. We would be glad, I would say on behalf of our commissioner, she would be glad to share more information about this proposal at the board if there's interest but this is just a quick snapshot of what is included in some of that. It's still very much a working process and this is, again, Anita's slide acknowledging the contributors to this project. Thank you. We'll have to go to the board questions and board numbers. So I'm the first one, I'm not the first slide and the fourth slide that the ultimate goal was to meet. I have a sense that there's a lot of data here but that I didn't see at the end here are the things that would reduce emergency department and so I'm wondering if you've spent any time at all with hospital folks to present this data and try to get some insight from them as to how this data might be best used because across our hospitals, emergency rooms, impact and under stress and if you have any good ideas and I'm sure that you might be one of them for us. Our commissioner certainly has been in conversation with the emergency departments. I don't know that we have presented this breadth of the data with them but that's certainly a conversation that we can have and some of the brainstorming that's happening around really trying to digest all of this and how does it inform our policy? When I talked about the workforce development components, there are some clinical interventions that we're also looking at, dialectical behavioral therapy. When we think about the self-inflicted, self-injurious behaviors, how do we address those and how do we have both practitioners as well as those other people around a child or a youth understand what's behind that and how do we respond to those needs? So that's another example of how we're trying to understand some of this, that the ADHD and conduct behaviors, we are having conversations about another, it all comes down to resources, right? So can we look at another roll out or a new iteration of training around parent and skill curriculum that we've done in the past, but it's been several years to ensure that as we have turnover of the workforce that the current folks are grounded in some of those good practices. What are your questions from the board? I just have a question on, during this time period, have there been changes in how medications are prescribed to patients, whether it's like Adderall or anxiety medications? And is that impacting the number of people that present themselves to the ER? Or just wondering if that's changed at all? Because at one point it seemed like there was a push for certain medications and whether that's still... I don't know the answer to that. What I can say is that there has been a group with psychiatrists looking at the prescription rates for children, but I can't speak to you without that information about it. Yeah, it just didn't happen. It's a good question. I guess there's one question about, again, partnerships, and I'm wondering with the Agency of Education, what kind of work can be done in schools around counseling, around prevention? I mean, this supports a lot of the anecdotal and other studies that I've seen about heightened anxiety, depression, some of the links in social media. There's a whole bunch of things that people were hypothesizing. But I just, I see it in my classroom, so I teach at the college level, and I see our counseling center that they have not handled the need out there, and I have obviously worked in the public school system, and they've seen over the last 10 or 15 years just real acute need for mental health services in our public school system. And I'm just wondering, this data supports this growing trend that I'm wondering, is the Department of Mental Health working with the Department of Education in ways that would get some of this in a really, really important way? Every time, I mean, yes. And some of that partnership is not just with Agency of Education, but also the health department, as well as EVM, around how do we work together through the multi-tiered systems of supports that the initiative that's happening with schools to really think about what do all students need, what do those students at risk need, what do those with the identified mental health issues or reading challenges need. And how to best leverage what we have available for school mental health at each of those levels. And we're actually in the process of putting together a legislative report because there's also some questions around how is school mental health structured in Vermont and how is it funded. We do have a funding mechanism under Medicaid called Success Beyond Six, but that's not the only way that schools address the mental health needs of students. It is a Success Beyond Six is a way that there can be a partnership between the school and the local mental health agency. But some schools might not go that route. They might do their own hiring or contracting with the private sector on that. So trying to understand all the different ways that the mental health needs of students can be addressed and not just waiting until they have the identified need, but really thinking about social emotional learning, thinking about how can all staff and professionals within our adults within a school building really have an understanding of creating that welcoming environment for students that really does help to establish a school climate that's responsive to kids' needs as they might develop more mental health challenges. Thank you. This is important. I appreciate it. Can you open it up to the public with comment? I thank you more and more of us very informative and very hard work. It's all the neatest work. Five minute break before we switch over to Kate and the data governance. So at this point we're gonna turn the floor over to Kate again to talk about data governance and stewardship. So I'd like to say thanks so much to our colleagues who shared with you all the good and important work that they're doing using the V-Cures data set. And I am looking to just share with you a little bit broader view of the Green Mountain Care Boards data governance and stewardship program under which data release and use of the data set falls. So the Green Mountain Care Board is the steward of the V-Cures and the hospital discharge data set. So a variety of resources and in this role the Green Mountain Care Board is responsible for a broad set of data management concerns. What you see here is not intended to be an all encompassing inventory but these concerns generally are attributed to these four categories. Risk management which is implementing and enforcing the most appropriate data privacy and security standards and practices. We look at data quality so that we ensure that we have data stewardship that promotes the highest possible quality and relevancy of our data resources. Program sustainability evaluating opportunities to optimize the sustainability and the revenue for our data stewardship program and also data release. So we looked at the results of data release but today where we support clear processes for the evaluation of data requests and the release of data to Vermont state agencies instrumentalities as well as non-state entities. There are a lot of... There are different ways to define data governance and there are different data governance frameworks across the state. Data governance means different things to different people but here at the Green Mountain Care Board we approach data governance in this way and we see it as the management of quality, usability, integrity, security and the availability of our data assets. Our governing body is our data governance council and that's a committee of the board and this council established a defined set of principles, policies and procedures as well as the operational resources to execute those procedures. So the data governance council currently is composed of seven voting members and these voting members are data contributors, data users, data experts or policy leaders and consist of two Green Mountain Care Board staff, one board member, two state agency representatives and two non-state entity representatives. The council meets about every other month in an open public meeting right here in this building up on the fourth floor and meetings are announced on the calendar of events. We have lots of resources on the Green Mountain Care Board website, I would just direct you there but two that relate very specifically to the data governance council and which have been recently adopted by the council are the governance and stewardship charter which outlines the processes and the procedures and the guidance for the council itself as well as data stewardship principles and policies. Currently the data governance council is addressing the V-Cures rule which is in need of update and we're looking at changing that from the V-Cures rule to a broader data asset rule because the Green Mountain Care Board has stewardship over more than just V-Cures and we're updating it and will propose a change to actually create two rules where one is a date submission rule specific to date submitters and the requirements that they will have and a data release rule which would be specific to how we ensure the safety and security and availability of the data outside of the use of the Green Mountain Care Board staff. We also address policy guidance such as data linkage concerns and opportunities. If data is linked together it can be much more robust in terms of the research that you're able to do with data but it also brings up some additional concerns around security and privacy and so we are careful about that and so those are some of the policy that we look at and structures for allowable data release based on intended use as well as pricing structures and why we're looking at that now is because traditionally and even today we have made V-Cures and the hospital discharge data set available to state entities and non-state entities who would roughly fall into academic research realm but we from time to time get applications for other intended uses for say outside of the state of Vermont private entities looking to establish a novel resource for price transparency for example and so we look at the policy considerations for using our data assets in that way and we're also looking at issues around healthcare data related activities at the state and the federal levels and by that I mean we follow what's happening at the state and the federal levels in terms of healthcare data for example there's a bill in the US Senate right now that addresses all pair claims databases and we're following that very closely and then the data governance council does look at specific data release applications and data linkage requests in part to understand better what requests come to us and how data is proposed to be used and also to understand what that means in terms of future policy guidance and to approve applications that have novel use that maybe we haven't seen before. So you heard from some researchers today their data use agreements are represented here on this slide but I wanted to share this with you because the currently active data use agreements that we have are many, the University of Vermont College of Medicine, the Vermont Department of Health and RTI International, also the Vermont Department of Health Access, Department of Disabilities, Aging and Dependent Living, the Vermont Department of Health Environmental and Public Health Division has a specific data use agreement related to the VUDS data set and hospital discharge data set and as well as the University of Vermont Medical Center Trauma and Acute Care Surgery Department. They're looking at the hospital discharge data for analyses, NMRC is doing an evaluation of the all-pair, the Vermont all-pair ECO model and they have a data use agreement and the Office of the Vermont State Auditor is examining the impacts of healthcare reform on healthcare expenditures. They're using V-Cures for that as well. We have a pending application that will go before the Data Governance Council at the next meeting, hopefully. And that's from Archway Health Advisors where they're looking to identify best performing providers for developing an episode payment market in Vermont. This is an example of a novel use that the Council and the Board itself hasn't really contemplated in any depth in the past and so this is something that will be a very interesting application to look at and to discuss in terms of its intended use here in Vermont or to benefit Vermont but may extend beyond Vermont. And I think that's it for me. Great, thank you, Kate. Are there any questions from Board members? Is there anyone from the public who wishes to stop at any times? Thank you, Kate. You said it, great afternoon. Thank you. Thanks for the opportunity to share. So at this time, we're gonna move to old business. Is there any old business to look before the Board? Yes, we're gonna mention forward and then we're gonna, of course, set it up. In the hospital budget process, we approved a 5.9% change in NPR FPP for UDM and this excluded the adjustment to reclassified payments for reform investments from expense to NPR. Although we had discussed these changes, we did not correctly make the adjustment for UDM. We did for Porter and CDMC. This change would result in a 6.4% growth rather than the 5.9% growth that we approved. Importantly, it does not change the total NPR for 2020 nor the change in charge that we approved for UDM. And also, it does not change the projected NPR growth for 2019 over 2018, which is 3.5%, or for 2020 budget over 2019 projection, which is 3.9%. I'll put forward the actual motion and then we can discuss if we need to. So the motion is to recognize UDM and CDMC's accounting adjustment to reclassify payment reform investments as NPR FPP deductions, which results in an effective NPR FPP growth rate of 6.4%. UDM and CDMC's from fiscal year 2020 total approved NPR FPP remains the same at approximately 1.3 billion. Is there a second? No, thank you. It's been moved and seconded. Is there further discussion? If not, all those in favor signify by saying aye. Aye. Any opposed? Motion carries. Is there other 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? Been moved and seconded to adjourn. All those in favor signify by saying aye. Aye. Any opposed? Thank you, everyone. Have a great rest of the day.