THIS TRANSCRIPT IS UNEDITED

DR. DETMER: I would like to call us to order. Welcome to all of you. It is actually a very nice, balmy February day. So, it is good to see all of you.

I am Don Detmer. I am professor of surgery and University Professor of Health Policy at the University of Virginia and chair of the National Committee on Vital and Health Statistics.

As is our custom, we will do is start by going around first the table and then the room and have people introduce themselves and then I will get into a review of the agenda and we will get to work.

MR. SCANLON: I am Jim Scanlon. I am here in HHS and I am the executive staff director of the National Committee.

[Introductions off microphone.]

DR. DETMER: Okay. Thank you very much.

We have a very busy couple of days ahead of us, as we typically do. We are particularly pleased to have some guests abroad to present in unfortunately too truncated a time frame, an enormous amount of interesting work that we see this really as an introduction to a lot of future opportunities.

Among the other items, we have a series of reports that we want to review and approve. As we typically do, we will talk about these today and then approve items tomorrow. We will have a panel discussion this afternoon. We will be talking about some other issues that we will want to move to approval on tomorrow, as well, we will discuss today.

Then we will break out for some work groups this afternoon. Tomorrow, again, some panels and then try to essentially reach approval on a variety of documents and then plan a 50th anniversary event for this committee before we adjourn.

Some of you may know -- I think all the committee certainly do, but this will be last committee meeting as chair. I have accepted a post at Cambridge University in England, as you say, Birmingham, England, starting in April. So, I, obviously, have some strong emotions today. This has been a really pretty amazing three years working with all of you and I think actually has represented really quite a unique experience for me.

I am very grateful to all of you for your hard work and interest and support.

With that, we typically begin with updates from the Department and there have been really a number of things to report on. So, I am going to turn to first Jim Scanlon and then Bill Braithwaite and then we will get into our panel from abroad.

So, Jim.

Agenda Item: Update from the Department

MR. SCANLON: Good morning, everyone.

The Executive Committee asked me to report on three specific items that have occurred since the last meeting of the full Committee and they all relate to data policy and I think they will be of interest to the Committee.

The three items are, first of all, a government-wide initiative on improving the measurement of discrimination. And I will talk about that in a minute. A second item relates to a provision that was included in the massive Omnibus Appropriations Act last fall for all of the federal agencies that will basically open up research, grant-supported research data through the FOIA process in general. I will talk about that as well.

Finally, I will talk about briefly a new policy that HCFA announced on its Web site -- Stewart can give us more details -- but relating to the use of the Internet for Privacy Act protected data and other personally identifiable information on the Internet.

So, with that, let me first of all start with the initiative on discrimination measurement.

Just a few months ago, OMB established an initiative that bills worked on by the President's Initiative on Race, which you heard about last year, and it is designed to document and further the nation's understanding of the role of discrimination in U.S. society.

The initiative falls under the auspices of OMB with the involvement of the National Council of Economic Advisors and its main purpose is to improve the current statistical capability and measurement capability to measure and track discrimination in five key sectors of the economy. Let me list those:in the criminal justice area, in the labor market area, in education, in health and in housing.

The goal is to expand existing knowledge on appropriate and credible ways to measure the presence of discrimination and to support empirical and technical studies that measure the presence of discrimination and to support empirical and technical studies that measure the scope of discrimination, using new and existing techniques and data.

Basically, this is meant to be an improvement in the methodology, a review of existing methodology and so on. It is not meant necessary to detect and enforce discrimination per se. HHS is responsible for the area of health care and in addition to HHS, there will be three other departments, as I said. Each one of those will be asked to look at the measures of discrimination in those areas as well.

Each of the departments involved will have to develop a work plan in terms of looking at the literature, looking at the current state of the measurement art and they will do this with the help of a research conference that is planned at the National Academy of Sciences in October. I believe that is the current planning.

So, the conference will basically set an overall theme. The individual departments will have to go back, obviously, and look at how they would approach this area from the point of view of the areas they are responsible for.

In HHS -- this is fairly new. This is just about a month old. So, our HHS Data Council has asked our Working Group on Race and Ethnicity Data to work on this issue within HHS. So, we are putting together a group and one of the first steps probably will be the review of the literature and a review of other unpublished work on how do we measure -- what is the statistical capability for measuring discrimination in the health area, at least for HHS.

Initially, the focus is on race and ethnicity discrimination potentially, but to some extent there is a somewhat open-ended nature so that it could later focus on Asian discrimination, sex discrimination, other possibilities as well.

So, I think the committee will want to follow this and provide some advice to HHS, obviously, as we move along.

Let me go to the next area and then we can take questions. This is an area, I think, many of you have already heard about and you probably heard from your colleagues about it. This is caused a lot of angst in the science community.

As I said last fall in the broad Omnibus Appropriations Act of the Federal Government, Congress directed OMB to amend the circular that governs grants to universities, hospitals and basically governs all research grants that Federal Government agencies award. They directed OMB to amend that circular, which is called Circular A-1-10 to extend FOIA, the Freedom of Information Act provisions, in a way that require federal awarding agencies -- and I am quoting here -- "to ensure that all data produced under a grant award will be made available to the public through the procedures established under FOIA.

FOIA, as you remember, is the law that allows anyone to request information that is in the possession or under the control of a federal agency. Previously, and from what I understand, data held by research grantees was not considered -- at least in HHS was not considered possessed by the federal agency and generally was not subject to FOIA.

This would make that kind of information subject to FOIA. There is another provision. If the agency obtaining the data does so solely at the request of a private party, the agency may authorize a reasonable user fee equaling the incremental cost of obtaining of data. So, there is a provision for a user fee associated with the request for this kind of data as well.

This requirement has caused a lot of concern in the research community. Those of you who look at Science magazine and so on have seen editorials and all sorts of concerns. I think that in general there is agreement that there should be procedures in place and an understanding of how research data is and should be shared.

There is great concern that too broad an approach would cause threat to the basic research enterprise and potentially some confidentiality issues and it would produce significant cost as well.

So, currently, this is already a law. OMB -- and OMB has been directed to implement the law. So, OMB is in the process of developing a notice of proposed rulemaking that would be issued shortly, I understand, and it would invite comment on this provision and how OMB hopes to implement it.

So, this would be an opportunity for the research committee and others to weigh in. Most likely, there would be a 60 day public comment period and, again, I think the committee here would want probably want to weigh in that one as well.

I might mention -- and I know several of you are experts in the FOIA process -- FOIA, the Freedom of Information Act -- has been in existence since the mid-sixties and it grants exemptions. In other words, generally identifiable information would not be released. There are certain exemptions that an agency can apply in terms of not releasing certain kinds of information, even when it is requested.

So, there are some protections already included in FOIA. I am not sure how much the science community understands about what exists already. Again, the Department has -- we have a working group to try to -- we have done some preliminary work and we probably will be developing a Department-wide approach to this, depending on how OMB comes forward with the language for implementing this.

Again, I think the concern is not so much the principle of sharing research data, but the idea that this may be a potentially blunt instrument and it may cause some downsides and I think there needs to probably be some fine tuning and I think OMB has clearly indicated they are interested in the research community's perspectives on this.

Let me finish up with the Internet policy and then we can take any questions.

This past November, our Health Care Financing Administration, which administers the Medicare program and works with the states on the Medicaid program, issued policy guidelines relating to the security and appropriate use of the Internet to transmit Privacy Act protected information, as well as other sensitive HCFA information.

The entire policy is on the HCFA Web site, WWW.HCFA.GOV, and I think we have got a copy for you at each of your places today. I am not going to say much more about this except that the policy states that -- well, first of all, it establishes the fundamental rules and system security requirements for the use of the Internet to transmit this kind of information and it states that it is permissible to use the Internet for transmission as long as an acceptable method of encryption is utilized to provide for the confidentiality and integrity of this data and to ensure that authentication or identification procedures are known to assure that both sides, sender and receiver, are known to each other and authorized to receive such information.

The policy provides detailed guidance and I think any organization desiring to use the Internet for this purpose has to notify HCFA with the formal notice of intent. But I think Stewart Streimer from HCFA actually was involved in this and Stewart can answer more questions.

MR. STREIMER: Yes. Let me just add to that that, first of all, again, it is only for HCFA data. Second, there are a number of implementation issues that we are working on right now. Very few organizations have approached us with any letters of intent, so to speak.

Right now, I think, primarily the use might be to exchange HCFA data with researchers at universities. We have a lot of work to do before we allow the Internet to be used for claims related transactions and right now we are exploring those implementation issues and I think that is quite a bit of ways down the road.

MR. SCANLON: Final note, our Data Council is, I think, in the final stages of the HHS data directory. This is basically an inventory of all of the data systems and analytic databases that HHS sponsors. We are about to put it up on our Data Council Web site and will probably publish it in hard copy as well for those people, like myself, who need to actually have a piece of paper to look at occasionally.

It includes virtually all of our HHS data systems, data collection systems. It includes detailed information about the race and ethnicity detail that the data provide and it also includes detailed information about how that information is available to the public and to researchers and so on.

Let me stop there.

DR. DETMER: Questions or comments?

Yes.

MR. GELLMAN: I just offer a comment or two on the FOIA thing. I gather that there has been some sense of panic in the streets or maybe in the labs. I don't know. I think the provision that passed is an incredibly blunt instrument and it is filled with questions. It is not clear how it is going to work.

I don't know what OMB is going to do with it. I suggest that anyone worth their salt will be able to delay disclosure of anything they want for about three years, no matter what the procedures are. So, I don't think this is that big a deal.

I do have, however, for anyone who is interested, I collected and compiled all the legislative materials, the committee reports, the floor language and the legislative language and I just snipped it all from documents on line and put them all in a document.

So, if anybody is interested in having that, there is not that much stuff there. If you send me an e-mail message, I will send you all the materials. It just may be useful to have it around if you are pursuing this.

DR. DETMER: Yes, John and then Lisa.

DR. LUMPKIN: As a remember reviewing it, Circular A-1-10 does not apply to grants to states?

MR. SCANLON: That is right. It doesn't apply to the assistance.

DR. IEZZONI: Jim, I was wondering whether the group that is looking into this for the government is thinking about changes that might be needed in informed consent procedures. You know, as people agree to participate in research studies, they are guaranteed a certain amount of anonymity and I just wondered whether informed consent language has been part of what you all were talking about.

MR. SCANLON: We have a group within HHS and it includes all the agencies. We are having NIH spearhead the effort. But the issue of potential confidentiality threats has come up.

Now, in general -- and Bob is an expert on FOIA -- in general, an agency doesn't release identifiable information with the FOIA request. That is one of the exemptions specifically. But I guess even generally the perception of that research information might be made available or even in an anonymous form, I think. You would probably have to include that in the informed consent material.

MR. GELLMAN: I actually doubt that. I mean, there is the protection for patient identifiable information in FOIA. It is very broad and very strong and very clear and I don't think there is any threat whatsoever under this that any patient identifiable information would be identified and, in point of fact, research in federal facilities has operated under the FOIA for 30 years and if anyone has got a problem, it hasn't come to the surface.

DR. MC DONALD: What about confidentiality for organizations? You know, if you try to get five hospitals to cooperate and give you information they wouldn't otherwise share, are they protected?

MR. GELLMAN: The answer to most questions is it all depends. There are nine exemptions under FOIA. A couple of them might apply, you know, and you actually have to work these things through, what kind of data, what are the rules and it is really hard to apply some of the FOIA principles to organizations that are not federal agencies. That is part of the heavy handedness of this approach. It is just a slapdash amendment, a one-liner, and it doesn't resolve any of these things.

Either OMB may address some of them or ultimately the courts, but potentially that kind of data could be accessible. There have to be, of course, federal funds --

DR. MC DONALD: They just won't play. I mean there are very risk adverse groups.

MR. SCANLON: Not only have federal agencies been subject to these provisions since 1966, but many of the state governments in their research universities have been subject as well. So, I think some of the states, in fact, I wouldn't doubt that Indiana, for example, the university, is probably subject to FOIA. But I am not sure it has been publicized or well-known. So, this may create a new --

MR. BLAIR: My question was going to go to Stewart Streimer's comments. So, it may be if you are still talking about the FOIA thing, my question can wait.

DR. DETMER: You are on. Proceed.

MR. BLAIR: Stewart, I apparently did not quite understand what the HCFA regulations were going to be in terms of sending information over the Internet with the initial announcements I saw in September and then again in December because you just indicated that sending claims over the Internet is not authorized. Could you please clarify that for me?

MR. STREIMER: Yes. November was when we issued this policy for data transmitted over the Internet. It is HCFA Privacy Act protected data and other sensitive information. We realized that while we have this policy out there, which has been responsive to a lot of industry requests -- and by the way, industry, as well as a number of the federal agencies have participated in developing the policy -- we knew that while we could make some of this data available over the Internet for like researchers much more easily than what we could do between the Medicare intermediaries and carriers and providers of health care.

We realize that the policy can span to that audience, but we realize there are a number of implementation issues associated with allow claims-related transactions to go over the Internet between Medicare providers and Medicare intermediaries and carriers. So, at this point in time there is no authorization for that to happen.

We are working on a pilot -- we are participating in a pilot, the Work Group for Electronic Data Interchanges is sponsoring with AFAC(?) -- AFAC is sponsoring a pilot to look at claims-related transactions over the Internet and until we have the results of that pilot, we will not be drafting instructions for the Internet's use for our intermediaries and carriers and health care providers.

Does that clarify?

MR. BLAIR: Yes. Can I prompt you -- you know, maybe you can't speculate, but do you anticipate that you might be looking for more than security measures like S-MIME(?) or -- gee, I forgot the other technique or encryption, where you would actually go to some form of authentication or electronic signature as well?

MR. STREIMER: Well, there are two requirements. Basically, you must have encryption. Okay? That is the first requirement. The second requirement is you must have authentication or identification. The authentication -- both requirements provide some -- in the policy some flexibility in terms of what is chosen.

For example, in authentication, digital certificates are one of the offerings that could be used. We do also have our HCFA's Senior System Security Officers, the point person on this, who might be able to help with any very technical questions that you might have --

MR. BLAIR: Okay.

MR. STREIMER: But I will point out again that there is some flexibility in terms of the choice of the encryption and authentication/identification options that are available.

By the way, the gentleman's name is Bill Pollak, P-o-l-l-a-k.

MR. BLAIR: Down at CDC at Atlanta.

MR. STREIMER: No, he is with HCFA.

MR. BLAIR: Oh, okay.

DR. DETMER: Kathleen.

MS. FRAWLEY: Yes. I just wanted to convey to Stewart from a lot of us that are on the committee and in the industry, our thanks that this has finally gotten out on the street because this certainly has been a long issue that we were struggling with.

So, it is nice to be able to thank the Department for something. So, this is a good first step. We appreciate the work of the staff at HCFA because I know they are very open to a lot of input from industry. I just wanted to convey that.

MR. STREIMER: Thank you.

DR. DETMER: Is that supposed to be registered as a thank you?

MS. FRAWLEY: Absolutely.

DR. DETMER: All right.

Bill Braithwaite.

DR. BRAITHWAITE: Well, as you know, we published four notices of proposed rulemaking last year to implement the HIPAA requirements for standards for electronic data interchange in health care; the transaction and code sets, NPRM, the provider identifier, the employer identifier and the security NPRM.

We received over 28,000 comments on those four and we are still in the process of reviewing those comments, analyzing them, grouping them and trying to figure out how to come up with answers that make sense for the industry and in the process we are trying to do the analysis in such a way that the drafts of answers to the comments form a draft of the final rule will be or at least part of it.

It will be coming out as soon as we can. The first rule I don't expect before, say, March or April of this year to be ready to go through the clearance process and God knows how long it will take to go through the clearance process, but certainly I wouldn't expect it to be published and on the street before July of this year and quite possibly later.

So, work is going on at a feverish pace on those four final rules.

In the meantime, we have other proposed rules that we are working on, the one for the plan identifier, which should have been published last year, but which had some problems and had to go back through another review and clearance process is continuing through that process. I am hoping that it will be published around the same time frame as the final rules for the others this year.

I am happy to report that the claims attachment standard NPRM has been drafted and is now entering the review and clearance process. So, that, too, is coming along. We were supposed to have that one published by February of 1999. So, we are not quite as far behind as we were on the others. I hope that that will get published in the same time frame, you know, mid this year.

Our first report of injury transaction is still on hold waiting for the industry to come up with their final consensus document that we can adopt, but I do expect that to be done sometime this year and we will get an NPRM out probably late this year.

The final one, the unique identifier for individuals, as you all know, when we held our hearings and got a somewhat negative reaction from the people who were concerned about privacy, given individual identifier. The administration asked that we hold on that and Congress passed an amendment to the budget bill saying that HHS won't spend any money in promulgating or adopting a final standard for unique identifiers for individuals.

So, that effort at the moment is on hold. The Department has not yet decided what activity, if any, will be made during this period before Congress comes up with a privacy law that would then allow us to proceed forward.

So, that is the current status of the standards rules.

DR. DETMER: Questions or comments for Bill?

Yes, Barbara.

DR. STARFIELD: Bill, I have a question not on what you presented but something else and that has to do with the race, ethnicity, social position data under HIPAA, the implications under HIPAA, what was the original thinking about standards for those kind of data and what was the thinking about where they would be obtained, in particular, enrollment encounter data and that kind of thing?

DR. BRAITHWAITE: Well, the general philosophy under HIPAA is that the Department will adopt industry standards and that the industry, including the government programs, like CDC and HCFA and so on that participate in those industry discussions from a government perspective, would reach consensus on what information is to be included in a transaction and exact specifications for each of the data fields and when and when not they should be used as a consensus process under one of the ANSI-accredited standards developing organizations; in this case, mostly X12 and HL7.

The HIPAA philosophy was not that the government would lay a heavy hand on this and specify things which were outside of the usual specifications required by various programs and that the programs, the people who participate in the programs would be participating in this industry consensus process.

As we have experienced, particularly with HCFA's participation, when a program requires data and the industry responds very clearly and cleanly to say okay, that is a response. HCFA requires this for Medicaid data. They are a major payer and player in this and the providers need to find a mechanism for providing that sort of information as part of the claims adjudication process.

That requirement, that business need flows quite quickly through the standards developing process and then that standards get adopted as part of a HIPAA standard that then everyone uses.

When it comes to race and ethnicity, there is a government requirement for a certain standard for collecting that that has come out of OMB and CDC has taken the point in presenting that business need to the standard-developing organizations and they are running it through the consensus process at the moment. There is nothing about the HIPAA adoption per se that would affect that.

DR. DETMER: Other comments or questions?

Just to make a comment, as you know, as chair of the Committee, I have been regularly attending, and if I can make it, someone else from the Committee, the Data Council meeting. Occasionally, I think what has come up are some of these other policy issues that do have implications for health information policy and I think that it has been a positive thing that we have worked so closely together with a number of staff in the Department and essentially have invited them to consider using us for hearings, where there are issues where there may be something that would be good to discuss that isn't just narrowly focused on HIPAAs per se, but certainly relate to vital and health statistics.

I think this pattern has really gotten established over the last couple of years since this legislation started playing in. It has really been a positive development. I think the Data Council is clearly finding its feet as well on this. So, it is just a comment that I think -- I hope we will see continue to develop as it has because a lot of these issues are pretty complicated.

There is one other departmental item that I want to give an update on before I go forward. And I would like to have Lynnette Araki please to come up here. This is a bit of a bittersweet piece of information actually.

I asked Marjorie Greenberg to help me pull together some information but it turns out that Lynnette is moving to a new position as a special assistant to the associate administrator for planning, evaluation and legislation at HRSA, Health Resources and Services Administration.

This very fortunate person will actually benefit as we feel the loss. But what I would like to do is read some of these comments and present her with a certificate and then ask your unanimous consent to put this into our minutes.

Lynnette joined the National Center for Health Statistics in 1987 and almost immediately began working with the National Committee on Vital and Health Statistics, eventually becoming the totally indispensable person that we feel her to be today. As early as 1988, she is listed as staff to the Subcommittee on Long Term Care Statistics. By 1989, she was staffing the Subcommittee on Medical Classification Systems as well.

Then in 1990, she became lead staff to both of those subcommittees and over the years wrote a number of important reports for each. She continued in these capacities until 1994, when a new lead staff was recruited for the Subcommittee on Disability and Long Term Care Statistics.

Marjorie tells me that the chairs of the two subcommittees virtually came to blows over which one would get to retain Lynnette as the lead staff. Meanwhile, in 1994, Lynnette also assumed the responsibility of team leader for the new NCVHS team at the NCHS, which continues to provide logistical and administrative support to the Committee.

With the reorganization of the Committee in 1996, Lynnette turned her considerable staffing skills to support of the executive secretary, the full Committee and the Executive Subcommittee. And I am here to personally attest really how valuable that work has been.

But Lynnette has always been generous in her willingness to help the subcommittee and work group chairs and lead staffs. An example is a significant effort she committed to assuring the success of the meeting this past summer with the Pacific Island insular areas, Puerto Rico and the U.S. Virgin Islands.

PARTICIPANT: She even cooked.

DR. DETMER: She even prepared delicious refreshments for both days to extend stateside hospitality to our guests.

In short, Lynnette's the consummate professional. For over a decade, she has used the respect, admiration and appreciation of the chairs of this committee and the membership, executive secretaries, executive staff director and staff. She keeps us all on our toes, keeps the trains running, makes us all look good and in the process looks the best herself.

We wish her all the best in this new position. I have a certificate. At any rate, this is the certificate. "The U.S. Department of Health and Human Services, National Committee on Vital and Health Statistics, a certificate of appreciation presented to Lynnette Araki for outstanding service, dedication and commitment to the National Committee on Vital and Health Statistics."

And I have signed this on your behalf. I would like a motion, please, to accept on this.

[Motion to approve.]

All in favor say "aye."

[There was a chorus of "ayes.]

Thank you very much.

Lynnette.

[Applause.]

MS. ARAKI: Thank you very much. I actually have really enjoyed working with all of you. It has been a wonderful experience and I think our paths are going to cross. It is not as though I am going to go so far away we are never going to run into each other. I am just entering another phase in my professional career, as is Don and as I guess all of us are during the course of our work life.

Thank you very much.

DR. DETMER: Thank you.

[Applause.]

Lynnette, I can assure you we will be staying in touch.

Agenda Item: Presentation from Work Group on National Health Information Infrastructure

I would like to move on to our discussion and presentation on national health information infrastructures and I would like to set the stage a little bit on this before I call on our distinguished colleagues from around the world.

If you look at the last page of Tab E, you see a draft charge for the Work Group on National Health Information Infrastructure and our four visiting colleagues. The U.S. has had an NII initiative, the National Information Initiative, but really the health component of that, I think, many of us feel is really quite inadequate.

It is not comfortable to admit that on the part of our nation, but I think that is the facts. To that end, this committee within the last number of months decided to make an effort to try to help in a public/private partnership to help create a bit of a vision and a strategy for our country in the hopes that actually it would catch the kind of fire that I think that we are witnessing actually already in Australia and Canada and the U.K.

So, the point is that what we have in front of us is actually a charge on not the concept paper itself, but we have created and approved a concept paper and this is a charge, a draft charge, for a Work Group on the National Health Information Infrastructure.

Basically, the areas that we identified in the concept paper related to a set of standards, computer-based health records, knowledge management and decision support, telemedicine and all of this under a umbrella of privacy, confidentiality and security as it relates to the management of this information with unique identifiers, as well as standards and population-based data.

So, essentially we sketched out things that I at least have seen in your models, but I must say not quite as eloquently and as comprehensively. So, one of the things that we will be doing is you will notice in this draft of the charge is Item 4, identify and analyze relevant models because we feel that we have frankly a great deal to learn and, hopefully, ultimately to share with one another because it isn't as though all of these things are totally unique to each of our countries and cultures. And to the extent that we could actually collaborate, I think ultimately the world will have better data as it relates to health and then subsequently one would hope better healthy.

So, in any event, this is a draft charge and I think it is reasonably straightforward and we will be coming back as a committee to address this charge and pass it, hopefully, tomorrow, but I just wanted to set the stage a little bit about that.

Our time is such this morning that I don't want to really spend much more time talking about what we have done to date because you have come an awfully long way to be with us so that we can hear what you have to say.

We had a subcommittee meeting yesterday on privacy and there was a gentleman in the group that his last name started with "Z" and we figured that the "Z's" always start last. So, I decided in a sense of fair play, I would actually ask Jennifer, our Canadian colleague, to start first and we will work back to then Laura Reece and -- Dr. Reece and Dr. Jones.

Let me just briefly introduce these colleagues and if you wish to say more about yourselves, please do, as you speak.

Jennifer Zelmer is senior consultant for knowledge development in the Canadian Institute for Health Information. She was previously involved in national standards and has worked actually in Canada, Australia and elsewhere. We are delighted she is here.

Dr. Laura Reece is currently information modeler for the Australian Institute of Health and Welfare in Canberra, the nation's capital. She is the former meta-data manager for the National Information Development Unit and Knowledge Base at the institute. She has a background in information management and systems modeling.

Dr. David Jones has spent the last past 20 years in health information, health infomatics. Prior to this, he spent 12 years in industry managing a medium-sized company. He is involved in policy and strategy development for a number of clinical professions. But his love is firstly analysis systems basically and then also gardening. You are welcome to tell us about your vegetables as well. Actually, there was quite an article today about Monsanto's genetic seeds in The Washington Post. I don't know if you saw that.

At any event, we are delighted to have you here. What we would like to do is have you, if you would, try to limit your initial comments to 15 to 20 minutes at most so we, hopefully, will have some chance to dialogue with one another.

As I told them before we began, this is really, I think, the first of I hope a lot of conversations. So, I will apologize that we didn't have a two-day meeting on just these presentations. But there will be more to come.

So, Jennifer, wonderful to have you here.

MS. ZELMER: Thank you very much.

I should thank you as well for recognizing the luck that we have had to suffer at being at the end all the day. And it is particularly good today -- it is especially good today because as I have read about what is happening in the U.K., what is happening in Australia, what is happening elsewhere, there are a lot of similarities in terms of what is going on around the world.

So, it is nice to be able to speak first so that it is not all new. In fact, I think for those private sector members in the audience, there may be a missed opportunity here in that we need some kind of an automatic turnaround thing that can do an automatic replace between Canada's plan and Australia's and the U.K.'s.

One of the things I was asked to talk about today was some of the unique things that are happening in Canada and one of them is that we got really tired of saying "Health Information Infrastructure" all the time. So, we have officially renamed it "Canada's Health Infostructure" and I am pleased to be here to talk to you about that today.

I am not sure quite how long this is going to remain unique because we had somebody from WHO, who came to Canada a little while ago and I think he is interested in using this word.

One of the other things that makes each of our health information infrastructures unique is that they are founded in our health systems. So, Canada's health system currently accounts for just over 9 percent of our GDP, mostly publicly funded and privately delivered, managed by the provinces and territories across the country.

But it is all within principles that are set, broad principles that are set within the national level, at the national level, around accessibility and portability of benefits and so on. Canada's health system, like our health information system, has been evolving over the last little while. Trends come and go. One of them is the aging of the population.

Another one is interest as in many other parts of the world in increasingly integrating our health systems. So, we are in a process of health reform, as are many other countries around the world. One of the key enablers for health reform is seen as health information.

We are making these big decisions. We are making big decisions about restructuring our health system to regionalize, to try and forecast future need, to try and look at major changes in how services are delivered, like the deinstitutionalization of mental health. And making these decisions without information is a bit like flying a jumbo jet without an instrument panel. It doesn't work very well. You may get where you want, but you don't want to hold your breath too hard to get you there.

But actually this isn't a new problem. In fact, many years ago, Florence Nightingale in some letters that she wrote, talked very eloquently about the value of information as a basis for making decisions about health and health care. So, building from what she said and what many other people have said over the years, there has been a lot of work in countries around the world.

So, today, we have made progress. We are further along than they were during the Crimean War. We have personal health records, most of them held by individual care providers at varying levels of automation and varying levels of comprehensiveness.

In Canada as well, we have standardized on some core minimum data sets in particular areas, primarily acute care, but also some others, some information that is pooled and shared for epidemiological and research purposes. We also have registers in areas like cancer. We are establishing a joint replacement registry and so on. That information is increasingly integrated with our population health surveys.

So, our latest national population health survey, one of the questions on it, as well as tell us about your risk factors, tell us about your health status and so on is will you give us consent to allow us to link the information that you provided through this survey to encounter data that we hold from our administrative records.

That is one of the areas in which we are moving forward. We also have stronger information about who is actually delivering care, health and human resource information and as you all know and as it is true around the world, the rapidly growing body of medical literature, evidence and multiple sources of consumer health information.

So, we have a lot, but it is not perfect. And what is wrong? Well, the data are incomplete. We know a lot about some things and absolutely nothing about some others. We know very little, for instance, in Canada about home care. We know very little -- I was interested to hear about you talking about discrimination and race and ethnicity. We know very, very little in Canada about the differences in health or health care based on race and ethnicity.

We have fragmented data. We have been collecting it in stovepipes. Different providers collect it and they group it for their purposes. As we want to look at a person and how that person uses health services, how their health changes over time, that is just not good enough. One of the problems in doing that is that it is fairly difficult to share data.

We don't always make the best use of the data that we have and the information that we need to make decisions is not always reaching the people who need it when they need it in the form that they can use it.

So, none of that is particularly new. What did we do about it? Well, about 18 months ago, the Federal Minister of Health established a National Advisory Council on Health Infostructure and one of the things that they did in conjunction with the organization that I work for, the Canadian Institute for Health Information and Statistics Canada, was we went across the country last year and talked to over 500 people from a variety of settings, government officials, non-government, hospital administrators, people on the street, as well, the private sector, health lobby groups about what kinds of information needs they had, trying to use this to drive our process as we move forward with our vision.

In some cases that has been done elsewhere. In some cases it hasn't. I like to think that that is one of the things that makes us a little bit different from some of the things that have been done elsewhere.

What did they tell us? Well, they told us about the importance of information that we need to know at a community level, particularly Canada's health system is regionalizing, as are many of the others, and that is a gap in our information.

They told us about the need for better information about cost and effectiveness of interventions, what works best, what doesn't and how we can make it better. They told us about the needs for outcomes information, for validated information for the public. That is not a new requirement. It is just it has reached a little bit more into the forefront because of the growth of the information over the Internet.

They also told us about the need not only information for consumers about their health but also about the performance of the health system. And to do all these things, they told us we needed common, consistent standards, both content and technical, that applied across the continuum of care.

We needed to be able to integrate our health information, our information about population health, about services, health expenditures and so on. We needed to do more comparative data and analysis, trying to identify best practices so we can use those as case studies to improve things. Better information about health and human resources and throughout all of this, a consistent framework for privacy and confidentiality.

But, of course, those things weren't out of the blue. There was a lot of work going on already and we are calling it the swirl in terms of what is going on. There are things happening internationally, like the G7 groups that are working in this area, things happening at a national level in Canada in our provinces and territories, locally and regionally.

I guess in the U.S. the context of the Canadian health system is a bit like having 13 HMOs, each of which have universal coverage in a defined geographic area, each of which have subcontracts with individual hospitals and care providers and it requires all of these people, just as it does in the U.S. So, we need partnerships with our provinces and territories. We need cooperation of our providers and our carriers and, of course, Canadians as well.

So, there is a lot going on. What are some of the common areas in which people are working? The first and foremost is a common framework for privacy, confidentiality and security. And there is currently a work group going on right now trying to harmonize our privacy legislation across the country. Health in Canada is constitutionally a provincial responsibility. So, some of the provinces have implemented privacy legislation. We want to try and make sure that works across the country.

There is also a variety of work going on in terms of security, PKI, those sorts of things. There has been a common recognition across the country of the need for consistent standards, standards not only for the kind of data that we want to exchange, but also the technical content.

We have been looking south of the border for the a lot of the expertise in that area, particularly because many of our hospitals, many of our care providers buy software from the U.S. and vice-versa. We also want to take better advantage of the physical infrastructure that exists, the physical networking infrastructure.

Canada is fortunate in that we have a strong telecommunications industry and we have a strong network that is in place. We want to use it for health, just like you want to do here. There has also been some interest in core software and applications, things like using telehealth technology to reach out to more remote and rural areas, things like using technology to try and help get evidence in the hands of care providers when they need it, processing through the mountains and stacks of information that exists.

And using all these different things to fill gaps in knowledge about how -- whether it is the gaps in your knowledge about your health or your providers knowledge about your health and the health services that you have received or whether it is operating at a community level or even a national level or International for that matter.

Where are we going from here? Well, based on what we heard and based on what is happening across the country, we have been working for the last few months on what we are calling a health information road map, a road map that will take us forward over the next few years.

One of the things that I was asked to comment on today is what is the purpose. Where do you want to go? And basically this is where we want to go. We want to modernize Canada's health information system so that it is respectful of privacy and secure, that we have consistent comparable information across the continuum, that it is produced in a way that is relevant to the people who need to use it at a time that they need it, that is integrable around a person, around a community, around an issue and that is flexible because we recognize that things are changing.

One of the things that has happened is that in areas that have built up their health information infrastructure in Canada at least early, some of the major care providers, for instance, they are now recognizing that, well, maybe it wasn't such a bright idea to establish these Fort Knoxes that don't talk to each other and so on.

So, we want a flexible system that is user friendly and accessible. Basically, at a national level at least, this is where we want to go. We want to be able to answer key questions, like how healthy are Canadians, what matters to their health, is their health improving and how healthy is our health system, what is the effectiveness like, what works, what doesn't work, what is the efficiency like? Are there ways that we can make it better? How is it responsive to the needs of Canadians? What are some of the accessibility issues, those kind of things.

I don't think I am giving too much away because it has been in the national press over the last little while, that the federal government is right now considering what to do in terms of moving forward and investments that are required. There is a federal budget coming up and there has been a fair bit of speculation about what will be in that federal budget in terms of health information.

Some of the areas in which I think we would all like to move forward and certainly the road map speaks to are continuing our consultative process, starting to work on filling some of the knowledge gaps that we have and working on enhancing our current data, being better able to integrate data -- we don't use the term "linkage" -- being better able to integrate data across sectors around an individual, improving our connectivity and our data exchange procedures and our analysis and dissemination.

Ultimately, starting from standards to develop consistent and comparable data that will move us forward as we start using that data into information and knowledge towards better population health and a better health system for all of us.

Thank you.

DR. DETMER: Thank you very much. I think what I will do is just let us hear from all three of our presenters and then we will just open this up because it looks like there will probably be some commonalities across all this.

So, thank you very much.

Dr. Reece. Laura, nice to have you here.

DR. REECE: I would like to thank the Committee for their invitation to come and speak on these areas and share some of the what we think are pretty exciting activities that have been going on in the area of health information management in Australia.

The first thing that I want to mention -- and I do want to echo Jennifer's sentiments about where we are all going for this. We all want better health outcomes for our populations. We all have information gaps. And what I will try and share with you today is how Australia has approached some of these issues.

This presentation is going to be grounded a little bit in some of the minutiae of how health information management is carried on in Australia. I hope it is not too detailed or too boring. If it is not the things you want to hear, I will be around for the rest of the day if you would like to hear some other aspect of it. I certainly can't do justice to any of the aspects of it in 20 minutes, as can either of my cohorts here. But feel free to approach me later on in the day.

The thing I want to mention, though, is that anything that I talk about today, whether it is the publications, the information agreements, some of our information models and so forth are all available for your perusal or downloading or ordering if you want a bound copy on the Internet at the Web site for the Australian Institute of Health and Welfare, which is the organization that I represent. That is WWW.AIHW.GOV.AU.

You can access any of our publications there. If you want to know about the actual information, you can access any of our publications there. If you want to know about the actual information agreement, information model, the data dictionaries and so forth, if you click on the triangular icon at that Web site, that is the national health information knowledge base, which is our on-line Medidata Registry, carries all the information available for you.

What I am going to try and do is cover Australia's health information infrastructure in what I call the four P's, four aspects of the information architecture: the plan, the players, who is involved, who is the audience, who contributes, the processes and the products. We have found all of these aspects of the information infrastructure to be critical to its success. So, I am going to give a small amount of time to each one of those.

The foundation of the framework of Australia's information architecture is the National Health Information Agreement. It was put together initially for five years in June of 1993. It has now been extended from 1998 to 2003. We have a robust infrastructure that involves people all the way from the health minister to local agencies.

The implementation and management, the processes by which we gather that information and revise and improve that information and feedback, I will give a brief introduction to and then I will give you some brief coverage of the National Health Information Model, the knowledge base, the National Health Data Dictionaries and some of the other products that have been derived from this process.

Don't be too intimidated by this stack of paper. I will get through it pretty quickly.

As I said, the cornerstone of this is the National Health Information Agreement and that is on-line, as I mentioned. I do have a copy of it here if anybody wants to have a closer look at it. Signatories to it are the commonwealth, the six states of Australia and two territories.

What that is is a commitment to support this process, to promote it, to provide for it and that is a significant of this process is a commitment to it. But it is a document only. It is supported by further documents. the National Health Information Development Plan provides a list of national health priorities in which work should go ahead and that is developed with the help of a couple of the players and input from a couple of the players that I will mention to you in a minute.

That is actually implemented by specifically identified project and those are identified in the National Health Information Work Program. Again, you can see details of all this on-line if you want to have a look. And it implements the work according to the National Health Information Agreement in priority areas.

Now, the aims of the agreement are multiple and I have got a few of them listed here. They are very broad. There are actually more specific ones included in the agreement itself. But basically they are the things that we are all about and I think are included in the vision statement that I have seen here.

They are to ensure collection, compilation and good interpretation of nationally relevant information, that there should be agreement on definitions, standards and rules of collection of information, agreement on guidelines for coordination of access to make this information available, interpretation and publication of national health information and to aid in improved access to uniform information by community groups, health professionals, academics, government and non-government organizations.

Some of the priority areas that have been identified for immediate activity and how these are implemented are specific projects in the work program, but the broad priority areas have been identified, at least in summary, are work with our indigenous peoples, aboriginal and Tory(?) Strait Islanders primarily, to improve their health and health service delivery, to develop a national health and welfare information model.

And I will make the point right now that although it is beyond the scope of this committee's consideration in all likelihood, wherever I am talking about an infrastructure in health, there is a parallel infrastructure for welfare community services. And we are always looking towards merging those or eliminating the overlap and that is one of the benefits of this approach because we all know how similar many of the items are and it is very arbitrary between countries, where the line is drawn between health and welfare.

So, that is one of the benefits of this approach is we can begin to identify that overlap and seek out some of the commonalities between those two areas. There is a parallel infrastructure for community services or welfare as well.

We are looking, as Canada is, at the ability to link health records that are provided from a variety of areas to identify the linkages that will result in the greatest community benefit. This is in process. What we are trying to do right now is look at the linkages, look at what is possible technically and in terms of privacy and then look at the outcomes and whether there actually a benefit from linking of these records.

To develop a plan to improve health outcomes information by developing clinically specific measures of outcome, everybody wants -- actually, the government wants to tie funding to outcomes, but we all want a measurable outcome that we can judge whether it is improved or not and whether it is really going in the direction that we want.

A couple more areas are to collect standardized information on the incidence, prevalence, consequences and outcomes of care of severe mental illness. Towards that line, we have been developing a minimum data set in the area of mental health, a minimum data set on primary and other non-institutional health care.

Our earliest data set, earliest data dictionary, started in 1989 with a minimum data set for institutional health care and the focus has been primarily on that. We have been expanding over the last decade to include a lot of other areas, especially non-institutional health care. We want to undertake a systematic review of major health data collections.

We publish a listing of the directory of national data collections in health, welfare and housing that is published on the site that I told you about. We are undertaking a review of the ones in all those areas, not just in health, in order to publish, not the data but publish the information about those collections and how all the data was defined and so forth for use by people using that data and to develop ongoing surveillance of potentially modifiable major disease risk factors and so forth.

We have had significant progress, I think, in all of those areas and I will talk about a few of the examples. This is just an example of what we have done with some of the data that has been obtained. We have specific data definitions for principle diagnosis or external cause. We have acquired that data by state and then it is for different principal diagnoses and then that has also been able to be combined in a fairly detailed manner for a national view.

That is the sort of thing that we would like to see in a lot more detail and for a lot more areas.

DR. STARFIELD: Excuse me. Is that institutional data, the primary diagnosis or not?

DR. REECE: All hospitals. Yes, it is all institutional data.

That is published in Australian Hospital Statistics and that publication is also available if you want to have a look at how that was undertaken.

That is the plan. It is the information agreement and the supporting development plan and work programs that underpin that.

Who are the players and processes in all this? The Australian Health Ministers Council is a direct link to the Health Minister, Dr. Michael Waldrid(?). It oversees the entire process and bears the ultimate responsibility for the directions that are chosen for these projects and these processes.

The National Health Information Management Group -- we deal heavily in acronyms, as do most governments, so -- and you will get very tired of seeing "National Health," but I can't get around it.

The Information Management Group is directly responsible to AHMC and it directs the implementation of the information agreement. It oversees direction of the information work program. It recommends these priority areas that I mentioned to your earlier, recommends those to AHMC based on what are perceived priority areas and interests in health.

It directly oversees the National Health Data Committee and oversees the National Health Data Dictionary and its development and publication. The National Health Data Committee is a group -- the representation on both these groups is signatory. It is a group of signatories to the National Health Information Agreement.

Two other signatories to the information agreement are the Australian Institute of Health and Welfare and the Australian Bureau of Statistics. They are the two primary statistical organizations in Australia that deal with vital and health statistics.

They also have representation on these committees as to non-government organizations, private hospital organizations and public health -- there is a public health partnership that has representation as well.

The National Health Data Committee is where a lot of the grunt work gets done, I guess. They review and endorse minimum data sets and specific data items that have been developed by expert working groups according to specific standards. They maintain and develop the National Health Data Dictionary, which was published since 1989, and the National Health Information Model, which is now in its second version. It promotes meta-data standards, which are standards about the data that we collect and I will talk a bit more about that later and it promotes sharing of information in the national health area.

Just a quick organizational chart to give you the lay of the land. Most of the grunt activity actually happens in the data committee. The National Health Information Management Group supervises that along with a lot of other activities, some ad hoc groups that they put together, depending on the activities that are happening.

All of that is overseen by the Australian Health Minister's Advisory Council and, of course, the Institute and the ABS participate as well.

The Information Management Group also has a New Zealand observer, I will just mention, in terms of international organization and a chair that is designated by AHMAC, the National Health Data Committee, again, has representatives from all the signatories, which is all the states and the commonwealth and the territories, the Private Hospital Association, the Department of Veterans Affairs, the National Center for Classification of Health, which is a collaborating unit of the institute and others as required, expert working groups that are involved in developing minimum data sets.

That is a quick view of the players and the processes. What are the products? Australian's health information products are grouped primarily in three areas. The National Health Information Model is an enterprise level model of what information we think we need to know.

First of all, you need to know what you need to know and you can readily identify gaps in the information that you have by mapping those individual data elements to a much higher level model like this. It is not a model that you can build a system on, but you can map lower level models to that enterprise level model and gradually get to a point where you can build a system.

But what we are interested in using right now is an enterprise level model to identify underpopulated areas in terms of data, areas that are changing and areas that we have a National Community Services Information Model we want to identify the overlap, so we are not duplicating effort in that area.

The National Health Data Dictionary is the vehicle for all national data definitions developed through the National Health Data Committee. We have done a lot of activity in making the National Health Data Dictionary ISO compliant in terms of meta-data standards. We have several representatives from the institute. Some of you have talked to them before, who are representatives on the International Committee for Standards for Data Development.

So, the Health Data Dictionary, which is now -- Version 8 is just being compiled. Version 7 is on the Internet now if you want to have a look. The National Health Information Knowledge Base is one area, which is our on-line Web site or meta-data registry, is the one product that has been getting a lot of interest recently.

It really started out just to be an electronic version of the National Health Data Dictionary, but we realized that it could be so much more than that. What the designers took the opportunity of doing is taking all of the health information products that the institute and Australia has and linking those on a Web site that you can go directly from the data dictionaries, which have individual relationships, they have specific permitted relationships with other data and data domains you can go directly from those to the health information model and see where those data items map to the model, where they are currently included.

You can see the definitions for parts of the model, which is actually the critical part of the model. I will show you a picture of the model in a minute. Don't get blown away by all the little boxes and things. It isn't really critical what the name box is. What is critical about it is how it is defined and examples that are given. This part of the model includes this. It is meant to include smoking indicators and not to include something else.

So, that is a critical aspect of definition of the model. You can link to any of the data items that were to be developed under specific projects in the work program. As I said, we have over a thousand data collections not on-line. We have the information on-line and as we get in now in the second, as we get in e-mail context or URLs for those data collections, again, not for the data collections themselves, but for the contacts or the responsible people, the data managers or data owners for those data collections, you can collect and go directly to those data collections, find out how to get access to that data, what is included in it and so forth.

Data agreements, the National Health Information Agreement is on the knowledge base. It is soon to include the community services information agreement and a source of terms included in all these definitions is soon to be on-line. It is not there right now.

I will put this up for a quick perusal. This is a draft version, too, and I can't get it all on here, of the National Health Information Model. It is not perfect. There will be a Version 3, but we are currently going through the first version, which I have got a copy here if you want to have a look, went on a tour of Australia basically, a wide consultation in Australia.

Version 2 has not yet done that, but we are preparing to do that sometime soon. This provides actually a graphic way to look at the information on the knowledge base, to look through the data dictionary.

Another aspect of the knowledge base that I didn't mention is it provides the ability to show data definitions or how data was defined, not just today but yesterday because we have still got data that was done in 1993 or 1989. This is especially important for groups that want to do longitudinal studies, want to know if that data is comparable. You have to know how it was defined. So, you can see the definition for anything; hospital waiting times, how that was defined in 1989, versus 1993 and 1998, see if there were any changes in that.

There are specific time lines about when that data definition was applicable to particular data and what part of -- what minimum data sets it might be involved in.

This is just a brief plug for a common template. I echo Dr. Detmer's sentiments that there is a lot of duplication in what we are all doing. Health can't be that differently defined in the U.S. from Australia, from the U.K., from Canada. So, although we might not all agree on specific data definitions, we can probably at least reach a commonality of a template of the information that describes a particular piece of data and then multiple sets of data.

If we are all collecting it according to the same kind of template -- and there is a lot of work being done by international standards organizations -- we are going to get a lot farther down the road about having data that we don't just share nationally -- that is our immediate interest in Australia, as it is here -- but ultimately we want to compare our health statistics with other OECD members, with the U.S., with the U.K., with Canada, and we can't do that if we don't know that the data is being collected the same.

So, it is just not in the national interest. It is in a global interest or international interest in getting better health statistics.

And I think that I will just leave it at that. I am sure I used plenty of time. And just say in conclusion that -- really echo what I just said that definitions -- we might not reach that in a very short time line agreement on definitions between countries, but if we can agree on a template to do that in terms of standards, we are a lot -- and even a health information model, some aspects of it -- we are a lot farther down the road to being able to get data that is comparable between countries, not just nationally.

Thank you.

DR. DETMER: Thank you. I think it is, obviously, to get around as much work as the country has been doing in the time and I think you did a wonderful job.

We will come back to that and move on at the moment to Dr. Jones.

DR. JONES: I don't know where to start really. I think Jennifer and Laura have looked at this information from two sides. I am going to look at it from the back, just a little different, I think.

I am an analyst by nature, by trade, by upbringing. I run a small team of data modelers, of analysts, of business analysts, and we work for the NHS Information Management Group, IMG, again. I always thought that was International Marxist Group, but -- and we work for the NHS in England, which is important. There is devolution in England, regionalization.

What I want to do is to talk about the background to the model that we have, the ways that it represents health care and how we perceive is the best way of going forward, the audience and the purposes, the benefit, the risks and eventually to talk about where the English Government is going with regard to information and its strategies.

In terms of the background it goes a long way. 1981, we looked to see is data important. I thought, yes, it is fine. That is really important, which is a step forward, but after about six years, we thought, well, that isn't the be all and end all. What on earth is the data for? That is when we started to think about a process view, an activity view, a clinical process view that data has and is necessary to improve the health.

So, we managed to convince the English Government that they ought to spend quite a lot of money, a few million, 7 million, over four years and to fund a lot of projects, which were looking at the detail of the health care process, that includes clinical, that includes administrative, that includes resource planning, service planning, planning organizations.

One project that we had, COSMOS(?), which maybe some of you have heard of, clinical process model, which is the first one we did in England, which used object oriented analytical techniques, well, we spent a lot of money and we abstracted from all those projects and we built what we called a common basic specification. That is a strange name.

We thought that all this originally was about building systems. We know better now. I will tell you about it. After the four or five years, the government said, well, have you got anything much? Is it worth it? Was it worth it? What can we learn from it? And we got 17 people, the great and the good from the clinical fraternity, from the executives of organizations, from the department, from the government, from academia, and they said actually it is not bad and what the government ought to do -- the government, the central organization should do as a minimum is support the building of the model.

That was a leap forward. It really was. It took us a long time to get that. That model should be freely available, a free good to everyone within the U.K. and internationally. We should not charge if there is something that we want to be used. And also, and I think this is where my looking at the elephant from the opposite end comes in, is exploring the ways of using that model to help development of policy and strategy.

What came out of that is that you get the wrong the name. So, we have now called our model the Health Care Model. Same model, really, but it has got a different name. And last year the government published an information strategy. I will talk about that a little bit later.

How do we represent the business of health care? We have three dimensions. We represent processes, the things that are done in the delivery of care. It is what is done. It is not who does it. It is not why it is -- well, in part, it is why -- it is not where it is done. It is not how it is done. It is what is done.

It depends on whose view you take as to what it looks like. So that what we have is a set of process views depending on whether you are a clinician, an administrator, a statistician and so forth. The reason for doing this is that users can understand that process view. They can relate to it. They can say, oh, yes, I do that. It is much easier to relate to that than to the data structures, which can be awfully complicated and very intimidating.

Because they understand it, then they can validate it. One of the greatest difficulties I have had in generating data models is in my heart of hearts believing that the users understand it and, therefore, it is right. I think that is a lot of con. It is so easy to convince someone that a model is right if they aren't real modelers.

If the users believe in that process model, then systems built on it will satisfy their requirements. This is what I do. This is what I want the system to do. Why should I worry about the data? I don't care about that. I want it to do the things that it needs to do for me.

The model has a very good class or data structure. It is in a built-in OMT. We use that not in place of but before UML came about and eventually we will move it to UML, but it costs a lot of money to move from one notation to another. So, I am leaving it as it is at the moment. It is generic. It is generalized. It is not specific. It doesn't say person anywhere.

It is about a thing we call a subject and it is about information you want to know about that subject or subject type and the activities that are there in order to maintain it.

From a patient care point of view, that subject will be for your patient. From an management point of view, the target will be an organization. You want to know the characteristics of the organization and to move it forward by undertaking particular activities and so on.

DR. LUMPKIN: Could you just tell us what OMT is since you are building on --

DR. JONES: Sorry. Yes. It is a notation for representing data, object modeling technique. I can't remember. It is the Booch(?) mechanism.

DR. DETMER: Dr. Jones, since he has interrupted just a moment, we actually need to interrupt just a second if we could.

Dr. John Eisenberg has just joined us and he is the chair of the Data Council that this group reports to and also administrator for the Agency for Health Care Policy and Research. Busy man and came in now. If you don't mind -- because it is a wonderful presentation, if we could just interrupt for just a moment and let John have the floor.

DR. EISENBERG: I am tempted to just stay and listen because I know I would learn something and I have plenty to learn.

I was just over in Alexandria where the group that is organizing the Quality Forum is meeting and the discussion of what we do with data to try to improve health care quality was a big topic of discussion. I left so that I could come over here because what I would like to do is to join you all in thanking Don Detmer for a terrific as the chair of the NCVHS.

I come on behalf of the Secretary and Peggy Hamburg, who is my co-chair on the Data Council. And I have this certificate. As you know, Don, we are not allowed to offer gold watches, Mercedes.

I want to thank Don. I want to read a letter from Secretary Shalala, who you all know is an old friend of Don's. They worked together at the University of Wisconsin and when she thought about who ought to chair the NCVHS, she thought about the differences between academics and government and realized that anybody who could make it through the University of Wisconsin alive could probably handle the National Committee for Vital and Health Statistics.

She often says, in fact, that anybody who wonders how she manages to deal with the United States Senate has never had to deal with the University of Wisconsin Faculty Senate. But Don having known her for a long time was someone who was a known entity and someone that the Secretary knew that she could count upon, someone who could really move us in the area of data and information.

As you all know, this is a person who has dedicated his entire life to better information for improving health and this group, you, are a major mechanism in which we get advice and counsel and ideas about ways in which government can do that.

One of my favorite quotes, though, is from T. S. Eliot, who wrote in a poem in the mid-1930s, which was called "The Rock." He had a line in which he said, "Where is the wisdom that we lost in knowledge? Where is the knowledge that we lost in information?"

Don is one of those people who understands that information is very important but information alone isn't enough. What we count on you to help us to do and we have counted on Don so much to help us to do is to find out what information is important and necessary but also how we translate that information into knowledge and then how we translate that knowledge into wisdom that is going to be in the interest of the American people.

Secretary Shalala and all of us in the Department are so grateful to Don for helping us to move this process along and for providing leadership for you. So, let me read this letter from Donna Shalala, which says -- it says,

"Dear Dr. Detmer,"

but then she crossed it out and said,

"Dear Don, I want to express my gratitude and appreciation for your outstanding as chairman of the National Committee on Vital and Health Statistics for the past two and a half years. During your tenure as chairman, the Committee has made major contributions to the advancement of health information and privacy policy.

"The Committee is one of the oldest and most prestigious advisory groups serving the Department. Its recommendations have helped to shape health statistics and data policy for our nation for decades. When I expanded the charter of the Committee three years ago and I asked you to serve as its chair, it was my hope that the Committee would become a broad-based national health information policy advisory body to the Department, as well as a bridge to the industry, to research and to public health communities.

"I am extremely pleased to say that under your leadership, the Committee has more than met my expectations. Your leadership and advice have been invaluable to the Department in addressing the requirements and the challenges of the Health Insurance Portability and Accountability Act for national data standards and privacy, in improving our population-based data systems and in promoting a vision for the national health information infrastructure that ties all of these elements together.

"I know that I speak for the Department when I say that we are proud to have had the opportunity to associate with you in this endeavor. Should the occasion arise, and we would like to feel free to call upon you for further assistance, we will.

"We wish you the very best as you move on to assume the Dennis Gillings(?) Professorship in Health Management in the Judge Institute of Management Studies at Cambridge University. Sincerely, Donna E. Shalala."

There is a certificate suitable for framing that is included with this, Don.

I just want to close by saying that, Don, for me, as for many of you, is an old friend, a colleague and one whom I have respected for a long time and one from whom I have learned a lot, both substance and ways of getting that substance implemented, a person who is really truly committed to getting better information, better data for the American people so that health care be improved.

What the Brits don't know is that when they were looking around for a new don for Cambridge that they found him. So, for our newest Cambridge don, thank you very, very much, Don. We appreciate it.

[Applause.]

DR. DETMER: This also happens to be my birthday of all things. At any rate, I am a bit speechless. That was very gracious. It also, as you can tell -- friends say nice things about friends and that is also very nice, but to the extent that we have succeeded, it is a very nice comment because, obviously, that is what I think all of us wanted to see happen over these last few years, really see the Committee's mandate and role and vision expand and to the extent that I played a role, it is my privilege, but I think absolutely the work that the Committee has done and the staff have done really is what is responsible.

But at any rate, that was awfully nice and I will cherish it. Thank you very much.

Now, let's move back to our good colleague, whom I actually hope to work with in the near future --

DR. JONES: Actually, it is also my birthday today. Surprise. Surprise. The Aquarians get it.

I will just skip this. That isn't very interesting. That is about the model that we have got and, in fact, I have got some CD ROMs with a model on if someone wants to take it away. Unfortunately, our Web site isn't quite up and running. In a couple of weeks, you will be able to get hold of it quite easily. But it is an HTML. It is the full detail of the model.

What I would like to go is go over the audiences and the purposes of the model and, in fact, the different aspects of the model have different uses, I think.

You are going to do it to me again.

DR. EISENBERG: I am going to interrupt you again. I forgot something in thanking Don, I got halfway down the stairs and realized I didn't remember to do something else, which is also very important and that is to say that we have asked John Lumpkin if he would serve as the acting chair of the Committee when Don leaves so that gavel will be passed down. I apologize for not having said that.

DR. DETMER: Let me congratulate John Lumpkin. I am absolutely delighted. Obviously, I have known John for many years. He is a friend and I think all of us have really admired his expertise and leadership of the Subcommittee on Standards and Security and, obviously, John, I really feel totally comfortable and confident that the Committee is in excellent hands. So, that is good news. Thank you. Congratulations.

Now, back to the birthday boy.

DR. JONES: Okay. The uses and the audiences, the process model, certainly from the clinical point of view, it specifies requirements. It is easy to grapple with from the furnishings point of view. That is, in our experience. It also supports the clinician's ability to reengineer what they do, to look at -- because what we have is a model of what needs to be done, then you can look at how it is done and where it is done to be more effective.

Importantly, the model is independent of technology and, therefore, you need to reassess and use the model to assess the current technologies. It helps the policy makers to appreciate the potential standards. That is all nice words. What I want to do is to explain that in a bit more detail by looking at part of our model.

From the clinical point of view, a clinician takes the history of the patient, including the medication history, the family history and all sorts of things. As part of that, there needs to be an appreciation of the guide. What questions need to be asked? Where does that come from? Is it in the mind of the clinician, who was trained 30 years ago?

Is it in a knowledge base? Where does it come from? I don't know. You may need to ask the question. How is it structured? Where do you populate it from? How do you validate it? Is it local? Is it national? A lot of questions.

Maybe you need to know the current properties of the patient. Where did it come from again? Is it distributed by different hospitals or different community settings? What is the structure of it? What, in fact, the person who is taking the history, what are they allowed to see of the patient record or the patient information? Where did it come from?

Classifications and terms. Barbara was talking about tell me about Reed(?), which I can do off-line if I may. We call it just clinical terms by the way now. We change the names to protect the innocent. What are they? How are they structured?

Once you have done that, then maybe you can ask the patient a few questions and get some answers back and store the information. But where do you put it? So, what this is is a way of asking questions, which lead to the desire, lead to, I suppose, another question, which says what standards do we need in order to allow a clinician effectively to take the history of the patient.

So, that would allow you to appreciate the potential standards. We aren't into data here. We are into -- at the moment anyway, we are just looking at the process of care. It will help to determine what is local and what is national. From a national infrastructure point of view, which is where I sit within my organization in England is the process model allows you to have a road map. It tells you where the different standards meet of the care process or, in fact, the administrative process or the management process.

Of course, you have an audit trail then that can justify why a standard exists. In the past, certainly in the U.K., we have developed standards mainly because someone fancied it, it was easy, for various reasons, but there was never always a nice, clean trail back to the clinical process or the management process.

You can use it to manage the projects, which you are developing national standards. It will help you to identify where standards touch and where they meet and where they need to integrate. It will help you to target research effectively and to understand the -- to assess the risk of data or not having it.

Let me go back to that, again, and say what is the risk to the patient if you don't have the all the current patient details, patient properties? I don't know the answer to the question. Is it life threatening? Do you mean it is quite likely to have to redo that clinical activity again? I don't know. But having a model like this allows you to make those judgments and, in fact, having talked to quite a senior transplant surgeon in England, he says, hell, I don't need the information. I will transplant the liver anyway. That is going a little bit too far, but I think there are risks to the patient if you lack information. And you can use this mechanism in order to try and judge that.

We have within the U.K. a data accreditation mechanism, which says mainly for institutions, mainly for hospitals, that says is the process where by you collect the data or is it bad and if it is bad, then you need to improve it. This has been a voluntary exercise so far. It will be mandated in a year's time and the hospitals will have to pay for it. It will probably cost about 3,000 pounds sterling each, which is good. I don't mean the cost is good. I mean, the fact that it is accredited.

It helps the system designers. Well, we know all that, but mainly to look at the integration. If you look at the -- if you use the process model as a road map, then the suppliers, like they do within COBAMED(?), they can look to encapsulate the systems, which is quite useful.

Our class model allows the data designers to make sure that the data structures are consistent and to assess the impact of new standards that are coming along. That is not very exciting. That is useful, the last one, that our model is -- the model that we have is independent of the organization structure, which is something that changes every five years when the politicians change in the U.K. It is independent of the technology.

The real benefits, the road map, that is what it gives you. It gives you an audit trail. You can justify the standards that you have developed. It supports different views. It supports the clinical view, the manager's view, the resource manager's view. It doesn't confuse the two. And that has been one of the problems in the U.K. up to date that we do have a set of data standards.

We don't know who owns each element of the data. We don't know who is responsible for maintaining the format or the size or the structure of it. It creates partnerships and that is one of great themes that we have that Tony Blair is instituting within information strategy that I will talk about in a moment.

This partnership, we are bringing the suppliers much more into the forum. We are bringing the technicians and the users all together and the process model allows that discussion to take place. It is not at the detail of the data.

Risks, if you don't have it, potentially inconsistent standards. We have got that in the U.K. We have got a couple of standards, maybe even more that are not totally consistent and it is going to cost us time and money in order to make them thus.

We have in the past developed standards -- "willy-nilly" is not quite the right word but we have developed the ones that aren't necessarily the most useful. Well, I suppose that is the same point really.

What is important, we don't have much resource. I mean, in terms of a modeling team of analysts, of business analysts, there are four of us. That doesn't go very far. That really doesn't and we are just one part of the information management group of about 400 strong, 350, 400 strong.

There are risks of having the model. If you are not careful, it will become a straight jacket. It will tighten. It will inhibit suppliers to implement individual ideas if you are not careful. Maybe if you are not careful it will be not responsive to change. It will not respond as the needs change.

Certainly we might have difficulty getting political and user endorsements. There is a difference between those users who can help you to validate the model and those that you need in order to gain political or user fraternity acceptance. They are not the same animal. You need to appreciate that and take appropriate steps.

That is a bit of a devil as well, competing models. We have a lot of models. We have -- I don't know how many there are; maybe up to a dozen. Competing models covering the same area and what a damn shame. I have no desire to be the maintainer of the U.K. model. What I want to do is to use a model that supports our requirements. I don't care who does it, who maintains it. We should pool our knowledge and our resources.

Where are we now? Where are we now, indeed? An information strategy for the modern -- yes, yes, yes. I think it is seven years because it took three years for the government to get around to doing it, to ten year -- they are hoping to be reelected in the year 2000, provided the bug doesn't screw them up.

I think that there is a commitment, there really is a commitment; 5 billion pounds modernization fund to improve the NHS, of which a billion is for IT. Okay. It is over that seven year period, but in our terms, I mean, we are only a little island. Good heavens. And that is a lot of money. That is a few bob.

And I think that it has been -- it is likely to be curtailed a little bit because we used to give the nurses and the doctors a few more -- an increase. But having said that, it is the philosophy and I will just give you two quotes from it. IT should be regarded as an overhead in the sense that it should top sliced. It shouldn't be up to each individual organization to do it. It should be top sliced.

And you can't afford not to do it. That is powerful stuff coming from the British Government and in terms of implementing the strategy, this is a quote from sort of -- this is a bit lower down from the top very strategic bit -- and it says you have got to have a data model. You have got to have one and you need a process model to understand what is going on.

That is what I have done. That is what I have talked about, in two or three lumps, bits. Thank you very much, indeed.

DR. DETMER: You had your share of interruptions.

DR. JONES: You can get hold of me on the e-mail. Unfortunately, as I just said, our Web site isn't up for a couple of weeks. We are having problems mainly because it has now decided we all ought to be consistent, which means slowing all the -- anyway. You have been there.

But what I was going to say, what I do have is a copy of new information strategies. Those exist really. You can get it on the U.K. Government's Web site and I can't remember --

DR. DETMER: We have actually referenced in our policy -- or in our vision statement.

DR. JONES: Saves me. Saves me.

But I think to conclude and to reinforce what both Jennifer and Laura have said, let us get together. Let us pool the resources and the skills that we have to get something that is of benefit to the whole of you.

I rest my case. Thank you very much.

DR. DETMER: Okay. I want to thank, obviously, each of you for coming and presenting and before, I guess, I open this up, I might call actually on Mary Jo. Do you want to make any comments? You have been helping staff -- why don't you come on up?

DR. DEERING: I want first and foremost to thank them for, at least two of them, on very short notice for coming to join us today; secondly, to alert people that we have captured them for a 3:30 session this afternoon, but more substantively then. I think that what is interesting is we got a sense of breadth from Jennifer. We got a sense of detail from the U.K. and Australia and I think the challenge for the work group will be to take what you have given us and see how we can map that to our goals and efforts. So, we are really looking forward to the dialogue this afternoon.

DR. DETMER: We have 15 minutes right now and let's go ahead and open this up. I am sure there are a lot of questions and comments.

John and then Barbara.

DR. LUMPKIN: I would like to thank all of you for a wonderful presentations, very thought provoking and challenging for the task that we have ahead of us.

I just had a quick question for Dr. Jones and maybe this is how we use terminology between here and the other place across the ditch, the big one, not the channel. And that is that you describe your model as a health care model. We frequently here differentiate between health, which incorporates treatment and what is called curative and the preventive area and use that as one model, where health care was just the curative component.

Is that how that term is used?

DR. DETMER: If you want, why don't you just go ahead and sit back here and then we can just dialogue among the three of us.

DR. JONES: Within the U.K., the health care spans the hospital, community and in part social care. It covers all that and when I use the word "health care," it is not the clinical aspects of it either. It is the management of it. It is the resourcing of it. So, it spans probably a broader scope -- has a broader scope than your use of the term.

DR. DETMER: Barbara.

DR. STARFIELD: I think it is correct to say that the United States was early on -- was very early into the development of health statistics and still is a model to the rest of the world in a variety of things, like the National Health Survey, and even in our hospital data, I think we are pretty far ahead.

But what dogs us is our fragmentation and we have very fragmented data and that reflects our fragmented health system. I think your health systems do a little better than our health system.

DR. JONES: Don't assume that.

DR. STARFIELD: You still have some way to go. But that brings me to the question and the question is we are struggling not only with data standards and that is the thing that is occupying us -- been occupying us a lot for a couple of years, but the whole issue of linkages across different kinds of data, the continuum of care, the social issue of terminal care and the unique identifier.

Are there any lessons you can give us as to how we ought to think about that in the context of an information infrastructure?

DR. DETMER: And I would really want all of you to weigh in on that, please.

You want to start, Jennifer?

MS. ZELMER: I guess so. The zeds get to go first again.

I think one of the things that has been very successful for us over the last little while is using our National Population Health Survey as a mechanism for requesting consent around data linkage. So, in several provinces now, the respondents to our National Population Health Survey, not the longitudinal survey, there are responses, which include risk factors, like smoking and so on -- they are used in health services and health status measures -- have been linked to all the physicians, any hospital care that they have received, long term care and a whole range of other interventions.

Because we got consent up front, we can do that and the rate at which people gave consent was very, very high. So, that is a very good resource for us now. There have been some pilot projects as well, looking at linking our census data with some interventions data, but that hasn't gone as far because we don't yet ask for consent as part of the census process.

We do have a unique identifier within each of our jurisdictions, I hasten to point out.

DR. STARFIELD: That means within the process?

MS. ZELMER: That is right.

DR. STARFIELD: I am sorry. If someone moves a province then --

MS. ZELMER: We are working on the national question.

DR. DETMER: Laura.

DR. REECE: Record linkage, I think, we are still very much in the investigative stage. We have projects going ahead to look at what record linkage might be advantageous, but we are still waiting on the outcomes of those and it will be an ongoing process, but I think it is something that needs to be done to make use of data that is already being collected or that we already have and not just consign it to the past.

I think what really triggered my interest was your mentioning the continuum of health care and I think that comes back to what I mentioned about the area of what we call community services, which is where in some cases health care is delivered, what is basically health care and not just drawing a line and saying this is health and that is something else, but being both outside the traditional definitions for health care.

And also it gets back to data standards, too. If you want to look at the continuum of health care, you need to know that the data that is being gathered at a very large city private hospital, the same data item, is going to be defined the same as small regional hospitals or, you know, we have a wide variety in health care delivery in Australia from the Outback to cities, to hospitals. So, there is health care delivery difference in different places and data standardization can go a long ways, too, overcoming those differences.

DR. JONES: Yes, we have been some there. We have an NHS number, which was mooted about five or six years ago. We went through the consultation process and it has been implemented and now has been -- well, it is supposed to be throughout the U.K. at the moment, but it isn't -- there have been one or two pitfalls. Do talk to the guys in that project if you are thinking of doing it because they have been down a few holes, political and other.

Yes, do talk to us about -- when you decide you want it. The other thing is within the strategy, within the last strategy, it identifies two types of record, what it calls the EPR, the electronic patient record, which is the one that is held by a hospital or by a general practitioner in a community practice and EHR, the electronic health record, which is a longitudinal health record of the patient, irrespective of where care is given.

It is the development of that latter, which I think gives you your linkages. The target within the strategy is I think 2002. So, we have got three years yet to set up a number of beacon sites, a number of prototypical environments within the U.K. to test the validity of the approach and whether, in fact, you can achieve it practically. It is a nice idea, but how practical is it and what mechanisms can you use to safeguard the data and confidentiality and all those sort of things.

So, there is a lot of interest and there is a lot of stimulus towards that link record and I think we will get there, but it will take a little while.

DR. DETMER: Other questions or comments?

Jeff.

MR. BLAIR: Could I follow up on that just a little bit?

We have slightly different health care models -- I am sorry. I am directing my question or comment to Dr. Jones. We have pretty much left to the private sector in the United States the funding and the decisions to implement information systems technology. We have tended to wind up having our public health sector wind up encouraging standards and we have private associations to do that.

In England, you have been more proactive. You have actually gone ahead and developed what used to be the Reed Classification System from Clinically Specific Medical Terminology and you call it now the NHS one. But you have set targets for implementation of electronic patient records and electronic health records.

One of the areas we struggle with in the United States, some institutions have moved forward to do that on their own here in the United States -- in some cases it is faith. We have an issue where we would like to be able to get some handle on the business case or a return on investment. Do you have any data that you derived in England, which backed up your drive or commitment that would be of value to us in supporting your decision to be proactive with electronic patient records and health records?

DR. JONES: Wow! I think at this stage it is very tentative. I think one of the reasons for looking at the pilot site or beacon sites, as they are called, is to try and get a handle on its value. We don't know, to be honest. I mean, we believe it has a significant value but it is not backed up by hard money.

So, I don't know. We do have a slightly different perspective on primary care; that is, care from general practitioners in that there is a funding mechanism for the computer systems and that is a significant driver. It does help. However, whilst it helps them to acquire computer systems, it doesn't necessarily help them to use them.

My local practitioner, he has quite a nice system, cost him a few bob, but he uses it primarily to generate repeat prescriptions because that is easy. And he has a large envelope with all my details in it by the side of the ut(?). So, one of the things that we have to do is not to move forward in technology but to move the culture of the environment forward and we should be very careful not to let our technological aspirations drive us too far down the road before actually -- it is like security. One needs to get the post-its with the password on off the screens.

You know, I mean, that is the sort of level that some of us are at at the moment. So, whilst it is very nice having political discussions about what and where, until you get to grips with the practicalities, then you might be sort of flying in the wind.

DR. DETMER: Do either of you want to comment on that as well before I move to the one?

DR. REECE: I know that there have been several -- what was the term you used -- beacon locations in Australia, as well, that have gone about implementing the electronic patient record. I certainly can't speak to the results that they found. I know that they are very enthusiastic about where that has been established and implemented.

Part of the problem is that it ends up being proprietary from location to location. Everybody has got their system to promote and what we want to get away from or get around and somehow -- people develop their systems, so we all get the same data and the same information to the EPR.

Australia has recognized the problem of proliferation, not just in full hospital systems, but general practitioners or primary care provider level and they have put forth a project that first of all will establish a general practice or primary care data model so that they can get a handle on what data set needs to be acquired, what information GPs need administratively, what information they need to provide health care, a wide variety of things.

That information then could be used by anybody to build a system that you can put on the GP's desk. Beyond that, going down the road, not to buy them a system but to pay them to collect the data and that, we think will make a big step. It is not a huge payment but when they know that it is going to be worth their time to sit there and make sure that that data gets entered and that it is entered in a standard mechanism down the road -- it certainly hasn't happened now, but down the road, hopefully, we will have a better set of data that is collected at the desk top of the primary care physician that will be useful.

DR. DETMER: I have Simon, Richard, Clem and Vince and then we will take a break and then we will see you some more at 3:30.

So, Simon.

DR. COHN: Actually, Laura, you were beginning to go where my question is sort of headed, which has to do with the issue of data quality.

First of all, let me say I am a very strong supporter of all the work that you are doing. Many years ago, when I took on a leadership position in my HMO, one of my early acts was to establish and sponsor our enterprise model and then our data dictionary, which is sort of our data model for the organization.

Over the years as we have tried to implement it, I see it is critical however but not sufficient in the sense that at the end of the day really what we are trying to do is to, I think, as Jennifer commented, move from good quality comparable data to information, to knowledge and maybe to some wisdom.

Obviously, I am sort of curious and from all of your views -- I guess, perhaps a small question to get started -- and maybe we will pursue it in the afternoon -- the issues of trying to get that good quality data to fill in that data model. I mean, one question, of course, has to do with just getting people to fill it in. The other piece has to do with what impact reimbursement models in all of your various environments may have on that ability.

I would certainly judge in the United States that reimbursement seems to somehow -- you may not get data but at least you get them to fill in the information. What can you tell us about that? And I would ask all three of you.

MS. ZELMER: Well, I guess, data quality is certainly an issue that we are trying to do some work on right now. We have some of the same issues. Some of our data is collected as a by product of payment mechanisms. For instance, that is how we get our physician data and that has its own hazards and risks, as well as benefits.

I mean, you do tend to get it because if you don't get it, they don't get paid. So, it does tend to be fairly comprehensive.

On the other side of things, we have this huge gap where we don't pay physicians, the non-fee-for-service physicians. We really don't have a clue what they do right now and we are working to fill that.

In terms of data quality, one of the things we are trying to do -- and this is fairly new -- is start working between our different data sets, looking at areas where one data set is compared to another and using that as a quality check. So, cross validation is selected.

DR. REECE: I think what we found is critical to this process of improving data quality is what I said at the beginning. It is a commitment by the major stakeholders to this process and for us that National Health Information Agreement. These states and territories commit to this, to collecting -- and this is all national data that I have been talking about. State-based data is something else and I will talk about how we are addressing that in a minute.

But basically the government sort of has the hammer. If they don't submit this data, you know, they don't get their funding. So, that is not necessarily a hammer that you have to wield here, nor is it necessarily a way that you may want to go about this. You can lead people that direction, lead stakeholders that direction.

One major way you do that is to see that they get something for what they do. What is the reason that they are taking the extra time or having their people take the extra time to input this data. Down the road, they need to get data back that they can use to make policy decisions.

So, part of it is closing that loop and making sure that your major stakeholders see something for their effort, see something for their money. The other thing that we are doing is -- again, this is just a state-based or national-based data.

There is a large amount of data that we haven't identified as of national importance that may be state or local interest. What we are allowing and encouraging and inviting people to do or agencies to do, which is all that the institute can do, is put the data standards for those locally-based definitions on the knowledge base and then when people -- any other agency sees that, they say, oh, they have already developed a definition for such and such. I don't have to sit down. I don't have to look at developing a minimum data set for whatever area for indigenous health. Somebody else has already done that.

I can use those. I can link to that. I can make those my definitions. So, you get a reduction in the amount of effort that groups have to go to and you get a fill in at the bottom of other data. So, it is not just national high quality data because you want all the data at all levels to be of improving quality.

MS. ZELMER: Can I maybe just jump in for one second. One of our big successes over the last little while is last year for the first time we published in Canada a version of the Times magazine, a national health report. It was quite an extensive report and it is amazing what that visibility had to do to make people pay attention to the data that they were putting in.

DR. DETMER: Interesting. That is maybe a very important message actually.

DR. JONES: I am just trying to find a stratagem in here that says just do it properly and I can't find it. It is in there, honest.

Let me take you back in terms of the U.K. to 1983. There was a lady by the name of Dame Edith Kerner and she was commissioned by the NHS to look at data collection in the health service. As a consequence of that, she came -- she advised the health service to collect particular data items, said go away and you can use those to manage your business, not necessarily take care of the patient, but to manage the business. That was the drive.

The problem was whilst she identified data, she didn't identify why, what the reason for that data was. So, even now, even though -- from 1987, that was a mandatory set of data that had to be collected or was collected centrally from all aspects of the U.K. The quality of it is questionable in some areas because those areas have not use for it. They themselves don't see a virtue in collecting it.

So, I would drive -- I would suggest that any collection of data is driven from a need -- any standards should be driven from a need from the user community. If it isn't, then your data quality may well be suspect.

DR. DETMER: Richard.

DR. HARDING: That was what I was going to say. I will pass.

DR. DETMER: Clem.

DR. MC DONALD: I had questions or requests for all.

You said you said you had it on the CD, the data model. Is that possible to borrow or steal?

Also, I really liked seeing your draft 2 model because it is one of the first that didn't look like a circuit diagram and it actually was readable because you nested it. That is not one of the standard mechanisms. Can you comment on that? Is that available, the draft 2?

DR. REECE: I have a publication that documents the first version of the National Health Information Model in Australia and that one did to some extend look like a circuit diagram. One of the major feedbacks that we got was -- we spent a lot of time training people on information modeling and as a result, you know -- not just at the institute but all over Australia at different levels and different areas and that has paid benefit in that they know a lot more about what we are doing.

So, I will just throw the thing in that education about this is very important as well but the lines are a problem for anybody. So, we took away the lines and what we found was that that didn't hamper its utility. In fact, it enhanced it because what we discovered is these models are very context dependent. You try to make these models as generic as you can to cover as large an area as you can.

Unless you control every aspect of that content, every line becomes a many to many. So, what is the point. Throw it out. When you need to look at a specific relationship, then you can get down in an area where you control or have intimate knowledge of the relationship and can limit it so that it is useful knowledge, then you can get back into that.

But that is one of the major changes that we made and we found that the feedback has been tremendous. We can read this. We can understand this. It still probably has way too many boxes and it is difficult on a two-dimensional thing. It is much easier electronically, so, I would encourage you to have a look at the knowledge base because there you start with the highest level of the nest and drill down and you are only confronted with the immediately related items or items on the same level or on the same part of the tree.

You are not overwhelmed with boxes and --

DR. MC DONALD: But you do have a hard copy here? It is awfully hard to make a hard copy draft that looks really nice.

Dr. Zelmer, I wanted to ask about -- Canada is really doing some -- I like your draft starting with standards because that is kind of what one of our subcommittees is all about and I think it really goes in that order because until we get it standardized, we can't get to those other steps.

But Canada is really quite advanced, at least by province in a lot of standardization activities and then some research projects or management projects, I think -- Ontario is planning on collecting all lab data in one master file and I think British Columbia already has all prescriptions that makes at least health researchers drool, but is there an effort underway to sort of blend those across the provinces with some of those?

MS. ZELMER: There is, indeed. We have a group called the partnership for health infomatics and telematics, private/public sector partnership. One of the things that they did was they started with the health information framework. We have spinning balls, not little boxes and circuit diagrams. But fundamentally, we are driving down from that to the National Health Data Model, but we are basing most of our data definitions on actually the HL7 data model because it was easier for everyone to agree to map to somebody else's data model.

Also, because it makes sense and because that is what many of the systems are built on and that is where we hope to be able to influence vendors. So, there is, indeed, an effort to map those data models.

DR. REECE: If I can just add one other thing, I don't mean to intimate that the National Health Information Model has no relationships. The relationships are still maintained at the level where the data elements map to the model. There is that hierarchical relationship, but also the data elements have very specific permissive relationships, permissive and exclusive relationships to each other, is calculated using or is derived from or whatever.

So, there is a layer of complexity that wasn't evident on that.

DR. JONES: We do have a spider diagram. Our model, I think, looked -- the whole is complicated. The view that we have taken, I think, is that generally the users perceive our model through the process model, not through the data structures. Because of that, it has made our job infinitely easier. It is like the person who drives the car doesn't need to know what thread the hood was bolted down. It is horses for courses and we need to understand the relationship between the car design and the car manufacturer, the car user and, in fact, the road maker.

Once we have done that, then I think standards will become a lot easier.

DR. DETMER: Vince.

DR. MOR: I am going to switch gears here.

In each of your countries there are very different sort of relationships between the public and private sector and Jeff alluded to this a bit in terms of his question about what is the incentive for people to make the investment in hardware, software, if a government is going to mandate something and make it uniform.

My question is a little bit different and comes from the unique U.S. perspective, which is highly proprietary structures and systems of data. Who owns the health data in your countries and what kind of discussion is there about data ownership and who can do what with what kind of data. I think that would be at least very helpful to me.

MS. ZELMER: That is a very good question in Canada. Usually the person who the data is about owns the data. However, it is held in the custodianship of whoever has written it down, for instance, the physician, and they own the record. That is the legal position. It gets mucky real fast.

In general, because of reimbursement mechanisms and other funding mechanisms, there are sometimes requirements to report things. Increasingly, there are voluntary agreements to report and share information so that it is not necessarily mandated by the government, but people have come together and decided this is a good thing to do. We are going to get together and do it.

So, increasingly, that is the way things are happening rather than an actual mandate.

DR. MOR: But are there prohibitions against what somebody who might use these data for clinical and potentially clinical/commercial purposes?

MS. ZELMER: In most provinces there are Freedom of Information and Protection of Privacy legislations in place, but we are trying to harmonize those across the country. There is also a brand new -- the Canadian Medical Association has a brand new privacy code for physicians that is very strong.

DR. DETMER: Australia.

DR. REECE: I can't speak to the specifics of who owns the patient record. I think it is the patient, in fact, in Australia because I know my home is filled with x-rays. I wasn't accustomed to having to take home my x-rays because that wasn't the case when I lived here. But the data -- it is my impression that the data is owned by the signatories to the Health Information Agreement. Most of this is laid out in the Health Information Agreement and I think that is a very important aspect.

There are segments in there that deal with privacy and ethics and data owners are different from data custodians, but it is -- each data owner, I think, establishes under the purview of the privacy act, which we also have, it is very, very stringent, who can have access to that data. That is also becoming part of the meta-data that we are starting to include about different data items at the level of data collection is who should have access to that data.

DR. JONES: I don't know the answer to the question unfortunately. I don't know who actually owns the data. I know clinicians have responsibility for it, for maintaining it and restricting access and view, but in terms of ownership, I have no idea, to be honest.

DR. MOR: In the U.K. there is a study in which about 5 or 7 percent of the GPs participate to put together the same long linked research data files for looking at medical practice patterns. I don't know whether those data are available to researchers or to companies that might want to develop medical management practice protocols or those kinds of things, which is what we do here in the States.

DR. JONES: I think anonymized data is used widely. It is where you can identify the individual's problem and, of course, some research needs -- whilst it needs to have the data anonymized, it may need to go back a few years later and pick the individuals in order to see the progress of that particular individual. I am not sure legally how that happens.

But can I say that there are changes within the structure of health care within the U.K., such that groups of general practitioners get together are getting together and forming what is called a primary care group and that has a responsibility for the patients within an environment and the people involved in that can see the data records of particular people within that limit. So, it is sort of broadening out sort of the responsibility, but may come up against the confidentiality wall. I don't know. Maybe not.

DR. DETMER: I want to break this at this point, but I am very grateful to each of you. I think this has been really terrific and we will pick up our conversation this afternoon.

Why don't we take a 13 minute break.

[Brief recess.]

DR. DETMER: Okay. What we need to do, you have gotten mailings of our reports and what we wanted to do is discuss those and while we essentially have three sets of discussion items here, one, these reports that we have going out and then a couple of letters that actually are coming from a couple of the subcommittees.

Agenda Item: Discussion of 1996-1998 NCVHS Report

So, the first is the 1996-1998 report and I will just open the floor for questions or comments. Do you want to make any before we get into this, Marjorie?

Tab I has the draft outline, but the actual report you should have gotten as a separate mailing.

MS. GREENBERG: It was sent with your agenda book. Susan, why don't you come to the table. This is our wonderful historian, reporter and Jill of all trades. Susan Kanaan put this report together and I want to thank her for that because it was a great deal of work and it also is really a new format for the Committee, much more an integrated -- in the philosophy of what we have been talking about of data integration and linkages and connections. It is a much more integrative approach than we have used in the past where we have dealt with the work of each individual subcommittee. But it does capture the subcommittee work.

Also, it isn't going to include nearly the detail the appendices, et cetera, because of our Web site, which also now makes it much easier to refer people to that. There are several pieces of it. There is Don's forward and then we actually included the NHII paper as kind of a special piece in the report and then, of course, a little bit of history and actually starts to take us to what we will be talking about tomorrow afternoon, which is the 50th anniversary observation or celebration.

DR. DETMER: Blowout.

MS. GREENBERG: Blowout, right. You can say that. You are leaving.

I think she did a splendid job at it. At the same time, she really needs your comments and critique. I know some people have already sent her some comments and we really do want to bring this to closure very soon to finalize and, hopefully, get at least tentative approval here and then we can incorporate comments.

Is there anything you wanted to say, Susan?

MS. KANAAN: Only to thank the half or dozen or so people, who have already given me very good comments. They have helped to move the process along, as well as a great deal of help from Marjorie.

DR. DETMER: This will be printed up as well as put on the Web site?

MS. GREENBERG: Oh, yes, we will do a real report from it, yes.

DR. DETMER: Okay. All right. The floor is open. As I say, we will act on it tomorrow, but we need --

MR. GELLMAN: I am going to make some comments that are directed at both of these reports. I actually had a chance to look at these reports because I basically find this kind of activity --

DR. DETMER: You mean the second annual report?

MR. GELLMAN: Yes. I don't find them worth much time. I don't think they get much attention or are really worth discussion. But it kind of fell open to the discussion of the patient identifier issue, which is one on which we have had some discussions before.

DR. DETMER: Which document are you --

MR. GELLMAN: Both of them. They both sort of have somewhat different but similar discussions and I wasn't very happy with what was in there. I think that there is a lot of history to this issue. I think this committee has made a series of substantive procedural and political errors, which have backfired. Some of us are happier about that than others.

I don't find that the discussion in either report fairly reflects what went on and I would like to read a short statement.

The discussion in the report, and this applies to both of them, about the unique health identifier for individuals offers a misleading impression about the activities of the Committee. The text suggests that the Committee was primarily concerned about the privacy implications of the identifier. It is true that the Committee's September 1997 recommendation stated that it would be premature to select and implement an identifier in the absence of legislation to assure the confidentiality of individually identifiable health information and to preserve an individual's right to privacy.

However, what the Committee really did in its 1997 recommendation was to decide that a unique patient identifier was a desirable goal. The Committee expressly voted on this specific issue and it insisted on affirming support for an identifier, despite opposition from some Committee members, who argued that any decision on the patient identifier issue was unwise and precipitous.

The Committee adopted its recommendation for a unique patient identifier in advance of public hearings and in the absence of any formal analysis of the costs or benefits of a patient identifier. Indeed, in its zeal, the Committee even voted to proceed with hearings on the issue before the Department's promised white paper was to be publicly available.

The consequences of the Committee's rush to judgment are now apparent. The hearing on the patient identifier sparked a national wave of opposition. It resulted in a promise from the Vice President to slow down consideration of the identifier issue. It also prompted Congress to enact a moratorium on the administrative adoption of a patient identifier.

The Committee's misguided and hasty actions on the patient identifier backfired in the end. A fair description in the report should make it clear that the Committee's primary interest was in having a patient identifier and that its concern about privacy was secondary. In addition, the report should take note of the sharp dissent of some Committee members.

It is my intention to submit this statement as dissenting views to both reports.

DR. DETMER: Okay. A couple of things I want to make sure I heard right.

You mentioned actually at the start of your comment that, in fact, the Committee was on record essentially saying the position of where things did come out, that, in fact, an identifier was not to be recommended until the legislation were passed.

MR. GELLMAN: Yes, but the Committee expressly -- we had a vote on this -- I actually went back and read the transcript -- to delete the recommendation for an identifier and the Committee insisted on supporting an identifier and I think that is a point worth making.

DR. DETMER: Okay. But I just wanted to make sure I also heard that the other point that was there.

Correct me if I am not mistaken, but actually the Committee was on record in the past for the social security number as the national identifier or --

MS. GREENBERG: Let me clarify that.

DR. DETMER: This thing has a history that certainly is more than this report.

MS. GREENBERG: The various core data sets, the uniform hospital discharge data set, the uniform ambulatory care data set recommended or recognized the value of a unique identifier for individuals and couched that language always with the concerns for privacy and confidentiality. But there was a recognition of the value in those reports and the core data elements report and recommendations that came out in August of 1996 extended the recommendations from those two data sets and stated that the -- actually, I do have it with me. I would hate to misquote it.

It was in lieu of a better alternative, the social security number was being recommended, but it was definitely equivocal in the sense of recognizing that there were problems with a unique -- with the social security number as a unique identifier. It says here, the personal unique identifier is the element that is the most critical element to be collected uniformly.

The NCVHS recommends the use of social security number with a check item, such as date of birth, while at the same time undertaking the study and evaluation needed to confirm this use or the recommendation of another identifier. More emphasis on the confidential use of SSN is essential. Standards groups should be consulted regarding setting criteria for recording of names.

DR. DETMER: All right. I guess actually the Privacy Subcommittee, as well as the standards and security have, you know, worked on this as well. I think the question that is in a sense before us do you have thoughts on how we might relate this. I guess we could take this section out of the report. We can add his letter to the report.

We could try to rewrite those sections with the subcommittee heads seeking to find some language that is satisfactory to the entire group or what are people thinking? What do you recommend?

Kathy.

MS. COLTIN: Well, of the options that you mentioned, I think I would prefer to see the latter option. I don't think that it should be dropped. I think this is clearly an area where the Committee invested and the subcommittee invested a good amount of time and effort and to leave it out of the report doesn't make sense to me.

I think to the extent that we can make certain that the description of those activities is accurate in everyone's opinion, I think that would be the way to go. I think we do need to check on the accuracy because there are some parts of what Bob read in his statement that I am not in agreement with; for instance, that the Committee never took testimony on that issue, that, in fact, going back before Bob became a member of the Committee, when we were developing the core data elements, we actually took a good deal of testimony around that particular issue.

I would agree that we do in doing so represent the needs of all of the stakeholders and all of the constituencies as well as we could have because it was a different process in terms of recommending the core data elements at that time. But it is not that we didn't hear from the field that there was a need for an identifier because we certainly did hear that.

DR. DETMER: Other comments? Kathleen.

MS. FRAWLEY: Just, you know, kind of reacting to your question, I certainly agree with Kathy in terms of some of her comments. My recommendation would be if possibly John Lumpkin and I and Bob can see if we can kind -- to work on the language because I agree with Kathy, I think this needs to be in the document and there is a history here in terms of where the Committee has gone -- and see if we can at least make some modifications to what we have got. Failing that, then, obviously, Bob can, you know, attach his statement.

But I think that we can at least try it without taking the entire Committee's time here to try and resolve it. I would offer that as a solution.

DR. DETMER: Okay. People happy with that as a -- okay.

DR. LUMPKIN: Just following up on that procedure, if we could do that tonight because the Standards Committee is meeting tomorrow morning to go over the report. Then we can further refine it at that meeting.

DR. DETMER: Okay. Other issues or -- yes, Barbara.

DR. STARFIELD: This is, I guess, a small issue but -- and I haven't read the whole report. So, I may be taking things out of context, but I looked at this table -- the pages aren't numbered -- the table on computer-based health records, there is nothing there about the sociodemographic accuracy of individuals and even our discussion about that. Maybe we ought to have that -- they don't have patient numbers. I know.

DR. DETMER: It is in the concept paper.

MS. GREENBERG: This, of course, is the paper that was sent to the Department.

DR. DETMER: It is already a matter of record in any event.

DR. STARFIELD: I don't know whether we can put something in the text or -- I guess we can't really amend the topic, but maybe somewhere in here --

DR. DETMER: Well, it was representative. It really wasn't trying to be exhaustive. But I hear your point.

Others -- we can talk to Mary Jo --

DR. MC DONALD: It would be helpful to have page numbers.

DR. DETMER: Right.

DR. MC DONALD: In the next draft.

DR. DETMER: I understand.

Other issues?

MR. BLAIR: Marjorie, the statement that you read, was that a statement of position on the identifier that occurred before HIPAA was passed or after?

MS. GREENBERG: Yes.

MR. BLAIR: Before HIPAA was passed.

MS. GREENBERG: I have always found it sort of interesting that the core data elements report was delivered to the Department the same day that the President signed HIPAA. Actually, I like coincidences. But, yes, I mean, that was --

MR. BLAIR: So, it really -- if it was before HIPAA, then it isn't in response to the HIPAA legislation that we investigate and decide on that.

DR. DETMER: No, I didn't assert that. I just made the point that this committee has discussed this issue in the past and has, in fact, been on record regarding it.

MR. BLAIR: So, that was August of 1996 or something like that.

MS. GREENBERG: Yes. I mean, the report was under -- in the making probably, you know, late 1994 to 1996, had a number of hearings, et cetera, that it resulted in.

DR. DETMER: Bob's comment focused on both reports. Let's move on to the second annual report document and see if there are other -- because the same issue will be dealt with with both reports. But are there other items on it?

DR. LUMPKIN: I had two comments. One, which may fit here and also in the annual report because evidently I didn't get it into my document, my meeting notes. So, I will look through that tonight.

That has to do with the time table for HIPAA and I think it is important to note in here that were a significant number of comments that we received about Y2K and that puts into context the time table in which the Department is promulgating the HIPAA regulations, which are not consistent with the Act, but I think are very consistent with the request of the involved public that they not coincide with the beginning of the year 2000.

So, if the rules are promulgated sometime this summer, that will give people at least 18 months after the Y2K deadline to implement the HIPAA regulations. And I think we should just note that that is fortuitous.

The second issue was on privacy and there is just the use of one word on page 25, which is that the NCVHS continues to believe the United States is in the midst of a health privacy crisis. And I am just a little bit hesitant of calling "wolf," and I am not positive, but are we absolutely certain that it is a crisis as much as it is a significant problem? So, I am just kind of raising that question.

MS. GREENBERG: I think that is direct language from the report that was --

DR. DETMER: Maybe circumstances have changed and we can talk about it certainly. I don't know how quantitative it is anyway, as far as --

DR. LUMPKIN: We said it. It is just that there are certain words that you use too much and then people don't believe they exist.

DR. DETMER: Are we in crisis about "crisis" or are we --

All right. Others related to the report?

MS. GREENBERG: I had actually sent an e-mail to Jim about this, too, and it is not a big thing, but I was going to recommend that we have at least a few sentences in this second annual report about the work that the committee has supported related to the workshop on the implications of HIPAA for public health and health services research because there is mention of public health and the committee was one of the -- did support the Centers for Disease Control and AHCPR in putting on that workshop. I can provide some language on that if nobody objects.

DR. DETMER: Yes, that sounds good.

Simon.

DR. COHN: Actually I had probably an issue similar to what John was raising on page 27 under "Security," and I unfortunately didn't have time to go back to our original letter, but I found myself underlining the second sentence, which said that "Security practices revealed an extraordinary lack of protection within and across health care organizations today." I don't remember what we --

MS. FRAWLEY: It is not in the letter.

DR. COHN: It is not in the letter. We identified there was an inconsistent level of protection and I thought "extraordinary" might be a word like "crisis" that might be -- just might be the wrong word for that. I just wanted to bring it to the Committee's attention.

DR. DETMER: How about the folks that were at that hearing?

MS. FRAWLEY: I was the author of the letter and I know the letter did not say that. So, of course -- I don't have it with it. Somebody must have a --

MS. GREENBERG: Well, it is on the Web site.

MS. FRAWLEY: That is not what it says.

MS. GREENBERG: It is on the Web site. We can just take the language from that letter.

DR. DETMER: We will check that because essentially what we are doing, John, on the other issue.

Okay. Others?

Obviously, before we move on then to the Iezzoni-Colton letter, a tremendous amount of work represented, obviously, in this document. I don't think any of us who have read it weren't certainly aware of the really substantial amount of effort that went into it. I want to pause for just a moment because I think the total collective effort that this represents is really quite impressive, both the past years, as well as the three years.

So, thank you.

Dr. Iezzoni, do you want to lead us into this next item?

Agenda Item: Consideration of Principles for Draft Purchasing Specifications Related to Health Data for Medicaid Managed Care Contracts

DR. IEZZONI: Patrice -- is Patrice here? Have we had a chance to hand out the draft Medicaid contract language from G.W.? Marjorie, we were going to try to get that to the Committee for today.

MS. GREENBERG: I may have fallen down on that in that I attended about four meetings yesterday and I may -- unless -- although I thought --

DR. IEZZONI: I think it is a little hard to talk about something that we don't have copies of. You will be getting a copy of the following. As part of our year and a half initiative on Medicaid managed care, one of the things that we heard was that states didn't know how to contract with MCOs about data and data reporting issues.

So, one of the things that we did from the subcommittee with George Washington and Sara Rosenbaum was actually ask her if she would work to draft potential contract language that states could actually use with managed care organizations around data and data reporting specifications.

This is simply draft language and states can take or leave it. It is meant to be helpful. What you will hopefully be getting is a draft of that draft language that right now we are going to be asking for your comments on, not today or tomorrow but over the next couple of weeks. If you could get back your comments by e-mail to Carolyn Rimes, who is the lead staff for our subcommittee and her e-mail address is easily rememberable. It is CRIMES@HCFA.GOV, C-R-I-M-E-S at HCFA.GOV.

But Kathy Colton was the subcommittee's participant in that and we are very grateful to her for having agreed to do this. It was a very arduous process. Sara Rosenbaum basically hosted conference calls with people from HCFA, from SAMHSA, from HRSA, from the CDC and Kathy participated in that as well. So, Kathy is going to kind of introduce to us what the basic principles were that led to development of this draft contract language.

The process to conclude this and report it to the full Committee in June will be as follows, that after we get your comments, which we hopefully will in the next couple of weeks, Sara Rosenbaum will take it out to vet it to a number of organizations and individuals, who we will suggest to her that we would like to hear from. The language will be finalized, hopefully, at a meeting that our subcommittee will have in May and then the final draft will be presented for your approval at the full Committee meeting in June.

So, that is the plan. So, Kathy, do you want to just walk us through the principles that guided development of this?

MS. COLTIN: Okay. What the subcommittee is looking for is feedback from other members of the Committee around the recommendations or provisions that are in the purchasing specifications, but also keeping in mind whether or not those provisions as they are written adhere to some principles that were agreed to around what was reasonable for a state Medicaid agency to ask for from a managed care organization.

We have endeavored to try to strike a balance there. I am not sure how well we have done and I think this is probably a document that deserves fairly broad vetting within the industry as well.

First of all, just to be clear what this is and what it isn't, it is a set of information and data purchasing specifications. So, it would be part of a broader contract that a state Medicaid agency would have with a managed care organization that would deal with a lot of other aspects of their relationship and the delivery of services to Medicaid recipients.

These provisions really have to do solely with how the managed care organization provides data to the state Medicaid agency and what is expected of the managed care organization and how it uses that data within its own organization to help manage the care that is delivered to the beneficiaries.

It is a set of specifications that these state Medicaid departments can draw from in developing their contract and the phrase "draw from" is key. It is not intended to be used in its entirety and, in fact, it contains some provisions covering the same data issues that are presented as alternatives. You could do it this way or you could do it that way.

So, the agency would have to choose one or the other. It is a menu of provisions among which the states may choose to best reflect their concerns and the nature of their relationship with the managed care organizations. While most of the provisions that are in here focus on data about Medicaid members of MCOs and the services they received, there are some provisions that focus on data about the managed care organization itself and about its provider network as well.

The set is intended to help meet the data needs not only of the state Medicaid agency, but also the needs of the state health departments and that is also key. You will see a lot of provisions in here that speak to types of data about notifiable diseases or laboratory results that are covered under state statutes, where there is an attempt here to put some of what exists in state law into the responsibilities of the managed care organizations around helping the health departments to gain access to that kind of information.

One of the principles that was used in developing these provisions was to try to avoid making the managed care organizations responsible for all of the inadequacies of state laws regarding public health data reporting, just because they represented the only organized systems of care delivery in the United States, while still holding them responsible for supporting those laws and for helping to compensate for contributing to some of the data problems that may arise due to the gaps in the state laws.

An example that was brought up is that there are requirements of clinical laboratories within a state to report results for certain specific tests to the state public health department. But if a resident of that state has that test performed by an out-of-state laboratory, there may not be a requirement for that laboratory to report the information to the state.

If the HMO or MCO -- I will get the alphabet soup right here -- is primarily responsible for causing many more of these tests to be done out of state because of their contractual relationships, then the principle says they own some responsibility for helping the state overcome that problem in some way.

So, that is one of the principles that was used. An attempt was made to try to recognize that the managed care organization is not always the appropriate accountable unit for reporting this type of information, that in the case of a notifiable disease, for instance, that accountability rests with the physician and not necessarily to try to make the managed care organization the police of the behavior of the individual providers in complying with state law.

Another principle was to recognize that managed care organizations are health care delivery organizations and not health data organizations and to propose reasonable limits on what data that they must provide within a capitation payment, a payment that is set based on the cost of delivering health services, not data reporting services.

For example, if the MCO provides complete encounter data to the state and these provisions do suggest that that be one of the requirements, then, in fact, under the BBA, that really has to occur, it would be reasonable to expect the MCO to then use its own version of those encounter data for internal management reporting and to support quality measurement improvement, as well as to submit established industry standard reports, such as the HEATUS(?) measures to the state.

However, they would not under these kinds of provisions necessarily be asked to develop and provide customized ad hoc reports that the state could be produce themselves from the encounter data that the MCO supplied to them without some additional financial compensation. So, the way the provisions are worded, they would identify those instances around what would be normal expected deliverables versus what might be negotiated for some fee.

Another principle was to try to recognize that decisions made by states to carve out certain services from their managed care contracts can impact the MCOs ability to manage both the utilization and quality. So, for example, if the state decides not to include pharmacy services as part of their contract and, in fact, in order to manage the care of certain categories of patients where pharmaceutical treatment is really key to managing both their utilization and their outcome, the managed care organization is really quite hampered in its ability to do that, not having that sort of information.

So, to put a provision in there that requires them to do something but not give them the data necessary to do it is also a problem. So, the provisions try to recognize that but don't always succeed. So, they may be asking for something that really can't be done without having access to that sort of information.

So, in all, what I am suggesting is that these provisions will succeed to varying degrees in adhering to these principles and some are still at variance with them. So, what we would like to have some discussion about is, first of all, how do people feel about those principles? Are they, in fact, principles that sound fair and equitable and should be used in evaluating what kinds of provisions go in here?

Then secondly, if that is the case, what would the comments be about some of these provisions in terms of whether or not they actually adhere to those principles? So, that is what we are looking for in the way of feedback.

Comments?

MR. SCANLON: Kathy, to some extent the analysis is based on a look at existing Medicaid state contracts.

MS. COLTIN: That is right.

MR. SCANLON: And then hearings as well in terms of what people thought would be reasonable or not. So, presumably a reflex, a current state of the art or the practice --

MS. COLTIN: Well, I am not sure that that is true. I think that there were a few model states that had these types of provisions in their legislation and there was a certain picking and choosing of provisions. This was really to say what is the complete range of all of these sorts of provisions that are out there in one form or another and any one state might have just a very few of these, but putting them together, it becomes a massive document with a whole lot of provisions that there is probably no state has all of these requirements in it.

MR. SCANLON: This is kind of a potential menu. There is no -- and it is voluntary.

MS. COLTIN: Yes.

MS. GREENBERG: Right. I might just add a little bit of history so you can see how the National Committee and its unique contribution here is that George Washington, the contractor, had been working with Centers for Disease Control, the Substance Abuse and Mental Health -- agency, whatever the last -- that "A" stands for, but in any event a number of organizations, public health groups, agencies in the Department and coming up with model language or possible purchasing specifications related to those range of issues, of prevention and mental health, et cetera. In each case, the group would then work on, well, what are kind of the information requirements, but there was no overarching effort going on to look across all the different types of services that might be provided by the managed care organizations as to say what are really the information requirements contract wide or managed care plan wide.

That is where the Committee came in and the subcommittee and supported this specific effort. It is my understanding that it will in some ways replace the individual data aspects of each of the other specifications and provide some recommended consistent overarching language on information specifications. So, they sort of filled a niche there that was or potentially are filling a niche that was recognized but there was kind of -- there was no stakeholder. Everybody had a stakeholder for their own little part, but not for the overarching and that is where I think the committee came in.

MS. WARD: I would just comment, I was struck by the fact that what we are proposing here was very relevant in terms of what we have heard from our illustrious guests, that you don't just have an information infrastructure that talks about just data standards, but you have process standards and that is what we heard from the states is total lack of understanding, what HCFA wants, why they want it. So, they couldn't standardize the data because they didn't know what the other sort of larger process standards are. This is what our committee was trying to solve and it is a big hole out there. We have got to standardize more than just data elements.

DR. DETMER: Clem.

DR. MC DONALD: Well, I would find it hard to respond. I mean, even though you have elucidated the principles well, it doesn't stand out without the document. Is that about to come out or is that just not going to be --

MS. GREENBERG: It is being copied. I am sorry I didn't have it.

DR. IEZZONI: Again, it is lengthy. You will look at it and you might fall off your chair but it is a menu and our goal is really not to have you give feedback immediately but to take it home, digest it, look at it and e-mail Carolyn.

Sara Rosenbaum is very, very interested in getting your feedback and will respond to it.

DR. DETMER: This isn't listed as an action item for tomorrow.

DR. IEZZONI: It is not. We just wanted to introduce your homework to you.

DR. DETMER: Are there other -- yes, Jeff and then Kathleen.

MR. BLAIR: I just wanted to make sure that I understood what we had heard. I don't know who it was -- it seemed to be across from me -- referring to process standards.

MS. GREENBERG: It was Elizabeth, Jeff.

MR. BLAIR: Elizabeth. Maybe you are right. I thought that Dr. Jones referred to process models that were working tools to help them work through how they wind up strategizing and implementing the information strategy for the National Health Service.

I tend to think of process standards as being prescriptive about how a health care institution might have to adhere to a specific process. At least my understanding of what he said -- and maybe I missed it because I can't see it -- I don't think he was referring to process standards. Maybe somebody else --

DR. DETMER: Well, it is our good fortune that he is still sitting with us.

DR. JONES: I don't think a process model should be prescriptive in the sense that this is how you do things. However, most of health care, there are processes that have to occur irrespective of how you organize yourself and how you do things. In other words, there are specific activities of taking a patient's history and so forth and if you can derive your data requirements based on those sort of principles of the activities that have to go on, irrespective of how you organize yourself, then you have got a chance of explaining them, having an audit trail and really being able to drop off data standards, when you would no longer do a process, which is something that we can't do at the moment.

Once the data is standard, it is standard for life and we found that to be a problem, too.

DR. DETMER: Kathleen.

MS. FRAWLEY: Lisa, just a point of information.

Can you give us some idea as to the time frame when you would like comments to Carolyn?

DR. IEZZONI: About two weeks -- what is the data around two weeks? Does anybody -- today is the 3rd. So, the end of the -- the Friday -- does anybody have a calendar?

MS. GREENBERG: I think we were saying the 22nd, I think it was, Monday the 22nd.

DR. COHN: I guess maybe I just need to understand the process a little better also. Is this going to be coming back in June as an action item?

DR. IEZZONI: Yes, it will.

DR. COHN: Okay. So, it will be two weeks for the next set of comments and we will have a chance for more comments.

DR. IEZZONI: Yes, for comments from you guys and then Sara will take your comments, revise the document based on your comments and then take it out to the field for broader vetting. If you have people, for example, Simon, that you think in your organization should see this and should comment on it, that is exactly what Sara needs to know, okay, because we, again -- we don't have a lot of money in the contract for her to go out and do a very formal, broadly-based effort to get input. But we do want to get input as broadly as we can.

MS. GREENBERG: She is also talking about much of this or most of it, I guess, has been developed through a series of conference calls that Kathy has participated in and then she is thinking in terms of like two 90 minute conference calls with this broader, you know, group.

DR. IEZZONI: And she will organize that and conduct that, yes.

DR. COHN: I presume that she would also like to have written responses. She might find that actually to be even easier, I would imagine.

DR. IEZZONI: Yes, although I think the dialogue has been helpful because she had HCFA on the phone, HRSA on the phone, SAMHSA on the phone and it might be helpful to her, I think, she is thinking during the vetting process to also be able to have people who can talk with each other as disagreements arise to try to get a sense of what the arguments might be.

DR. DETMER: Kathy.

MS. COLTIN: I was just going to say that the process that was used by the conference calls group, whatever they were called, clearly that group had representatives from all of the various interested federal health agencies. A lot of these principles that we began with in the first draft were drawn from state documents so there is some representation of the states and also of work that Sara has done with many of the states, but there hasn't been, you know, broad-based opportunities for comment from state Medicaid departments across the country, but, in particular, my concern was that I was the only managed care organization representative on the call. I felt very strongly that this needed to be open to comments from the industry widely and clearly even, you know, speaking for myself, there are things in here that I would take issue with.

DR. MC DONALD: Again, without the document, I -- we have had a lot of discussion about Medicaid's data requirements in the context of some attachment discussions, which were overwhelming, I mean, the range of requirements and the breadth of data, like ranging from, you know, attach a copy of the green card to this piece of paper because they want to have that. So that I just want to -- this may not be dealing at that level, but the Medicaid --

DR. IEZZONI: We can safely say, but you will be interested by some of the things on like knowing about language competency of providers and language is represented in certain care areas. So, there is a lot of diverse types of information in there.

DR. MC DONALD: The Medicaid groups, they get really -- by state, they get really excited about those particular variables that they are hot about or they have had in their laws or whatever. They typically get these kinds of things all mixed together about what the data -- so, I imagine if you do go to the Medicaid people, you are going to get a lot of different kinds of comments coming back about how do I make sure they give me this variable.

DR. IEZZONI: Sara Rosenbaum is a great contractor for us because she knows the Medicaid states cold. I mean, she is probably the best in the country.

DR. DETMER: All right. Do you have another comment?

MR. SCANLON: A question. Does the scope of the recommendations cover mental health, the principles and guidelines as well, or is it -- in terms of the menu, the requirements --

MS. COLTIN: SAMHSA was on the call and there is something in here, but there isn't a lot really.

MS. GREENBERG: But that does relate to the carve in, carve out issue also.

DR. IEZZONI: Let me just remind people that there will be a second document coming to you for the June meeting for your approval, which are recommendations for the Secretary, based on our two year exploration of Medicaid managed care and carve outs, such as mental health, are featured prominently in the draft recommendations that we are beginning to specify.

DR. DETMER: Okay. Well, it is coming around such that you can take it to lunch with you and digest it with your meal.

We are scheduled to start at 1:00. If you could at least try to make it in 40 minutes, I would appreciate it. Thanks

[Whereupon, at 12:35 p.m., the meeting was recessed, to reconvene at 1:20 p.m., the same afternoon, Wednesday, February 3, 1999.]

AFTERNOON SESSION [1:20 p.m.]

Agenda Item: Comments on NAIC Model Privacy Legislation

DR. DETMER: I think I will call us back to order and we will start with Kathleen Frawley and the letter that you have got.

MS. FRAWLEY: Thank you.

At all of your places -- and I did not put it out, but, hopefully, you will find a letter dated February 4th, to George Rider(?), who is the president of the National Association of Insurance Commissioners.

What I want to do is just give you some background. You are not going to be asked to vote on this letter today, but so that everybody knows what we are doing and why we are doing it. The Subcommittee on Privacy and Confidentiality has a number of issues that we have been tracking as part of our work plan and one of the things that we have been watching is what is going on at the federal level and also the state level.

In September of 1998, it came to our attention that the National Association of Insurance Commissioners had adopted the Health Information Privacy Model Act and initially the language was not readily available. So, we decided as part of our work plan that we thought it would be helpful at our November meeting to have a briefing on the model language and ask some of the staff from the National Association of Insurance Commissioners to come forward.

For those of you who know nothing about the National Association of Insurance Commissioners and I will give you some background and then I would have to punt through to Kathleen Fyffe, who certainly knows more about this or even probably John Lumpkin, who maybe has a little bit of knowledge in terms of his role in Illinois.

But the National Association of Insurance Commissioners represents the 50 states, the District of Columbia and all FOIA territories. There are actually 55 people who are insurance commissioners on the National Association of Insurance Commissioners.

The easiest way to explain this is that their focus is in terms of regulating insurance and health insurance. So, to give you an idea in terms of volume, there are about 40,000 pieces of legislation that are introduced each year in the states on insurance and health insurance issues.

So, you are talking about this massive kind of thing. Some of the insurance commissioners are elected and some of them are appointed. So, we have a real mix in terms of who is on the NIAC. The way they work is their structure follows our structure. So, there are 55 of them with staff support. They meet by committees and working groups four times a year. One of their major focuses is developing model acts. They have model acts that they have promulgated in a lot of different areas.

On September 14, 1998, they passed their Health Information Privacy Model Act and then we had picked that information up out in the trade publication. They actually had a Health Information Privacy Working Group, who worked for about four years and a number of people who are on the committee at some point had interaction with some of these folks over the years.

But probably most of their activity escalated between 1996 and 1998 into response from the things that were going on in industry. The scope of their model act is strictly covering protective health information and activities of insurance carriers. That is important because when you look at the letter -- Jeff Blair had raised a good point, you know, from our subcommittee -- when you start to think about this, well, why doesn't this affect employers, why doesn't it affect this? This is a model act that the insurance commissioners now can go out and try to enact through their state legislature.

I am trying to make it as simply as possible for those of you that may not be familiar with this group. What we decided to do is the staff from the NIAC came in November, did a briefing and after the subcommittee meeting, what we decided was important was to take that model act and go back and look at our reports to the Secretary that we released in June of 1997 and then look at the Secretary's recommendations to Congress.

So, we were actually comparing three documents. We were looking at our initial report, the Secretary's report and the model act. I do have to acknowledge Gail Horwick(?), who is the lead staff to the subcommittee, who did a lot of the analysis for us and certainly Bob Gellman and John Fanning and other subcommittee members were very helpful in identifying some questions.

We then had a conference call in January to kind of articulate some more concerns and did some e-mail conversations and circulated a draft, which we agreed to yesterday. I am surprised that we were able with two minor changes to do it in almost under ten minutes, which I think for our subcommittee might be a record for a change.

The letter basically, the way it is structured, is just to thank them for the opportunity to have their staff come forward. The NAIC is actually in Kansas City, Missouri, but they have a Washington office. So, it was their Washington staff, who came to do the briefing. And we explained to them what we did and one of the things that is a problem for us is we cannot endorse their model act, because we have a lot of problems with a lot of the language in different sections.

But we have two really serious concerns and one is the whole issue of redisclosure of protected health information because they are not following all the recommendations that the committee has placed on record and the Secretary has presented to Congress.

The second issue is the whole area of -- let me just get to the next page -- health services research and the use of non-identifiable information and, again, the whole issue about would it be appropriate to have an IRB and the other concern, which is our mantra is that wherever possible, aggregate data should be used, non-identifiable information.

So, what we decided to do was just go with the key issues and not get caught up in their definitions and their drafting errors and a lot of other problems, but there are a lot of other procedural issues with their model act. The subcommittee made a decision and it is our recommendation to the full Committee and you will vote tomorrow is that we were concerned if we sent a letter to the NAIC, it wouldn't get out to the other 55 insurance commissioners.

You will see that our plan is to send it to the president and then in turn send it out to 55 people that we know currently are their membership plus the staff that came to do the briefing. So, that just sets up for you what the letter says. I mean, you are not being asked to do anything more this afternoon, other than, hopefully, get a chance to look at it before you have to vote tomorrow afternoon.

I do have a copy of the model act for those of you who want to take a look at it, but basically just to give you a real quick rundown on the sections so you just know what is in the model act. They have a title section. They have a purpose section and as I pointed out, the purpose is to regulate the use of protective health information by insurance carriers. It doesn't employers. It doesn't cover all the other cast of characters in this situation.

There is a definition section and applicability and scope. The requirements for health information policy standards and procedures, the notice of health information policy standards and procedures, the right to access protected health information, the right to amend protected health information, the list of disclosures of protected health information, the authorization for collection, use or disclosure of protected health information, the collection, use or disclosure of protected health information without authorization, the collection, use and disclosure of protected health information without authorization for scientific, medical and public policy research, which is a hot part for us, the unauthorized collection, use or disclosure of protected health information, the right to limit disclosure, sanctions, regulation, separability and the effective dates. And there are 18 sections.

It is not a very long model act and there are instructions to the insurance commissioners, if they want to drop this into their state legislature -- there is some additional little drafting information, but the substance is those 18 sections that I have kind of rattled off for people. So, if someone has a compelling need that, you know, they want to look at this a little bit more closely, I have -- we have some subcommittees, yes, who participated in the process can help or, you know, you just have to look at the model acts of -- that is what we are presenting this afternoon.

We will come back tomorrow afternoon and ask you to approve the letter because we are anxious with Don heading off to England to try to get this off our plate and get some feedback to the NAIC.

So, I will throw it open. Clem.

DR. MC DONALD: Just to comment on the second page -- and this is maybe too specific, but the second page, the fourth paragraph, the biggest paragraph, the last sentence says that the researchers should be required to remove personal identifiers and to provide IRB with assurances that the information will not be further disclosed.

IRB already requires that. I mean, I think that this may be -- get them confused and you may just take that sentence out. As long as you get IRB, you should be covered there.

MS. FRAWLEY: That is helpful because I know when I was reading -- going back to the transcript, you raised a lot of the issues about the IRB process. So, that is helpful to us.

DR. DETMER: Vince and then John.

DR. MOR: The only question I have is whether -- since I have not read the act and probably wouldn't be able to understand it that much if I did -- is whether focusing just on these two issues, if those -- are these the only two issues where there is a problem?

MS. FRAWLEY: No.

DR. MOR: I don't see any language in here to suggest that these are just the worst of the problems we see or maybe I missed it.

MS. FRAWLEY: We were trying to be very diplomatic yesterday afternoon in our subcommittee meeting.

DR. MOR: My only concern is if the implications are that if you just fix these two, it would be okay.

MS. FRAWLEY: No. What we decided to do if you look at the second paragraph, the second sentence, it says the NCVHS would like to call to your attention some of our major concerns regarding the model act.

DR. MOR: I just raise the question of whether -- and I don't know whether by implication it means that if you fix these two, it will be okay.

MS. FRAWLEY: You raise the exact discussion we had yesterday at the subcommittee meeting because there are a whole bunch of other problems with it. There is their definition of "insurance functions" is way too broad. There are drafting errors, some of the language in terms of "shall deny," some of the device issues are complicated.

DR. MOR: You just may want to say in addition to a variety of other issues are major concerns.

MS. FRAWLEY: Yes. I don't want them to think that these are the only two problems we have.

DR. MOR: That was my -- I agree.

DR. DETMER: John.

DR. LUMPKIN: Having seen now this letter for the second time and based upon your introduction actually thinking about the individuals who this letter is directed to, they are not going to particularly care what the NCVHS thinks. Basically, by and large, when you work at the state level, you are concerned with what your governor thinks, what the people of your state think and what your legislature thinks.

I think the strength of our argument is that we have conducted hearings and based upon those hearings, we have gotten very clear messages from a number of groups around the country. Based upon that, we are making the recommendations. So, we may want to enhance the introduction just a little bit to speak to them.

MS. FRAWLEY: Thank you. That is a good point.

I agree. The thing that we felt important as a subcommittee is we just didn't want them to go back after they had done their briefing and thinking that everything was swell. So, we wanted to kind of say to them -- because I agree. I mean, that is why we want to send it to each of the individual commissioners because we figure if we just sent one letter, it would just get buried in someone's file.

DR. LUMPKIN: It is just a matter of adding on that we based this upon our hearings that we have done and what people told us.

MS. FRAWLEY: Great. Thank you.

DR. DETMER: Barbara.

DR. STARFIELD: I want to support the suggestion that Vince made but I want to suggest that it be made at the end rather than at the beginning because it is -- so, at the end we ought to say we have additional concerns we would be glad to share with you at the end.

MS. FRAWLEY: Thank you.

DR. STARFIELD: And second of all, I need to ask about the second complete paragraph on the second page, the one starting the NCVHS notes that this provision, I have trouble -- I don't understand what we are trying to say there. Is this bad, good, what? You make a statement but I don't know what it means.

MS. FRAWLEY: This isn't good.

DR. STARFIELD: You have to make it clear that it is not good.

MS. FRAWLEY: This is not something that we -- this is a major concern. This is some new information that we have uncovered. So, in light of some of the information that John Fanning and Bob Gellman and some others have provided, we are really concerned about the fact that this would actually allow health information to go to the MIB, which is not something that we would that we would ever support.

Our laundry list in terms of our working group, we keep putting the MIB back down to the bottom of the list, but there is a whole host of issues about what information is going there in terms of the Medical Information Bureau. It is not good at all.

We were trying to be as diplomatic as we could.

DR. STARFIELD: You have to make it clear.

DR. DETMER: John says it is "Men in Black"

that --

Others? All right. I think you have got your suggestions and I think they are good ones. So, that is good.

MS. FRAWLEY: Thank you.

MS. GREENBERG: It is coming back tomorrow.

DR. DETMER: Yes, this will come back tomorrow in the revised form, coming to a table near you.

I don't know if all of our panelists are here. I hope they are for our panel discussion. It is 1:30. Are some of you here?

Well, let's see. Let's relax for about five minutes here. Sorry I rushed us back from lunch.

[Pause.]

We will go ahead and start. This morning was very -- quite linear, except for the interruptions also we had then. There is just something about the month, but at any rate, we have two experts here. We are delighted that you are here and I think what we will do is go ahead and get underway.

Agenda Item: Panel Discussion on Data Requirements for Medicare Risk-Adjusted Payment

This topic of risk adjustment has been a longstanding concern and interest to the Committee and also to some of the members on the committee, in particular. And it has become especially relevant as many of you know with the Balanced Budget Act that has a requirement that managed care organizations provide encounter data to HCFA for risk adjusted capitation rates.

We thought it would be timely to hear from HCFA, as well as several researchers about the rate adjustment methodologies that they will be using and that are potentially available as well and essentially what are some of the benefits and risks of this and how -- what challenges, I guess, we will face as this sort of rolls forward.

We have David Knutson here from Minnesota and we are pleased to have you here, and Dick Anderson from Kaiser on the West Coast. So, I don't know which of you gentlemen want to go first.

MR. KNUTSON: Well, if Cynthia were here, she probably would say -- well, let me start by saying thank you for inviting me. I am very happy to be here.

I had assumed I would be the third or fourth down the line. So, I was going to avoid redundancy and be a niche player.

Let me briefly describe what HCFA is going to do and I mean very briefly, just in case there are a few of you, who aren't familiar with the context for the comments that Dick and I will be making.

DR. DETMER: David, you might say just a little bit what you told me, too, about how you come to these data, what you do in Minnesota and so forth because I think that will be of interest to all of us.

MR. KNUTSON: I direct a health research organization affiliated with a large multi-specialty group practice and hospital system, but the organization I am part of is an independent research group. We do research mostly related to health policy issues, consumer information, chronic illness and organizational development. And in this case -- and mostly what we do is work on risk adjustment issues. I have a contract with the State of Minnesota to develop the risk adjustment method for the Medicaid managed care program and do a lot of that kind of work, have worked on analyzing risk of some of the Medicare choices sites.

So, I would say that I am very actively involved in the technical part of risk adjustment R&D and in implementation. I also was going to speak from the perspective of someone who has also been in HMO administration for a number of years before getting into research and talk about what it is like inside the black box trying to produce these data and what it would be like going forward as HCFA implements its new risk adjustment method.

And then also a bit about what I know as a person who does research on these new methods, what some of the implications we think these methods have for diagnosis coding. But HCFA is going to implement a diagnosis base, risk adjustment method to replace the AAPCC for the most part and it will implement in the year 2000, if plans go as well as they hope.

Because they are concerned with the availability of ambulatory diagnoses, they have picked a model that uses only inpatient diagnoses and primarily the principle inpatient diagnoses. Nobody is happy with that including HCFA, but they have made a decision that the data is just simply not there. And I want to talk a little bit about that. I am sure Dick wants to talk a little bit about that.

So, that system is the diagnostic cost group system but more specifically what is called the PIP or the principal inpatient diagnosis model of that system and only uses principal inpatient diagnoses to classify enrollees in a plan, beneficiaries on the basis of hospitalizations and then predicts their expenses, not just the ones hospitalized, but the whole population's expenses in a subsequent period.

Because it is hospital only and it requires an admission, there is a concern that it -- even though it is a significant improvement over AAPCC, that it is also introducing a perverse incentive. For example, with CHF doing well in the ambulatory setting may get you no credit at all. But the developers have tried to make the system a little bit robust to those kinds of problems, but I just needed to say that that is a controversial issue right now.

Maybe I will leave it at that and I am sure many things will come up. Hopefully, Cynthia will arrive and there is a lot more about what they are doing than I am able to tell you now. But what I want to focus on is the data issues that I think are associated with this implementation.

When fully implemented, health-based payment using diagnosis codes to risk adjust capitated payments will change the incentives for managed care, hopefully, the desired incentives, reduce the financial rewards and penalties, based on an early risk selection quite significantly, we hope.

But the new approach also creates strong incentives that will change diagnosis coding practices and data availability. More specifically, I want to talk about the implementation of the full diagnosis models, which involve ambulatory diagnosis data and that as these systems are implemented, the timeliness and error-free transmission of these codes from physicians and other professionals to HCFA or any purchaser will improve.

Actually, HCFA currently enjoys that arrangement on its fee-for-service site. But that more specifically related to codes that we will also see incentives that will result in more diagnosis codes per encounter in the ambulatory setting and more specific coding, more modifiers and just in some ways you could say better coding.

In almost all cases this better coding would result in a higher risk score for a plan that is able to do this better coding. So, the incentive to improve coding is quite significant.

Overall, the reliability and validity of ambulatory diagnosis will improve probably similar to the DRG-driven changes in diagnosis coding that occurred for hospital coding under Part A. But now we find ourselves in a dilemma. Currently, the data needed to implement the full diagnosis model, namely, ambulatory coding, is variable in quality and the data are not flowing in a timely and air-free path between providers and the purchaser, in this case, HCFA.

But even inpatient data, where the data quality is relatively high is not flowing smoothly across these multiple layers of administration that we have in the U.S. health care system. The dilemma is in that we have found, I believe, that unless payment is imminently -- I don't mean down the road, but imminently riding on the availability of complete and accurate diagnosis data, there does not appear to be the collective will among any players, including purchasers plans and providers to seriously and creatively address the data flow problems, let alone the data quality problems.

I have come to realize that even the demands of purchasers, employers and others for utilization reports from plans and even the scores on HEATUS quality indicators, even though generating significant improvements in the IS capabilities of managed care companies is not doing the job and that dollars really need to be at stake before the problems are seriously addressed.

Now, HCFA has deferred that because it believes that the data availability limitations on the ambulatory side are so severe that it needs to implement the inpatient-only version and then develop other incentives, other than payment to try to develop the capabilities there. Some of those will be effective, requiring that the CFO or CEO attest to the completeness and accuracy of the data and the fact that they will be actually getting ambulatory data and looking at it.

But I want to talk a little bit about why are these data flow problems -- why do they exist and why do they seem so intractable, but because I had thought Cynthia would talk about the flow between the managed care companies and HCFA, an area which she is probably the expert, I wanted to talk about something I know more about and that is data flow between providers and plans and back and forth, the first and probably the most critical path in the flow of these data.

Now,, let's start with the idea that most clinics, labs, hospitals have some fee-for-service patients and routinely generate HCFA 1500 and UB92-like claims. Even traditional staff models with a few notable exceptions -- and Dick can talk about that world a little better than I -- but because they develop self-insured products and they have had to produce employer-required reports, have developed the capacity to generate a claim.

So, the codes are there somewhere. Even if it is one ICD code per encounter, it is out there somewhere. And at the MCO level, the information systems, the claims processing systems have evolved from very crude simple systems with renewal buckets for encounter data because that was thought to be one of the savings of managed care in the old days, to two fairly good systems.

Now, what I want to talk about, though, is something else, another trend that I think should cause us to think about whether we want our information flow to be conditional upon the dynamics of organizational rearrangement in the health care system. Now, under increasingly complex relationships between providers and plans, providers are taking more risks and in many cases are even taking on delegated responsibility to process claims. This is happening everywhere.

Now, in these complex relationships, we have fertile ground for breakdowns in errors in data flow between the providers and the managed care company, which influences, of course, the ability of the managed care company to produce the data needed for risk adjustment. The plan may not be receiving, editing, even processing the data it receives from providers with delegated responsibilities.

Now, there are three types of arrangements. The first is that the plan processes all the claims that you would see then in a traditional IPA. The second is in a fully delegated model where virtually all if -- and almost all the claims are processed by the provider. The plan doesn't know what is going on unless the provider sends it to them.

But also what is most common is a mixed situation where the plan pays some claims. The provider pays some claims and they need to compare notes to figure out what is going on with the population. Now, there is no good business reason or excuse why that should be a justifiable barrier going forward because to admit you can't get data together is to admit you can't manage care.

But that is the situation right now and it is becoming more complex. These three types do not simply exist across plans, across markets, but more commonly these types exist within the same plan and the same market. There will be an IPA for which the plan is paying all claims. There will be a delegated provider for which they are paying no claims and there will be others with a mixed mode in the same market, same plan, same product.

Now, another issue related to the organizational machinations of managed care inside that influences the sharing of these data is that these data are often of tactical use by either the provider or the plan in the negotiations regarding next year's capitation contract. Withholding, slowing down, speeding up is a way of either showing you are not doing well, you need more money or showing -- or not showing someone else that they aren't doing well and so that they don't ask for more money.

It is very common that information is used in this way. So, we have the additional complexity of the fact that this information is of tactical use for other than everyone trying to figure out how to manage health care for a population. But even with these problems within the black box of the data flow between managed care plans and providers, there are both public and private examples of diagnosis data, including ambulatory diagnoses being used for risk adjusting payments to plans.

For example, Colorado and Maryland Medicaid are using diagnosis data. Other states are just starting. Minnesota will be starting in 2000, as well others. And the data flow has not been trouble free, but the point is payment is being made on the basis of these data right now and in some cases since 1997.

Now, on the private purchaser side, there are also a few examples of diagnosis-based risk adjustment that relies on full diagnosis data. One that I am close to and have done research on is the Bias Health Care Action Group of the Twin Cities. This is not a very good example in total for how -- you know, lessons for HCFA because there is so much that is different about the environment, the conditions of purchasing of the Bias Health Care Action Group and HCFA, but I did want to use it as an example to raise a point.

The Bias Health Care Action Group is a purchasing alliance compromised of most of the big companies, either based or operating in the Twin Cities, 3M, Northwest Airlanes, and they offer a joint product that is a self-funded product, to where they offer over 20 some provider-sponsored managed care organizations in the Twin Cities to all the employees who participate.

Because it is a self-funded product, there are limits, significant limits, on how incentives can be transferred -- efficiency incentives can be transferred to the managed care organizations. But they were able to introduce capitation-like incentives by using a diagnosis risk-adjusted PMPM budget target and I won't get into that. That is a story unto itself.

But more pertinent here is they decided to eliminate the problems of running data through HMO systems and a lot of these providers really have a natural affinity, in some cases are creatures of an HMO, but instead to have every physician, every lab, every DME supplier, every hospital, bill the purchasers single administrator directly. That administer pays the claims on the basis of the MCO specifications in their contract, for example, to referral providers and then delivers back monthly to each MCO its own utilization.

The Bias Health Care Action Group thinks that the cost of this is not sufficiently greater than any other system and it really thinks it is a significant advantage of its approach, should recognize that they do cash flow on a fee-for-service basis. That is also a very significant incentive for data flow, obviously. But they really believe that the direct submission of data from provider to purchaser has solved a lot of problems for them.

It also does not mean -- it means that a new entrance into the market, new PHOs, for example, don't need to go out and shop for a claims system. That will become another source of error in the transmission of data to the ultimate purchaser.

Now, inside the black box is a myriad of reasons that we could list about why data stops at different points from provider to purchaser. When you have been in the black box, you recognize that these are often odd things that happen, surprising, mysterious, silly, trivial and some very fundamental. I have in my history been involved in a lot of data analysis and a lot of -- working on a lot of claims data sets. So, I would be happy in a Q&A to talk more specifically about all of those very strange things that happen when you added up mean -- the data isn't flowing, but in some cases they are trivial, even banal in their -- and I don't think we have -- I think as we look one black box to the other, we tend to overestimate the fundamental nature of the problem. We think it is more fundamental than it really is, but it is intrinsic to hand offs and of many, many, many handlers.

But if we want to talk about some of those specifics, I would be happy. The category I am most familiar with is silly mistakes and I have done a lot of those.

Next, I would like to talk a little bit about data quality and what we can expect under Medicare's new risk adjustment system. Well, we certainly as I said earlier, can expect more ambulatory ICD codes per encounter. We can expect more specific codes with more modifiers. We can expect also that where the group, or in this case, DCGs, but it could be true of ACGs or DPS or some of the other diagnosis groupers, where they are particularly sensitive to coding change and where payment is riding on it, that vendors, that certain more sophisticated providers will figure that out early, and the coding training that they will implement will be strategic in the sense it will focus on those codes.

At the same time, where clinical indications of when to diagnose a condition or how to code it or not hard, the developers of these systems will continue to do what they have always done and try to make the systems less sensitive to those kind of coding options. There are a number of examples. If you think of the -- like the DCG system, it is a combination of hierarchical boxes related to certain kinds of conditions, related to organ systems and if you get a more serious condition, like cystic fibrosis, which has pneumonia often part of it, you won't get credit for pneumonia. You will get credit for cystic fibrosis. That is to prevent proliferation of codes and all the complications that go along with some costly conditions.

So, that is one area where they have attempted to solve that problem, but there are other cases we know where a coding change related to coding a more severe version of a condition will move you from a low to a medium to a high payment category within that hierarchy.

Also in these models, not everything is hierarchical. Some is adequate. So, for example, you can have a problem in the endocrine area and you can have a problem in cardiovascular. You could code diabetes in endocrine and also get credit for cardiovascular disease or you could get credit for a higher payment in diabetes with complications.

In other words, there are some ways that the system is a bit vulnerable, but mostly with coding rules and just training on how to code legitimately, the data quality should improve with this incentive. I would recommend that somehow training be provided for plans and providers on how to get up the learning curve altogether. So, reliability is preserved to the extent possible during this learning curve.

I would hate to see the positive incentives of risk adjustment that we -- the incentives we intend mutated by the code arms race, so to speak, where bigger clinics with coding experts and more resources learn this strategy faster and for the first few years we have variation in coding driving payment more than illness burden of their population.

So, some massive training program that their chance to understand how to legitimately code in a way that will get everybody maximizing their revenue together and then, of course, we would recalibrate the models. And this is going on in some states and some places -- I know, Colorado in their Medicaid program will be introducing a training program on coding for plans and providers coming up this spring and there are other examples.

Let me finish by suggesting a couple of things that I have talked about earlier. One is that there be some attempt by organizations that care about data for the public good to try to really understand the process engineering aspects of data flow from providers to HCFA rather than just that black box just sort of speculating on what is wrong in the other black box and pointing fingers.

I think that because data issues are so intermingled now with other strategic and tactical issues related to provider plan relationships, between plan and HCFA relationships, that the flow is confounded for reasons other than data and there may need to be some consideration -- I heard earlier about a process orientation versus -- that is somehow more anchored in a world where the organizational structure in what managed care is is changing so rapidly and creating so much complexity because data have to ride that, that bucking horse.

If that study hasn't been done, do a study that looks at data flow from the provider to HCFA, to Medicaid, to anybody, to try to understand and separate out those more fundamental problems from what I think are many trivial problems that all add up and are just kind of problems and process control.

Then finally look at training programs and monitoring to understand and improve the reliability and validity of coding so that we don't leave it to accident. It will happen eventually but if we can get there faster, I think it would be an improvement.

DR. DETMER: Very good.

MR. KNUTSON: I will finish on that note and be available for questions later.

DR. DETMER: All right.

Hi. I am Don Detmer. It is nice to see both of you.

We went ahead and got underway, as you can see. David has just completed, as you could hear, and I think I will go ahead since we are already to do this and do this, call on Dick, Richard Anderson from Kaiser. He runs the policy center out there.

MR. ANDERSON: This is absolutely backward because I have been trying to think of what I would say if I was going to be representing Cynthia. I have developed a great deal of respect for her over the last year. We have had a number of conversations about the new requirements.

What I want to do in my part of this -- and it really should be last, I guess, is to discuss some of the things that have made it difficult for Kaiser Permanente and other plans to respond to the data requirements for Medicare. It will be a lot of this stuff that David has just discussed, too.

I also have some suggestions for the National Committee about their role in this area, but before I start, I guess it is helpful to have a little bit of philosophy. Kaiser has long supported improved methods for risk adjusting payments in Medicare and in other settings. And "improved" here means, implies that there has been a risk adjustment method already in place in Medicare.

The actuarially-adjusted per capita cost is a very crude method for doing that. Everyone knows that it is crude. It explains very little of the variation in cost as a predictive model. So, what this is all about is to try and deal more fairly with the fact that when you have choice of plans, and some plans get higher risk persons than others, it is not fair to pay the same amount. It is not fair to ignore differences in health status.

We got into this subject many years ago in the private sector when employers began to realize that the selection was getting in the way of effective choice and their solution was just to go to exclusive replacement arrangements, kick out all the plans and not have to deal with adverse selection by fiat.

So, facing up to that, we began to realize that it is good to support risk-adjusted payments even if it might in many cases result in reduced payments to Kaiser.

I am going to talk about five barriers to successful submission of data in Medicare. And here are sort of the headlines. First -- and these are all intuitive -- substantive change is never easy. The second is that new Medicare risk adjusters do have varying effects on different organizations and providers and I want to talk a little bit about the implications of that.

Third, Medicare did face a catch-22 on whether or not to go ahead and implement an improved method for risk-adjusting payments. A new method could create harm and not adopting a new method could create harm is basically the catch-22. So, I will talk a little bit about that.

Fourth, organization should be expected to behave in a financial self-interest on this subject and another way of saying that one is that this has become very highly politicized and that is going to get in the way of the data agenda among other things.

Finally, it is not clear to me that we have national agreement about this broad topic generally, about vision and direction for risk adjusters, not only for payments but for other applications.

Let me talk about this change is never easy part. Kaiser is a big, complex organization with many moving parts and when we were first faced with new requirements and we had a fair amount of notice -- this came out in the Balanced Budget Act of 1997 -- it still took us a long time to capture the attention of managers, sort of make the business case that we had to do some new things, commit new resources, establish new processes and create new relationships with suppliers and intermediaries.

It was difficult for us to convince hospitals that they who formerly were the vehicles for submitting data to HCFA, now it was our responsibility to do this. And you can imagine the issues of trust that came up, especially for us with non-contracting hospitals. Some of the hospitals balked. Some said, look, we have problems of our own in systems and data. Others said, you know, we have got incompatible systems. We have got missing records. Capturing data retroactively is going to be a problem and so on.

Then, of course, we had to establish new systems of our own and on short notice, we had to change to a new fiscal intermediary, which still hasn't been fully ironed out. I sent Cynthia a couple of days ago a copy of an e-mail. It represents the sort of types of issues that we are still having with our fiscal intermediary in HCFA as well.

Responses to our requests for help are too slow. We can't reconcile what we have submitted. Their reports are impossible to use. HCFA has rejected our claim on the basis that the hospital is billing incorrectly but the hospital is telling us that they are following the correct procedures and won't resubmit.

Records resubmitted come back with the right HCFA number but a different name on it. HCFA has rejected our claim on the basis that the beneficiary is dead and we knew the patient was alive when we provided the service and so on.

[Laughter.]

So, we are still trying to sort out, you know, accuracies in code and underreporting issues.

Now, let me not forget the Y2K problem. Anyone here that is involved with a health care organization or almost any kind of organization is struggling with this issue. I don't know, Simon, my partner over here, urged me to mention that we spent probably a couple hundred million dollars on this issue already and it has become the priority for now. It has had to be given attention first and foremost. So, it is an issue of competing priorities.

The punch line here, if there is one, is that most people thought that doing the inpatient stuff was going to be easy. We have been talking about full encounter-based system for risk adjustment and I think the problems there of the type that I have described are going to be much more difficult to resolve quickly.

The second area that I talked about was varying implications to organizations and providers. Now, first of all, let me talk about the Kaiser Permanente perspective. Our physician, medical groups are not paid on the basis of what they do or the kind of treatment they provide or the kind of procedure they provide. They are not paid directly, specifically, according to that.

So, one of the issues for us is the concern about creating a new system of coding and reporting that will have the right incentives for them to code fully without affecting what they do, the incentives, the behavioral incentives to provide appropriate care.

We have also had sort of different extremes within our organization. We have some small plans with small Medicare risk populations, which can literally almost deal with the new HCFA requirements, if you will, on the back of an envelope. But when we have large, mature programs with volumes -- in California, a hundred thousand inpatient admissions a year for our Medicare population, legacy systems, enormous investments and things that are working reasonably well, long lead time is for investment in new information technology, then you have a different kind of a question about implementing something quickly of large magnitude like this.

Other types of organizations -- and these are what Dave called the more fully delegated or mixed models, where providers are much more involved in the claims activity -- are problematic, too. It is not atypical for some of the other plan types to have contracts with large numbers of physicians, many of whom have affiliations with other organizations. They have some interesting contracting arrangements, percent of premium arrangements, where if the payment to the organization, to the health plan is reduced, that flows directly down to the providers.

You can imagine what the implications are to the providers. First of all, they have been resisting providing data because of the fear of intrusion, but from the health plan in any event, and now there are questions about how to ensure that they can maintain long term viability and payment arrangements. They don't want to enter into long term percent of premium contracts to the extent that this will drive their payments down.

Also, providers are becoming aware more broadly of the whole issue of risk-adjusted payments. In the State of California, not too long ago, a bill was submitted that would have required if a plan gets a risk adjusted payment, to drive this all the way down to the provider level. I think the best take that I have at the moment is providers are confused about this and becoming increasingly wary about the whole subject.

Now, this catch-22 that I talked about, I think I will leave it to Cynthia and others to kind of draw that out a little bit. The concern that we have had is should we wait to implement an improved method of risk adjusters, a full encounter-based model that may take years and years to implement or do something that is directionally correct at the moment. My own view is that they made the right decision. But flat out that HIB DCD method that relies solely on inpatient data is biased and it promotes the wrong incentives.

HCFA, I think, has made a valiant effort to try and deal with the incentive issues by hulling out discretionary hospitalizations and even not counting one day hospital stays. But those changes have been very controversial and I think there is still a remaining question about whether the method is biased in favor of fee for service, which you don't want to have a biased method.

Also, it doesn't explain and predict quite as well, which is a problem. So, a lot of discussion has been going on, how quickly does Medicare move forward in trying to collect, capture full encounter data? My only comment on that is move forward swiftly. Give us plenty of notice, engage us in the communication and dialogue so that we can help to make it all happen.

I also mentioned that organizations behave according to their financial self-interest. I just want to reinforce for you the high stakes in this payment issue. People have estimated the amount of overpayment to health plans in Medicare to be as high as 10 to 15 percent. A lot of people have done these measures. I am not sure where it is going to end up when it gets officially scored. I think HCFA was at something like 7 percent recently

But you can put this in a different kind of a context. All are concerned about increased uncertainty, increased unpredictability that may occur especially during a transition time. We are very gratified that HCFA has adopted a modest phase-in of this new method of payment over five years. And all of us are aware that a single unreported inpatient encounter can have a value, can be translated to potential underpayment in as much as three to -- more than $25,000.

So, you can imagine the motivations to be sure that everything is counted correctly. The bottom line here is that many organizations, because of the uncertainty, because of the potential for significantly reduced payments, that ultimately translates to reduced benefits or beneficiaries, have increasingly become strident on this issue, have increasingly supported deferring any implementation at all or even outright repealing this issue. I think this political opposition is going to complicate this whole discussion of how to move forward with a full encounter-based model.

I would be remiss if I didn't at least mention the broader context for risk adjustment and there are many applications beyond payment; performance measurement, resource allocations, clinical applications, evidence-based health care and quality assurance. If there is another message here, it is pay attention. Perhaps these are the most important areas for the future evolution of methods for risk adjustment.

You know, what variations in use, cost of services and outcomes exist after controlling for differences in risk?

Some suggestions for the National Committee. I guess, number one would be to help ensure a focus so that the data requirements are limited to what is essential for the purposes in hand. There have been a lot of discussions within the industry about HCFA collecting more information in UB92s and in HCFA 1500s as that evolves than would be necessary to run adjusters. I guess, you know, HCFA as a purchaser and a regulator does have other needs for data, but help to ensure that data is collected for specific purposes that there is coordination among the different users and that unnecessary data is not collected.

Factor in the implications to providers; physicians are the ultimate decision-makers about the type of care that is provided and what is entered into records and they have a very important role in ensuring that the records are complete and accurate. So, help them develop strategies to encourage proper and complete coding with the right kind of incentives to produce the right kind of care.

Also, help to create the environment so that standardized definitions, standardized methods of electronic transmission will be adopted and then that data reporting will be more feasible for them. Help to place confidentiality and privacy in the proper perspective.

I notice that you were referring to a letter before this discussion that dealt with the issue of protected data. Fundamentally, risk adjustment can't occur unless the risk adjuster is using individual level data. Now, that doesn't mean it has to be identifiable, but there are many third parties, not only Medicare, but in the employer community, business coalitions and so on that will be increasingly using this information.

As the context spreads into other applications, especially clinical applications, evidence-based care, someone needs to help set the boundaries about protected data and access so these things can occur. And it might be tempting to create boundaries that are too stringent so that they couldn't be done effectively.

Finally, help to shape the national agenda for risk adjusters. I don't think there is a vision, a clear vision, for this broad context. So, help to explore applications in other areas beyond payments. Help to sharply define the common denominator data elements and then help to build consensus about the means to effectively support and use the risk adjusters.

DR. DETMER: Thank you.

Cynthia, nice to see you.

MS. TUDOR: I am sorry I am late. There were a lot of communication problems.

DR. DETMER: I am sorry. We were responsible for that.

MS. TUDOR: I have overheads and someone is supposed to be making copies. I can do either.

DR. DETMER: Overheads are fine.

We have a little lapel mike over there.

MS. TUDOR: We wouldn't need risk adjustment if Medicare beneficiaries were distributed randomly among the plans, but for whatever reasons, both due to the behavior of our beneficiaries and the behavior of plans, they are not, which means that we need risk adjustment in order to determine the level of accurate payments to plans.

We are using risk adjustment to set payment levels accurately for plans. Studies done by a variety of organizations, Congressional Budget Office, the OPPRC, the original Mathematical Policy Research Study of the evaluation of HMOs and HCFA's 1996 HMO study all suggested that plans were enrolling healthier people in the average Medicare beneficiary. They suggested selection bias in the area of 10 to 14 percent, which means that we were substantially overpaying.

Finally, the ultimate goal is that we want to compensate plans accurately and appropriately if they enroll sicker people. To review the mandate, the BBA required us to implement a risk-adjusted payment methodology by 1-1-2000. It also explicitly gave us the authority to inpatient data, but by specifying a date for the collection for inpatient data, pushing out the date for the collection of other data and locking in the implementation date for risk adjustment, we were essentially locked in to implementing a risk adjuster based only on inpatient hospital data.

It allowed us to collect other data beginning July 1, 1998, but because of the variety of reasons that Dick has pointed out, we didn't move forward at that point to doing it. The regulation that went out June of 1998 basically said that we would not collect anything before 10-1-99.

I will speak some more towards sort of the time line for implementing any new risk adjustment mechanism and how long it takes. Our implementation schedule, on January the 15th, we published our 45 day advanced notice of methodological changes in the payment rates. We essentially gave people the methodology we were going to use.

We released the software onto the Web. We released a look-up table that allowed you to basically track diagnoses into PIP DX groups and we released a file of county level risk factors. On March 1st, 1999, we will release our report to Congress on the accounting of actuaries evaluation and the risk adjustment methodology.

We will basically explain the assumptions in the March 1 notice and finally we will release a letter to plans that tells them what our estimated impact on their payment rates is using list adjustments. The current date for the ACRs that are adjusted community rate proposals that are due for 2000 is now May 1, 1999. They are trying to get that date extended.

January 1, 2000, risk adjustment becomes effective. The basics of risk adjustment, for those that don't know it, are that we currently use a demographic basis for our payments; that is, we multiple a demographic factor by a county level payment. In risk adjustment, we are simply supplementing these demographic factors with health status factors. In the slide and the papers you are now getting show you some examples of this.

There are lots of risk adjustment approaches that people can choose. They can use surveys. They can use encounter data. They can use actuarial approaches. They can use a concurrent model, which looks at current costs. They can use a prospective model that uses diagnoses from one year to predict for the next year.

HCFA selected a prospective model based on individual level data. The payment is determined by each enrollee's risk factor, rather than an average for the plan. We are basing risk adjustment initially on the principal inpatient diagnostic cost group, abbreviated PIP-DCG risk adjuster and we expect to move to a comprehensive risk adjustment mechanism, which has not been selected, by 2004.

I am sure everybody is quick to point out what are the disadvantages of inpatient and did a good job of doing that. The advantages of inpatient risk adjusters, especially from the perspective of someone who has struggled over about 18 months to get the data in is that this is the only place we could have started. There is absolutely no evidence that plans were able to give us anything more as of July 1, 1997, than they did.

So, we started at the only place we could have started. The diagnoses were easier to obtain and that wasn't easy. They are easier to verify and we will look towards some verification later. We expect them to be more accurately coded because hospitals are accustomed to providing good bills to us.

We also believe that inpatient admission is a proxy for greater severity and, therefore, this is where you would want risk adjustment to be. Finally, this helps us transition to a comprehensive risk adjuster.

This is a good first step. The inpatient model essentially risk adjusts only a small percentage of individuals and that is 18.6 percent of fee-for-service beneficiaries are hospitalized and about 12 percent of these, two-thirds of these are placed in a PIP-DCG category other than the base. Twenty percent of the dollars are moved in a PIP-DCG and that is in a demographic model, all dollars are paid based on demographics.

In a PIP model, only 20 percent of the dollars are moved away from demographics into the higher level. Under a comprehensive model we move a lot more dollars around. Many more of the dollars are paid based on the health status, rather than the demographics of individuals.

I am sure that people like Lisa can answer a lot of questions much better than I can on a lot of the mechanisms that went into developing the PIP-DCG model, but essentially you begin with grouping diagnoses that are clinically alike. Then you place those diagnoses into DX groups, essentially based on the expected costliness of those diseases in the following year.

Each person under this methodology is assigned to a single highest cost group. The PIP-DCG model has approximately 15 DCGs that trigger increased payment. The demographic variables in the model include age and gender. Originally disabled is added on. The current methodology uses current disability.

When someone who is disabled aged into Medicare, they basically lost their bump up for being disabled. By using the original reason for entitlement, we sustain a bump up for the originally disabled through their aged years.

We basically have maintained a Medicaid add on. Currently, we use a concurrent method for identifying people under Medicaid. We now use a retrospective actually looking at Medicaid eligibility in the year prior to payment. And, finally, we use a separate model for individuals that don't have a diagnostic history. We call this the new enrollee model and they basically are adjusted using a different methodology than we use today, but also a different methodology than someone who has a diagnostic history.

I have given you a couple of examples in your handouts. The first is an example, a male, 75 to 79 with no hospital admissions. That individual gets a payment of about $4,600. He gets no increment for health status, no increment for Medicaid or originally disabled.

So, the total predicted expenditures in 1996 dollars would have been 4,625. We are basically making this a factor by dividing by the average for Medicare fee-for-service, which is estimated to be about $5,100, which gives him a factor of .91. We then multiply that factor by the county capitation rate adjusted for risk.

Continuing with this male, he has now developed congestive heart failure. He gets a bump up of $12,435. He has suddenly become originally disabled for Medicare and Medicaid and gets additional bump up of 4,000, for a total expected payment of $21,000. We again divide by 5,100 for a factor of 4.14. We multiple again that by the county capitation.

While this method is not as good as a comprehensive model and I would be the first in line to tell you that, it is an improvement over where we are today. When we evaluate how well these models do, we not only look at the percentage of variance explained, but we look at how well these models do for atypical groups of enrollees.

One way we do this is by diseases and that is in the bottom half of the chart and the other way is by expected cost. We ranked individuals according to whether or not they were in the top versus the lowest 20 percent of cost and looked to see how well the AAPCC-like model did that we have today versus the PIP-DCG model.

Right now for individuals that are in the lowest 20 percent of cost, the AAPCC overpays by about 1.66 times the expected costs. The PIP-DCG on the other hand comes close to lowering that payment to only about two times.

In the highest cost categories, the AAPCC drastically underpays at about a .44 ratio. That is related to actual cost. And under the PIP-DCG, we raised that a little bit. If the numbers for a comprehensive model were up there, you would see us coming much closer to 1 on both the top and the bottom 20 percent of cost.

We also do better for disease groups. As you can see, we improve the payment across bias groups composed of people with these diseases.

One of the sort of last minute changes we made to the 45 day notice was to implement a reconciliation for late data. Dick's voice rang in my mind as we wrote this. Basically, this allows plans that don't make the September 10th, 1999 deadline for encounter data to get that diagnostic bump up recognized at least later on.

Basically, it is going to be a one time annual reconciliation. They will have until about June of the following year to get the data in and then sometime between March and June of the following year, after the payment year, we will go back and reconcile, determine who should have gotten more payments and who should have gotten less, if that is, in fact, the case.

I want to talk a little about this year and how we did this year in collecting data. We basically collected data between July 1, 1997 and June 1998 on discharges during that period. As of about November the 5th, which is when we made our first draw, we had approximately 1.2 million encounters.

We have gone back for the final draw and it is up to about 1.4 million. The majority of the data were abbreviated UB92s and that is a much smaller data set with relaxed edits that allowed the majority of plans to get their data through in a timely basis.

We assessed the relative completeness of these data, basically looking at the number of discharges that we had relative to the number of enrollees. We have found this average to be about .22 encounters to enrollees. This is in contrast to what we see in fee for service, which is about .36.

This is a histogram, which shows you the ratio across all the plans submitting data for the first year, basically with a big bump up in those people in the left, the 0 to 0.3, with some ratios as high as .54, much higher than fee for service.

We basically excluded everybody below .12 in our initial impact estimates and I would like -- I want to make another comment about the large number of people on your left. There were about 35 plans in there as of November 5th. When we went back for our second draw, we were basically down to six plans in that very low band.

The rest of the plans have attested that those data are complete. So, we feel like we right now have very few plans that what we deemed as very low.

When we were assessing the completeness of the data, we did a lot of univariate statistics, a lot of bivariate, looking at distributions by gender, by age, by diagnostic category. We looked at the month of discharge. We looked at sort of everything we could in a massive way to make sure that we were fairly complete.

Another thing we did was to compare plans' numbers that they submitted for HEATUS to the numbers that we had for inpatient encounters. Basically it shows you that we are much higher than the HEATUS number that have been submitted, suggesting that we are close to what they -- close to the encounter rates that they believe they have.

When we were determining the data set for the impact analysis, we limited it to 195 plans, deleting all those plans that were below a .12 ratio, deleting terminating plans. That was pretty much it. We had no minimum enrollment for our impact data analysis.

In the analysis, we compared the actual payment for September of 1998 to what a plan would have gotten if risk adjustment had to be fully implemented in that same month. The average weighted impact if the PIP-DCG had been fully implemented is a decrease of 7.6 percent in payment, again, assuming no transition. We found that the impacts did not vary substantially by region or by plan size.

Because of the transition and the 2 percent minimum update, we believe that no plan will receive a reduction in total payment relative to what they received in 1999.

I think I have another histogram here. This may be out of order in your pack. This is a histogram of the impacts of risk adjustment. It basically shows that the range was from about a negative 16 to a positive 5. This, again, was based on the November 5th data pool. We expect these numbers to go up for the final estimate for 3-1, and that is because for the same period of data, we simply will have a greater volume, which means no plans number can go down. They can only go up.

So, we would expect these impacts to actually lessen with this final set of numbers.

Payment in 2000. The reason to start as early as 1997 to implement 2000 was actually to give the plans time to get accustomed to sending in data, to working with it and to provide plans a year ahead of what their estimates might look like. Basically, plans are now submitting data that they will be paid on in 2000. The data will come from the period 7-1-98 to 6-30-99. Plans must get these data in by September the 10th, 1999 in order to be incorporated into the payments for 2000.

The reconciliation of the 2000 payment that I was talking about earlier will be taken in early 2001.

I think I had one more chart actually to show you enrollment size and impact analyses. Basically these are just selected plans, which shows you that we had wide variation and impact, but this did not essentially vary by plan size.

In order to sort of figure out how to end this, I wanted to talk some about implementation of the comprehensive model. Certainly when we implemented the inpatient model, we were forced to start essentially 30 months ahead of the date that we wanted to implement. That allows us to have an estimate ready 12 months prior to the payment time. When you move to a comprehensive model, you have to expand that window. You not only have to provide us more time to analyze the data because the volume is so immense, but you have to provide the plans the time to give us those data when they may not be used, acknowledging that the quality of the data may not be great.

So, our estimate of how long it will take us to implement this model is anywhere from 39 months to four years, which means that we must move towards the collection of outpatient data sometime in 2000, which means we should adopt it sometime in 1999.

I think when you look back over to what was a failure and what was a success about the current inpatient model, the failure is that we had to collect data retroactively. The plans were not given sufficient notice of their expectations and enough time to get them set up.

The success, I think, is that we provided multiple ways for plans to send us the data; that is, plans could use hospitals and some national plans actually used the hospitals to send all the data. The second is that we designed an abbreviated version that plans could use. This abbreviated version has helped tremendously.

We cut off a number of edits that we normally apply to the UB92 and that enabled massive amounts of data to get through fairly quickly.

The third success, I think, is that we did everything we could to try to ease the burden of plans. We basically intercepted between plans and FIs in order to work out as many problems as we could. We faced immense Y2K problems in making any systems changes, but we still managed to get the data in.

Hopefully, we will be able to resolve as many of these outpatient problems on a timely basis and set up a more efficient system.

Finally, I wanted to make one comment about the privacy issue. I am certainly concerned about the privacy of the factor. We basically have had some conversations with privacy people in the department about issues related to this. We can protect the factor. The factor looks like a 1.3, a 5.7. We provide it back to the plan only after the plan has completed an enrollment of an individual.

Even in the estimates that we are providing, we are giving them only on a plan basis and that is because individuals that were enrolled in September of 1998 may not be enrolled in the plan any longer when they get their estimate. So, we are being careful even now to protect the privacy of this factor.

From a factor, you cannot determine the diagnosis that placed that individual there. A factor like 4.4 may tell you that the individual is sick. You may even be able to figure out which PIP the individual has to get them in there, but you cannot figure out the diagnosis that got them in that PIP.

I will be glad to answer any questions.

DR. DETMER: I think we probably will have a number, but I think we might as well go ahead and hear from Dr. Boesz and then we will open this up.

You are with Aetna and if you don't mind --

DR. BOESZ: My name is Tina Boesz and I am with Aetna U.S. HealthCare. I have been with Aetna for about six months. Prior to that, I was with NowCare, which is an HMO that was owned by New York Life. And three years ago, I was with HCFA. So, I have come at this from several perspectives and delighted to be here.

Cindy and I were actually at a previous meeting together, talking about some of these same issues and it is always nice to follow Dick. It saves a lot of energy on my part. I think it deserves a couple of minutes to go back and talk a little bit about the history of encounter data and some of the issues that have cropped up over the years.

The first time I got involved in encounter data was with Dr. Denson(?) at the Harvard School of Public Health in the early seventies. At that point in time, encounter data was very important. So, you could assess the risk of your population in order to figure out what kind of productivity you needed to have in your medical components and things of that nature because at that point in time, everybody was going to look like Kaiser.

So, you had to have this lovely data set and collect information. I remember working very hard on coming up with a lot of comprehensive data and I was a statistician, still am. I thought this was really great. The problem was nobody wanted to pay for collecting all of this data, even within the business part of running the HMO.

When I joined the Federal Government in the late seventies, I got engaged, again, in an encounter data project and this time it was with the National Association of Insurance Commissioners with HCFA and actually then the Public Health Service with federally-qualified HMOs, trying to figure out what kind of encounter data did we need in order to regulate HMOs, regulate managed care companies to be sure that they were providing all the care that was necessary and making certain judgments over their operation.

What came out of that exercise basically was a substantially reduced amount of data that companies would have to report into the Federal Government and state governments to make these judgments because it was very, very difficult and costly.

So, I would characterize the seventies as encounter data for the business of understanding how to manage the business, the eighties in terms of how to regulate the business and now the nineties in how to pay for the care that we are supposed to provide.

Throughout all of this, I have heard a similar refrain from the business people and that is what is the minimum amount of data that you need to make these decisions. That is consistent no matter what the purpose has been. As I have gone through the transition of my career as this refrain has happened over the last, you know, three or four years, the same kind of question, do we need to collect this level of data on individuals in order to accomplish appropriate payment?

And I don't think we have answered that question yet. We certain as a company feel very strongly we don't need a full UB92. We certainly don't need that information to pay claims. We don't need that amount of information to even start to do health risk assessments. It doesn't really work in that area. We use different entry points. We might do some little screening from it and we certainly don't use UB92 data to do any kind of patient management in terms of care management. We have to get much more specific and don't throw it off of that kind of a database.

So, really the warehouse of this amount of data is very costly and is very labor intensive to assure one thing and that is some kind of appropriate payment from the government. We don't use this kind of data to deal with our private pay business. We do a much more simplified version.

So, the business folks get a little bit nervous when they hear all this. And Cindy and I have had many debates on all of this and she knows my feelings. The question comes, what kind of stability in all of what the government wants, when will that settle out and what is all this comprehensive stuff going to do for us? Is it going to lower our payment? Now, that is the business question.

And I go back to them and I say, well, Cindy has got a very good slide show. Trust her. She is here to help you. Then, of course, they read that HCFA is projecting a 7 percent reduction in payment, that it was a general understanding that this somehow was supposed to be budget neutral and money was just supposed to shift around, but suddenly it is disappearing as we go through all of this. So, it raises some very real questions in terms of whether or not companies can stay in this business and the reasons that drive that decision could vary from a Kaiser's interest to an Aetna U.S. HealthCare interest and anything in between.

We went through last summer a very difficult period of time in deciding whether to remain in certain markets or to exit certain markets. I think all companies face that decision. Part of the driver was what kind of business partner is the government going to be in the future and what does all this risk adjustment mean and what does all this collection of encounter data mean when coupled with the challenges that we all have with Y2K, with understanding all the requirements of the Balanced Budget Act, with employers trying to understand their obligations to retirees and what their commitments are in the future in terms of liabilities to retirees and also in terms of basically what do stockholders of for profit companies think is important in terms of results and what -- I would imagine what you would think if it is not for profit in terms of remaining in the business.

All this came to bear very quickly last August, September and while I can't sit here and tell you that encounter data and risk adjusters was a total driver, it certainly was a factor in my company's decision to exit or reduce certain markets and part of the issue is the fact that it is easier to develop and maintain certain products in certain markets than it is others and if this product, namely, the Medicare product is so labor intensive and burdensome administratively, we need to think about whether or not we can stay in this market long term and it is better to perhaps exit a market now than to fail in that market sometime later.

This included relationships with providers. We pay physicians, hospitals, IPAs. We have every kind of arrangement possible from global capitation to individual claims payment and we take this relationship very seriously. Some of our providers push back rather vigorously in terms of saying that they did not feel very comfortable with a long term strategy of collecting all of this data so that we could reduce their payments when we got reduced payments from the government.

So, that became part of the equation. Now, none of this would be a big problem if we had time if we could work through these issues, but the timing is such that you have to make from a business perspective some of these decisions rather quickly, particularly given the implementation of the Balanced Budget Act, where we made certain pricing decisions to get into the market and we felt as the fall approached, where we couldn't make certain changes because the agency had its challenges in terms of getting information out, that it was better to exit a market quickly than to strangle everybody in that market.

Now, the reason I mention that, all of this, is we are not over this yet. The whole issue of how risk adjusters and counter data is going to affect business decisions is still very much with us. While we are hoping that there is some relief in terms of how we have to pull together our pricing proposals and get them submitted to HCFA -- right now the date is May 1st -- that this would be somewhat delayed.

We have had some relief in that there is now going to be a blend, a phase-in, if you will, where next year we will be paid 10 percent based on the heath risk status adjuster, 90 percent on the old methodology. That helps. But what we are now hearing from employer groups, who are saying, gee, that is good for next year, but what about the year after.

If we decide to enter into a partnership with you and the following year when that risk adjuster shifts over to a 30 percent -- I believe it is the second year -- are you going to stay in a certain market? We cannot give any assurance because we are not quite sure yet how all of this is going to play out with the party that is paying us; namely, the Federal Government.

So, there is tremendous market instability at this point in time. The one thing I have learned in the past few years is that people making the business decisions do not like instability. I think you alluded to that. And I am starting to certainly feel it.

So, it is not just the cost of collecting the data and warehousing it. That is real. It is the challenge of Y2K. That is real. And I will be the first one to admit we are so successful at meeting these challenges. I was explaining to Cindy that last week we did a test to see whether or not we could mail ID cards to Medicare beneficiaries in a Western state, a small plan that we have, and we were so successful we actually mailed them ID cards effective February 1st, 2000.

We forgot to tell the people not to mail them. Now, my law department called and said what do we do about this and he said we have proved something. I don't know what, though.

I think that is indicative of the technical challenges that we all face. No matter how hard we try to make something work, there are just going to be these glitches. We have had some real challenges with the fiscal intermediaries. We actually, both NowCare Net and U.S. HealthCare used the same one. So, that was helpful.

But in the NowCare setting some of my programmers basically switched a birth date. They switched a month and day and when it got to the fiscal intermediary, the transactions were accepted and then rejected, but the files were set up. So then we corrected it and tried to resubmit, it was rejected as already duplicate individuals and the intermediary could not make a mechanical fix to this and had to change every one of something like 20,000 records manually, which took all of November and through December into January. So, then when we resubmitted our data and finally got an edit on some of the problems we had on diagnosis, we missed Cindy's date of January 15th.

So, we know the data is not complete that they are working from and it is not her fault. It is just a complication of some of the technology. And I know I am preaching to the choir here when I mention those challenges, but they are quite real for those of us that have to explain this to the business side, explaining, you know, why this is so important that we do this.

The product is becoming too difficult in many levels for the Medicare product that is, at many levels for the business side. They see the -- they understand the notion of assuming risk. They understand the notion of appropriate payment. They are two concepts you deal with constantly on the commercial business. So, this is not alien.

The idea to get paid that you have to be so -- such precision on getting paid on an individual basis is an alien idea in terms of understanding risk management and the push back I get is when -- so, we do all of this and so then back to the year 2010, we are back to claims payment, fee for service. Haven't we already been there before and if this is the direction we are going in, why don't we just stop all of this and move there now.

You know, I hear that time and time again from -- unfortunately, that is a part of the business.

There is also a fear that with this warehousing of data that somehow it will be used for purposes that are not being discussed at the present time. There are accusations that the data could be used for integrity issues and it is not that my company is afraid of the question. It is when you are submitting data with all these attestations for one purpose and then suddenly it switches to another purpose, you don't understand again all of the reasons.

By integrity, I am talking, you know, the fraud and abuse issues of whether or not this data will be used to build a base of some type to say, gee, we would have expected X number of encounters. You had Y and so, therefore, perhaps you are underserving that population. It is a legitimate question, but the fear is where do you get that data and how do you address it. So, it is not the question. It is this whole issue of what kind of protections if you are collecting data for payment purposes do you have being used for other purposes.

I have said a lot about program stability and as long as -- we thoroughly agree with everything that has been said in terms of needing more -- if we are going to do risk adjusted, we need better data. The fact that there still remains this question going out into the future still creates instability in the program and it creates instability with our ability to negotiate and deal fairly with our providers.

When we can't -- some of them do want long term contracts and then we can't figure out what we are being paid, it is pretty hard to figure out how to pay them and as the risk adjuster flows, we may get paid a more appropriate amount from HCFA, but depending on the luck of the draw for the provider unit, they may or may not have even yet a different problem.

So, we are into risk adjustment at that level. So, we have sort of lost the notion of risk management in large numbers. Now, none of this is new. Some of you might recall the early days of the federal HMO program in the mid-seventies, when employers also were concerned when they had to offer federally-qualified HMOs as an option to employees. They had many of the same concerns that the Federal Government has now. They said HMOs were skimming. They were taking the good people. The sick people wouldn't change doctors. Only the well people would switch over to an HMO. Then there was the accusations of shadow pricing.

Eventually, I think what has happened is as the volume of individuals has shifted into a new model, if you will, of risk sharing, a lot of that has settled out. I can't help but wonder if that is not what we really need to do with the Medicare program is concentrate on incentives that will get more volume into the managed care programs and have some of this settle out because of large numbers as opposed to trying to fine tune and pay on the exact status of an individual.

I think there is also one other fear from the business perspective and that is if the government is successful in implementing risk-based adjuster payment, will the employer community want to move in this direction? Then what have we done in terms of our abilities to manage care and use those kinds of economic incentives that we think might work, not that it is a perfect system, but going back -- we think it is a step backwards if we move too far in that direction.

As far as this committee goes, I would echo some of the things that Dick Anderson said in terms of what I think the contributions you can make and that is try to do some kind of standardization for data reporting that -- or encourage that type of thing, so that when we deal with HCFA, we are -- whatever the minimum data set is that we have that. When we are dealing with Medicaid, we have some consistency there in dealing with the employer community.

There have been private efforts at this through some of the HEATUS data and some of that has been very successful. What we see is that, by and large, the government has not yet accepted some of that. They are starting to move in that direction. The process of deeming and accepting what accreditation bodies does is a very slow process. I think that needs to be encouraged in terms of some of the data efforts.

The confidentiality issue is very serious. It is something we take very seriously and would encourage you to be proactive in that area. So, I can be brief because so much of what I would have said has been said. And thank you for inviting me.

DR. DETMER: I want to thank each of you. We have about ten minutes, slightly more than that, for discussion. Obviously, this is a tremendously important area with a lot of both dollars and health care in play, frankly.

So, Vince, let's open it up here.

DR. MOR: I have two questions, one for HCFA on the factor. That is an individual factor?

MS. TUDOR: Yes.

DR. MOR: I am assuming the statistical properties of that factor in any particular sample of people is going to be extraordinarily skewed. If 20 percent of the people have any data, the rest are going to be whatever because they haven't had any events. Is that going to be right? Most people have whatever, a zero factor or a 1 factor or whatever you have it and then it is going to tail off way down like that. Is that right?

MS. TUDOR: Well, the basis for the factors were developed in the 5 percent Medicare sample. So, the weights assigned to each disease and to the demographics are accurate and reflect fee-for-service cost.

DR. MOR: Of all the Part As, of the hospital side.

MS. TUDOR: No. They were developed on total cost. So, the weights are accurate. The weights are based on a huge sample. So, there is no instability that way.

DR. MOR: Then they are applied per individual in a given area.

MS. TUDOR: They are applied per individual period and the area comes in when you multiple it by the county rate.

DR. MOR: Now, I guess the question is why would you ever want to give that information to anyone because all that is relevant about it is its aggregate quality?

MS. TUDOR: No. The plan --

DR. MOR: Maybe I didn't understand you.

MS. TUDOR: The plan is paid based on the multiplication of the factor times the county in which that individual is living in. So, the plan needs to know that for Mrs. Smith, they got a factor of 1.3 times Baltimore County's rate of $550. That is necessary in order for them to understand that they got the right payment for Mrs. Smith.

DR. MOR: Now I understand Dr. Boesz's concern. This really is your individually pricing in that sense and every new person that walks in the door has some kind of factor invisibly tattooed, just like we now do with DRGs. I just wish there was some other way to do that without having it individually identified.

Now, let me ask Dr. Boesz, would you imagine or would it be anticipated that Aetna and other MCOs would pass down this risk in some proportionate manner to its risk sharing subgroups within its network? That was implied a little bit by one of your --

DR. BOESZ: Yes. I mean, I think that is -- we would retain some risk. We pass them on, depending on what the arrangements are with the partners.

DR. MOR: Making the maldistribution of that already skewed distribution even more complicated.

DR. BOESZ: That is the concern. It is more complicated.

One of the things that I was going to say that -- and it is just another technical glitch, but one of the factors now that is used to adjust payment is working age status. While I appreciate the effort at reconciliation -- I mean, you can imagine -- on working aged alone, we do a survey. When the person enrolls, we ask them whether or not they are working and we get a lower payment if they are working because the presumption is, you know, Medicare is secondary. So, something changes in that person's status and we do surveys. We all do the same thing every year to try to upgrade that. We do our best.

Then these people file returns with the IRS. Within approximately three years after the fact, the IRS records come in and modify the HCFA records and our payments are then adjusted accordingly and we have over the years built up these huge reconciliation issues that are millions of dollars. And with all due respect, we go back to the agency and we say, well, what can we do to adjust this and the general attitude is, well, we would have to pay you more money. So, that is a lower priority. This is a continuing fear.

And it is just on the working aged. We ought to be able to figure that out.

DR. DETMER: Dick.

MR. ANDERSON: I wanted to, I guess, raise a concern about the implication that this will be pitching us headlong into sooner or later fee-for-service reimbursement or claims-based reimbursement at the individual level. A lot of the effort here is to try and design an adjustment system that is incentive neutral at worst and that, of course, is what a lot of the controversy has been about, just focusing on this inpatient-based model.

But I suggest that there may be kind of a paradigm beyond where we are now. In a way if you pay more if somebody is treated than if you don't treat them, that sort of has wrong incentives and that is what is sort of happening with any of these models. But maybe down the road the next stage would be to define ways to adjust payments for improving the health of people or slowing the rate of decline of disease.

There are people that are working on variants of that and I am hopeful that that will evolve.

DR. MOR: But if you think that the data requirements are big for what we are doing now --

MR. ANDERSON: You could do it on a piecemeal basis.

DR. DETMER: Paul, Kathleen, Simon and Clem.

DR. NEWACHECK: My question is for Cynthia.

Cynthia, we are all interested in understanding the impact of this new risk adjustment system on costs and quality and plan participation and a whole variety of other areas. I am wondering if you can tell us a little bit about what HCFA's plans are for internal and possibly external evaluation of the new risk adjustment system?

MS. TUDOR: Well, initially, these numbers will go to the plans and I think we will have feedback from the plans certainly on how good the estimate is or how good the estimate is based on what we have. We are going to look at the impacts on plans of risk adjustment. Part of the M Plus C Evaluation will look at changes, try to benefit packages, to, you know, what counties the plans are located in, to the level of benefits. And they will determine -- they will try to determine, ferret out, what affect risk adjustment is having apart from the other M Plus C changes.

We will -- I think our concern continues over small plans, rural plans, is there a differential impact? We will look at that in the next several years as we move towards greater and greater transition blending.

DR. NEWACHECK: Are there any plans for a formal external evaluation?

MS. TUDOR: Well, the M Plus C Evaluation I am talking about is an external evaluation.

DR. DETMER: Kathleen.

MS. FRAWLEY: Yes. The first question I have is to David. When you were talking about the training for coding for plans and providers, I wasn't sure when you were talking about providers who you were defining as a provider. So, I just wanted to make sure I am clear on that before I make my next comment.

MR. KNUTSON: Okay. The idea here is to -- for example, the program in Colorado I mentioned that I will be involved with will be trying to train plan medical directors, anyone in the plan who -- so the medical director can take back the message that there is something to pay attention to and then also someone who is technically more proficient in coding issues, but then also to invite key provider groups.

The key there is to try to get at as many layers as possible to start with so that the translation, say, from a medical director in a plan to someone who can carry out a program to the providers is a little less daunting.

MS. FRAWLEY: I guess I -- I don't want to be dense, but when you are saying key provider groups, are you talking physicians, nurses, allied health --

MR. KNUTSON: I am talking primarily physicians and possibly the -- some clinics will already have coding experts in house, but all hospitals will, of course. So, at that level, that will be more of the responsibility of plans to go deep into that area, but I think if we can get some of the key influence leaders within the structure, that would be the intent.

MS. FRAWLEY: The reason why I raised this because actually both you and Richard and then Cynthia made some comments that were very helpful. I just wanted to respond a point because I don't know any of you and I work for the American Health Information Management Association and we credential coders.

I am the only person sitting at this table who is credentialed in coding and there is one person in the audience who I know is. My concern is -- and I am just really directing this a little bit to Cynthia and just also to some of the folks from the community is we are struggling right now, trying to, you know, work with the hospitals through some of the fraud and abuse initiatives in the Office of Inspector General. Now we are struggling with the APCs and how we are going to work DRGs and APCs when the notice of proposed rulemaking has all kinds of craziness, which led me to write a letter back saying has anybody heard about HIPAA, DRGs, APCs or whatever.

Now, I am struggling with my oasis(?) problem. So, the one thing I just want to let Cynthia know is that the association is available because we have 38,000 credentialed professionals. About 50 percent are working in hospitals. What we have noticed over the last two years is 50 percent of our members are now not in acute care. They are in managed care.

They are in physician offices and whatever. So, I was very happy to hear some of your comments because we have been struggling for years on the hospital piece. The other problem I just want to raise because we brought this up in hearings two years ago in the administrative simplification is that the big problem that we have is that the FI as in a lot of the payers don't follow the official coding guidelines that HCFA, NCHS, AHIMA and the American Hospital Association promulgate. I mean, those four entities have an agreement about how the coding guidelines in this country get developed.

Our second problem is that we have got payers, who don't use the current version of either ICD-9-CM or CPT-4 or HCPCs and one of the problems that we have been running into recently -- and I raised the fraud and abuse piece -- is that we have seen some of the proposed data sets coming out of HCFA, lots of different offices, that do not accept five digits for ICD-9-CM, don't accept V codes when that is the appropriate diagnosis if you follow the official coding guidelines.

My only point here is just -- I am staying away from confidentiality for a change and going back onto coding -- is that, you know, we are out there. So, I am just am very pleased to hear some of your comments, but I am just troubled in terms of -- and Cynthia and I can talk off line -- that there seems to be a major disconnect and I am running around trying to figure out who in HCFA I need to reach to say remember, HCFA, you know, we are not going to have a UB92. We are going to have this change in this coding system. We are going to have APCs.

So, I just wanted to get on my soap box for two seconds there. Thank you.

DR. DETMER: Simon.

DR. COHN: Well, first of all, Dick, I want to thank you for doing such a great job representing Kaiser Permanente, especially on such short notice. I am not biased, mind you.

Cynthia, I also am with Kaiser Permanente, for your information.

Now, I actually have two questions for you, Cynthia, and I guess I just need to understand a little better about some of the data requirements and I will ask about data requirements, rather than privacy and confidentiality or anything else.

Now, first of all, I was actually struck by your comments about the abbreviated UP92 and successes this year and I was curious even before we start talking about full encounter data, based on those successes, is that being extended as the methodology by which you will accept data in the future since it has been so successful? That is question No. 1.

MS. TUDOR: Currently, the abbreviated UB92 will be accepted for discharges through 12-31-99 and can be submitted to the fiscal intermediary through 3-31-2000. We are concerned about use of the abbreviated UB92 once the requirements for HIPAA are released and our ability to continue with that form under HIPAA.

So, we are looking at whether that is going to be possible. Another thing to bring up is that -- and it sort of talks about some things that Dick said were that this form contains those elements necessary to run the risk adjuster and certainly some necessary to develop some monitoring.

It does not provide sufficient data to price the managed care data at fee-for-service pricing. If we ever want to develop risk adjustment methods that are based on either managed care data alone or a combination of fee for service and managed care data, we must have information that we can price, which means that we have to be able to fit these data into something that Medicare prices currently.

That issue certainly continues as we -- and becomes even more acute as we obtain the rest of the data because it is only through, you know, having the SNF(?) and the home health and physician and outpatient that we can even move towards incorporating managed care data into the calibration of risk adjustment models.

DR. COHN: I guess if I can just ask a slight question on that before I jump into my next one.

Obviously, I think you are aware that in many of the hospital environments, especially those that deal with integrated health care entities, and I will probably look at Kathy Coltin as I say this -- obviously, the pricing may be a weak point in their ability to deliver data.

MS. TUDOR: I understand.

DR. COHN: Let me just ask the same question about this issue about full encounter data. I think we all think as -- at least intellectually, a better methodology, risk adjustment. Now, we talk about full encounter data, which I think really means diagnostic data and yet we also talk about requirements for HCFA 1500 documentation, which includes a whole lot more.

Now, what would be the rationale and what would be the timing as you look between now and the year 2004 around what sort of data requirements, I mean, knowing that this isn't set in stone yet. But what would you see as the requirements down the road and what would be the rationale for that full -- the 1500 data.

MS. TUDOR: Again, we have to take into account what the HIPAA requirements are. They are standardized forms. We have to comply with the standardized forms under HIPAA and we will have to make sure that whatever format we use for the 1500 or its equivalent under HIPAA are complied with.

When we speak of -- you know, when I said what we have done right is the easier format. I am certainly looking to see whether we can have an abbreviated format and comply with HIPAA. It is not clear to me that we could, so -- even in the abbreviated format for the 1500, at least for a short term.

DR. DETMER: Clem and then Kathy. Then I think we are going to have to move forward.

DR. MC DONALD: I must say I am impressed by how immensely hard everything is, the more I hear about these things. I picture this very well-oiled machine, which suddenly arbitrarily makes a new gear and jams in there and then we hope it all works.

MS. TUDOR: You don't know how true that is.

DR. MC DONALD: This has been given a lot of thought and some very careful thought. I read Lisa's paper and the thing that struck me is the 8 percent R squared we are explaining, which usually, you go, oh, shoot, I can't publish that. It is not really a big hitter in terms of effects. And the goal of this was to get rid of this potential skimming.

So, the question is when you interact with the choices people make in the various communities today, they make already because everyone doesn't -- this is a volitional thing. Where these things are available and particularly these -- will it change that? Is there still room for that given all this complexity?

MS. TUDOR: Will this change people's behavior or will it change --

DR. MC DONALD: Well, I mean, I have heard one of the extreme cases, they invite Medicare patients on a second floor elevator to a dancing class and, of course, they are more fit.

MS. TUDOR: Kaiser would never do that.

DR. MC DONALD: I know, but that is one of the extreme stories you hear. But when you look at this, you have got the big categories like HIV and brain tumors. Well, brain tumor patients aren't going to sign up for Medicare the day before they first found out their brain -- they are going to be dead in six months. Those are $129,000. So, is it really going to -- has anybody kind of gone back and analyzed how this stuff falls out, if it spreads, will you really make a big difference? Will it really be anybody who is in Medicare because you are doing

-- you did run up those stairs and the next day, boom, you had this brain tumor, you know, and you are not alive three years later to really kind of cause long term costs, but you did that one year.

MS. TUDOR: Well, I am not sure how many people we are going to find in PIB 29, though we have --

DR. DETMER: Do you have a comment, Dave? Oh, I am sorry. Finish, please.

MS. TUDOR: Though we have when we have run the impacts, we do have individuals coming in PIB 29. I think that we will change two kinds of behavior. If plans can use risk adjustment to develop networks that treat sicker people. Then we will have individuals joining those plans to get access to those physicians.

I think the general aging of the population will push us in that direction, too. On the plan side, I think that plans will determine that they can afford to at least select slightly sicker people under the inpatient model and begin to move towards changing their behavior under the comprehensive model.

DR. DETMER: Dave and Dick.

MR. KNUTSON: I just wanted to briefly address the issue of the 8 percent R square. A lot of variation in utilization in a subsequent year is just not predictable using any information that we easily can obtain. So, it is really an asymmetry of knowledge issue between the supplier and the purchaser. In this case, the plan can use utilization data and we have chosen not to in the risk adjustment because that is fee for service. So, you get paid for what you have done versus you get paid for the diagnoses.

Although that is not completely outside of the control of the provider, obviously, it is still closer than procedures. So, when you look at how well you can predict going forward, given information, including procedures and all the services rendered, the best you can do is about twice that or a little more than that maybe.

So, it is not like you are closing the gap to a hundred percent. The rest of it is apparently random, at least given the current systems and that is insurance and that is just a lot of large numbers you can handle unless there are some strange bias issues related to very acute things that have no sequelae, like maybe births, you know, as a classic case.

So, you are not trying to close a huge gap. You are trying to close some gap. Is it adequate? Nobody knows yet.

DR. MC DONALD: If this is correct, you have got something to skim it down by 12 percent. I mean, that is not the R squared. So, we are dealing with two different kinds of numbers. But it seemed like they weren't doing bad on no information, trying to figure out how to bias it. So, how in the world can it happen? Or is that inflated and it is not really fair at all?

MR. ANDERSON: Can I common this?

I had a different point to make about the 8 percent and I think it has, in part, to do with what capitation is all about. This is not predicting payments for individuals that we are really worried about. It is predicting payments for populations that are treated by health plans and that is the number you want to get right on average.

Now, if you have systematic bias, that is a problem and that is the case with the PIP-DCG model. But I am not as concerned only 8 percent explanatory power if I can have a predictable stable capitation payment that gets close enough, well past rough justice. I don't think we have dealt in any of this discussion with the catastrophic unpredictable stuff and there are folks -- Joe Newhouse(?) at Harvard, who strongly believe that some so-called partial capitation system needs to be overlaid on top of the risk adjusters payment, some form of reinsurance, for example, or payment based on episodes that are really high cost.

So, that is another discussion.

MR. KNUTSON: If I could just make a follow-up comment, to think back on Cynthia's slide where she showed

-- when she created skewed groups on the basis of risk artificially, but, you know, and tried to see how close the payments came at a group level to those to paying the right amount. You will notice that the PIP score was much better than AAPCC and she said that if she had the ATC or the ambulatory database model, it would be very close.

So that for low cost groups it would be close to 1 predictive accuracy. For medium, it would be close to 1 and for high cost groups it would be close to 1. Generally speaking, the predictive accuracy with these new models is significantly better than any old actuarial model and even if they slightly still underpay on the high end and slightly overpay on the low end.

DR. DETMER: Care to comment?

MS. TUDOR: It is true.

DR. DETMER: Kathy.

MS. COLTIN: I have a question for you. When you calculate the factor for, say, Mrs. Jones in fiscal year 1999 and then the following year, you go back and calculate factors, can Mrs. Jones factor go down or only up?

MS. TUDOR: Mrs. Jones factor from one year, the X could go down or up. I think one of the criticisms of particularly this model is that someone could be hospitalized for a catastrophic disease like brain cancer and let's say they do live and they don't get hospitalized again, their payment in the year following could go down.

My short answer is that this is the only thing we could do. Plans had difficulty producing one year of data. Now think what is going to happen when it takes two years of data to run your model. You can certainly move towards developing multi-year models and people are working towards multi-year models and I think multi-year models certainly are important and we have to think about them in the comprehensive, then you need to see if they add anything.

MS. COLTIN: The reason I asked the question is that, you know, under HCFA's QSMC(?) program, you are trying to build in incentives to improve quality of care. Lots of the programs that plans are evaluating to try to improve quality of care have costs associated with them. You know, disease management programs like those for CHS, which is one of your examples, are clearly a popular one right now for plans because the cost of investing in the disease management program can be offset by the savings in keeping people out of the hospital and keeping them healthy, but there are real costs.

The pharmaceuticals go up. The case management costs go up. The outpatient visit costs go up. What you are saying now is the reward for that would be to get paid $12,000 less the following year.

Well, I am wondering, you know, why would a plan want to invest in --

MS. TUDOR: That is another reason the transition is at such a slow rate, that we aren't putting much emphasis on risk adjustment during the next several years. We are trying to get it up to the comprehensive model where you won't have this disincentive and I think that is a valid criticism.

DR. DETMER: Stewart, do you want to make any comment or --

DR. BOESZ: I have one --

DR. DETMER: I am sorry. I didn't see you.

DR. BOESZ: If I may make just one comment to further complicate this problem. It sounds like there is an assumption that this person also stays with the plan over a period of time and I think as you all know, at the present time, that it can change every 30 days.

Now, while there is some attempt to lock in downstream, this person is hospitalized this year and one plan pays the expenses, they could move -- and we find this typically happens many times. People who do get -- the elderly, who do get ill and frail tend to move sometimes back home with children and things of that nature and then they enroll in a different plan and not necessarily yours. So, it isn't as simple as -- I wish we could figure out how to do all this selection.

DR. DETMER: Lisa, final question or comment?

DR. IEZZONI: [Comment off microphone.]

DR. DETMER: Okay. Well, this has been very useful and I think actually -- I think the committee will continue to track standards, quality, the issues of confidentiality and such, but it is also very apparent that we need to keep an eye on this particular issue as well, among a raft of them, it seems, but I think you have really helped us a great deal and I thank each of you very much for your participation.

We will break at this point.

[Whereupon, at 3:35 p.m., the meeting was recessed.]

THIS TRANSCRIPT IS UNEDITED

National Committee on Vital and Health Statistics

February 3, 1999

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

PARTICIPANTS:

Committee Members:

Liaison Representatives:

HHS Executive Staff Director:

Executive Secretary:

Staff:

TABLE OF CONTENTS

P R O C E E D I N G S [9:05 a.m.]

Agenda Item: Call to Order, Welcome and Introductions, Review of Agenda

Agenda Item: Update from the Department

Bill Braithwaite.

Agenda Item: Discussion of 1996-1998 NCVHS Report

Agenda Item: Consideration of Principles for Draft Purchasing Specifications Related to Health Data for Medicaid Managed Care Contracts

AFTERNOON SESSION [1:20 p.m.]

Agenda Item: Comments on NAIC Model Privacy Legislation

Agenda Item: Panel Discussion on Data Requirements for Medicare Risk-Adjusted Payment

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