 Good afternoon and welcome back to the Green Mountain care board meeting. We're going to start with the executive director's report. Susan Barrett. Thank you, Mr. Chair. I'm scheduling announcements and updates. So for the month of February, our schedule will be posted tomorrow on our website. So I'd encourage folks to check out our press release on our calendar page. And just as a preview for the next Wednesday, February 3rd, the board will be hearing from our partners at DEVA on the Qualified Health Plan Standard Plan Designs. So just like that, the rate review season is starting. That's all I have to report. Thank you. Thank you, Susan. The next item on the agenda are the minutes from last week's meeting. Is there a motion? So moved. Thank you. It's been moved and seconded to approve the minutes without any additions, deletions, or corrections. Is there any further 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 that the minutes were approved unanimously. So next I'm going to turn over the meeting to Michelle Degree to introduce the team from Mathematica. Michelle. Good afternoon, everybody. So today our agenda is a discussion on the peridifferential reporting as required, the all-paramodal agreement, section 10, if I'm remembering correctly, although it's not right in front of me. And so today we're going to talk about three of those reports, the annual report, the peridifferential assessment report, and the peridifferential options report. As Chairman Mullen noted, we worked with Mathematica Policy Research on this work. And they're here to walk us through that today. And I believe with me, I saw Shule lock on. I believe we're also expecting Vincent and Carrie-Ann from their team. This work was conducted sort of throughout the summer. Dealing with quite a few data delays, as we're all aware of that have been happening as a result of the public health emergency. And we worked closely with the Agency of Human Services and Diva on the production of this report. So with that, Shule, if you are able, you're welcome to share your screen. I do have the slides on deck if you need me to handle that for you. But with that, I will turn it over to the Mathematica team. Shule, you may be muted if you're speaking. Can you hear me okay now? Yes. Double muted as usual. Good afternoon, everyone. Thank you for the opportunity to present the presentations on the three reports. Before I get to it, I wanted to acknowledge and thank the team, Carrie-Ann and Ms. and Paul work with me on this report, as well as on point health data, who is our subcontractor provided the data that we needed on the APCD. In addition to Diva, commercial payers also created and provided us valuable information to get to a stage where we feel confident with the results of the report. In addition to Michelle, Sarah Lindberg, Alina Derube from the staff, appreciate their feedback as well. So this is really a big and important project for us and very complicated topic as you're gonna see in a minute. So with that, I'm going to share my screen. So I do have two screens. I wanna make sure I have the right one before we get to it. Actually, all right, let's see. Okay, do you see the slide there? Yes. All right, now if I do the presentation mode, I think you are seeing both of the slides. So I'm gonna make the next slide smaller. Is this a good size for you all? Yes. All right. As Michelle mentioned, we have three reports. So at the end of 2020, we were able to complete all these three reports. So we have a Torah picture of the payer differential. The agreement specifies that we focus on all payer, ACO benchmarks in all these three reports. So one thing to note is that we are really looking at the ACO benchmarks and the differences between payers who have contracts with the ACO. So it will be the commercial Medicaid and Medicare ACO benchmark analysis in all these three reports. We changed the order of the presentation a little bit and we wanted to start with the assessment report and then we'll look at the annual change report. And then finally, I will summarize the options report. We spend a lot of time on the assessment report, developing the methods. So we have apples to apples comparison between ACO benchmarks and that kind of triggered down to annual change report as well as informed our considerations for the options in the third report. Moving forward, we are obligated to report the annual change to CMMI. So that's gonna be an ongoing reporting that we would do, but the methods will remain the same as we reproduce the change over time for the annual report updates for the CMMI. So with that, I'm going to the methods for the assessment report. So as you all know, we have differences in ACO benchmarks and the first step is thinking through what the ACO benchmark rate is gonna be. So for the payer differential, we picked Medicare as our reference. So in this slide, what you are seeing is the denominators as we are looking at the payer differential is the Medicare fee for service ACO benchmark rate. So what we compared in these reports is what we call payer differential. So that would take the Medicaid ACO benchmark rate compared to the Medicare fee for service ACO benchmark rate. And for commercial, the same that applied. So we are comparing the commercial ACO benchmark rate to the Medicare fee for service ACO benchmark rate. As we went to the next stage, we needed to align the way the ACO benchmarks were calculated. So in this case for the Medicare fee for service, the benchmarks are set on paid amounts. So the benchmark numbers that we received from the commission is excluding the co-pay and coinsurance that is on the patient side. Conversely on the Medicaid, as you all know, Medicaid do not have co-insurance and co-pay. So we had to make a decision on how we are gonna compare the Medicare and the Medicaid benchmarks. Even though both of them are on the paid amounts, we are missing that co-insurance and co-pay on the Medicare side. Because of those differences, we calculated an allowed amount version of the Medicare fee for service benchmark. So if you're familiar with the benchmark numbers from other reports, keep in mind that our numbers may look different compared to what you are seeing in other reports. The other point of difference was on the Medicare side. There is a prospective benchmark set, but at the end there is a resettlement and that benchmark is recalculated based on the performance of that year. On the Medicaid, as you all know, it's a prospective per month per member rate that the Medicaid pays prospectively. Again, because of those differences, we adjusted our benchmark numbers. So we are looking at what I would call the end settlement rate that takes into account all those adjustments that happens throughout the year. So again, we are looking at apples to apples comparisons when you look at the benchmark rate. Finally, this was a simple math. Some benchmarks are published as the per member per year versus per member per month. So for simplicity, we converted them all to a per member per month rate. So our numbers are looking similar to each other as you are gonna see in the next slides. The third step that we needed to do is to consolidate some of the rates that were in non-Medicare hairs. On the Medicare fee for service, we have a single ACO benchmark that we are using for non-ESRD patients and stage renal disease patients. So that's the most aged Medicare beneficiaries. For Medicaid, we have three rates. We have an adult rate. We have a children and ABD age disabled and blind rate. So we needed to come up with a methodology to combine all these different rates into a single number so we can compare it with Medicare adult rate. For this math, what we decided to do is we created a weighted average of these three rates because if you think about the way the Medicare is constructed, it is the population of all the elderly that gets into that rate development. Similarly for Medicaid, we aggregated these three rates using the population numbers that are provided to us to come up with a single Medicaid ACO benchmark which is weighted by the population distributions for these three groups. On the commercial side, we do have different contracts for each commercial payer and it didn't make sense for us to weight them based on the attributed lives for each commercial plan. So for commercial plan, what we are looking at is a straight average for the commercial rate as we compare the commercial benchmark to the Medicare fee for service. Finally, a big work that needed to happen was to think about the covered services and additional adjustments that we needed to make. So I already mentioned the patient cost sharing that was part of the Medicare and Medicaid comparisons that we needed to do. In addition, there are some additional payments that are included in the contracts like administrative fees, additional fees for coordination. So we made some adjustments to the additional fees to make sure that the benchmarks exclude those additional fees that were only specific to a special contract for Medicaid or commercial payers. So they are all, again, on the same medical expenses benchmark targets. In addition, there were some adjustments that we weren't able to do, which is the specific coverage that the commercial plans have that would be different from the Medicare covered services. In general, our assessment of the ACO contracts is that the covered services are very similar. The additional coverage that was in the commercial plans were removed if it was a big, significant amount, but the detailed coverage and some of the requirements around inpatient services, our patient services obviously wasn't part of the detail that we were able to get. So there are still some design differences that are still in these numbers. On the Medicaid side, as you all know, the Medicaid contract is also based on Medicare covered services for hospital and physician services. Medicaid excludes long-term care, which is gonna be a big one. And as we're thinking about the options, I will take that back again to consider the implications of the services not covered by ACO contract, but provided by Medicaid or commercial payers. Finally, as we're looking at the comparisons, the what we call a health status risk adjustment is an important factor. The utilization rates will depend on whether the patient population is sicker or healthier. And to control for those differences across three payers, we chose a single common risk adjustment methodology for all payers. This is a methodology developed by the Johns Hopkins group called ACGs, a just clinical group widely used by many payers, including Medicaid. So we use this algorithm and calculated the scores for entire population in the ACO attributed lives in our all payer claims database. But because it still has some limitations, we excluded and stage renal disease population from Medicare because that population is a very high utilizer, high risk patient and we didn't feel comfortable to have them in the calculations as we're looking at the average ACO benchmarks. The risk scores are calibrated to predict future cost. So they are prospective in the sense. So what we are trying to estimate with the scoring is their healthcare costs in 2018, looking at their historical costs in 2017. So with that to be calculated an expected score that would adjust for any cost variations due to their disease burden. We used a nine month continuous enrollment. So we have enough data from the claims for us to be calculated a reasonable risk score. And as I mentioned, our data source for risk scores is the APCD, recures medical claims. We did not use the pharmacy claims, which did not have a significant impact based on our literature review on the risk scoring. And finally, as we looked at the 2018, we did not make any exclusions for the ACO enrollment. So if the ACO aligned members dropped coverage during the year, it is part of the calculations that we have in 2018 estimates. Am I going too fast? Should I slow down a little bit? Just a hair. Okay, I tend to speak quite fast. So I'm doing my best, I will. So maybe before I get to the results, let me summarize what we did, right? So we take the ACO benchmarks, we adjusted it for differences in clinical services. And then on the population, we adjusted for health status differences between Medicaid, Medicare and commercial plans. So that's kind of at a high level, two significant adjustments that we needed to make. Once we adjust for the risk scores, we didn't find everybody in the data sets, right? Because we are looking at the 2017 claims, and in this table, you are seeing our verification results to verify that in our analyses, we have the majority of the ACO aligned beneficiaries in our analysis. So in the first column, what you are seeing is the count of member months that we included in the payer differential report. So it's about the million member months that are aligned with the ACO. And when you look at the V-Cures, when we have those ACO alignment at the claim level, you see that those numbers are slightly higher. So we lost about 2% of the ACO aligned beneficiaries because we didn't find historical claims for us to calculate their risk score. And you'll see that in general for Medicare and Medicaid, those numbers are quite small, but you do see for commercial that the loss, the percent lost in this analysis is about 5% of the commercial ACO aligned beneficiaries. So instead of 283 member months, we were able to base our estimates on 267. Talking with the commercial payers and the other folks who are familiar with the attribution, we think that this is a result of changes in the commercial plan coverage in 2017 and the way that they attribute the members. Still relatively speaking, that number is quite small. So we felt confident to move forward and look at the commercial ACO benchmark results with the statistics in mind. Then the second validation, we looked at the risk scores. Again, risk scores are meant to adjust for expected healthcare costs because of the health status. And as you expect, you would see that the Medicare risk score should be higher given the disease burden for the Medicare population compared to Medicaid and commercial. Overall, in general, what you have is the first column is the ACO aligned members. So when you aggregate all the ACO aligned members, what we are seeing is that the ACO aligned population is 2% less healthy than the general population in Medicare. So ACO has somewhat high risk patients in their attributed lives at 2%. When you look at Medicare, the Medicare is 48% higher overall. And here, we verified it by looking at Medicare only and dual-eligible. And you'll see that the dual-eligible score is 90% higher than the state average. And this is, again, reflection of the high disease burden on dual-eligible Medicare if you for service beneficiaries. And for the Medicare only, again, what you are seeing is a 42% compared to our state average in the health score. Moving on to Medicaid. Medicaid ACO attributed lives are healthier than our average in the state. So it's 0.84, converting to a percent. It's 16% lower than average health score in the vehicures. Again, here, you do see an expected variation with adult population to be sicker, 20%. The children being very healthy at a 0.39 health score average. And then the ABD, which is the adult, age-blind and disabled, you see that their disease burden is twice as much as an average risk in our vehicures population for the state average. Going to commercial. Commercial has a healthier population and their score is at 0.72. So all these findings was in line with our understanding of the populations and the way that the attribution algorithms worked. So after this verification, we went to adjusting the ACO benchmarks using these health risk scores to come up with a comparison report. In the results, so in this table, what you have is our statistics to calculate the risk-adjusted benchmark rate, which is on the fourth column. You'll see the member months that we included as a reference and the average rescaled risk scores that are in the previous slide. So using these two statistics, what we have is the ACO benchmark rate, and that's adjusted using the average risk score to come up with a comparable risk-adjusted benchmark rate for peer differential. So if you follow the Medicare fee-for-services example, their ACO benchmark rate is 806 per member per month. When we adjust for the disease burden, which is 48%, you'll see that the risk-adjusted benchmark score is now at 645, right? You reduced it because of the disease burden for the Medicare fee-for-service population. And similarly, when you look at Medicaid, Medicaid benchmark went up because they have a healthier population. And for commercial, their benchmark also went up because of the healthier population that they have in the ACO-aligned member months. Once we compare those numbers, what you see is what we call the peer differential ratios. So compared to Medicare, Medicaid has 0.45, again converting to a percent. Medicaid ACO benchmark is 55% lower than the Medicare fee-for-service benchmark. And conversely, the commercial average is 5% higher than the Medicare fee-for-service benchmark. So this is our final result for looking at the peer differential, again, focusing on the ACO benchmarks only. We also did some subgroup analysis. And here you see that the impact of different beneficiary groups, I am not gonna go into detail, but in this slide, you could see that the Medicare-only population is 5% higher than the Medicare average. And what you are seeing on the Medicaid side is that they are similar to each other because Medicaid is setting the ACO benchmarks for these individual groups separately. So they take into account the risk factors for these three groups and their ACO benchmarks are adjusted for those separately compared to a Medicare where Medicare is using a single average for both Medicare-only and dual eligible population. So after the results, we did some additional analyses and I'm gonna talk about the limitations of the numbers that we calculated. The ask and the scope of the report was once we calculate the peer differential, the consideration was how is this impacting the ACO and ACO financial benchmarks? So the question here is, as we calculate this differences, is that something about the way the Medicaid set ACO benchmarks that explains that difference or is there something else that is explaining that 55% difference that we are seeing in the benchmark analysis? And the second one is because the healthcare costs differs by geographies, are there other factors that we need to take into account? It could be low cost areas, it could be socioeconomic factors that may also explain the differences that we are seeing between the ACO benchmark for Medicaid versus the Medicare or on the commercial benchmark. So I will share with you two analyses on these two topics. The first one is the question around how the ACO benchmark is calculated. Here we looked at the 2018 cost and tried to test whether these benchmarks were similar to the expected cost given the fee for service schedule that the Medicaid has or commercial payers have so that we can have some assessment on whether the differentials that we are seeing in the benchmarks are a result of the way the benchmarks are set or is it coming from something else such as a fee for service or to claim historical costs that we have in the benchmarking methodology? So this is somewhat a complicated view. I would like to freeze this next table in this context that this was a data point for us to take a look at. Definitely the use of this data is limited because when you look at the cost in 2018, there are multiple factors impacting 2018 cost estimates, right? The big one is the ACO performance. If ACO manage these populations differently and their expected cost in 2018 is different than the ACO benchmark, we haven't taken into account those type of factors in this analysis. Secondly, we didn't take into account any of the membership changes during the year that may bias this analysis as well as some of the additional impact on the cost. For example, the mortality and the percent changes in the dual, eligible, et cetera, that is not part of the analysis. So what we wanted to do is let's take a look at what the 2018 cost look like and evaluate the benchmarks from the cost perspective. And here by cost, I mean the claim-based cost. So this is if they were not in ACO, what would be the PMPM payment amounts for the aligned population, right? So with those caveats, what you have in this table is the risk-adjusted PMPM costs. So these are the claim-based calculations that we did for 2018 in the first column. And the next is the risk-adjusted benchmarks. So we expect that those two would be related to each other. If they are not, then we could conclude that the differences that we are seeing in the Medicaid and commercial ACO benchmarks are related to something that they are doing in the ACO benchmark methodologies. Here, the important numbers are in the last column, the cost to benchmark ratios. And what you are seeing is a one to 2% differences between the claim-based cost versus the benchmark, which made us to conclude that the differences that we calculated as peer differential is a function of the fee-for-service claims and the payment differences. It is not a function of the way the payer said their benchmarks differentially. So this is kind of an important and difficult concept. So I'll be happy to answer the questions after the presentation if you have any. The next table, we looked at the geographies. Here, the hypothesis is that if certain payers have more attributed patients, members in certain geographies, would that bias the statistic that we are calculating in the ACO benchmarking? And in this table, what we are presenting is the similar cost to benchmark ratios by the hospital service areas. So in Burlington, that ratio is 0.924 Medicaid and 0.944 Medicare. And in this case, if Medicaid percent ACO members is 24, if that was higher, that would reduce the cost of benchmark ratios for Medicaid. In this case, what you are seeing is for Medicaid is 24% of the attributed lives versus 42% for Medicare. So it didn't have much of an impact on the differential that we were calculating. We did do additional analysis taken to account this variation across the different hospital service areas and our conclusions around peer differential did not change. In other words, the 55% differential we are calculating for Medicaid is not coming from any geographic distributional differences in the Medicaid population versus Medicare fee for service. After those two initial analyses, we wanted to outline the limitations to put these numbers into context. ACG risk scores are commonly used in a valid healthcare utilization measures, health status measures, but they still do not take into account all the variations. So we might be still having some measurement limitations around health status. We didn't add additional factors such as mental health. ACG is supposed to take most of it, but there might be additional factors that we may consider in the future. The second one is we didn't examine differences in service sites. So you could consider some of this variation coming from heavy utilization and expensive services such as hospital-based outpatient clinics versus more physician-based clinics. So that might be a factor explaining the benchmark analyses that we did on peer differential. And thirdly, as I mentioned, the apples-to-apples comparisons are limited because of the way the plan designs are working for each commercial payers as well as Medicaid and Medicare. And finally, as we are thinking about the peer differential, we really took a very limited view for the peer differential and we did not consider some of the variations coming from the attribution methodologies and some of the additional funding that the Medicaid is providing for high-cost services such as long-term care. All right, so this was the assessment report when we looked at the differences between payers in their actual ACO benchmark. How did the benchmarks change over time is our second question. Here we are looking at the growth rate in the benchmarks. The same risk adjustment is applied. And here we have two growth rates. One is just actual benchmark growth rate. So we're looking at what did they set in 2018, what did they set in 2019 and looking at that change over time. The second one is we calculated what we term update factor. This is looking at the change in attributed live costs. So in the first one, because the attribution change, network change over time, the statistic that change in the annual growth rate has a function of changes in network and changes in population, as well as changing in the financial expectations from each payer, right? So it does have a limited view into how actually the payment amounts are changing over time. So to get gauge on that payment amount question, we also calculated what we call update factor. In this case, it's the same attributed population looking at their base year estimated cost versus what the ACO benchmarks are. So the first one is year over year. Just the ACO benchmark change here, you do see a drop in the ACO benchmark rates. But the drops are similar in both payers and we weren't able to calculate the commercial payers because of the lack of data for multiple commercial payers. So in this case, looking at the 2018 to 2019 on a weighted basis, Medicaid ACO benchmark dropped by 2.4%. And on the Medicare, it is dropped by 2%. So they are very similar. When we factor into the changes in network and attributed beneficiaries, I believe this is the more relevant data to consider. So when we look at the 2018, when the payers set the 2018 benchmarks, compared to the total cost in the baselines, Medicaid benchmark was 5% higher than the estimated cost. And for Medicare, it was 3.5%. When you look at the 2019, the similar number is now at 2.2%. And I believe there were some work done on the Medicaid increasing fee for service payments for Medicaid. That may factor into that 5% growth rate for Medicaid as well as some additional administrative fees, other funding that may have been in the 2018 numbers. The 2019, we do see the alignment between Medicaid and Medicare around the 2.2% what we call the update factor. All right, so now we looked at the assessment, we looked at the growth rates and now finally where we are with the payer differential and the options. So for this report again, we focused on the ACO benchmarks, but we wanted to think a little bit broadly what this differential mean. And we did not analyze additional things, but you're gonna see some ideas around to take this work further as you're thinking about the payer differential and what the next steps are to reduce that differential. Here, we discussed the options with the GMCB staff, Diva and the commercial payers to get their feedback and come up with a three options that was included in the report. So the first one is thinking about the broader aspects of the ACOs and factoring into the changes in the attributed lives. And really if the differential is considered both in terms of the benchmark and the impact on the population, how many people are enrolled, how many people's care is coordinated would be an important factor to consider. So that's the number one, shifting the focus to the scale target conversations. Number two is thinking about the benchmark. Our analysis concluded that the differential that we have seen is coming from the fee for service based claim payments. The option could be to think about the benchmark calculations differently as you are preparing for the second phase of the model and evaluate different alternatives for constructing benchmarks across different payers. And third, there is an interest in the policy to look at the payer differential in a wider context. So ACO was the significant portion but could we look at additional analyses to look at cost shifting or are other benefits that are covered by the Medicaid and Medicare and assess the payer differential from the full benefit perspective. So this slide is providing you with a summary of the scale targets. So the scale targets measure the percent of ACO line beneficiaries. In this case, you have in the dotted lines are the scale targets established by the agreement. So the top one is the old payer, sorry, the Medicare target, the red one. And the below is the old payer target in the yellow one. As we look at the 2019 and 2020, what you are seeing is in the green line you see the progress the Medicaid has made in 2019 by changing their algorithm and including geographic attribution in their ACO contract. With that, the reach of the ACO program to the Medicaid members, eligible members is exceeding 50%. When you look at the old payer, we are way below the 50% target in the yellow lines. The Medicare fee for service is close to 50 in 2019 but you do see a drop in the 2020 and when we prepare this table, we did not have the numbers from the Medicaid and we will be updating this with the Medicaid numbers in the future. And as we are looking at the commercial, commercial payment rates were similar to Medicare but you do see the limited reach to the commercial population averaging around 10% of the commercial members in the state with ACO alignment. The other way to think about the stale target is and the payer differential is how is this impacting the provider participation? So we outlined some ideas around to look at if there are differential provider participation in the ACO and concentrating on Medicaid dominant providers or Medicaid specific providers and evaluating the proportion of those providers in the ACO network as additional information to understand the implications for the payer differential. The second option which is somewhat more major change for you to consider in terms of the phase two development for the ACO program is to rethink about alternative payment models. So here I term the alternative payment under the shadow of historical fee for service cost estimates. If you think about the ACO benchmarks, right? We use the historical fee for service that includes the payment rates as well as utilization estimates from the fee for service legacies and then we estimate the expected growth. We could put efficiency estimates, we could put expected growth because of the medical inflation but at the end of the day we tie ourselves to the historical fee for service legacies. As you are thinking from the payer differential perspective there might be alternative approaches to think about this population based payment method where it is truly about estimating the cost of efficient high quality care that is on a PMPM prospective basis that has those incentives that you are installing in the system but on the financial side creating those cost-based estimates where the payer differential would narrow as a lot of this is coming from the traditional fee for service payment mechanisms. And finally, the last option is to consider the payer differential in the larger context. Here graphically we tried to put a visual to show you the limitations that I mentioned earlier as we were looking at the ACO benchmarks because it's centered around the Medicare covered services. It does include significant portion of the Medicare covered services, part A for hospital services, part B for the physician. It doesn't include the pharmacy, part D which is covered by Medicare but not part of the ACO and as you all know, Medicare doesn't cover any long-term care services so that's at zero. When you look at the Medicare, this is totally illustration but an assessment of how much of this payment and services covered by ACO versus not covered by ACO and incorporating those long-term care services, mental health, et cetera into the equation might give you a better sense of where we are in terms of the payer differential. In addition, there might be questions around the source of this payer differential and thinking about the cost-shifting. Here you could think about looking at the fee-schedule differences and utilization differences to unpack the understanding around payer differential. Site of service I mentioned before, I believe could be an important factor to analyze in addition to the overall PMPM estimates. As we think about this from a wider policy perspective, do we need to think about some cost estimates for providing services which will be looking at the cost of providing services? Not there the payment side but the salary supplies, et cetera to estimate the cost of providing services and looking at the payer differential from cost versus payment continuum and looking at what percent of the cost is covered by Medicaid, Medicare or commercial payment rates. So those are ideas in terms of taking the analysis to the next level to help you to think about the options for reducing payer differential that we calculated on the ACO benchmarks. So that ends my slide deck. I appreciate the time and I know this is a lot but I hope this gave you the overall picture which is more comprehensive than a single report. So with that, I think we could open to questions. Super, thank you. We're going to go in alphabetical order and we'll start with Robin. Okay, I'm not sure how that's alphabetical order but fine. Jessica, I'm sorry. Okay. Well, thank you so much. I appreciate it and all the hard work here. I know that it's a very complicated analysis and I appreciate all the complexity and also your presentation. What struck me at the end was as you're talking about maybe shifting to a population-based payment system that's based on the cost of delivering efficient care and you talked about estimating the true cost of providing service. And I'm wondering how do you go about doing that? Wouldn't it be the first step at really truly understanding that the true cost of delivering a service not linked to what the fee for service payment system says it is but the actual cost of the supplies, the salaries, the equipment, whatever that might be. You know, I attended a couple, I think a couple of years ago there was a health affairs national meeting around healthcare costs. You know, they do annual analysis of the cost nationwide. And I remember one of the renowned health economists that we don't know what the cost of healthcare is. And I think it is a difficult question but I do think there are places where we could start. You know, the hospital, like facilities have cost reporting. I know it is hard to analyze but there are places where we could start turning that conversation and really think as a CFO of these organizations, right? So they do have estimates on what it takes to run an office, right? How many staff that they need to have. There is not a good uniform data set like APCD but there are places where these costs are estimated and calculated on the provider side. And as the board, you may have some initial analyses on the hospital side, on the physician side to get a average estimates at the high level to start picking on that cost question. But how do you, I mean, any inefficiencies and the delivery of care will be baked into those cost estimates, right? So how do you start to unpack what is, what is a cost-effective price? I'm using too many terms here. But, you know, how do we ascertain that this particular cost is efficient? However, it's calculated through the cost reports or whatnot. I mean, how do we start to say this is the true, this is what the cost should be of delivering that particular service. A hospital may be delivering it at above or maybe below, but how do we start to say this is a reasonable cost to deliver a particular healthcare service without using embedded costs already in the system? Right, so I mentioned first, you need to have the cost information. Then your question is then, how do you know whether it's an efficient cost or not, right? And there, two things. One is from the organizational behavior side. You could do management type of analysis to estimate, which is gonna be very narrow, but at least will give you some sense. But more importantly, in efficiency calculations, usually you look at the variation, right? So you do get a sense of the variation across different providers and have some assessment of, there will be outliers. So you may not find an efficient one, you'll find the outliers in type of variation analysis, and that would help you to think about what your efficiency is. There are, that people have used for this kind of analysis that you're aware of? Not established. No, the short answer is no, but there are some work that is happening in other states in Maryland and in some other places where it is happening. What I've seen is you are probably familiar with the Medicare reference pricing, right? So without doing the cost analysis, a lot of work right now is happening around just benchmarking against the Medicare fee for service and see what the variation look like. Again, it is not getting into the efficiency, but it is giving you that benchmark that is applicable to multiple payers. Thank you. Thanks for the presentation and all the smart work. Thank you. Okay, now we'll go to Robin. Thank you. Yes, thank you, Julie. It was a great presentation and the report also was very comprehensive and thorough. And so we really appreciate your work on this. Kind of following up from where Jessica was headed, I was curious, this issue around rate setting comes up in a broader context for us from time to time in terms of differentials in the fee for service area. And so I can see how you can use the Medicare cost reports in a facility for hospitals, for example, and that there are probably similar types of reports for F2HCs or other entities, but how do you tackle it with independent providers where, quite frankly, we don't have any data? It seems like if we wanted to really tackle that, we need to start collecting something similar to a cost report from those providers. My own face there, I'm totally like throwing this in your direction without any prep, so feel free to say. You don't want to comment, but I was just curious about that. I think there are probably the approaches to be targeted and find good examples of type of providers that you want to look deep. And there is the technology piece to it, right? So if you are an independent provider, I think there are significant cost items that you can get your head around, but not a comprehensive analysis of old independent providers. I think what we call more of a qualitative approach where it's more case studies to understand the main cost drivers and assessing whether that's a typical provider or is it a very unique provider that you want to take into account? But I think the other thing on the rate setting, Robin, to think about is we are moving on to a PM-PM type comprehensive approaches and where does that leave us to estimating this cost on a unit, still on a unit base, right? So that is another consideration that you need to think about. The analogy is on the car, right? So we are kind of trying to put the car speed dials and everything else. Do you really care what is under the hood? Yeah, so that's the other place of thinking about what level of cost analysis that you want to do to get a sense of what that benchmarking would look like on a PM-PM basis. Yeah, great, thank you. And that was really my only question, but thank you for your presentation and the options. Thank you, Robin. Tom. Making sure I'm not on mute. So there was a lot there and I'm just wondering as obviously, I'm in no position to replicate the analysis from a technical mathematical point of view, but I'm just wondering that there were so many adjustments in the calculation process as to whether or not there's a margin of error around the results. That each one of those adjustments must have a margin of error and cumulatively, they compound each other to some extent, maybe as you go through the analysis. And I'm just wondering if there's any significance to that. Great question, another great question, Tom. Definitely there is a margin of error if you think about the approach apples to apples comparison, we did our best to get to apples to apples, but we were limited because of the way the plans are designed. So for example, the commercial average is 5% higher than the Medicare. With the margin of error, it could be similar to Medicare. So I think there, the difference that we found could be explained by that margin of error that you mentioned. On the Medicaid side, the number is big, 55%, right? If you think about just the ACO benchmark and then the margin of error around ACO benchmark is not going to make up for that difference. So there is, right? So the 55%, you could kind of throw some estimates, but at the end, that difference is gonna stay significant no matter what the margin of error is, given all the adjustments that we made. The challenge with the Medicaid number is we know Medicaid is covering more than what we measured. And how do you think about that as you're thinking about the payer differential? And that's more of a policy question in addition to estimating that number, right? That number by design wasn't part of the payer differential because we're looking at ACO only, but from your perspective, from the policy, maker perspective, you do wanna know that number as well. Thank you. So this is just a, I mean, I know that this is just a one year window, 2018, relatively 2019, but during that time, this may be more of a macro question. The Medicaid caseload actually dropped in Vermont. And the Medicare caseload, I think, increased because of an aging population. And I'm just wondering if that kind of differential trend, even on a one year basis, has any significance to the analysis. At a high level, you're talking about the annual changes, right? At a high level, we took account the changes in the risk. So if we had more healthier population joining Medicare roles and leaving Medicaid, risk adjustment took into account that. So I think we are controlling for the changes in the alignment and the roles. What we are not controlling, however, is if there was a differentially pent up demand, so for the Medicare, if more people joined Medicare in 2019 and they were waiting for certain services to get it from Medicare, that is not controlled as part of the whole health status adjustments, right? So there might be some potential impact of that dynamic in the 2019 numbers. So I just turned my camera on. I thought I was worried about my muting and I forgot to put my camera on. Just a couple more questions here. Wouldn't over time, I mean, this is a very narrow window of analysis, but over time, if the ACO model is effective in terms, wouldn't you see, begin to see a divergence in terms of the actual experience that's occurring versus the benchmark, which is based on still the rear view mirror of the claims that are in the rear view mirror. So wouldn't over time, you expect this gap to be growing and which would be evidence that the ACO model is working and becoming more efficient and effective? Very good question. And I'm happy that you took that table to cost the benchmark ratios exactly the way that I hoped you would. Those differentials should grow over time as ACO becomes more effective. What we calculate from the fee for service claims should be lower than the ACO benchmarks and differentially it should be different for Medicare and Medicaid. If you think about the population needs, we haven't talked a lot about what we call avoidable utilization, right? If you improve care coordination, patients are gonna use less ED, less inpatient. And when you look at the estimates, there is more on the Medicare population because of their health status compared to Medicaid. So that differential should increase much more on the Medicare side than the Medicaid or the commercial side. One last one is assuming all things being equal in your analysis except one variable. And that variable is that Medicaid did not increase any of its reimbursement rates during 2019. And I ask that because Medicaid, the Medicaid budget for 2021 was presented to the legislature and adopted by the legislature with no rate increases except for those that are federally mandated in Medicaid reimbursement rates. So I'm just trying to get a sense of if that were true in the period of this analysis, that there were no Medicaid rate increases, how would that show up in the analysis, if at all? So, sorry, can you help me understand the timeline? So we looked at the 2018 for a differential, are you referring to 2019 or 2020? Oh, well, I'm taking a real world experience for the 2021 budget. And we were told and the legislature was told and they adopted that there would be no reimbursement increases in the 2021 budget over the 2020. So I'm thinking if in, I'm just wondering how that might show up if in 2019, there were no Medicaid rate increases relative to the 2018 profile. So, I guess in terms of the peer differential, you need to look at what the ACO rates are changing the, what we called update factor. We do see an increase there and it could be not the rate increase, but the utilization factor, right? So they may not change their fee schedules, but how much growth that is gonna be in the ACO benchmark compared to the estimated cost. So you do wanna look at that to see if there is gonna be an impact on the differential. Being everything equal, let's say everything is equal and they're gonna reflect that to the ACO benchmarking, right? So no increases for ACO benchmarks. Then I think then you would look at the Medicare increases and I don't know the detail well enough, but on the Medicare world, you know, there is a 2% adjustment factor for all IPPS inpatient outpatient hospital services, right? So Medicare fees grow, I don't know what the ACO benchmark is gonna be for the Medicare. So I think at this point, you wanna think about the ACO benchmark rate increases that may come in front of you and evaluate it from that perspective. Yeah. Well, thank you. Just one last question floating around in my mind that wouldn't going looking forward option two in terms of your option recommendations, the one that talks about uncoupling the benchmark from fee-for-service wouldn't be that where we would wanna be going anyhow just because that's the whole point to a great extent of what the ACO was there to do is to decouple from the past. And so if we're always looking forward based on the past, then that those benchmarks might not be helping us get, if they're based on the past, helping us get to where we wanna go from a policy point of view. But thank you very much. This is, it was pretty dense. Very dense. I wish my best, I apologize. I wish we had like one report per meeting so that way you can delve into the detail. For your last comment, you know, Vermont has been trade blizzard in multiple places. You are the only state who is doing all PR ACO, right? And if you can crack that knot and figure out how to set the benchmarks outside of the fee-for-service legacy, you are gonna help not only your state, but everyone else who is watching Vermont on the alternative payment models. Yeah, thank you. Thank you, Tom, Maureen. Well, thank you very much for the presentation. I think going near the end, I actually don't have any more questions, but it was very informative. Thank you. Thank you. Thank you, Maureen. At this point, I'm gonna open it up for public comment. Is there any member of the public who wishes to comment at this time? And I see that Rick Dooley has his hand raised, Rick. I do, thank you so much. I'm always struck by these presentations about the complexity of everything involving healthcare in general, but certainly the benchmark setting and the ACO, it's astounding to me. I just wanted to, I appreciate both Jessica and Robin bringing up the, you know, sort of the idea of the site differences. I know that was in your presentation as well and how complex it is to sort of determine the difference in cost between it with independent providers in the mix. And I don't have any easy answers, but I was intrigued and we've kind of circled around this and, you know, we've talked to Greenmont Care Board numerous times and we always come back to the same thing, which is that there's this sort of, you know, gap in data from the independence because they're all so unique that there's not like a representative practice. But I'm intrigued by this idea of having, you know, maybe a, you know, sort of a template that you can provide to a sampling, a statistically significant sampling of practices across the state to get at some of the, you know, some of the data that we've been trying to get at for a long time. I also wonder how much can you extrapolate based on just reimbursement rates? You know, now that the hospital's reimbursement rates, you know, become public after the first of the year, you know, we have a sort of a marker. How much can you extrapolate? Because presumably if you look at the reimbursement rate for an independent practice and there's still a float with that, you look at the reimbursement rate for the hospital, I know there's always some caution if they're still a float with that. How closely would that match to, to cost the providing service, do you think? Does that make sense? It does. I guess, you know, the first one complexity, I agree with you. And the reimburse, I would be very careful using the reimbursement rates because at the end of the day, it's great improvement that we're going to have more data transparency. But in terms of making sense of that data is going to be a challenge. One is like, you could compare a price of a test, right? That's comparable across no matter whether you are a Medicaid patient or not, but once you are in an institution in patient care and they are providing you the service, the level of effort, right? The complexity is not going to be easily seen by looking at the reimbursement rates, right? So that's where it's going to get very complicated in my view to understand the payer differential just by looking at the reimbursement rates. So I would be very careful in kind of teasing out that a little bit. On the provider side too, you know, independence, right? It's their salary. Like I think I agree with you. There might be a good way to get a sense of the cost for the independent providers if they're office-based. But again, there are so many things to think about. So I would take very comprehensive approach initially to make sure that you get what you get. Like for example, in this analysis, we were constrained to look at ACO benchmarks, right? We didn't look at the site of services by design because we didn't set up our systems and analysis to look at the site of service. We just found one initial analysis. Now it's the next day. So being nimble and thinking through the analysis would help to get to a kind of a good analysis and good numbers that you can use. And I think part of the strength of the GMCB is to data-driven approach, right? So you do want the data to be able to make a rational and sensible decisions moving forward. I agree, thank you. Thank you. Other members of the public? Any public comment? No, seeing none. Julie, I want to thank you. I always learn a lot whenever you spend some time with us and today is no different. It's a lot to really decompress and try to figure this all out. So I'm glad you slowed down a little bit. I was struggling at the beginning. I'll be happy to come back next month if that's what you'd like and kind of do another one at a high level. But yeah, I think it's a challenge to explain the complexity and have you get with the take-homes, right? So I hope at least you understand. My goal was to get you understand 60 to 80%. So hopefully we are somewhere around that range. Okay. So thank you and your team. And with that, is there any old business to come before the board? Is there any new business to come before the board? Hearing none. Is there a motion to adjourn? Some move. Second. It's been moved and seconded to adjourn. All those in favor signify by saying aye. Aye. Aye. Those opposed signify by saying nay. Let the record show it was unanimous. And again, thank you everyone. And Michelle, you'll have to give me the version for dummies. We'll work on that together. Yeah, on the way back. Thank you everyone. Have a great rest of the day. Bye. Have a good day.