Transcript of July 21, 1997 meeting (morning) of NCVHS SUBCOMMITTEE ON POPULATION-SPECIFIC ISSUES

I would like to maybe start by each of us in the room introducing ourselves and so we get a sense of who is here and then we will get started with our morning presentations.

I am Lisa Iezzoni. I am from the Beth Israel Deaconist Medical Center in Boston.

DR. AMARO: I am Hortensia Amaro and I am a professor at Boston University School of Public Health.

MR. VAN AMBURG: I am George Van Amburg from the School of Public Health Institute.

DR. MOR: I am Vince Mor from Brown University.

DR. NACHIMSON: I am Stanley Nachimson from the new Center for Medicaid and State Operations in the Health Care Financing --

DR. IEZZONI: The new center.

DR. NACHIMSON: Yes, from the Health Care Financing Administration. We are about a week old.

DR. BLOCK: Rachel Block. I am now the director of the Data and Systems Group in the Center for Medicaid and State Operations and immediately previous to that, I was the director of the Medicaid Managed Care Team at HCFA.

DR. HERZ: I am Lisa Herz with the Medstat Group.

MS. GREENBERG: I am Marjorie Greenberg. I am the executive secretary of the committee.

MS. RIMES: I am Carolyn Rimes, HCFA.

MS. WARD: Elizabeth Ward from the Washington State Department of Health.

[Further introductions off microphone.]

DR. IEZZONI: Good. Thank you.

I would like to first start out by thanking Carolyn Rimes for putting together this program on a very quick turnaround and to those of you who will be presenting to us this morning also for responding in a like and timely fashion.

Basically, the context is that over the next year, hopefully, the Subcommittee on Population-Specific Issues of the National Committee on Vital and Health Statistics will be looking at how the nation is going to monitor the impact of Medicaid managed care.

Today is our kind of kick-off, where we are just trying to learn a little bit about what the issues are, what the data systems are, for example, that can support that kind of evaluation. So, we are very grateful to have you all here and we would like to get started then.

Stanley, you and Rachel are -- I guess, Rachel, you are first off. Is that how you would like to organize this? Okay.

Agenda Item: Overview of the Medicaid Managed Care Efforts, Description of the Medicaid Data Systems and Status of Encounter Data

DR. BLOCK: And I am going to be pretty brief because I think that Stanley, some of my other colleagues, Lisa and others, are going to go into more depth. I just wanted to paint a very quick, broad picture of what is going on out there since it seems as though our ability to stay up-to-date on developments in this area at a very basic level are as challenging as some of the more sophisticated and complicated issues that you are interested in.

The handout that I have brought, Medicaid Managed Care Overview Trends, I am not going to go through this. This is sort of a standard presentation that I often do for groups, but let me just quickly run through some of the numbers and trends that we have seen recently. Then I am going to just make a few observations that I think are broader and will help to focus some of your deliberations over the course of that process that you just described.

The first thing is I think it is fairly obvious to everyone that there has been a very significant increase in the number of Medicaid beneficiaries, who are enrolled in Medicaid managed care and that, at least, from a HCFA perspective is directly correlated to the fact that we have a tremendous increase in the number of waivers that we have been asked to review and approve.

If you flip past the first few pages, which provide some definitions that you can go back and look at later, what you will see with the first bar chart that looks like this is in 1996, this is a point in time annual enrollment figure that we gather every year. We estimated that there were 13.3 million Medicaid beneficiaries enrolled in one of the wide variety of the forms of managed care that HCFA recognizes.

And you will see, in particular, that that represents a very significant increase even just from the one year before that. From 1995, we saw a very significant increase and that this increase has occurred in a very short period of time overall. If you go back to before 1992, you would notice that the numbers were hovering around that 1 million, 2 million mark for a good decade, right up until the early nineties.

So, not only has there been significant growth, but it has really occurred in a very recent time frame. The next one just shows you the same figures in terms of percentages. The next one is a new table that I decided to include this year, since we try to adapt our reports to the number of different questions that we get, particularly from the public, and this is now what I hope to become the famous top ten list.

What you will see are a couple of things that I think are interesting. One is that there is certainly a wide variety of the types of states that are represented here. The middle column indicates the authority under which they are implementing their programs and, here again, there are a wide variety of authorities that states have used and often these authorities are in place at different parts of the state because of the unique geographic characteristics that have driven how states have implemented these programs.

You see that the No. 1 state and the No. 10 state, there is a pretty significant difference in terms of the enrollment between those two states and also that these ten states represent slightly more than 60 percent of the total. So, what that would signify to me is that while there has been this enormous increase and it clearly is a significant trend that we anticipate continuing, we still see that the bulk of the activity, both in terms of populations and the extent of the managed care implementation, varies widely and there are a few states that really still represent a very significant portion of the population now enrolled.

The final chart that I would quickly describe here is -- and here is where those earlier definitions come in. I will try to briefly characterize this. Basically from the largest bar on down, these are HMO programs, either state plan defined or federally qualified and the HIOs represent a unique managed care feature that exists only in the State of California. There are several counties, which by law are allowed to take on risk directly and organize and finance the delivery system.

What you will see here is that now significantly better than half of Medicaid beneficiaries in 1996 were enrolled in some form of HMO or reasonably comprehensive risk kind of arrangement. The bar immediately above that, PCCM, primary care case management, had been the predominant form of Medicaid managed care up until, again, quite recently.

So, here again, we see a pretty significant shift in terms of not only the amounts of enrollment, but also in the characteristics of the programs that beneficiaries are enrolled in.

So, that is my really quick update on the statistics. The reports and the specifics for each state and the plan level enrollment for each state are part of our annual enrollment report. It is available on our Web site. Of course, I didn't bring the Web address with me, but I am sure that we could provide that if people were interested.

I think that there are three issues that I would like to highlight that I think will complicate your deliberations, but I think also have to be part of them and some may be obvious than others.

The first is that, clearly, as you look at these very general characterizations, the first thing that you notice is that there really isn't only one type of managed care arrangement that Medicaid beneficiaries are involved in. This is somewhat distinct from, I think, the commercial population and I think it is particularly distinct from the Medicare population, where now, in effect, we really only have one type of a thing that we call managed care and there are fairly specific limitations in terms of the types of entities that can qualify to get contracts.

Now, obviously, that might change with some of the new legislation and we have a few demos here and there that push those boundaries out, but the fact is that I think that Medicaid represents a far more complex and varied number and range of types of managed care arrangements. That is something that I think needs to be considered as we look at the underlying needs and capabilities of developing and collecting reasonably useful data and information about these programs and their impact on the population.

Also, a corollary to that, which I think probably is obvious but I want to restate, if you have seen one state's Medicaid managed care program, you have seen one state's Medicaid managed care program. There may be some very basic ways in which these programs on paper look alike, but each state's experience with them is really quite different.

I think maybe some of you have seen some of the recent reports that have been produced by Mathematica with funding from, I think, the Kaiser Family Foundation and it describes some of the political history associated with these programs that, again, is just one additional dimension that makes Medicaid unique.

Another issue that I think is critically important in terms of evaluating the impact of managed care -- and I think this would be true even if we weren't talking about managed care, but just focusing on the population -- and that is, by definition, we are talking about populations that have certain kinds of characteristics that make them eligible for these programs. Income, obviously, is one. Family characteristics may be another, disability status and so forth.

As you look at the whole complex web of eligibility requirements and ways in which people can become eligible, including through the 1115 eligibility expansions, which the formerly ORD folks will be addressing, you get the general picture that Medicaid eligibility is quite complicated and an understanding of the importance or the role that medical care might play to these populations needs to consider a much broader web of social and other kinds of programs that probably have at least as much to do if not more to do with health status and health status improvement as anything that the medical care system or the managed care system may be able to directly intervene with, unless there are more explicit connections with some of those different kinds of social and support services in the community, many of which represent other streams of funding from both the federal and state governments.

And it seems to me that because most of those dollars, certainly not all, but many either go directly through states or at least have some connection to some state management or oversight responsibility that states are in a unique position to organize their data in a way that would give you a much fuller picture of not only the impact, which, again, is a pretty sophisticated analysis, but even just broad characteristics of expenditure patterns, population characteristics and so forth.

An example of this, just so you get a better sense of what I am talking about, I think, is the HIV Information System that Maryland developed. They are not the only ones who have done this, but it is a particularly model example of the type of thing I am talking about, where they have a database that integrates basic vital and health statistics information, including their public health HIV surveillance data, along with their Medicaid expenditure data, along with the data that comes from a variety of different programs that the state either administers or puts its own money into.

As you look at that much richer array of information, I think it tells you a lot more about the characteristics and, hopefully, about the outcomes for the HIV infected population than if you were just looking at the Medicaid data alone.

A related but I think separate issue is we are starting to see states now really interested in how to bring managed care models to their long term care systems, either as a separate and distinct managed care financing approach or more likely in some way integrated with the acute care system. That brings up, I think, two important issues that I don't think anyone has really invented the wheel on yet. So, you will be there as the cutting edge is just starting to take shape.

The first is it, obviously, brings up issues about Medicare and Medicaid dual eligibles and the degree to which the different rules of those programs and the limitations that that places on some of the types of models and the state's ability to implement new and innovative programs in that area. While there are barriers, I think we will start to see some very interesting and different models emerging, many of which will be at the state's initiative, some of which will continue to be driven by HCFA's either demonstration interests or, again, possibly other means that Congress might deem worthy of considering.

The second is that, again, obviously, when you are looking at long term care, you are looking at a different set of services, different types of providers, different kinds of information and the way in which you might want to evaluate the impact of managed care for that population may, in fact, be qualitatively different in terms of the types of measures that you would want to consider from those programs that are serving the low income women and children, who still to this day represent the vast majority of beneficiaries enrolled.

Having said that, I don't think we can afford to lose sight of the women and children who are enrolled in Medicaid managed care. One of my big concerns is that everyone has just sort of assumed that we have figured out how to do managed care for those populations. And I don't accept that assumption.

I think that there continues to be evolution and we have seen some indications where in certain states you can see some kinds of things that we assumed were being done well, if they aren't subject to rigorous monitoring, may well not be done as well in the future, specifically they have both prenatal care and so forth, that we just assume are or should be the bread and butter of managed care plans at this point.

There are so many dynamics and so many other factors driving how plans serve these populations that I don't think we can afford to lose sight of those basic issues of evaluation at the same time that our interest is obviously drawn to special populations, long term care, the disabled and some of the other things that I know are more explicitly on your agenda.

So, that is my really quick version of the broad landscape and hopefully a few things for you to consider.

DR. IEZZONI: Rachel, that was a really wonderful introduction for us.

What I would like to do is maybe ask Stanley to do his presentation and then we have the committee begin to talk with you more directly. But let me pause. Are there any clarifying questions for Rachel before we ask Stanley?

No. Okay.

DR. NACHIMSON: Thank you. Good morning.

I wanted to talk a little bit about the basic Medicaid data and the basic Medicaid systems that are now existing in the states. I think it is important to remember that most of our Medicaid data is produced from claim records and eligibility records, which are primarily administrative records, not designed for research purposes.

For example, when someone is enrolled in Medicaid, there is not a health status taken. There is no history taken. If a pregnant woman comes in and enrolls, there is not necessarily a record of the number of prenatal visits that she had before she came on to the system. So, she just enters Medicaid basically as an unknown.

Encounter data in the Medicaid world is usually collected through the use of what we call dummy or shadow claims. It generally contains the same information that is contained on a fee-for-service claim, most of the time without a dollar charge. Some states have instituted sort of a proxy charge that a provider would put on this encounter claim that guesses what they would have been reimbursed under Medicaid had this been a fee for service claim.

And most importantly, HCFA has not yet mandated the collection of encounter data in every state Medicaid program, although there is data for the 1115 waiver programs that is supposed to be collected by the states that have those waivers.

That is sort of the broad outline of Medicaid data, what we kind of don't have. What we do require for every state Medicaid program is an annual summary data report for each state. We call it our 2082 report. Just as an aside, people have asked where does the number 2082 come from. It actually came from the room number in the HHS South Building of the person that put together the report. There is a room 2082 over there. That is why we have the 2082 report.

DR. IEZZONI: Could you tell us more about this report?

DR. NACHIMSON: Yes. We actually have an alternative to the 2082 report, which I will also get into in a little bit more detail, which we call the Medicaid Statistical Information System. It is more detailed data. I will talk about that in a few minutes.

As Rachel mentioned, there is an annual managed care enrollment report that we get from every state. That is a point in time estimate or count of the number of folks that are enrolled in managed care programs in the state.

The 2082 report is, unfortunately, an annual hard copy summary report at this time. It is, oh, about 44 or 45 pages, about 4,000 or so cells of information that comes in or is required to come in from every state. It has summary data on eligibles, number of eligibles, number of recipients, the types of services that are provided, the expenditures broken down by some of the demographic characteristics, like age, race, sex, the program affiliation, whether you are a pregnant woman, child, disabled, SSI recipient, CAPS(?) recipient and those types of things and the type of service.

This data is summarized and we publish it every year. It is also available at the HCFA Web site. That is WWW.HCFA.GOV. And you can sort of wind your way through that Web site to get to the exact piece of information, whether it is managed care enrollment or HCFA 2082. There are some various other things that are on there.

But this is what was started out with. Obviously, it is kind of old technology. We still require it, but we have moved a little bit into the future and you have an alternative for states that don't want to produce a hard copy summary report and that is the Medicaid Statistical Information System. And it does collect detailed person level data on eligibility utilization and payments.

We have 30 states now participating with more on the horizon. On the good side, we have large states, like California and New Jersey and Michigan. On the negative side, we don't have some of the big states yet, like New York and Texas. Again, it is optional. At this point it is not required.

What happens in this system is basically every quarter we get a carbon copy on tape of a state's eligibility file and their service records or their claims records for those previous three months. MSIS also receives the record of a managed care premium payment. So, that comes in. We know for most states the individual premium payments that are made by person and to the particular plans.

The MSIS system also does have the capability of receiving encounter data, these dummy or shadow claims. There are three files that we generate from these tapes that come in every quarter from the states. It is called the valid tape file. Basically, after the tapes have gone through some set of editing, we put them on a database and we have the detailed records, detailed claims and eligibility information for each Medicaid beneficiary in the state.

There is what we call a personal summary file, which for each person is an annual summary of their eligibility records and the payments by broad service types, so we know during the course of a year a state paid x dollars for long term care, x dollars for physician services, x dollars for hospital payments for an individual person.

There is also the 2082 file. One of the advantages for states to join the MSIS is that all they have to do is send us these tapes every quarter and we will produce actually on a quarterly basis this summary report for them. So, that requirement is waived for those states and they like that.

There is also a spinoff of the MSIS system that is called the SMRF or the state Medicaid research files and this was an attempt by the agency to make the MSIS files much more useful for researchers. There has been a little bit more quality assurance done on the data and it has sort of been transformed and structured by calendar year date of service as opposed to the date of payment records MSIS is really set by.

So, the SMRF files will allow researchers to track an episode of care. You can go to the SMRF file for 1994 and look at all the services that a person has received that were paid for by Medicaid and received in 1994, as opposed to the MSIS files, if you looked at them for 1994, you would see all of the services that Medicaid paid for a person in 1994. Those may have been received in 1994 or 1993 and, interestingly enough, sometimes the claims come in two, three or even four years later.

The SMRF files also have some indicators for delivery, ESRD, tuberculosis, so you can identify some populations fairly easily.

One of the weaknesses of the 2082 is that it really has not been changed to accommodate the shift to managed care for Medicaid programs. It was fee-for-service based and tried to measure services that were actually paid for. So, it has not really collected information on managed care premiums or encounters. Since a managed care premium is sort of a prepayment and does not really represent any services received, those services were not included in the service of the payment side of the 2082, although the eligibles were still there, there was no record on the 2082 of the fact that there was a managed care premium or any encounter services received.

We are making some shifts in accommodations. For fiscal year 1998, on the 2082, there will be data on premium payments and PCCM payments with service categories. So, we will -- this will allow folks using the 2082 to do some demographic analysis of these payments. We will have an age distribution of recipients who had premium payments, program distributions, raise distributions, those things.

But, unfortunately, there still will not be any service information, no encounter service information on the 2082 for folks in prepaid managed care.

Fortunately, the MSIS system does allow us to collect some of that encounter information. It already collects premium payments and collects the PCCM or the primary care case management payments as a service type and it does collect encounter data information from those states that have been willing to give us those encounter records.

We can summarize encounter data separately from fee-for-service data. So, if you identify a population in managed care and identify a population in fee-for-service in the state, we can do some comparisons, either over time or at a point of time to see if those populations are being treated any differently.

The MSIS system right now has the capability of analyzing premium information and plan membership. So, we can do things like graphs of managed care premium payments versus total Medicaid expenditures. We have done things like managed care enrollment by county, by gender and by race. So, some of those numbers are available.

We are getting encounter data only from a few states right now. For some reason, Iowa has been the leading state in providing encounter data to us. We have got a number of years worth of Iowa encounter data that we have looked at and actually looks pretty good. New Jersey has been sending us encounter data in California, especially for their dental plan has been sending us some encounter data.

We haven't spent a lot of time to date looking at all of this data for the accuracy and completeness and the validity. We are sort of moving into and trying to figure out how best to make sure that we are getting all of the data, that the data looks decent and is really worthwhile looking at for research purposes.

And, again, with encounter data, we should be able to make plan comparisons with the MSIS system. If you have got a state with three or four different HMOs, we should be able to identify the members of each of the HMOs or each of the managed care plans and make service and utilization comparisons among those plans.

Now, on the MMIS or the systems side of the house, it is the Medicaid Management Information System that really is producing all this data in every state. That is the basic set of requirements the states are supposed to have to process Medicaid claims.

We are working on revising the requirements for managed care. To date, the instructions to the states and the requirements of the states have not really addressed what they are supposed to have in their system to really oversee managed care plans. We are looking at both the data that they should collect and how they should manage that data and what they should be doing with encounter data as a requirement to managing their HMOs

DR. IEZZONI: Okay. Good. Thanks, Stan.

Members of the committee, any comments, questions for Rachel and Stan at this point?

MS. WARD: I guess, Rachel, maybe the question is directed to you. You made a point that each state is very different. From that perspective, do you feel there is an evaluation capability across the country or, in fact, is your recommendation that state by state should decide whether it is working in that state?

DR. BLOCK: My ideal view of the world is that it would be a combination of both. I think that there are some widely accepted national health status kinds of goals, whether we are looking at Healthy People 2000 or some areas that may be represented by widely accepted clinical practice guidelines and so forth, that, at least stated as goals, should be reasonable ways to look at where is a state relative to that goal.

I am not sure that I have a lot of examples beyond those -- you know, from those two areas, but I think that that is a pretty healthy starting point. I also think that as we get more information from standardized instruments, like HEATUS(?), like the CAP surveys, like particularly the functional status survey, which Medicare is going to do for all of its managed care members, as states start to pick up on some of that kind of methodology, what you would get there, I think, is at least the beginning of establishing some baselines and then you could see, in fact, how different are states in terms of their characteristics from an outcomes perspective.

So, my observation was more that the design characteristics and the environment of each state's program is highly varied, but that probably isn't all that different from what we know are the underlying variations in medical practice patterns and stuff like that.

So, I think it is a matter of acknowledging that, saying what do we think, generally speaking, are reasonable in terms of broad national goals. One of my hopes is that HCFA will continue to improve its capability to work with states to set their own goals, which might be specific or relevant to some unique population issues that they have, which may not be in common with others and then have sort of simultaneously some basic measures that we would be conducting at a national level or at least where we have information that could be out there for a discussion and dialogue at a national level at the same time that we encourage states to work on their state specific and population specific goals.

So, I would like to visualize those as not mutually exclusive and, hopefully, supportive of each other, particularly since in addition to just having the actual measures and outcomes out there, clearly, what we are talking about are not only being able to measure these things, but being able to influence either characteristics of the delivery system or provider behavior that -- and consumer behavior -- that are likely to contribute to some of the variation in those things. And that is a pretty complex technology and I am not sure how you envision intervening, except at a pretty local level to really influence some of those things.

So, I think there is a lot of different levels to the system but I think it is very reasonable to have national goals. The only other thing I would mention is at least under current requirements for Medicaid, and I don't think there is anything that Congress is considering that would waive this, we do have some other areas in which there already are some national standards.

EPSDT is probably the best example. It may not be the best example of an outcome measure, but at least it is a national standard that has been out there for awhile that has served, I think, a very useful purpose from the states' point of view in terms of shooting towards a goal of how to improve their performance in that area.

DR. IEZZONI: Can I just follow up on this from taking our very parochial perspective?

Rachel, this afternoon, we are going to try to come up with a work plan for how we are going to figure out what our role will be in looking at this and one of the options that we are thinking about is using case studies of maybe two states, possibly three, but probably more likely two, to kind of give us examples of various states are doing.

Would your advice to us be to try to find states that are at the cutting edge of information systems or are there any general things that you might be able to say to us about how these states are functioning that could help us think about how we might choose states for case studies?

DR. BLOCK: I think that there are a couple of ways to slice it. One would be that you probably could come up if you looked at very basic population-based or demographically-based characteristics of states, you might want to come up with some sort of a model to support your selection of those states that include at least a representative range of states in terms of the types of populations or some of the socioeconomic variables in those states since the socioeconomic environment is a very important determinant of who is in their Medicaid program.

So, I would advocate that you look for a broad selection based on some basic socioeconomic kinds of characteristics that you could select. I also think that it probably would be important to at least consider the difference in their progress towards managed care. I am not sure if there is a real scientific method to do that at this point, but, again, if you review some of those mathematic reports, you can start to see how the characteristics of those programs really have evolved very differently over time.

The one other thing I guess I would suggest is that to the extent that you can utilize some of these or at least a selected number of what are now increasingly becoming standardized measurements, whether they are from HEATUS or from some other source, I think that that would be particularly valuable because among other things we would have a better basis of comparison in terms of the underlying characteristics of how the data were actually collected and reported.

DR. IEZZONI: I think I should just clarify kind of what we will be focusing on. We are not actually going to be evaluating the performance of Medicaid managed care. We are going to be evaluating data systems that will assist people in doing that kind of evaluation. So, we are very interested in different models for kind of data gathering.

One of the comments that you made that resonated especially on this was your plea that we consider other social support programs that might be funded, not only by the Federal Government but also by state and local governments and the information systems that they might have to also bring to the table to be able to perform a complete evaluation of Medicaid managed care will be important, too.

DR. BLOCK: And also if you can get out of that some basic ideas or recommendations for future directions in terms of where in that system is the best place to try to bring about that kind of link or common set of standards or specifications around those systems.

I haven't seen -- my bias is that I think that you have to get down to a pretty local level to really make that kind of sharing of information really work, but, again, so much depends on sort of the characteristics of each state system that it may be hard to generalize. But I guess I would say at this point if you even had out of your case study approach or some of your other work, the ability to identify some basic characteristics or issues and perhaps a continuum of different kinds of models that are currently out there, that even just documenting that in a sort of concise and systematic way, I think, would be very valuable.

One of the reasons I say that is because in my new capacity, one of the areas I am now going to be involved in more is states MMIS systems and also many states are implementing these integrated eligibility systems that link up their eligibility and Medicaid data. I think it is fair to say that states aren't real satisfied with the value that they are getting for their expenditures in that area and I am not going to comment on how satisfied HCFA is with the outcomes either, but just suffice it to say that we have heard enough from states, I think, to say that they are not real happy either with what they are getting out of those systems or maybe even what their available choices are.

So, I think that any suggestions you could come up with, even if they are very basic ones that would help facilitate the states' design decisions in purchasing or managing your information systems would be incredibly valuable at this particular time. Then also, obviously, anything that you could do to help us better sort through the issues associated with encounter data, again, would be very helpful just because I think that in the next year or two, we are going to be getting around to zeroing in on some standards in that area, whether they are mandates and requirements or some other form, but I think that that is inevitable at this point.

Obviously, there are lots of considerations and concerns there. So, I think if there were a way for your case studies, given this particular focus that you have described, could help states understand their information system needs and capabilities better and also deal with some of the issues that we and others might want to consider about encounter data.

Those would be two things I think would be particularly valuable at this point.

DR. IEZZONI: Good. Thank you. That is very helpful.

George.

MR. VAN AMBURG: In developing the encounter data systems and the advice to the states, how are you taking into consideration what some states are carving out now in special services, like children with special health care needs?

DR. NACHIMSON: I am not sure I quite understand the question.

MR. VAN AMBURG: Well, some states are going to have a separate contractual arrangement for those services. DR. NACHIMSON: Right.

MR. VAN AMBURG: How are you going to integrate those?

DR. NACHIMSON: I am not even sure we have gotten that far to even think about that.

DR. BLOCK: Although if you have got Iowa, then you should have that -- or there should be data in the system addressing that, since they have had basically three programs now; their main program, if you will, their mental health carve out and their substance abuse carve out under a managed care arrangement. So, if you have any recent data from Iowa -- the two carve outs have only been up for two years, but if you have got any recent data from them, you should be able to -- we could look at the Iowa data.

That is actually another issue I would have brought up in my broad characterization thing, whether those carve outs are done through additional capitated kinds of arrangements or remain in a fee-for-service environment and then how do you deal with not so much carve outs but what are some specific areas in which out-of-network use is allowed or encouraged or required. And different states are going to come up with different -- and I think you would want to look at that issue. As you just describe your case study, be very careful to look at those issues because, again, each state has very different kinds of requirements in terms of how tight or loose they try to make that contractual arrangement with the MCO.

And also, they may have very different degrees of how specific or explicit they are with the MCOs in terms of their tracking of out-of-network use, either that which is required, such as for family planning. Again, if I were going to make a pitch for a special area that I think is of critical public health importance, and we really have no idea what is going on out there, it is in that area.

DR. IEZZONI: Family planning.

DR. BLOCK: Yes. It is, in part, I think, because of the federal requirement to carve out which applies in all except for a couple of the 1115s, where that requirement has been waived. So, the multiplicity of different out-of-network uses and then as well as those which are formalized as carve out programs, I think, are important to look at.

DR. MOR: The structure of the system that you have described, the non-systems from state to state suggests, however, that it might be possible to come up with some way of coding up the characterizations of the different state approaches. I know that in the private sector people have come up with all these different paradigms for classifying managed care contracts and arrangements and so on and so forth in the Medicaid -- your description of it in the Medicaid world at the state level.

It is a little more complicated, but are there data that HCFA has now characterizing the current state contracts and structures and are they in a codified form as opposed to just applications?

DR. BLOCK: I am not sure -- I flinched at that last point.

DR. MOR: Data.

DR. BLOCK: Yes. Very limited aggregate. It is more program information at this point than actual data.

DR. MOR: Where program information is basically narrative text.

DR. BLOCK: Yes, or, at least, we do have in a somewhat more automated or systematic form some basic characteristics of the programs. For example, I could run a report for you today that said how many of the state programs were fully capitated versus partially capitated versus some other arrangement. So, they are descriptive in that sense but they are not limited to narrative and we can actually manipulate that information reasonably easily today.

Also, one of the areas that is weakest in our current system that we hope to improve next year is being able to break out those broad eligibility characteristics, but, again, if you have got a managed care state that is in MSIS, where we have some confidence that that basic data would be reasonably clean for you to look at, possibly from some other SMRF experience as well, then I think you could start to patch together some of the different data that HCFA has that could start to come together in a little more systematic form.

I think, though, I would also mention that there is probably very useful information to be had from some other sources, not the least of which is some of the evaluation contractors that we are now working with. You may know, The Urban Institute now has some huge study they are doing on the new federalism and looking at welfare and Medicaid policy in particular. They have the capacity to bring some data and they have noodled around with Medicaid data for probably as long as anybody else has at this point.

So, I think that there might be some other sources, besides HCFA, that you could go to, but I think we have -- what you would end up with if we could patch together some of our different systems is probably a reasonably good sense of the framework of states and how those characteristics would vary from one state to the next.

But, again, for some of that more qualitative or population-based data, unless we can get it from MSIS, we would probably have to go to some other sources, including the states themselves.

DR. IEZZONI: Richard, you had a comment?

DR. HARDING: I apologize for being late. U.S. Airways, you know, you are always glad to make it.

[Laughter.]

DR. IEZZONI: No, I disagree with that. U.S. Airways is my favorite airline.

DR. HARDING: It is about the only one I have. So, I use it. It was only 20 minutes late, so I can't gripe much.

DR. IEZZONI: Richard, could you just introduce yourself for everybody in the room.

DR. HARDING: Sure. I am Richard Harding. I am a child psychiatrist from South Carolina and work in private practice and in academic settings in administration there in hospital settings.

I apologize, again, for being late and you may have covered something that I -- but George brought up a point that I had thought about and, of course, being a child psychiatrist, that issue of mental health and substance abuse and so forth in children especially -- the issue is standards of how to evaluate a state's performance in the area of mental health is a tough one. I am not suggesting that you all have the answers or anything, but just wonder if -- how you are going to look after that issue.

In South Carolina, we just had a kind of a fiasco, where we had kind of a managed care setting in a department of juvenile justice area and the company who was providing that gave wonderful reports. It sounded like the best thing that ever happened in our state. Then when others went in and looked at it carefully, it wasn't quite that way.

I know that that is going to be a problem of accepting reports back from the states or from a carve out of the state to the Federal Government. Are we going to have developed in the future adequate guarantees that it won't just be a report back, a paper report, that we can really feel pretty confident with the things that we have contracted for, I guess, if that is the word, will be carried out?

I am sorry if that has already been --

DR. BLOCK: No, no, I don't think so and I think it is fair to say that the whole area of behavioral health and Medicaid managed care is one in which we have basically been applying a broad set of standards and tools that really emerge from a very different type of program that we had grown very accustomed to dealing with.

I think this is particularly true in mental health, but I think it does have some bearing on other special populations, as we get again to some of those services that may have been covered by Medicaid but not included in managed care and often in these programs we have states using some of their savings in a very explicit way to provide some additional services that they couldn't cover under Medicaid before. That is another data thing that I think you would want to look at if you look at some of those kinds of programs.

So, there is basic, you know, outcome and evaluation type of standards and I think that we are, along with everybody else, just at the beginning of trying to sort through the wide array of accreditation standards, data and other things that that community has generated and I hope there will be more sifting and consensus around those. In part, that is what makes it hard for us to sort of say this is a standard that makes sense to hold all states accountable to. Also, it partly goes back, I think, to your question, Elizabeth.

But having said that and what that means is just is there a national standard in that area that is meaningful and broadly accepted or is our job to try to work with states to ensure that they are coming up with better standards that are home grown and unique. I think that mental health is a particular example where that may be the more likely place that will be in the short term until some of those national standards start to gain more widespread acceptance.

I think there is another issue, though, that you are raising, which is important and that is that one of the things that we have insisted on in our approach to quality is that the basis for studying and evaluating the outcomes of programs from whatever different perspectives, there has to be a varied range of approaches to it, including that there have to be externally generated independent, validated kinds of studies. We have had in the basic HMO area for a long time and still will have under the new legislation a requirement for annual independent external review. Now, that has some shortcomings because they are only going to be as good as some of the things you ask them to do, but, nevertheless, you at least have the ability to have some independent assessment in addition to whatever it is that your contractor is providing to you.

I think that we think that that is important and I think Congress has basically agreed that they think that is important to maintain.

That requirement doesn't apply to the carve out programs because they aren't -- they don't fall into that full risk HMO category. So, that has been an area in which we have not had that requirement in place, but, again, I am hoping that some of these national bodies that are dealing with accreditation standards and some other standards in that area will come along and start to fill in, at least, in terms of being accepted as -- and we would start to use those as guidelines and possibly standards as more of that took shape.

So, I think that the need for some independent external review clearly is called for. We asked states to conduct a wide variety of studies and other ways to evaluate their programs, but ultimately -- and I guess this would be sort of my closing comment that I would hope you would include somehow in your report or whatever form this takes, that you could have the best data system in the world. If you don't have the appropriate purchaser and regulatory oversight of the data system or the system overall that you are purchasing, it won't do too much good, other than to provide fodder for other kinds of parties, who are interested in the stuff.

So, I really do think that the purchaser's definition of their role relative to that system is critical and I would say that states are all over the place in terms of to what degree they develop the skills, capacities and systems necessary to do that work.

So, again, I think it is very timely to come up with some recommendations in this area because it will help provide a foundation that they can work on.

DR. IEZZONI: Well, I think, Richard, your question actually has a really complex core to it on the data side because it suggests that we need auditable data, which suggests that we might need individual level data as the atom from which the larger fabric of the reports are built and that begins to get into issues of privacy and confidentiality. How we protect, you know, individuals especially in mental health and substance abuse programs, goes back to what our committee, you know, has been debating about for months on that level.

So, I think that these issues are going to be extremely complex, especially given what you have just told us about the carve out programs not having some of these external, independent, oversight functions mandated.

DR. MOR: Vince, a quick question.

DR. MOR: Okay. The last question -- you mentioned the notion of actually beginning to develop systems that could integrate across a data outside the traditional sort of health consumption or health utilization world; housing information, food stamps, you know, those kinds of fun things.

Are there any examples you know of of states that have actually, other than the HIV system I am familiar with in Maryland, that have tried to do that in other domain areas, either Stanley, you, or Rachel?

DR. NACHIMSON: South Carolina very much so. They have done quite a job of integrating data from a wide range of programs. They have set up a system so that you can sort of track individually what is going on with any one person across the wide range of social services.

DR. MOR: And this is the population of individuals who are Medicaid-eligible at some point in time?

DR. NACHIMSON: As well as other programs. It doesn't center --

DR. MOR: On the Medicaid.

DR. NACHIMSON: -- Medicaid. Right.

DR. MOR: I see. Any public program?

DR. NACHIMSON: It is a number of public programs. They didn't stick the DMV in there, did they? I thought they were even talking about some driving record stuff.

DR. HERZ: Yes, they have highway accident information, criminal justice system information.

DR. MOR: This is all at the micro level, at the individual level?

DR. NACHIMSON: Yes.

DR. IEZZONI: In Iowa, your state that seems to be your benchmark for Medicaid data, is that a similar situation that they also have data from other -- from the DMV, for example, and the other --

DR. NACHIMSON: Not that I am aware of.

DR. HERZ: Illinois is another state that has an integrated data system across social welfare, nutrition and Medicaid, as well, and that system has been put together by the University of Chicago, Chapin Hall for Children -- the Chapin Hall Center for Children. That is a very similar system, I think, to what they have in South Carolina, probably a good model for an integrated databases, as well as South Carolina.

DR. NACHIMSON: There was a foundation, I think, that was funding a number of these projects and I don't remember which one it was unfortunately.

DR. IEZZONI: Robert Wood Johnson.

DR. NACHIMSON: Okay. Robert Wood Johnson was funding in a number of states these types of integrated data approaches.

DR. BLOCK: And on a smaller scale, I think you would find even more states -- again, Maryland comes to mind just because I saw a presentation that was done recently, where they at least have come up with some pretty good links between Medicaid and WIC eligibility and participation and were able to generate some pretty interesting reports, I thought, in terms of county level analysis and where they could figure out that there were good links going on as opposed to less good links going on.

DR. NACHIMSON: In one of the New England states

-- I think it is Vermont, I am not sure -- it has a registry of children that encompasses not only immunization data and a number of other items that they are able to kind of track the progress of --

DR. BLOCK: They have actually incorporated education data and education outcomes, as well as some others now. What I don't know is what is the underlying data system that they have used or if they are just sort of slapping pieces of things together or if they really have an integrated database that supports those outcomes measures.

DR. NACHIMSON: They do have an integrated database. We got some documentation on it from the project officer.

DR. HERZ: ASPI(?) has put together a really impressive registry of state level efforts to integrate data across these different systems. It is on their Web site, which, unfortunately, I don't have the address with me,

but --

DR. IEZZONI: Our committee is on ASPI's Web site, if I recall.

DR. HERZ: It is a very impressive inventory of activities at the state level.

DR. IEZZONI: That is excellent, a very good clue.

Do any of the committee members have any final comments for Stanley and Rachel? This has been really a wonderful start for us and I really thank you.

DR. BLOCK: Well, thank you for inviting us and, Carolyn, thank you for inviting me. I am going to have to leave in a few minutes, although I think Stanley is going to stick around for awhile longer this morning.

But if there is anything that we can do to help in the future when you get down to trying to select some states and if you would like some feedback on that, we would certainly be happy to help you out with that.

DR. IEZZONI: Okay. Great. Thank you, both of you.

I am going to get a reputation for an unsympathetic chair because I would like to just continue right through this morning. You will notice that we don't have any breaks listed, but we have so much to hear from -- we might stand and stretch, maybe after the next panel, but let me just say that it is socially acceptable to get up and leave the room at any point when you want to do so. Okay?

So, with that in mind -- are you James or Jim?

DR. HADLEY: Jamie.

DR. IEZZONI: Jamie. Okay. Another alternative.

Jamie, you are -- do you have a new role at HCFA, too? We just learned that --

DR. HADLEY: New in name, I guess, because the office has changed from the Office of Research and Demonstrations to Office of Strategic Planning. Well, I shouldn't say it has changed. A lot of people who were in ORD are now in OSP. But it is not an identical kind of office because the demonstration part of ORD, the "D" in there, has been distributed, also a "d" word, throughout HCFA basically.

So, now, the focus is a little bit different. Obviously, the research is still there. There is intramural research, the Office of the Actuary has been integrated into it. The focus has shifted, but I am doing basically the same thing I was before. Life has not changed for me day to day.

DR. IEZZONI: Okay. We will hear something about that in a minute.

We have -- the two of you are talking, I guess, about monitoring evaluation effort for looking at Medicaid managed care. Are you comfortable with the order in which we have you listed? Yes. Lisa first? Okay. Great.

Agenda Item: Monitoring and Evaluation Efforts for Medicaid Managed Care and Discussion of Issues and Recommendations

DR. HERZ: I think the order makes sense because I think my comments are pretty directed at sort of the nitty-gritty of data collection and information systems more, I think, from a state monitoring perspective and my guess is Jamie will have a lot more to say about the evaluations that HCFA has funded. So, I think it is probably a good order.

Just for those of you who are not familiar with the Medstat group, I just had a couple of introductory remarks to say about our organization. We specialize in health information and application of that information for improving the quality of health care and we basically do that through data systems, consulting and research.

We have done a lot of work with a federal and state government agencies, particularly with respect to Medicaid and Medicaid managed care and also on what I call the commercial side of our business, we also build decision support systems for state Medicaid programs and I have listed a number of them that we are working with now.

Based on my conversation with Carolyn Rimes in preparation for this meeting today, she basically asked me to talk about three things: to describe the types of technical assistance on encounter data issues that Medstat is currently giving to state Medicaid programs, to identify from that experience some of the common encounter data issues that are faced by states and health plans and then, finally, to try to -- for this committee to try to translate some of those lessons learned into some recommendations that you might consider.

The project that I want to focus on we call the State Health Care Reform Monitoring Project. Its purpose, again, is to provide technical assistance to HCFA in mostly 1115 waiver states and defining, obtaining, validating and then using encounter data for program monitoring and evaluation purposes for Medicaid managed care. It is a three year project, which we are a little over halfway into and our funding is through a HCFA Master Task Order Contract.

Now, that contract is under Rachel's group in the Center for Medicaid and State Operations. Our technical assistance to states is what we call showing, not doing and that is mainly so that we can provide them a lot of technical assistance to as many states as possible. An example would be developing a data validation system.

Within the scope of our contract in that context we would do things, like describe the key components of the process of data validation, outline some of the steps in a comprehensive plan, share the experiences of other states, which is of great interest to individual states and review and comment on the strengths and weaknesses of a state's existing plan for data validation.

Outside the scope of our contract includes things like medical record collection and abstraction, analysis of collector encounter data and also preparation of reports describing those findings, just to give you a sense of what we try to do in the context of this contract.

The next handout just lists for you the states that we have given technical assistance to so far. There are ten of them and we have provided varying levels of technical assistance to these states, mostly depending on the states' immediately priorities at the time that we are talking to them about technical assistance.

These states, I should mention, are different for the states that I listed in that earlier overhead that we provided some decision support services for.

Just to say a little bit about the kinds of technical assistance that we have provided to states, I have tried to break them into two categories here; uses of encounter data and also other technical issues involving data collection and measurement.

Performance measures for special populations, we have done a project for the State of Hawaii, trying to provide them with an inventory of clinical performance measures for the SSI population. They were very concerned that they weren't doing a good job understanding what was going on with their SSI population and, in particular, they wanted to have us identify for them some existing measures to look at mental health and mental health impairment.

So, we provided them with some information on some specific areas of mental illness, as well as -- they also had an interest in Down's syndrome and the HIV/AIDS population, which is also usually part of a state's SSI population.

Those we really used existing information systems to provide that information, which I am assuming you all are aware of the Conquest database. I think those are important systems to access when you are trying to identify measures for these populations, measures that have been used or that are under development. It is really quite a comprehensive system that I think might be useful to this committee.

Also, we have had some states ask us about access of service use in minority populations. Minnesota, in particular, was interested in having us help them come up with an analysis plan for their Asian and American Indian population. They were very concerned that these folks were not having the same kinds of access to services as other Medicaid beneficiaries.

So, we are trying to help them understand how best to measure race and ethnicity, which, of course, we are keeping them abreast of the activities of the committee, and helping them try to develop that analysis plan.

In terms of satisfying HCFA reporting requirements with encounter data, the more immediate example I have is, again, with Minnesota, where they are trying to understand how to complete the HCFA 416 report, which is a special report for looking at EPSDT services. It also documents referrals for treatment for problems who had been identified during EPSDT screens, which might be of interest to this committee.

But states are struggling with trying to determine how to use the encounter data that they are getting and fold it in with the fee-for-service experience to be able to accurately report on this form. So, that is a common request that we get to help states understand how to do that.

Then I don't want to go into a lot of detail on the rest of these, but just, again, to provide some examples of the kinds of technical assistance that we have provided to states.

Let me spend a few minutes talking about, based on our experience, some of the encounter data problems that states face and I think this will echo a lot of what has been said already and provide a little bit more detail about that. States don't always know what encounter data to collect. They are not sure what they need for ongoing monitoring and evaluation purposes. And they are not quite sure whether what they need is the same or different from what they had under fee-for-service, although I think the typical model is to start from what they had in the fee-for-service system and try to build on that for encounter data.

States also don't always know how to process encounter data. There are longstanding MMISs or Medicaid Measurement Information Systems have a lot of Intertet(?) editing checks that go on records when they enter the system and early experiences of many states were when they tried to put an encounter record through the MMIS, they were just being rejected left and right because of the lack of financial information.

Some of those very practical system adjustments need to be made and sometimes those are very complex when the systems are very big, very longstanding and given very limited state resources for doing this sort of thing makes it a difficult process.

The other point I would like to make about encounter data problems that states face, states may receive encounter data from plans months after waiver implementation. Why is that true? Usually for a couple of reasons in our experience in talking to states. Their initial focus is and needs to be very operational. They need to make sure that their enrollment systems and their payment systems are in place and working and sometimes that takes awhile to get that in place.

We mentioned earlier some contractual requirements on data. I will be saying a lot about that today. Plan RFPs and contracts may be inadequate. That might be the reason why they are having a difficult time getting encounter data from plans. Often, those contracts have limited or no encounter data requirements, especially with respect to very practical things, like the timeliness of when data needs to be submitted, the completeness of data when it is submitted, the medium on which data needs to be submitted.

In fact, George Washington just released a very comprehensive report on a lot of contractual issues, mostly focused on sort of programmatic issues, but they also had a component on data requirements that they reviewed in I think 35 or more states. They also identified a pretty serious lack of encounter data submission requirements in most managed care contracts in states.

When there are data requirements in contracts, they are often not enforced. I haven't quite figured out exactly why this is true, although I suspect, at least in some circumstances, especially in the initial stages of a waiver, states are hesitant to put a lot of pressure on plans about data, especially if those plans are sort of the only game in town and they don't have a lot of choices in terms of getting plans to take on Medicaid business. It is a very practical issue for states that they have to deal with.

Let me say a few things about the encounter data problems that plans face that I hope will be helpful for the understanding of this committee. Information systems have limitations for both the providers and the plans. On the provider side, they often have limited resources for data processing and submission. They also have trouble submitting data in an acceptable format, especially if they contract with multiple plans who have different data requirements.

This is a tremendous burden on these plans. This is very common sensical, but it makes a huge difference. For plans, they often trouble processing non-payment transactions, which is usually what an encounter record is all about. They often have very complex, multiple systems in place that were set up for administrative purposes that don't communicate with each other in an integrated data system type of way and, therefore, they have trouble reporting to the states the kind of information that the states need for their own monitoring purposes.

This is all getting better, but it is definitely still an issue out there for both plans and providers. A couple of other provider issues I wanted to mention, their willingness to report the plans. Under managed care there have been these early promises about no more paperwork and now we are not saying that anymore and it takes a lot of effort to get them to sort of change their frame of mind and say, okay, well, we can do this. But it is a tremendous obstacle to overcome with a lot of plans or providers, I should say.

Compensation arrangements, again, this is probably pretty obvious, but I think worth mentioning. If a provider is being capitated, they are not being paid to submit an encounter. So, incentives to submit encounters are not very great, unless they are being paid to submit the encounters, which some plans have chosen to do to get the data from those providers.

In terms of plan issues, I think they are particularly relevant to Medicaid managed care and particularly relevant to the carve outs that we talked about earlier when they are trying to put together managed care programs for special populations.

Many states when they want to do that, they want, too, involved in those managed care programs the traditional providers of care to these special populations. These traditional providers don't often participate in standard managed care and, so, the states try to set it up to allow them to form their own managed care organizations to be able to continue to provide services.

They are great at providing care. They are not always so great about running a managed care business. That is sometimes where the data issues then will come in in terms of providing that information back to the states. They need to understand how these programs are working.

Also, there are a lot of commercial who are new to Medicaid and their problems -- so, in a sense, they are also kind of a start-up and they have particular problems dealing with eligibility in Medicaid and the on/off nature of enrollment. Their eligibility and enrollment systems have a very hard time dealing with Medicaid and they are not quite -- haven't quite resolved those issues yet.

We mentioned -- this, again, is in the context of carve out services. Data requirements and provider contracts are often inadequate. Plans, when they subcontract out things like lab and x-ray, behavioral health, pharmacy services, if they don't -- I mean, they are contracting with those vendors. If they don't have in their subcontracts very specific data requirements, you will often see when a plan transmits information back to the state, big holes in these carve out services because they can't get it from their vendors.

Then, finally, I just wanted to mention leverage over providers varies by plan type and the comment I want to make here is mostly with respect to the exclusive nature of contracting with providers. If a plan has exclusive contracting with a set of providers, in other words, if those providers can't contract with any other managed care plan, then those plans have quite a bit of control, not only over the kinds of services and the treatment protocols that those providers are implementing, but also data requirements.

So, the standard staff model HMO does a pretty good job of getting the kind of data that it needs from its providers as opposed to situations where it is non-exclusive contracting with a provider, where the providers -- and many providers do contract with eight, nine, ten different plans and the individual plans control over that provider in terms of data submission can be pretty weak. That is a very practical issue that affects the transmission of data to states.

These state and plan issues won't disappear with national data standards, I don't believe, but I do think coming up with data standards will go a long way towards reducing some of the burden that plans and providers feel by these many different standards from many different purchasers and I think in the long run it will be a help. It will improve the quality of the data and the completeness of the data to have more national standards than we currently have.

I am often asked to give presentations where I talk about all these problems and I always hate to end on that note because there is actually a lot of states, who have made a substantial amount of progress in terms of collecting and using encounter data. I just wanted to provide some brief examples of some of the progress.

These problems are not insurmountable and these are a few of the states that I think are definitely on the road to being where we want them to be. Minnesota, for example, they have an 1115 waiver for a very long time, but it has actually only been quite recently that they have been encounter level data from their plans. In the past, they have been monitoring services under Medicaid managed care through aggregate reports that the plans had submitted to them.

But this year, they decided to make a very concerted effort to get the data in and they have started to do that. So, for the past six months, they have gotten a little over a million records that they have accepted. They have received more than that, but they have accepted a little over a million records.

Just as a practical issue, I think what they found in the past was a lot of resistance to encounter data submission from the plans, much along the lines of what I described earlier. They found that a letter from the Medicaid director to the plan CEOs emphasizing the importance of encounter data appeared to have made a lot of difference and that is why they are getting a lot of information in now.

There are plans for analysis. I mentioned a couple of them earlier, looking at minority groups. They also want to go beyond the HCFA 416 report that is required for EPSDT services and look at other kinds of preventive care that are being given to children under Medicaid in these plans.

Just to skip down to Tennessee because Tennessee has had a very rocky history, but they appear to be doing -- they have made great, great progress, at least for their own internal monitoring purposes. Maybe you can say something about what they have submitted for the evaluation. But they have data back to the beginning of their demonstration. They have over 117 million records to date. They get monthly submission from their plans now. So, they really seem to be on the road to being able to monitor and evaluate their demonstration.

They have lots of plans for analysis. They are producing key indicator reports on enrollment and utilization and expenditures under TennCare. They have done an infant mortality study, where they link their encounter records with birth certificate data to look at the effects of TennCare on the trend in infant mortality over time. They do a very targeted quality improvement studies that are identified by their medical directors, issues that they want to focus on. And they talk to their plans about these a lot to get that information.

They have a very nice Web site where you can find a lot of these reports. They have some infant mortality reports on their Web site. So, they have made a lot of progress.

Why don't I stop there. I just want to point out things are happening. There are some success stories out there, which isn't to say there isn't more work that needs to be done. Obviously, there is. But it is happening.

So, finally, in closing, I just have a couple of areas that I would like to recommend that the national committee and the subcommittee take on or pay attention to.

I think articulating the importance of encounter data to monitor care for special populations is very important and I think the importance here goes beyond just meeting reporting requirements. If you want to get state and plan buy-in to the data standards that you would like to propose, these data standards need to be relevant to that. They need to address their own monitoring issues not just -- sort of bigger evaluation purposes. And I think that is possible to do.

I think a lot of what I have seen put out by the national committee is things that the states and the plans need, but sometimes they need to be told why they need it and how it can be useful to them. I think if you all can focus on that, especially in terms of special populations, I think that would be a big help.

Also, to establish encounter standards for monitoring care for special populations, I just had a couple of comments I wanted to make here. I think, what I haven't seen so far and maybe it has been addressed, I think it is important to add a family identifier to the core data elements that are being considered for standardization and I say that because my context of a special population is children. I think there is a continuing need to look at, for example, prenatal technologies and other kinds of interventions and their effects on newborn health, as well as more work on the clinical and behavioral risks that mothers have that can affect newborn health.

It will be very difficult to look at those kinds of issues and questions if there isn't a reliable way in which you can put those moms and babies together. I think the identifiers that are proposed now, I don't believe, do that and I think -- we had a discussion earlier -- especially if you want to talk about integrated data systems where you are moving beyond just health care delivery information to social welfare, all of those issues, I think there might be some lessons that you can take from AFDC and TANAF(?) for example.

I tried to find out what they are doing about personal identifiers and family identifiers through the TANAF program. I wasn't able to get a specific answer in time for this meeting, but I have to imagine that they are thinking about this and trying to address it. So, maybe some conversations with them perhaps would sort of flesh out logistics of coming up with a family identifier, which I think might be important for lots of program evaluation purposes.

The only thing I want to say about indicators of health status and outcomes and race/ethnicity measures, I know that these things are being considered and included. I just want to reaffirm that I think that is the right direction. These are not traditional parts of information systems that plans on the provider level go. So, I would just recommend that very complete and very detailed specifications be provided for these things, otherwise you will be getting a lot of apples and oranges and you won't know what to do with them.

Finally, I drew a line here in this last handout because I think the last two things are probably not in the official purview of the committee, but these are things states ask us about all the time and they would greatly benefit from some attention either by the committee or maybe by the Department of Health and Human Services. If somebody could take this on, it would be a big help.

Standardizing methods for assessing and ensuring encounter data quality, states are -- much of our work focuses on that under the contract I described earlier, but some standardized official methods for this, I think, would be helpful and then, finally, creating a national repository for performance benchmarks. States are always asking us, you know, what is a good number, what should be trying to hit here and what are other states hitting on these numbers.

So, I think that would be a big help as well.

I will close there.

DR. IEZZONI: Lisa, great.

Before I move to Jamie, I just would like to follow up on the family indicator because I saw a number of heads nodding and while we are hot on that -- Richard, you had a comment on that?

DR. HARDING: I think I know what you mean, but when you say "family," what do you mean?

DR. HERZ: Well, I guess the easiest example is the one I gave before, when you are trying to determine the relationship between maternal behaviors and newborn health. So, I guess at sort of the smallest level, I think, some kind of identifier that permits reliable linking between mothers and children, but I think there are bigger questions than that, which we touched on earlier. And Illinois has done this in the integrated database that they are focusing on. In fact, their efforts began with trying to track children in a foster care system. And that foster care is, of course, focused on a whole family.

So, they do -- I don't know the specifics of the identifier that they use, but I think it is important to be able to tie together, I guess, parents and children. I guess that is what I mean, in terms of a family identifier, so that you can look at the obvious questions about prenatal care in newborn health, but I think there are a lot bigger questions that that kind of indicator would facilitate analysis of.

Right now, I think what the committee is working with is unique personal identifiers and I believe there is another indicator that says the relationship of that person to the insured subscriber, I believe, but that is not quite going to get you a family identifier in the sense that I am trying to describe.

Now, the logistics of assigning a family identifier, I think, require a lot of thought and I don't have a lot of good answers about that now, but I would be very interested in having conversations about that. That is why I was trying to mention before about under TANAF and formerly under AFDC -- under the former AFDC program, they did have case identifiers, which were basically mothers and children linked together. That case identifier was often imbedded in the beneficiary ID, for example, in many states, but I think using that at least as a model to start from might be a help.

DR. IEZZONI: Great.

Marjorie.

MS. GREENBERG: Just following up a little bit with that, if you could, you know, solve all the different issues about confidentiality, et cetera, but I think in having the -- and you had a unique identifier for each person -- having the mother's identifier on the child records go a long way towards achieving what you want or are you thinking of something more complex?

DR. HERZ: I think that would help in that really small focus, the need for a family identifier. But I think if you want to do more of the sort of integrated service analyses, where you are trying to look at families who are users of multiple support services, especially publicly funded support services, I think you would need to move beyond that concept to a family identifier.

So, I guess I am thinking of sort of a nuclear family. I guess that is what I had in mind.

DR. HARDING: Pardon me. Like a family would have like a social security number with some additional thing that would say I am -- this number means I am a Harding and then another digit or something would put me in my place in the Harding family, so to speak?

DR. HERZ: I think that would do it, but somehow there has to be some common piece of that -- some common piece of that identifier that would be carried by the families or the members, that you would want to keep together. I mean, the logistics of this, I think, are more complicated than the personal identifier.

DR. IEZZONI: But that is a great issue to bring to our attention because kids is one of our special populations that we are very interested in and finding out who they are living with, which may not be biological relatives might be very important.

DR. CLARK: I guess the Medicaid analogy would be in the HIC and the BIC. The HIC is the health insurance claim number and then BIC is the beneficiary information something or other. One of the big problems -- and I am not an expert in the use of Medicare data, but I hear that one of the big problems in using Medicare data is the BIC is constantly changing. So, you know, just defining what -- number one, defining what the family unit is --

[Multiple discussions.]

-- and then trying to keep up with all these changes.

DR. AMARO: It might be easier to really track the mother/child, you know, to connect that into a broader family identifier because that is going to be constantly changing and --

DR. HERZ: Yes, that is going to be a much bigger challenge. Right.

DR. IEZZONI: Let's move on. Are there any other questions of clarification for Lisa at this point?

DR. HARDING: Could we get a copy of that George Washington study?

DR. IEZZONI: Well, there is actually -- Richard I have been making notes. There are a number of reports that have been mentioned this morning, the G.W. report, you know, there is a number of them. Let's keep track of those and then this afternoon, as we begin to talk as a group about where we are heading, let's make sure we get all these reports.

MS. GREENBERG: Just to mention for future reference of the subcommittee that the Centers for Disease Control and Prevention is working with G.W. on developing contracting specs for getting public health and prevention types of goals into the -- and services, into managed care contracts, very much emphasizing what you were talking about. A lot of this is not in the contract and that just kind of -- it all goes from there.

One of the elements is actually developing contract specs for data requirements, not only individual data requirements, but information kind of infrastructure requirements. So, that is something you might want to --

DR. IEZZONI: We do need to know about that.

Elizabeth, did you have a comment?

MS. WARD: No.

DR. IEZZONI: Hortensia?

DR. AMARO: You mentioned creating a national repository for performance benchmarks. Is that in relation to benchmarks around performance of the managed care system or, I mean --

DR. HERZ: Yes.

DR. AMARO: Is anybody working on that right now?

DR. HERZ: Not that I am aware of. I think, what I had in mind, for example, if HCFA, for example, ends up requiring a subset of HEATUS measures for Medicaid managed care plan, having a repository that shows what those values were on a year-to-year basis by state and perhaps by program type within states, I think, would be helpful for plans and states to understand, well, sort of what is the range of experience. Where do I need to focus my efforts to try to have better performance, given what other state plans are able to achieve.

That is sort of what I had in mind. And that is what states often ask us about. They don't know what is a good number to try to hit. You can take the Healthy People 2000, for example --

DR. AMARO: That is what I am thinking, whether anybody is thinking about developing any further, next 2020.

DR. HERZ: Yes. Not that I am aware of but that is sort of thing -- I think, that is sort of what I had in mind. So that there -- but that is sort of numbers that we want to hit. I don't know that it is so much experience in achieving those kinds of outcomes. So, I think what states want to know is what are other -- how is this working in other places.

DR. IEZZONI: Jamie, you seem to be having a technological fight there on your hands.

DR. HADLEY: Yes, I am.

[Pause.]

Again, I am Jamie Hadley from the Office of Strategic Planning at the new HCFA. And what the old Office of Research and Demonstration was charged with and what OSP is charged with is one of their tasks is evaluation. So, when the 1115 waivers started on being granted, one of the jobs we had was to figure out what we were going to do to evaluate these states.

We had a couple of choices, I guess. One option would have been to go with the 1915 B model, where we had the states, you know, contract out and basically do their own evaluations. And I got in a little late, but I think as Rachel probably mentioned that those have variable results. Some of those are good. Some of them are mediocre. Some of them are really awful.

Especially in the beginning, the 1115 states were -- we considered them very research oriented. We really wanted to see what was going to happen in these states. So, we wanted to be sure we came up with a good, high quality, comprehensive evaluation of each of these.

Another option would have been to do it in house, but we really didn't have the staff to do the surveys in the scope that we needed to do. So, the option we chose was to contract out with some large research firms and we made the decision at first, at least --

[Technical difficulties.]

Okay. Anyway, our first thought was that would go for some deficiencies and group the states as they were coming in. They were coming in at a pretty high rate of speed in 1994 and 1995 and it just seemed as if they would be too expensive and too time consuming to award a contract for every single state. So, what we did for the first evaluation was to -- we decided we will group five of these together and make an award. And in September 1994, we made an award to Mathematica Policy Research and their subcontractors, Urban Institute and Systemetrics, to look at the impact of five of these demonstrations.

At that time, we had Hawaii's Health Quest, Rhode Island's RiteCare and Tennessee's TennCare, three of the first states that they were granted waivers and were going operational. Later, we added Oklahoma's Sooner Care and then Maryland's plan, which didn't have a name when I made up this slide, I think, has a name now, but I am not sure what their newest name is.

And at the same time, we awarded a separate contract to evaluate the Oregon demonstration that -- well, let me cover, I guess, the second five state evaluation first, since they are both on the slide. We also have what we call the second five state evaluation because it was awarded a year later, which was awarded to The Urban Institute and their subcontractor, Mathematica Policy Research and they are looking at also five state demonstrations, Minnesota, the great state of Los Angeles, Vermont, Illinois and actually New York. On the slide it says one other state to be identified, but recently New York was approved.

We have decided to add New York into that evaluation group. There were some states, though, that just weren't easily grouped. Oregon ended up getting its own evaluation. We thought that they deserved a separate contract since the demonstration involved, at least at that time, an employer component and a priority list, features that really weren't in common with any of the other states, plus we wanted to take a much closer look at quality in Oregon because of the priority list. So, they got their separate evaluation.

Delaware's Diamond State Health Plan got a separate evaluation for a different reason. We already had an evaluation going on in that state, a children's program, and thought an efficient way to handle that would be just to convert that evaluation to an evaluation of their 1115 study.

To give you an idea of magnitude of these things, the two five state evaluations are both roughly in the range of about $6 million over five years. It seems like a lot of money until you think of looking at five states and doing a really comprehensive evaluation of those five states. It is surprising how that money doesn't go as far as you would like.

The Oregon evaluation is about four -- right into four and a half million just for Oregon, again, because we are looking in much more detail at quality. Medical record reviews cost a lot of money. Doing a different series of surveys there and it is more expensive.

Also, we have got a -- HCFA is a primary sponsor of these evaluations, but I should also mention that ASPI is putting a significant amount of money into the first five state evaluation, primarily Tennessee, and into the organ evaluation to look at disability issues in those two. So, we have some special surveys in there, some special case studies that were going on.

Also, SAMHSA has put money in to look at mental health in much more detail in the first five state evaluation and they may be interested in doing some funding in the second five state as well there. Doing some interesting things, not only looking more in depth at encounter data from those states on mental health and substance abuse issues, but at least in the first five state evaluations, in Tennessee, we are doing something a little interesting. We are doing some longitudinal focus groups and we are looking at people over time to see what kind of services they use and what their experiences are not only within the range of mental health services that are covered by Medicaid because that is all we would end up getting from encounter data, but we are also very interested in what kind of services might they be using outside of that, because especially with mental health there are so many other kinds of agencies that provide services.

So that if we see no increase or a decline in utilization, it could mean a lot of things. It could mean they are getting services somewhere else. It may not be the managed care impact.

In all the evaluations we have some things in common. We are looking at two primary impact issues. What are the impacts of the expansion of eligibility, which has taken place in most of the states, not all, but in most? And what are the impacts of the increased use of managed care?

Obviously, we are having to use different kinds of comparison groups to look at each of those. When we looked at the expansion of eligibility -- well, we looked at increased use of managed care -- that is an easier one to start with, I guess -- there it is a little more of a before and after comparison. Most of the states that didn't have extensive managed care experience, we can look at what -- you know, what the experiences of people were in the pre period and then post. When we are looking at expansion of eligibility, we have got to do some post comparison groups, as well as the pre post comparison.

So, those are the two basic areas. Then within those we are sort of looking at a standard set, using cost of services and that is in terms of the state, Federal Government, out-of-pocket costs and it impacts on providers, access to care, satisfaction, quality issues, outcomes, health status, morbidity and mortality, process of care, continuity of care, timely hospitalization, hospital admission rates.

Now, that varies a bit from state to state, as I mentioned. Doing a lot more in Oregon with quality and looking at medical record reviews. Most of the other states we are looking at quality primarily through the encounter data.

We are doing case study interviews in each state with the evaluators going into each state interviewing people in the state government, different provider groups, individual providers, advocacy groups, anybody who has an interest in the demonstrations basically. They occur over a period of time. There is the implementation one that, you know, we are doing right after the -- probably about a year after the waiver is granted for each state and then later on doing updates for each of those.

In the first five state evaluation we have done -- in terms of reports that are in, we have a report right now on basically the first set of case study interviews for Tennessee, Rhode Island and Hawaii that looks at the implementation of the waivers in those three states.

We don't have any startling findings yet and things like that, but it is a good basis, I think, for anybody who is going to be following these evaluations along. It is good to know what kind of changes the state had to make, reactions of providers in the state and the problems the states have with getting data systems on line. It gives you a basis of knowledge to understand the other reports that will be coming along.

On each of the states we are also doing focus groups. They are used partly for primary data, partly to guide the construction of survey instruments. The encounter data is really the guts of these evaluations, state MMIS systems, surveys of each of the evaluations, I think, with the exception of the Delaware one. We will have surveys of beneficiaries and providers and/or providers, I should say. Not everyone has providers but everyone has a beneficiary survey. And then medical record reviews in some of the states.

For encounter data, when each of the waivers were approved, there is a special term and condition for each of them that required that the states submit encounter data. The special term and condition didn't specify what encounter data had to be submitted, what form it had to be submitted in, who it had to be submitted to any of those details.

Afterwards, each of the states were supplied with a set of minimum encounter data requirements. That is basically ambulatory and inpatient basic kinds of information. We kept it very, very minimal. There was a lot of discussion about how extensive to make it. Our minimum requirements really were minimal. We thought we could do something with this if we got this from each of the states.

Each of the states fortunately plan on collecting a lot more extensive data than what is up there. Now, the problem is between planning and actually doing and submitting data to us. At this point, we have good encounter data from Tennessee and as you mentioned, not only are they using it themselves, but they have supplied us with a couple of years of what looks like good usable data and we are optimistic.

Rhode Island, I think, will be providing us with some good data eventually. Rhode Island is in a unique situation. I guess this is the first time they have actually had an automated data system.

PARTICIPANT: That is true. It never existed before.

DR. HADLEY: Yes. They have had a double struggle, not only with the encounter data and, you know, just struggling with the whole idea of getting data from capitated plans, but just automating their system. So, they have been working very hard on it and we are really helpful we are going to get some good stuff from Rhode Island.

DR. MOR: You will get it. My fellows are already out there.

DR. HADLEY: Okay. Well, you probably know how good it is better than I do then. But, I mean, you had some very ambitious data plans. So, we are hopeful there.

Hawaii, we are a little unsure of right now. Each one of these states is a little different. Hawaii is in sort of a strange position now that they -- they had some labor negotiation problems with their contractor for their data and my understanding is our encounter data is actually sitting on somebody's desk in Hawaii and has been for about six or eight months now because it was ready to be shipped out within a couple of days at the time the contract negotiations fell through.

Everything stopped and our data is still there somewhere I guess. I have been trying to work out a trip to Hawaii and get that data but I haven't convinced anybody that that is necessary. So, I don't know when we are going to get Hawaii data.

Oklahoma, we haven't asked for anything yet really. They have just started. It is too soon. Maryland is too soon from the first five state evaluation. That is the evaluation I am most involved with. I know we are getting decent data at least at this point from Oregon. That is also very good. That was on your list.

Minnesota, I would imagine for the second five state, we are probably starting to or going to get some data soon, but I don't know about the other states. So, it is -- I don't feel great about encounter data because we are three years into a five year evaluation and we are just starting to get a trickle of encounter data and I don't know yet how we are going to end up handling the evaluations. We are, obviously, not going to have at least two years of encounter data for each state that we planned on having for our studies.

We are going to have to make some adjustments to that. In fact, one of the things we have discovered in scheduling these evaluations is that it probably doesn't make sense to group states together because our idea, I think, was that the states would come on line at about the same time. They would develop it about the same time. Things would flow, you know, kind of in groups.

It hasn't. Each state is very different. So, we are not sure what we are going to do from here on out to evaluate other states that come on line. I know Stan's group is working on coming up, you know, ways to help the states out with being ready to and able supply encounter data. We are looking into some possibilities of contracts that would, you know, work with someone to just perhaps build an encounter database, you know, for future evaluators to work with or even in house researchers.

It doesn't seem practical or efficient to do full scale evaluations of every state that comes on line. First, not every state is doing something shockingly new and really interesting. At this point, all of them are starting to look similar. And we may not learn that much more and it may make sense to have the states start to do some of their own evaluations from here on out.

But we are thinking through a lot of possibilities at this point. We are not really sure what direction we are going to go. I think our big focus now is trying to get a good core set of encounter data so we can do something because whatever we do, we are going to need that.

We are still very much in an exploratory phase with that. We are looking at the data we are getting from the states, none of which is really comparable from one state to the other. We can't really combine it across states. I know Stan's group is looking into that. There are some people in the office -- some researchers in OSP that are starting to look into that, too, to see if we can integrate that into the MSIS kind of format. But we don't really know if we can squeeze this encounter data into that fee for service type format yet either.

So, a lot of work remains to be done and it all revolves around encounter data. Something that all of you are interested in is the data group, too.

DR. IEZZONI: Okay. Jamie, this has been sobering, but informative.

Are your contractors, Mathematica, Urban, Urban, Mathematic, Share, RTI, RTI Share or HER, I guess, not Share, are they instructed at all to look at data systems within the states to support quality?

DR. HADLEY: There is money in each of the contracts to provide technical assistance to the states. Some of the states overlap with the work that Medstat is doing and some of them -- the contractors are doing that technical assistance without Medstat.

DR. IEZZONI: If we wanted to ask Mathematic, for example, to come and speak to us about data systems for evaluating --

DR. HADLEY: They would be willing to do that.

DR. IEZZONI: They would have the knowledge based on the activities that they are contracted to do for their evaluation. Okay.

DR. HADLEY: Sue Donns(?) is a good contact person from Mathematica.

[Multiple discussions.]

DR. RAJAN: I would probably go to Sue before --

[Multiple discussions.]

DR. HADLEY: Yes. She has worked with all the states in detail and can really help you out with a lot of individual elements and where they are having problems.

DR. IEZZONI: Okay. Good.

DR. HADLEY: But each state is -- it is unique. It is hard to generalize about the kinds of -- and I guess something nobody has really mentioned is that these states are really running the demonstrations on a shoestring staff. In a lot of cases, they just don't have people to do everything they need to do. Data is important and I think they know it is important, but they need to get it up and operating and monitoring the thing and it is not always --

DR. IEZZONI: Can I just -- you keep calling these demonstrations. We are interested in Medicaid managed care. Are these demonstrations of Medicaid managed care that will go away at some point? Or is this -- is what you are evaluating Medicaid managed care?

DR. NACHIMSON: The experiments that were allowed the 1115 waivers are basically where certain standard Medicaid requirements were waived so that states could implement these broad managed care initiatives expanded eligibility in services and things like that.

Technically, they had a five year --

DR. HADLEY: I can't imagine them going away and saying -- I can't imagine that we looked at and said, you know, I don't think managed care worked very well there. Let's go back to fee-for-service.

DR. NACHIMSON: Technically, they had a five year life and the demos would need to be renewed. Arizona, I guess, is a perfect example where they initiated --

[Multiple discussions.]

-- demonstration that started in 1982 and continues to this day.

DR. CLARK: They did get congressional legislation for that.

DR. HADLEY: And there are probably other bits and pieces of what we are waiving now that will eventually be legislated, I think, and won't be waivers in the future. So, it is hard to say what form it will take, but, no, I can't imagine these just ending in five years and going back to something.

DR. IEZZONI: But what these are are basically the managed care programs in these particular --

DR. HADLEY: These are the new programs.

DR. IEZZONI: Okay. So, we will get rid of the word "demonstration" for what our committee will be doing. Is that okay?

DR. NACHIMSON: But there states in which there are not these 1115 for demonstrations that have managed care programs.

DR. HADLEY: And what is important to realize, too, is in the non-demonstration states, there aren't any requirements for data in those and at least -- even though we are getting some trickles of data, you know, some of the states Stan mentioned are just ones that happened to be providing good data, but we haven't been working with -- pushing them.

DR. NACHIMSON: Well, there is no requirement for evaluations in these other types of --

DR. IEZZONI: So, the demonstration in states have the requirement for evaluation and then other ones don't.

DR. HADLEY: Right.

DR. AMARO: But they didn't receive any resources?

DR. HADLEY: They are not doing their own evaluations. I mean, they may, but we didn't fund them to do their own evaluations. It is an external evaluation in each of the states funded by HCFA.

DR. AMARO: But are they -- you were mentioning that they are impacted in that they have to provide -- in providing the data that requires some person power and --

DR. HADLEY: There are no separate funds, as far as I know, set aside for them to provide that data. It is just part of the program. In the past, they provided claims data. So, we have asked them to provide encounter data instead. From the state's perspective, I think they are still providing data. Unfortunately, it means a new data system and a new way of doing it in a lot of states. It is not actually a new task.

For the HMOs in a lot of cases, though, it is a new thing that they are doing, where they weren't providing data on those people to anybody else, perhaps.

MS. GREENBERG: Just, for example, in the top ten states that Rachel gave us in her handout, only Tennessee and California, I guess more recent, are 1115 waivers. So, I guess, because they went into managed care on their own -- well, all the rest have 1915 waivers, but those -- there is no requirement for encounter data in 1915 and no evaluation requirement either.

DR. HADLEY: Well, there is an evaluation requirement. 1915Bs have to do a yearly -- is it every year or every two years?

DR. NACHIMSON: I am not sure -- it is not really a -- it is not an evaluation requirement in the same sense.

DR. HADLEY: Right. They have to justify their waiver each time. It is not as extensive --

PARTICIPANT: Isn't the Congress about to make a change --

DR. HADLEY: That might happen.

PARTICIPANT: You are describing the situation as of today, which may change in 30 days.

DR. HADLEY: That is right. That is correct.

DR. CLARK: Actually, in California, that is not a statewide --

DR. HADLEY: No, actually the California one is just Los Angeles. It is a little different animal from all the rest.

DR. IEZZONI: Okay. We are running a little bit behind on our schedule, but, Richard, you had a -- okay -- yes, George.

MR. VAN AMBURG: Can you give us an idea of what measures of evaluation is -- you say they are going to measure access to care. What are they going to use for measures?

DR. HADLEY: Well, some of that will come from surveying the beneficiaries themselves, you know, asking about -- you know, for access, a lot of the standard kinds of measures, you know, waiting times for appointments, you know, distance they have to go, access to primary care specialists. Have you ever wanted a service that you couldn't get? You know, those kinds of things. So, there is a whole set of basic access kinds of questions we always use in the surveys.

So, that kind of information will be -- wherever we can we use encounter data to look at use of specialist care. It is always -- whenever you are evaluating managed care, though, it is always hard to sort out things that have changed because you have got managed care now and, you know, what was the necessary and what was the unnecessary care?

So, we learn an awful lot on the survey data, I think, for access, probably more than the encounter data, asking questions. Then again comparing that where we can to a fee-for-service kind of equivalent.

DR. MOR: One question on the surveys. In all the states will there be special designated oversampling of at risk populations or is this just going to be general?

DR. HADLEY: I would say there will be some in each of the states, but --

PARTICIPANT: I think it would depend on the demonstration.

DR. HADLEY: It varied, though, in the state which populations we are going to be looking at. Rhode Island is, obviously, unusual, too, with their population. I don't know if any of them are typical now that I think about it. But we are looking at different kinds of populations, I think, is the safest answer.

MS. GREENBERG: Can you answer a very quick question about encounter data? The states that you are hopeful you are going to get some encounter data, are you expecting that minimum data set or --

DR. HADLEY: I think it will be well beyond the minimum data set.

MS. GREENBERG: -- was the McData(?) data set shared with them, which is --

DR. HADLEY: Well, actually the minimum data set is the McData set. The first two columns in the McData set basically, it is the inpatient and outpatient, not the pharmaceutical and not the --

MS. GREENBERG: Well, but like the minimum data set has one diagnosis and things like that, which the McData set had more in it.

DR. HADLEY: The McData set, I think, only has two diagnoses because there was a lot of discussion I know with Sue Dodds saying that, gee, even two is pretty unacceptable to do much of anything with. So, I think from the states where we are getting diagnosis, it varies from state to state, but I think it is more like, you know, we are getting three, four -- you know, we are getting -- the data is pretty decent where we are getting it.

MS. GREENBERG: Okay. More than that minimum --

DR. HADLEY: Yes, right.

DR. IEZZONI: Any other comments from the committee at this point?

[There was no response.]

Well, again, sobering but informative.

DR. HADLEY: Probably you or someone should be on the list to receive reports as they come out.

DR. IEZZONI: I have made a list of those as some of the reports and I think we will be wanting to hear directly from the investigators, some of the investigators involved in those reports.

Okay. Great.

Rachel was actually quite eloquent in talking during her presentation earlier this morning about our need to look at the dual eligibles and how that kind of a group that has fallen between the cracks and been in a bit of a nether world. Nobody has really looked at them, except, I guess, we have two people here who know something about this.

Could you just pronounce your first name?

DR. RAJAN: Shruiti.

DR. IEZZONI: Shruiti and William or Bill or Will or Bill. Okay. So, why don't we during the end of our morning session hear from the two of you. I guess, Bill, you are listed first. Is that okay?

Agenda Item: Dual Eligibles: HCFA Initiative, Status and Discussion of Interrelationship of Medicare and Medicaid in Selected States

DR. CLARK: Okay. I will be brief.

I guess, I will go over what I intended to do and see if that sounds like what the committee might want to hear. I was thinking about just talking a little bit about the dual eligibility issues, talk a little bit about some of the major policy issues that are sort of under discussion right now and then talk a little bit about some of the data issues that we are looking at down the road, if that sounds reasonable.

I just don't know how much background -- I know that some of you are from the New England states where this is very hot and -- but I guess that is what I would like to do if that works out for everybody.

DR. IEZZONI: This has been an educational morning. So, I think that that sounds good.

DR. CLARK: My name is Bill Clark. I am with the Division of Health Systems Research in the Office of Strategic Planning in the new Health Care Financing Administration.

I am project officer right now for a number of states that are requesting waivers for Medicare and Medicaid under a demonstration authority to test integrated approaches to managed care for people who are dually entitled for Medicare and Medicaid.

Just by way of background, really there are now only two what I would call comprehensive models of care for dually entitled beneficiaries, one of which is ONLOC(?), which is what those characters say. If they are oriented correctly, which I am not entirely clear, since I don't speak Chinese -- they are? Very good. Thank you.

I think I finally got that right. I have had several presentations where people came up to me later and said you have got it upside down and backwards. And I really didn't know.

Then the other one actually started this last April, which is the Minnesota Senior Health Options Program. So, there are really two models out there probably serving maybe 2,000 people total nationwide, maybe 3,000, that are consciously designed models serving dually entitled beneficiaries.

Now, ONLOC -- the reason I like to use this overhead -- it has two meanings, one of which I learned on an airplane somewhere, but the first one, which many of you may have heard in one context or another is peaceful, happy abode and the other meaning is the total absence of financial worry. While that may have marketing draw in Chinatown in San Francisco, I think it is actually a very interesting kind of goal for the kinds of systems that we designed for dually entitled beneficiaries because we know that these are individuals who are, by definition, impoverished and the issue of financial input is key to their ability to use care.

So, the interesting thing about ONLOC is that really combines three aspects in making a successful model. First of all, sort of on the financing end, it combines Medicare and Medicaid funding through two separate funding streams and then when it is in the plans' purview, they basically are able to launder the money any way they want to provide whatever services they want, as long as they are by Medicare or Medicaid certified providers.

So that if you are getting money on your capitation to cover hospitalization, if you have lower hospital utilization, then you can put that money into community support services, transportation, whatever else you might want. But the three sort of service components of the model that are interesting and I think are unique are that they focus on adult day care as the core service site for bringing in the physician care, the dental care, the occupational therapy, all the therapy services.

Everything is brought into an adult day center for the most part. They also have home-based services as well. Then, secondly, is this concept of multidisciplinary team, which involves a weekly meeting of all the professionals involved in the care delivery, including the physician, the driver, the day care, social worker, all that. And, finally, the part that is not generally talked about very much, but I think is really key is the housing itself is also part of their overall concept.

In many of the sites, the assisted living or whatever you call kind of housing support is available to members. Some people have called this model life care for the poor as opposed to the CCRC type of movement. I think in many ways that describes it.

The Minnesota model I am not going to really get into right now, but I wanted to, as a way of rounding out our sort of background on the dual eligibility issue, this appeared actually just after the Kennedy-Kassebaum law was passed and I thought it just really showed a lot about our current perceptions of dual eligibility or our current perceptions of long term care in America. To me, it showed that the inevitable spend down process to access long term care, the high cost of access, the drawbridge effect, the heavily institutionalized bias of our system -- notice the castle -- and the total dependency of our beneficiaries as seen by the nurse helping somebody walk and then someone pushing a wheelchair and then the concept of no return. Once that drawbridge is up, forget about it.

What it doesn't show is that behind this lies an incredible entrenched bureaucratic tangle that has to be sorted out and in my personal belief, it cannot be addressed until statutes are changed because I don't think waivers can really get at the bottom of this problem.

But it doesn't show beneficiary confusion over health insurance coverage in a transition from private insurance to Medigap to Medigap with Medicare and then eventually -- at least, if you are institutionalized, then the total conversion to public financed health care. And it doesn't show the kind of provider breakdowns that occur in the whole continuous process of care or the payer fragmentation that results from having all these different sources of financing.

In spite of this medieval concept of long term care, I think that there is some room for optimism as we work with -- as HCFA and the Department work with states to develop new approaches to serving dually entitled beneficiaries.

Basically, this is what I like to use as a way to describe the relationship between Medicare and Medicaid in serving dually entitled beneficiaries. I am getting a little heavy on the Asian symbolism, I realize, but it sort of works for me. If you think of Medicare as having -- the Medicare and Medicaid as complementary and mutually symbiotic, I guess, might be one word, programs that serve in a totality of the population that are entitled for both and in one program, Medicare, let's say, starting with Medicare, Medicare has some services that might be considered long term care kinds of post-acute kinds of services that one might actually see more often in the other program, the Medicaid program -- that would be the little circle inside the big half.

And then in the Medicaid program, you would see generally the long term care kinds of services or the support types of services, the non-medical kinds of services with the inside of that being prescription drugs, co-payment for Medicare, the medical acute services, things like that. So, within each program is a little bit of the other program and if you look at it in terms of where we are in the dynamics between states and Federal Government, I see that there is a sort of a large -- a question of who owns this population. I think states will come to us and say, gee, this is really our population and we are going to serve them.

The Federal Government may have the same perspective in the opposite view. But, actually, the population of dual eligible beneficiaries was about six million in 1995 nationally and they represented about 16 to 17 percent of each program's total enrollment, but represented about 30 to 35 percent of the total cost of each program, according to work that was done by our division using the Medicare CBS survey.

By the way, my apologies to Bruce Lee. I picked this thing off his Web page and there is an interesting little paragraph about martial art and what I would like to do is substitute the words "health policy analyst" for "martial artist" and you basically have got it.

What I would like to do -- by the way, I don't really know how much time I have to --

DR. IEZZONI: A few more minutes.

DR. CLARK: Okay. I guess I would like to list some of the major policy issues that are under discussion. The last one is not a policy issue or maybe it is. And maybe just try to give you a little sense of where some of the different models fall on this chart.

The first issue really is the issue of continuity of care across the systems of health care. For example, the ONLOC pace model, which I have briefly described is only available for people who are dually entitled or Medicaid only. If you are Medicare only, you could buy into the plan, but it would cost you $1,400 a month or so.

So, it basically involves people at a certain point in their illness of making a conscious decision to go from one kind of provider to an ONLOC or a pace provider. So, there is a break.

Now, the Social HMO, which in theory would involve Medicare and Medicaid together, but, in fact, does not, would be an example of a continuous model in which you could go from fee-for-service as a Medicare-only beneficiary, enroll in a SHMO and then as you needed to have care from both programs, you could receive that from both, but, in fact, none of the current SHMOs, I believe, have a contract with Medicaid.

Many of them, dating back to 1985, never had. The Minnesota Senior Health Options is a model that is only for dually entitled beneficiaries. However, their major contractor right now -- they have, I think, an enrollment now of about 460. Their major contractor, which is expected to come on line in the next month, has a Medicare risk contract as well. So, presumably then you could switch plans from your Medicare risk contractor to your dual eligible plan at some point.

And that the model service, not only just dual eligible, but I believe it also serves Medicaid only beneficiaries. The concept developed by the New England states, Integrated Service Networks, also involves right now, I believe, for dually entitled beneficiaries, may be reaching up to a higher level of eligibility in terms of income for some -- for people who may be at risk of institutionalization.

But for the most part, from what I have seen so far -- and this is very much in development -- the target population is exclusively the dual eligible population. I think that is a major issue to consider because for those -- for long range program savings for Medicare and Medicaid, how much money can you really save if everybody is at an institutional level of care to begin with? Or if someone is low income to begin with and the state is always going to be held to their cost-sharing requirements.

It seems to me that states would do better to work with HCFA in targeting a population that is more generally at risk of spending down, at risk of institutionalization and then working accordingly. That is where I think the long range savings could be for our programs if these things were done correctly.

The second issue -- I don't have a watch, so I don't really know, but beneficiary choice is a major issue. ONLOC is strictly a voluntary model. Other models may have Medicaid as a mandatory feature of enrollment and Medicare is a voluntary feature, making systems very difficult to administer and making it very confusing for beneficiaries.

The third issue, administrative oversight, this is where we get into who owns this population. Should the states administer these programs or should the Federal Government administer these or should we do it together and then if we do it together, how?

Finally, the financing, most of the models that we have approved to date -- all the models we have approved to date are based on some variation of the AAPCC as a method of payment. However, we are hoping that the newer risk- adjusted methodologies, such as the diagnostic cost group, hierarchical coexisting conditions, methodology or maybe the disability payment system may have some value as future risk-adjusted models that could be applied.

Finally, I get -- this is something -- I don't really know. Your committee is dealing with the HIPAA(?) data requirements and I think that we need to really think when those become implemented how that will impact on the design of these systems because presumably if there is some sort of standardized encounter data reporting, then we should be planning for that now when we design our waivers and when we make approvals to these waivers.

Maybe somebody will repeal these provisions of HIPAA before that happens. I don't really know anything about it, but I guess we need to start thinking in terms of coordinating those requirements in the future.

I just want to talk a little bit more about the kinds of data that we are -- data activities that we are involved with with some of the states. We have merged Medicare and Medicaid claims for about 12 states right now. Each one is at a state level of the data construction and, so, we are working towards assessing the strengths and weaknesses of these individual data sets and to what extent that they could be made -- they could be merged to be more uniform for purposes of analysis, so that people like Shruiti could go to town on it.

In addition to that, the claims level data, there are additional data systems that would be really interesting to get into this, including the MDS system, the Oasis Home Health System and some of the home and community-based waiver data that is out there. Furthermore, there is all sorts of state and local level data that would include function status information and other things that should be also considered in terms of how can you merge these databases.

I am wondering actually in the context of this discussion today whether we need to be thinking more broadly in terms of bringing in mortality data, other sorts of public health data systems that states maintain that we haven't really thought about at all.

Finally, we need to keep ourselves thinking about those studies which we have done in the past, like the Framingham Study in Massachusetts, the EPI Study in East Boston and New Haven and surveys like the HIS and the Special Disability Supplement and the National Long Term Care Survey and all these different things. When we think about implementing these kinds of projects, how can those systems be used creatively in the context of these larger merged data files?

Just in closing, I guess it is a little frightening to hear the experience of our AFDC 1115 waivers and that our encounter data is not uniformly reported or that we don't require uniformity and it is not even -- and then it is not uniformly reported and then to know that we don't really know the impacts of these programs three or four years after we have implemented them.

I think with the dual eligible population, the stakes are sufficiently high with the vulnerable populations that were being served here, that we need to design these programs correctly from the get-go, make the requirements that we want and try to make them uniform across the applying states and then enforce those standards that we develop.

Finally, a handout I circulated lists a grant solicitation that is out on the street to states, which lists a number of priority areas, which I hope you will take the time to look at. And at the back of the solicitation is a chart book that was developed by our office on dual eligibility. So, I hope you will take the time to look at that as well.

Thanks.

DR. IEZZONI: Thank you, Bill.

Brief questions for clarification for Bill?

We will move on to Shruiti. Okay.

DR. MOR: What are the 12 states?

DR. CLARK: Well, in various years and --

DR. MOR: For public use now.

DR. CLARK: Nothing is public use right now. Actually, the caveat to all this is privacy and how we deal with many of these issues in terms of trying to get this stuff utilized. As some of you may have worked with HCFA's data system before, it gets very complicated very fast with transferring data from one party to another, but right now, these are not even close to being publicly available.

We are only talking right now about whether we can actually use this data at all in a merged context. So, if it turns out that even at the state level that it is worthwhile, I mean, Rhode Island is an example of a state that has not yet submitted its updated requests for some of it, but it has leaked data from 1992 to 1994.

This is six New England states: Colorado, Wisconsin -- then we have some additional states from another project that Medicare -- the Medicaid Drug Utilization Review Demonstration, which includes Iowa, Maryland, Georgia and Washington State.

DR. IEZZONI: That is great. Let's move on.

Shruiti.

DR. RAJAN: The dually eligible population, as most of you know, has been receiving a lot of attention lately from both federal and state policy makers. This really isn't surprising because the duals are a sizeable, very vulnerable population. As Bill mentioned, there are about six million dual enrollees and that is institutionalized and non-institutionalized.

In Table 1, I have put together sort of a brief profile of the duals and, as you can see, in comparison to other Medicare beneficiaries, the duals are disproportionately for non-white and female. I think it is also important to note that the duals are disproportionately made up of individuals who are under age 65 with disabilities. They tend to have, you know, stronger health care needs.

From a positive perspective, the duals are a challenging population because of their health care needs but also because of the lack of coordination between the Medicare and Medicaid programs in serving them. The lack of coordination is problematic. Both from a quality of care standpoint, as well as from a cost standpoint, there is going to be a lot of cost shifting between the two programs and interactions between the two programs are important to consider when state and federal policy makers are trying to limit spending in both programs.

So, the paper I am going to present today is a result of work that Genevieve Kennedy and I have been doing at The Urban Institute under a grant from HCFA, on which Carolyn Rimes is the project officer and Corben Moon(?) is the principal investigator.

The focus of our research has been on home health for the duals and more specifically on the interactions between the Medicare and Medicaid programs in providing home care to this population. It is one of the few service areas where both programs overlap. We have approached this project basically in two phases. The first phase was to look at spending patterns in Medicaid and Medicare over time and across states and we supplemented that analysis with case studies in five states trying to understand why we are seeing these spending patterns.

The second phase, we have explored the issue further through a multivariate analysis using the Medicare current beneficiary survey. Today I am going to focus on findings from the first stage of the project and I would like to conclude by looking at some of the data needs.

Now, we were motivated to research this topic by many of the reasons I just mentioned. The duals are a very vulnerable, diverse and sizeable population and they are also disproportionate users of the Medicare home health benefit. There has also been very little recent research done that has considered both programs together. And sort of the last factor that spurred our interest in the topic was the growing interest in all the innovative programs that Bill just mentioned, the MSHO project in Minnesota and the New England states recent interest.

What kind of relationship did we expect to find when we first started this project. We expected there to be a positive relationship between Medicare and Medicaid spending on home care across the states. Our thinking was that states that had more restricted nursing home policies and more generous home and community-based waiver programs, not just waiver programs, just home and community-based care programs in general would have left proportionately more individuals in the community who could access both Medicare and Medicaid uses.

In addition, since Medicaid is a payer of last resort and because states have a strong incentive not to pay for services that would otherwise be covered by Medicare, we felt that there really wouldn't be that much substitution between the two.

Finally, both programs have very different histories and objectives. So, we thought there would be minimal tradeoffs between the two programs.

Before discussing the spending growth patterns that we saw in these programs, I guess I would just like to say a few words about the structure of the Medicaid and Medicare health care benefits. As many of you may already know, the Medicare home health benefit is a medically- oriented, post-acute care benefit and in contrast the Medicaid home care program tends to be more of a chronic care benefit and it is made up of three different components, the home health benefit, the personal care program and home and community-based waiver programs.

The home health benefit is mandatory and it most closely resembles the services that are provided under Medicare. That is home health aides, skilled nursing and therapy services. The personal care services are optional. About 30 states offer those now and they are typically designed to help individuals with activities of daily living, such as eating, dressing, bathing.

And the third component, home and community-based waiver programs, are meant to substitute for more costly institutional forms of care for individuals who are at high risk of going into nursing homes or other institutions.

In Table 2, we have overall Medicaid expenditures and expenditures for each of the three components under Medicaid, as well as expenditures for Medicare home health between 1987 and 1993. Between 1988 and 1993, Medicare spending increased fivefold. That is an average annual growth rate of about 38 percent.

And Medicaid home care outlays tripled over that same period. So, at least from an aggregate spending standpoint, it doesn't appear that Medicare spending was offset by lower Medicaid spending levels. In other words, both programs are seeing high spending growth on these services at the same time.

In Table 3, we put together aggregate and per enrollee spending on home care services in Medicare and Medicaid for 1993. I would just like to point out two observations here; first, a concentration of Medicare home care spending is in New York. Almost 37 percent of all Medicaid home care spending was in New York in 1993.

In fact, if you look, New York's Medicaid spending levels are almost four times as high as its Medicare spending on home care. The second point is that we see some real outlier states in terms of Medicare home health outlays. Tennessee, Mississippi and Louisiana stand out as having overall Medicaid home care expenditures that are substantially lower than their Medicare.

So, we used these current enrollee spending numbers from Table 3 to classify states based on their spending relative to each program's national average and the results can be seen in Table 4. You can see from this table that a majority of states, 30 in all, conform to a pattern, which suggests that there is a tradeoff between the Medicare and Medicaid programs. That is, you will see that many states have high Medicare and low Medicaid spending or vice-versa.

It is also interesting to note the geographic patterns that show up in this table. For example, all six of the New England states, which have had a strong tradition of home care through visiting nurse associations, are in the high Medicare, high Medicaid category, while many of the states in the Southeastern part of the country wind up having -- wind up in the high Medicare, low Medicaid category.

So, to get a better understanding of the spending patterns that we were seeing, we collected five states for closer study through telephone interviews and the five states we selected were Connecticut, New York, Minnesota, Louisiana and New Mexico. The states were selected based on the four different categories. We wanted to select one from each category. But we also wanted to get geographic variation and we felt that we had to include New York, since its Medicaid home care expenditures are so large and the program is unique.

So, we gathered information through telephone interviews with three to five individuals in each state and generally we spoke with Medicaid staff and other state officials involved with home care, as well as with home care provider association representatives.

In terms of the topics we covered, we wanted to learn more about state policies and practices, as well as provider behaviors that might be affecting home care use. So, for example, we asked states about the presence of Medicare maximization or third party liability programs. Those are programs in which states are trying to recoup as much Medicare and private funding for home care as possible and limit inappropriate billing to Medicaid.

We were also interested in how other state practices might be affecting home care. So, we asked about the capability of state MMISs to track duals and to identify the appropriate pair.

And the other area that emerged as an important factor was provider practices, particularly a provider's knowledge of Medicare and Medicaid regulations and their relations with the Medicare fiscal intermediaries. So, the three major findings I would like to point out from our case studies are, first, all three states with large Medicaid home care outlays -- that is, Connecticut, New York and Minnesota -- have initiated Medicare maximization programs and these programs have a number of different components. A typical one is provider education.

States are trying to educate providers about Medicare and Medicaid billing and eligibility regulations, but it appears that the key to the effectiveness of these programs was the state's ability to identify dual enrollee claims and route the claims to the appropriate payer. Usually, that is done through the MMIS. In fact, the two states with low Medicaid spending that we studied, Louisiana and New Mexico, both had MMISs that enforced the requirement that providers bill Medicare first on behalf of dual enrollees.

The second finding is that the two high Medicaid, low Medicare spending states -- that is Minnesota and New York -- have personal care programs that often relied on providers who weren't necessarily certified to provide Medicare services, so that these providers had no incentive to refer dually enrolled clients to Medicare even when appropriate because they couldn't reimburse.

The final point that was actually surprising to me was that Medicaid-financed home care was sometimes preferred by home health agencies because its eligibility criteria could be interpreted more broadly. Reimbursement seems more certain under Medicaid in a lot of cases because there are strict Medicare FIs. And in some cases, reimbursement rates are actually higher under Medicaid than under Medicare.

I guess I would like to step back a little bit from our research to conclude and talk a little bit about some of the data gaps that exist for the dually enrolled populations, these 12 states that Bill mentioned were actually new to me because I hadn't heard about that.

I think we really could benefit from uniform individual level Medicaid expenditure and utilization information for the dual population. While MMIS systems might contain this information, they are usually not uniform and they are very difficult to work with. I guess Nirvana would be merging Medicare and Medicaid claims for this population so that you could really look at the continuum of care. You could look at acute through long term care and I know that has been discussed and it is being done. I think Maine has done that on its own, right?

But it is really not done and it is a large data gap. I think it would be a very valuable tool for states, as well as for researchers.

The second data gap really concerns policy information, more detailed information for all 50 states on program dimensions that could be affecting interactions between these programs would be very useful. For example, we might want information on the existence of the Medicare maximization programs across all 50 states or the ability of state MMISs to track duals and how states are using that to route dual claims.

DR. IEZZONI: Okay. I see the pre-lunch kind of people are feeling a little bit tired, but those were interesting presentations. So, let's see if there are some questions.

Can I just start out, though? You know, what we are beginning on for this particular year is a focus on Medicaid managed care. I just wonder from your point of view -- Rachel spoke about it briefly this morning, but from your point of view, where are the dual eligibles in this broader issue of Medicaid managed care that we are going to be focusing on this year?

DR. RAJAN: I know probably most about the MSHO demonstration and I think, you know, they -- Minnesota had a very difficult time trying to figure out what do you do with the institutional long term care piece. That is only -- I think that the plans are at risk for the first 60 days of institutional care and they are not -- it goes back to fee-for-service after that for the duals who are in this integrated managed care.

[Multiple discussions.]

DR. IEZZONI: -- two programs that Bill referred to?

DR. CLARK: I am sorry. I think it is for people who are actually institutionalized after a certain number of days that the fee for -- it is a pass-through of fee-for-service.

DR. IEZZONI: Okay. So, people who are in Medicaid managed care and then get institutionalized for --

DR. CLARK: They are actually still enrolled -- they would still be enrolled in the plan presumably, but their payment would change.

DR. RAJAN: Right. And then they go back to having fee-for-service.

DR. CLARK: Right.

DR. RAJAN: I guess the point I was trying to make is that Medicaid managed care for this population, I think, is very complicated because you are getting into an area that states and no one really has a lot of experience with. You are getting into capitating long term care and I think plans are very reluctant. I am not quite sure what the agenda would be in terms of what you want to look into.

DR. CLARK: Just to follow up on that, I think the relationship may be that if the states recognized that the AFDC population -- I think it was a strategic approach that states took. They realized that -- because if you look at the cost structure, I mean, the old -- the 80/20 split, where the AFDC is higher enrollment in terms of total populations, but lower costs overall per capita.

So, that was maybe the easier challenge for states to embark upon in terms of bringing people into managed care. They also greater control because it was only Medicaid to worry about. But if you look at the duals, it is much higher cost per capita plus it involves -- it is an order of complexity, which is greater than Medicaid only because you have got Medicare.

But I think that the relationship is that the AFDC waivers really were sort of the first wave of Medicaid managed care and now that the duals represent sort of the second wave and then having sort of gone through a dry run, if you will, and learned how to report encounter data and to implement health plans across states and things like that, that kind of experience now, they are bringing to bear in serving this other more difficult complex population.

But I think New England is actually -- the Massachusetts waiver request right now, which is under review is, I think, the first one that really includes all institutional care under a capitated arrangement. It is certainly the first one statewide and I think other than ONLOC, which does include all Medicare or Medicaid services, including institutional care -- this is, I think, probably the big one in terms of that institutional component.

I think there are going to be a lot of questions about the capacity of providers to absorb that risk in serving that population.

DR. RAJAN: And I guess I would just like to say duals, I think, in Minnesota -- other people may know more about this than I do, but I think the duals have been in managed care for the Medicaid wrap around for a number of years in Minnesota for the prescription drugs and the other pieces that Medicaid fills in for Medicare for this population.

DR. CLARK: But then Minnesota is pretty unique, too. I mean, because that was through a previous demonstration waiver that started in 1980 or 1982. Are you talking about the P-MEL(?)?

DR. RAJAN: Yes.

DR. MOR: What is the relationship between the EverCare Program, which is apparently just a Medicare thing, although it could be a Medicare and Medicaid, but it is not, and the dual -- because all of those people largely are dually eligible and they are living in nursing homes. That is happening in multiple states. What has been the rationale for a non-dual-eligible formatting under that circumstance?

DR. CLARK: I am not sure if there is a rationale. I mean it was --

DR. MOR: It just happened that way?

DR. CLARK: Yes. Somebody came in with applications. Hey, let's try managed care -- Medicare managed care for nursing home residents. Okay. Let's try it.

DR. MOR: And everything Medicaid stays the same.

DR. CLARK: Yes. I think probably the original applicant thought, well, why would we want to mess around with however many state agencies and worry about getting the Medicaid piece of this when what we really want to do is to try to work managed care in a more limited context. But I think that there are probably pretty -- a number of pretty severe limitations to what has actually been tested in terms of the conceptual model, in terms of EverCare.

DR. HARDING: Your statement earlier on where you said 16 percent are spending 36 percent or something along that line. I take it that is in the area of nursing homes and community care homes. What is the disproportionate share?

DR. CLARK: It is actually -- it is right here in this slide, but it is of total Medicare expenditures, 16 or 17 percent of the population is costing about 30, 35 percent of the total. It is right on the first --

DR. HARDING: The double is the dually eligible?

DR. CLARK: Right, the dually eligibles. Sixteen percent of the Medicare -- let's see, where is it?

DR. RAJAN: Sixteen percent of the Medicare beneficiaries got 30 percent of Medicare program spending in 1995.

DR. HARDING: What are they? What are they expending all that money? What is their problem?

DR. CLARK: They are sick and they are poor for one thing. So, they were being hospitalized a lot. They have high physician costs. They have high home care.

DR. HARDING: So, it is all of the above, not

just --

MS. RIMES: It is A and B, Parts A and B.

DR. CLARK: It is total Medicare, but Medicaid has the same experience. I mean, it is 16 or whatever, 17 percent of Medicaid and was it 35 percent of total Medicaid expenditures. So, we are talking about a very sick, old and frail, people with, you know, severe disabilities and, you know, they have high costs. So, there shouldn't be any real surprise that, you know, these people are going to have costs, given their health status and disease experience, I guess.

DR. HARDING: You said that there is another layer right above that group that is at high risk that isn't expensive yet. Can you just say a word more about that?

DR. CLARK: These are the notorious QMBI(?), SLIMBIs, LBBYS(?), whatever -- the BWI -- BWDI -- qualified working disabled individuals, that is the QWDI, which I don't really understand the eligibility constraints, but it is a little different than Medicaid dual eligibility status. Then there is the qualified Medicare beneficiary, which is a hundred percent of the federal poverty level, where the state must buy in for Medicare cost sharing. And then there is the specified low income Medicare beneficiary -- that is the SLIMBI and that is actually very -- that is really key in terms of the current policy debate or the current legislation on the Hill because that is where remember the Congress is talking about having the Federal Government or the states pay more for the SLIMBY cost sharing. These are people up to 300 percent of the federal poverty level, where the state buys in for the Medicare co-pays and deductibles as I understand it. Am I right?

DR. RAJAN: 120 percent.

DR. CLARK: 120. Oh.

But these are sort of eligibility groups above the basic dual eligibility level, which really are not really being targeted by most of the states because they are not really on their plate right now. On the other hand, as I was saying, it seemed to me that these are the people who for one reason or another are probably -- might be worth thinking about in terms of folding in to these approaches.

DR. IEZZONI: Any other comments from the committee, questions before we break?

DR. MOR: The comments you were making about the HIPAA and the -- did I gather from your comments that this issue is not actually integrated or taken into consideration in your review of the existing plans or the waiver requests for dual eligibles or anyone at this juncture?

DR. CLARK: I think it has to be. We haven't gotten that far along with the reviews to assure that it will be. I think that depending on what certain -- there are certain implementation guidelines; 22198 for the transaction data sets, unique identifiers and security standards, that certainly falls within the window of the implementation plans of the New England states. Nobody plans to implement before that.

So, one would think that these standards should be applied to these waivers at that point in time that they are determined. Then the other one is adopting standards for claim -- I mean, it seems to me that the best way to evaluate these programs and projects would be to assure that if the requirement is anything that is electronically transmitted, it has to be uniform.

I think that is what the HIPAA requires, isn't it, that since almost all these plans will be electronically reporting things, that we should try to build on that under whatever the requirements are that HIPAA is coming up to it with.

Does that make sense? I am not an expert in this at all, but there are a lot of people in HCFA who are working like dogs on it right now.

DR. IEZZONI: I understand there is a question behind me. Could you introduce yourself again?

MS. SMITH: I am Elise Smith from the American Health Bureau Association.

Bill, I just have one question on your comment about what these demonstrations might -- the population that they might try focusing on. You were talking in terms of QMBIs and SLIMBIs, but correct me if I am wrong. Isn't the New England demonstration, at least within the context of their stated framework, aiming at a soon to be at risk or a population that is not right at this moment at risk of institutionalization. Wouldn't that be an example?

DR. CLARK: I think it is a variation across the states, under approach -- well, I think, Massachusetts has that as part of their intended goal, but I am not sure about some of the others and some of the others, I just don't know about what, for example, Rhode Island's thoughts are about this.

DR. MOR: Rhode Island hasn't submitted it yet.

DR. CLARK: No, no. But then Maine, I think, does not -- is not reaching out particularly and Massachusetts is not -- I am not sure how well defined that might be.

DR. IEZZONI: Okay. Thank you, all the speakers. This has been a very informative morning.

What I would like to do now is have all of us break and go, I guess, upstairs is our only option, and buy lunch.

What we will do kind of while we are eating lunch is go over the draft of the charge. Okay? So, this will, hopefully, be low intellectual energy. I think, Gracie, that you have copies of the draft of the charge for the committee.

I would like to see if there is any way we could approve the charge and the work plan as outlined in this document, so we can get that off of our table for what the committee has to do and be able to report to Dr. Detmer that we are done with that. Okay?

And then we will start our discussion around 1 o'clock or so about -- we will come back here to eat.

[Whereupon, at 12:25 p.m., the meeting was recessed, to reconvene at 12:52 p.m., the same afternoon, Monday, July 21, 1997.]

[THIS TRANSCRIPT IS UNEDITED]

National Committee on Vital and Health StatisticsSUBCOMMITTEE ON POPULATION-SPECIFIC ISSUES

July 21, 1997Morning Session

Proceedings By:CASET Associates, Ltd.10201 Lee Highway, Suite 160Fairfax, Virginia 22030(703) 352-0091

TABLE OF CONTENTS

Morning Session

Afternoon Session

P R O C E E D I N G S [9:30 a.m.]

Agenda Item: Welcome and Introductions

Agenda Item: Overview of the Medicaid Managed Care Efforts, Description of the Medicaid Data Systems and Status of Encounter Data

Agenda Item: Monitoring and Evaluation Efforts for Medicaid Managed Care and Discussion of Issues and Recommendations

Agenda Item: Dual Eligibles: HCFA Initiative, Status and Discussion of Interrelationship of Medicare and Medicaid in Selected States

Afternoon Session

National Committee on Vital and Health Statistics
SUBCOMMITTEE ON POPULATION-SPECIFIC ISSUES

July 21, 1997
Morning Session

Proceedings By:
CASET Associates, Ltd.
10201 Lee Highway, Suite 160
Fairfax, Virginia 22030
(703) 352-0091