Hubert H. Humphrey Building
200 Independence Avenue
Washington, D.C.
Workshop on Future Needs in Health Statistics
General Surveys and Population-Based Data
JEFF BLAIR
SIMON COHN, MD
KATHRYN COLTIN
DON DETMER
KATHLEEN
FRAWLEY, MD
MICHAEL FITZMAURICE
KATHLEEN FYFFE
DANIEL
FRIEDMAN, PhD
MARJORIE GREENBERG
CLEMENT MC DONALD, MD
ROBERT
MOORE
PAUL NEWACHECK
JAMES SCANLON
BARBARA STARFIELD, MD
DR. STARFIELD: We are under some considerable pressure here to get done a little bit ahead of schedule. We will try to get done at 5:00 rather than 5:30.
We have three items on the agenda, once I get finished with my opening remarks. First is the workshop on future needs in health statistics. The second, we are supposed to have Chris Schute come and talk with us.
He is not coming, but we will replace that with some thinking about the general surveys and the population- based data that is not clinical data. We will particularly take advantage of Paul Newacheck's present to lead us off on the thinking.
The third is the claim attachments following on the hearings that we had last month.
So, I will turn it over to Dan Friedman, who has done a lot of thinking about this workshop and a lot of discussions.
Agenda Item: Workshop on Future Needs in Health Statistics.
DR. FRIEDMAN: Thanks, Barbara. At the November committee meeting, Ed Sondik stressed his desire to identify long-term goals for health statistics and health statistics data needs for the year 2005 and 2010, and the specific implications of those needs for NCHS.
Then thirdly, was his desire to work out some kind of mechanism where the committee could function in a continuing advisory capacity to NCHS.
As an initial step, Ed suggested the possibility of the committee working with the National Academy of Sciences and NCHS in order to organize what was initially conceptualized as two workshops.
The first basically was a broad agenda on future health statistics needs. The second workshop was specifically focusing on the implications of those needs for the national center.
In mid-January, Barbara and I started discussing this and Marjorie and I and Jerry Hendershott and Ed Hunter and Ed Sondik.
From subsequent conversations with Marjorie and e mails, it has become clear that, not surprisingly, the discussions -- and I think productively -- have continued within NCHS.
So, what I am going to say now is based upon a somewhat dated notion of what we had been talking about all of two weeks ago.
As we had been talking about it, the initial workshop was going to focus on health statistic needs for policy making and program development in the next century.
It was going to basically focus on two central questions, with the first central question being, how do we evaluate health statistics systems.
In discussing the evaluation of health statistics, one of the things that is sort of troubling to me is that there really, at least in the United States, there doesn't seem to be a literature identifying or defining health statistics as an endeavor as, say, apart from public health surveillance.
There doesn't seem to be really a literature identifying the purposes of health statistics systems or setting forth criteria for evaluating health statistics systems.
In fact, in a lot of ways, I think that the only definitions of health statistics are essentially legislative definitions.
This really stands in contrast again, for example, to the public health surveillance, where CDC has really spawned, you know, a small but very, very helpful literature on what is surveillance, what is the purpose of surveillance systems, how do you evaluate them, how do you plan for them.
For those of us who are responsible for surveillance systems, and for statistics, that type of literature has actually been extremely helpful and extremely important.
So, the first central question would be intended to start remedying this gap, and could focus on such issues as evaluatory criteria for health statistics systems, definitions, boundaries and differences between health statistics on the one hand and surveillance, perhaps health services research, on the other.
The second central question that we had discussed essentially revolves around how can health statistics systems meet policy needs in the next five, ten, whatever, 20 years.
As somebody from a state which has had a 150-year string of collecting mortality and natality data, and had a 100-year string of collecting communicable disease data, sort of the time perspective that I think of in terms of the systems is, you know, more than five or 10 years.
In fact, those data, while sort of quaint, having the historical data continues to be very useful.
In any case, inherent in the first question is really what issues will health statistics need to address. What data needs will exist.
We have talked about it and gone around and around in terms of the breadth of this. We keep focusing back on the need to identify needs in population components, health behaviors, risk status, resiliency, events, et cetera, insurance status, transactional components, and then finally, service delivery components, data elements that would need to be included, methods of linkage and then finally, and really extremely important, private and public partnerships and private and public provision of data.
I remain troubled by the breadth of what we are talking about. Sort of the message I got -- my interpretation of the message that was sent from the January executive subcommittee meeting was think big.
I think that we will try to do that. At the same time, I think it is also going to be important to anchor the discussions in some retrospective looks at what we have learned, and what we have learned in two or three areas, one area being a number of recent reviews of health statistics and related systems, like the 1992 committee on national statistics report, which was very valuable, a recent Mathematica report.
A second area is taking a look at health statistics systems that have continued over time and have continued to provide valuable data, and data that, at some 10,000 or 20,000-foot level, continues to be comparable over time, despite changes from ICD-1 to 2, to 3, et cetera, et cetera, but still, some really valuable data.
Third, also a look at major, at least one major attempt in health care policy making, and what were the data needs that went unmet at that time.
An example is health care reform. I think it would be helpful to look back and say, what did we need to know that we didn't.
That is basically where we are. The workshop or workshops remain very much nascent in the morning. We are going to be meeting with Ed Sondik to discuss it further.
One of the things that Marjorie has said, and I think it sounds like a very good approach at this point, is to step back a little bit from the workshop idea in and of itself, and to focus more on what we are trying to find out, rather than how are we going to organize a workshop.
The idea is that there could be a great deal of prior work that could help nail down some of these things before a workshop and then, hopefully, focus the workshop somewhat more.
DR. STARFIELD: Any questions or suggestions for Dan? Marjorie.
MS. GREENBERG: I was just going to say that we are meeting in room 440-E tomorrow morning with Ed Sondik. Anyone who is interested in working with the center on this, and Dan, you are more than welcome.
DR. FRIEDMAN: Absolutely. Help would be appreciated.
DR. NEWACHECK: How would this be jointly done with the Committee on National Statistics. Do you have a sense of logistics on that?
DR. FRIEDMAN: I have no idea.
MR. SCANLON: This is a metaphysical exercise, Paul, but we would have to deal with the bureaucracies over at the Academy as well as here.
But what we would do -- and we have, in fact, started on this -- we have an HHS planning committee to help plan the structure of the questions and possibly participants in the workshop.
When we are ready, the Academy would put together a small planning group from CNSTAT as well, probably only one or two people, and we would combine forces and agree on an overall agenda, participants and so on.
So, we will have to deal with that. They will put together a small planning group as well, since that has to agree, obviously, as a committee over there. They have to agree to do this. I think they basically agreed that they would do this.
I would emphasize one other thing. While the impetus for this inquiry was from Ed Sondik, I think Ed knew that he needed a fairly broad -- he needed advice from the best thinking.
This was meant to be health statistics needs in general, not just NCHS from the start. So, we have had as a principle from the very beginning that the HHS planning group was really drawn from all the agencies across HHS.
The content of the first workshop, in terms of what the needs are, would be much broader than just NCHS and much broader than just government, in fact.
This issue of roles and responsibilities, I think, is a critical one.
DR. NEWACHECK: Jim, can you comment a little bit on the different roles that the Committee on National Statistics plays versus our committee, in terms of the policy side of things? It seems that they are competing in some ways, and complementary in others.
DR. STARFIELD: We are not picking up all the comments. Could you speak into the microphone?
MR. SCANLON: Well, the Committee on National Statistics at the National Academy of Sciences, as you all know, the National Academy of Sciences is chartered by Congress and it goes way back to prior to the Civil War when President Lincoln asked the Academy to provide scientific advice, I guess it was, to the government.
Since then, it has been a source of advice to federal agencies that ask for it, and the industries and sectors at large.
The Committee on National Health Statistics was created, as you remember, Paul, in the early 1970s, and it was to extend this kind of capability to federal statistics, but national statistics for policy and programs as well.
So, they serve as an independent capability that can establish panels, bring together experts, convene workshops, on a variety of issues in terms of national health statistics.
We certainly, in HHS, have used them many times to help us with workshops or panels on topics that have a largely statistical or data component.
An example, in 1996 HHS asked that same group CNSTAT, to convene a two-day workshop on how do we assure the statistical capability to monitor changes in welfare reform and health reform and so on.
They did a very nice workshop report. I, to some extent, envision that as a model, to some extent, for what this other workshop could be.
There is a danger to over-analyze, about over-specifying exactly what you want this to be. At one point you have to just sort of stop and let creative forces of the participants take hold.
You always learn a lot from these. Otherwise, you never take the next step because you are analyzing too much, what about this, what about what.
CNSTAT, at any rate, Paul, is sort of this government-wide, actually national resource that provides advice to anyone who asks for it.
We deal with them by contract. They establish, typically, expert panels or workshops. I think they are largely seen -- like the IOM and the Academy -- as sort of the highest level of advice on scientific matters that one can get.
This committee is more of an advisory committee to HHS and to health agencies generally. To some extent, this committee has provided advice on health statistics to the various agencies in HHS, but it is not -- obviously there are a whole variety of other issues that this panel deals with.
The academy can take -- they can take one or two issues. They can devote a lot of attention to it over time and they can come up with recommendations that often sort of bear the acceptance of the community at large.
They are able to get, as well, the very top researchers or participants, just because of the Academy association.
They have a much broader view in terms of national statistics generally. They deal with immigration statistics, transportation statistics, education statistics and so on, welfare reform and poverty measurement.
This group advises HHS specifically on health data.
DR. COHN: I actually had a comment, but more of a question, Dan. I think what you are doing is very interesting and I guess I want to provide some sort of input.
Unfortunately tomorrow at 8:00 a.m. is also the subcommittee on privacy --
DR. DETMER: 8:30, I think.
DR. COHN: Actually, what I was going to ask specifically, as this is moving forward, is it possible to see anything on paper that we could comment on?
I think we all have sort of potent visions of the future of health data. I don't know exactly what the difference between that is and health statistics. That, to me is one of the key issues.
DR. FRIEDMAN: No, neither do I. I would guess that if we go around the room and ask, what is health statistics, we would probably come up with 20 different answers.
DR. COHN: Exactly, and that is why I want to I guess potentially provide some input. I guess maybe I can make it at 8:00 o'clock.
MS. GREENBERG: They are working out some kind of a document.
DR. FRIEDMAN: There are various iterations. They obsolesce so fast, but yes.
DR. STARFIELD: I am on record as saying that I think we have to be really very careful. This is a potentially high risk venture, I believe, because you could be so general about this sort of thing that you don't move yourself forward.
I do think we have to get it down to exactly what it is that we are after.
MS. GREENBERG: That is part of the reason why we kind of stepped back, also. I think in our conference call, you provided some of those cautions. I think we will get it together.
MR. SCANLON: Again, I will emphasize that it is possible to be too cautious and let the whole input slow down. I just wouldn't want to see us slow down and get bogged down in over-analysis. There is some middle ground, I hope.
DR. STARFIELD: I wasn't suggesting we slow down. I think I was suggesting that we speed up.
MR. BLAIR: I know so little about this area that this is very general. I would think that in terms of what we hope we will be able to get in terms of clinically specific outcomes data from computer-based patient records over the next five to 10 years, that we will be getting information across a wide population spectrum that we have never been able to get before.
It would seem to me that there are certain diseases that we know are very, very costly over a long period of time to treat and reimburse, are major parts of our Medicare and Medicaid budget.
We know what those diseases are and we know that there have been clinical trials for a long time, trying to identify procedures and treatments and medications that are effective.
It would seem to me that we could look at the clinical trials that have been done in the last few years, where we just haven't been able to either afford large population samples to get down to refining them, or where you have multiple factors that you are considering, and you couldn't do that and get meaningful information with 500 or 1,000 or 3,000 patients over time where, with computer based patient record systems, five to 10 years from now we would get information where we could correlate a disease to two or three or four different things for the very first time.
From a public policy standpoint, I would think that we would focus on those, and try to provide some incentives so that, as we move toward computer based patient record systems and this data gathering, and making sure that we are gathering that type of data that would wind up allowing us to get better outcomes information on the highest expense diseases.
DR. STARFIELD: I think that is definitely something that is going to be under consideration. This workshop is what is the potential for adding other kinds of information than the health statistics that we have customarily been thinking of.
MS. BALL: I am pleased to hear that the health services research is being explicitly being connected to health statistics, and health statistics to health care policy. Is there someone from AHCPR who is participating in these discussions?
MS. GREENBERG: We have formed a planning group but we have not convened it yet. There is definitely someone from AHCPR on the planning group.
I think there was some discomfort level with convening with -- I think the members of the data council were asked to name someone.
MR. SCANLON: Those are our usual suspects.
MS. GREENBERG: We may want additional people beyond the ones that have been named. I think there was just some discomfort even moving ahead with the planning committee when things weren't more spelled out.
I think Jim also gives -- you are going to come to the meeting tomorrow morning; right? I think he gave us a good caution, though. I mean, I think we could spend the next few years trying to think this out.
We also formed an NCHS group but we haven't convened them either. Obviously we want people to be involved.
I would just say that I see -- that doesn't mean that everything is going to be done by NCHS or any particular agency -- but I think health statistics -- and part of what we have been talking about is that people's eyes glaze over when they hear health statistics.
Actually, I see everything that this committee does is related to health statistics, because that is my vision of health statistics, as really being health data, as being information on, about, for health, and not just the medical model and everything else.
Then, of course, you have to look at what can you actually do within any particular agency. But I would caution you not to be too -- if your eyes glaze over when you hear health statistics, to try to counteract that.
It gives a very kind of specific sort of sound, but it really can be extremely broad.
MR. SCANLON: As an example, again, the workshop that HHS sponsored, I guess it was December 1996, Dan, I think I gave you the report of that. Again, this was a similar example.
We asked the National Academy of Sciences at the CNSTAT workshop. It was meant to bring the best thinkers in policy -- welfare and health policy -- those who believed they understood what the changes were and what the needs were, as well as the welfare area, together with the data users and producers.
Far from being dull, it was actually very well attended and it included the director of CVO and other agencies, as well as OMB folks.
I wouldn't start off assuming that it is a dull topic that people aren't interested in. What differentiates health statistics from health data may be the quantitative aspects, as opposed to the transaction.
This is a pretty large range, and I think we are thinking about, you know, a perspective that includes the traditional vital statistics, public health uses and functions.
It goes into the administrative, clinical improvement, and health sector, health services research as well.
MS. GREENBERG: The first workshop is really supposed to be relevant to the department at large. Ed Sondik wears a second hat on being the senior advisory to the Assistant Secretary on Health Statistics. Obviously you have to have AHCPR and HRSA and CDC and everyone involved in thinking this through.
DR. MC DONALD: There is another difference between individual data and what you are describing as health statistics. I don't think it is a very descriptive word. So, I don't like it as a word. I don't think it tells anybody what you are thinking about.
It is the requirement to collect a defined sample, which is where all the work is.
MS. GREENBERG: What is a sample?
DR. MC DONALD: Whatever you want to call it. You have got a denominator, so you are doing something like a sample, but you have got to get all of them.
It is the problem of hospital data reflecting outcomes. Well, if they go somewhere else to die, you are not getting the outcomes.
So, what you are calling health statistics are just the individual data all pulled together. But the pulling together has this huge challenge, often, of tracking down those missing ones, the hard ones, or finding something to make it meaningful. On top of that, you have got the analysis component.
MR. SCANLON: It may be that the population focus is what differentiates this from the clinical side.
DR. MC DONALD: Whatever the word is, you have got to have a denominator that means something and that is the hidden work in all these things. It either means extensive data collection, tracking, follow up, something.
DR. DETMER: I do think that with the rising pressures on just the aggregate budget and the shift toward managed care, and the emergence of much more sophisticated information systems, or more ubiquitous, part of what I think we are really going to want to see is both a more value driven health care system, but also one that tracks the health of populations, not only as aggregates of numerators, but actually as populations and tries to build incentives to see populations health improved.
That starts to put in a health of communities kind of mentality that has been somewhat different, I think, than necessarily other kinds of populations as well.
The 2010 data sets and so forth, I do think that we are seeing some shifts and some emphasis that really do see a need to relook at this; as you are saying, though, not losing track of the past, because that is also very important if you are going to go forward. You don't want to lose track of the stuff in the past that is certainly there also in continuity.
I think there are some changes here. There are some boundary kinds of sets of conditions that then could drive the thinking.
DR. STARFIELD: I want to shift gears a little bit.
MR. SCANLON: Is there a time frame, Barbara, before I leave here.
DR. STARFIELD: A time frame for what?
MR. SCANLON: The workshop or are we just sort of putting it --
DR. STARFIELD: Do you mean, is there a date that we are targeting for?
MR. SCANLON: Yes, a month or --
DR. STARFIELD: I guess in the winter.
MS. GREENBERG: Ed Sondik was interested in the spring.
DR. STARFIELD: Which year?
MS. GREENBERG: It is sort of like the HIPAA regulations. We were going to have this workshop last October.
DR. DETMER: This doesn't need OMB clearance.
MS. GREENBERG: Then I think both Dan and Barbara felt it was important to commission background papers. Everyone agreed with that, so it seemed like the fall.
Now, tomorrow morning, what we are going to talk about somewhat is, are there things that really need to be done before you have a workshop.
There was too much focus, it seemed on the workshop and not enough on what are the goals and how are we going to accomplish them. A workshop might come more at the end of the process and we have to do some other things, some smaller things before. So, I would say that it isn't going to be before the fall.
DR. STARFIELD: I have got to switch topics, because we are going to lose Paul if we don't do that. It has always been the mandate of the committee, and also the interest of this committee to deal with the survey data. Paul has done some thinking about survey data. Ed Sondik, from time to time in the last year or two, has identified things that he thinks are important. There are additional ones, I think, that we probably think are important as well.
In a memo that Paul wrote me last week he identified three things: first of all, timeliness of data; the second is state data, things that are useful for states, particularly with a lot of these reform efforts which are state based.
The third one is, you know, what is the role of the committee in terms of survey integration that the data council has been dealing with.
There are a couple of others that we might want to throw in the hopper. One is what is the relationship between the facilities based surveys and the HCFA efforts on post-acute care.
The last one is the monitoring of specific programs like welfare reform and more specifically the children's efforts. Paul, why don't you give us the benefit of your early thinking on that.
Agenda Item: General Surveys and Population-Based Data.
DR. NEWACHECK: I have been using a series of surveys from the National Center for Health Statistics for the last 15 years or so, particularly the population-based surveys, like the National Health Interview Survey and the National Health and Nutrition Examination Survey for various purposes.
Barbara asked me to think a little bit about what I considered to be some of the problems, or the potential areas where we might improve some of those surveys, particularly those surveys that are based in NCHS.
As she pointed out, there are really three areas that, in my initial thinking, I identified as areas where I think there are good prospects for improvement.
The first of these is the timeliness issue. Typically the surveys that come out of NCHS are made available about two to three years after the final data is collected. That process requires some editing and there is some internal work that goes on before the data are released.
The problem with that is that the health care system is changing so quickly now, with managed care and everything else going on, with state policy changes and such that Barbara described, that data that is three or four years old, as it often is, is often not very useful.
We can use health insurance as I think a good example. Right now in the National Health Interview Survey there is a supplementary series of questions about health insurance that are very good.
They are good questions. They allow you to link people's information on health insurance to all kinds of other variables like health status, health care use, and the like.
The trouble is that right now we only have 1994 data. The 1995 data will be available soon, but we are talking about data that is almost four years old, that we are working with.
The Census, or the Census Bureau, in its current population survey, is able to turn their data around in six months. So, when people talk about the 40 million or the 41 million uninsured people, they are always using CPS data, never using NHIS data, even though the questions on health insurance are far superior in the health interview survey compared to those in the CPS and even though the National Health Interview Survey provides the opportunity to link data to all kinds of other domains of health.
So, one of the areas that I would like to see, or at least think about whether we could help out as a committee is improving the timeliness of those data releases, so that we can be competitive with CPS, so that people will start using data from the center more than they have in the past, and the visibility of the center will increase and the quality of the information, in the eyes of the public, will heighten.
I really do think -- I don't mean to bad mouth NCHS in any way here, because I am a big fan of the agency and I think they are a great resource for us -- I just think they are under-utilized and partly it has to do with these delays.
Some of that will be corrected when the computerized personal interviewing is fully implemented. I still believe that there will be gaps, because there have always been gaps between when data are released from CPS and from HIS, you know, over the years before there was computerized interviewing. Even now, I think there will continue to be after.
Some of that has to do with budget and staffing. A lot of the agency or business heads are absent, or there are staff that have left that have not been replaced in the agency and the agency has seen some budget cuts in real dollars as well.
All those things, I think, have contributed to lower morale and to a slower release of information. So, I think that money could help out. That is one direction to improve things. Staffing could be improved in terms of filling vacancies, that sort of thing.
I think there are probably other ideas that we might think about that would be helpful to speed up the release of information and put NCHS back on the map in terms of policy relevant and useful information in this changing health care system. So, that is one area.
The other area, or the second area, that I was thinking about has to do more with state level data needs. Almost all the data that is collected by NCHS is designed, at least in the surveys, for national purposes, national estimates or regional estimates.
Very few people use the regional data. The national data obviously has a tremendous amount of value. But what we hear over and over again is, well, what do you have at the state level.
There really isn't too much out there at this point. We could do synthetic estimates where we take the national regional estimates and make state level estimates, but it is synthetic, and it is a problem.
Now with welfare reform happening, with this new balanced budget act program with expanding health insurance for children, with all kinds of things going on at the state level, we really need to have this state level data.
Some of this might be dealt with by the 1995 redesign of the National Health Interview Survey. Originally, the health interview survey was designed in a way that precluded doing real state level estimates, because some of the sampling units crossed borders and such, which made it very difficult to do state level estimates.
That has been changed now, but the sample size is still too small in most states to present valid estimates. Even if you combine two or three years, you still don't have a good solid base to work from.
So, the question is, is there some way to improve our capacity to get state level data from surveys like the National Health Interview Survey, again, so NCHS could be put on the map as being a real resource.
Right now, it is CPS. It is the current population survey that everybody uses when they want to know what is happening at the state level. But this current population survey has virtually no health data in it and no health care utilization data at all.
So, I am not sure, again, what the answer is here. It is expensive to add sample size, particularly in small states.
There might be some possibilities for doing some telephone supplements, and in fact, I think that has been on the table as a discussion point under this new Balanced Budget Act, but maybe in an even more integrated way with the health interview survey.
In any case, that is another area where I think it would be very valuable, or we might be able to make a very valuable contribution to increasing the value of NCHS data.
Then the last point that I raised with Barbara was the issue of integration and linking of survey data.
This is an area that people have complained about for a long time, that we do all these surveys in HHS on health related issues, and there is overlap, there is different questionnaire wording on the same topics, that sort of thing.
So, we end up with different numbers. If we look at, say, for example, the number of people without health insurance, we can get one estimate from the survey of income and program participation, another one from CPS and yet another one from the health interview survey, and the like.
That is changing now, and as Jim pointed out this morning, there is this survey integration plan going on now within DHHS, in particular to use the national health interview survey as a sampling frame for many other surveys including the medical expenditure panel, possibly the survey of family growth, other surveys.
There is some integration that is now occurring and I think that is very positive and will be very helpful, and probably deals with some of the past problems that we have had in this area.
The area of how do you integrate the provider based surveys, like the nursing home survey, the hospital based survey, the ambulatory care surveys, that is a whole other area that probably needs some additional work.
Then there is the whole question of, well, what if you go beyond the purview of the data that are collected at the NCHS, for example, out to census and other places. Are there ways to integrate those data sets in ways that would save money and perhaps produce better data.
DR. STARFIELD: In particular, the HCFA provider based ones, as we heard yesterday.
DR. NEWACHECK: Yes, so that is another big area. So, those are three areas that, in my thinking, are ones that if we could make some improvements, we would be much better off in terms of having more valid, representative and current data to look at for policy questions.
I am not sure how this all fits in with the rest. I think it might relate --
DR. STARFIELD: It does all fit in.
MR. SCANLON: I think it relates to the workshop issues, to, what are we trying to accomplish, what are the issues and what are the needs. Can we restructure what we have.
MS. COLTIN: I was just going to say, another thing to consider is new data that is going to be becoming available, particularly through HCFA, in terms of the consumer assessment of health plan study data, where HCFA is going to be surveying 140,000, I believe is the number -- 136,000; I rounded -- members not just of health plans but also fee for service samples in six markets around the country. So, there would be some comparative data as well.
Under the Balanced Budget Act, they will also be collecting from health plans, beginning this year, hospital encounter data for all the Medicare beneficiaries and then, within another year, outpatient encounter data as well.
DR. STARFIELD: We heard about that this morning.
MS. COLTIN: So, I mean, the potential exists actually to link those data sources and think about whether some of the patient reports of quality, in fact, coincide with objective measures of quality that might be defined from some of the administrative data bases as well.
MS. GREENBERG: I was just going to say that everything that Paul mentioned I think really falls under the rubric of probably population based data, of which there was a work group that, like our planning groups, was on paper but really hadn't been energized.
DR. STARFIELD: We don't even have a work plan for that work group yet, so we need to do that.
MS. GREENBERG: No, we don't. What was decided was that this workshop or this work with NCHS and the NCHS director around the division for health statistics would be the first activity of this population-based work groups.
There are two work groups within this subcommittee. The K-2 data standards one has been very active, to the exclusion, really, of really being able to focus too much on the other.
So, everything you mentioned, I think, and more falls under that potential, and Dan agreed to co-chair that work group. So, I think there is a lot of opportunity.
DR. FRIEDMAN: Paul, I do think those are three fine issues for NCHS to address. In terms of the state based thing, one of the things that we are going to have to deal with is that it seems that no matter what level you are generalizing to, the urge is always going to be to move one level down.
I mean, when we have state level data, we are told it is useless. We need to move to city level data. When we get city level data, often there is a need to move to neighborhood level data.
I do think that relates back to that comment you made about can we find -- this is not what you said -- but can we find cheaper ways of providing data that may not be quite as reliable but still we may be able to aggregate up to where we can have the geographic levels of comparison.
I do think, especially for NCHS, that is going to be a very important discussion to have, largely because of the expense of the methodologies.
DR. COHN: I was going to comment that I think actually this subcommittee may have some ideas about cheaper ways of getting data. In fact, it is most of what we have been working on for the last year around the data standards piece.
To more immediately comment about what Paul was mentioning, I guess I hear sort of two areas that I think need to be addressed.
One, this issue of 1994 data, data that is four years old only now being released, I find myself sort of shaking my head a little bit.
I am just thinking, that may be a letter from the committee.
DR. NEWACHECK: It has been out for a while, though. It has been out for about a year. The 1995 data are due out in another couple of weeks. So, it is not quite like it is just that 1994 just came out.
DR. COHN: Okay, well, maybe it isn't so bad, but if there is an issue of timeliness, it seems like a letter from the committee to NCHS --
DR. STARFIELD: Ed Sondik is very well aware of the problem. So, it is a matter of coming up with good solutions for him.
DR. COHN: In that case, it really does sound like part of the sessions that Dan is leading, to sort of try to figure out focus.
DR. NEWACHECK: At least as a starting point, you know, to bring it up in that context. It may require a longer term strategy, but I think it certainly fits into the workshop concept.
MR. SCANLON: It may be, after pressing for the workshop so vociferously here, there may be some more interim way of advising NCHS without the full blown -- I mean, I would continue with the workshop planning, but there may be three members of the committee, for example, who could think about this issue.
Maybe we could reserve half a day for kind of a round table at NCHS, just because some of you guys have done a lot of thinking about this.
I am just worried that we barely spend $130 million in all of these surveys in HHS and this is a trillion dollar sector of the economy.
We are having trouble holding onto that amount. I think we are basically losing ground every year. As Paul brings up the need for state data, other groups bring up the need for minority data and state data.
There are so many different horses on HHS and other agencies, I just think sort of thinking through this and getting some thinking from outside like this at NCHS, I think, would help.
You could ask other agencies as well. It doesn't have to be NCHS alone, but I am just worried that we are going to lose what minimal investment we have now, with the press of other business in the health sector. We are going to lose some part of this investment in the surveys and on population-based data.
A year from now, to be honest, I think it is just too late. So, we can continue with the workshop planning, but I think we need some -- I think some advice is needed even before that.
DR. STARFIELD: Now that we have got an enriched committee, I think we are in good shape. Thank you, Paul. I want to turn now to the claim attachments.
I didn't really realize that John was not going to be able to make it to the meeting. I guess he had an emergency, a legislative session or something like that. So, Simon, could you bring us up to date on where we are? We had a wonderful set of hearings at the beginning of February.
Agenda Item: Claim Attachments.
DR. COHN: Actually, I am not going to try to bring people up to where we are. I am actually going to try to run this part of the meeting around -- this is really stuff with the K-2 work group, in terms of trying to make further progress on this.
What I want to do is just remind everybody where we were, at least in my view, at the end of the last session, and probably ask Clem to talk about what has happened since the last meeting, as well as clarify any misconceptions I have, to the extent that he can.
Finally, I think really the big issue is trying to come to some sort of agreement about what next steps we need to have.
I wrote down here the issue of coming to closure. I don't know that we will come to closure, but I think we can figure out a plan for next steps that will eventually get us to closure.
I think most everyone was at the last session. I think you could just be reminded that we spent two days talking about claim attachments, heard some very, I think, aggressive and potentially exciting plans from HCFA and Steve Barr about testing this fall -- actually, testing this spring 17 different document types being done by a combination of a work group that includes X12 and HL7 participants and others from industry.
There is a plan this summer to develop an NPRM that was going to be written and finished sometime by the end of this summer.
I was sort of struck by the fact that I didn't know what we were writing an NPRM about. I think that may be an overstatement.
I found myself being a little concerned about both the pace as well as the definition of exactly what we were specifying.
Subsequently we have had further discussions that haven't necessarily gotten me down to ground level on that. I am actually hoping that Clem can further elucidate what has happened so far and we can talk about what is next.
DR. STARFIELD: That is only three weeks, Clem.
MS. GREENBERG: I was just going to mention that Phyllis De La Varis, who is the co-chair of the claim attachments team is actually out there and she might want to join us and might have something to add.
DR. COHN: We will ask Clem to add and also ask Phyllis if she has something to add.
DR. MC DONALD: I didn't make slides and it is fairly complicated. The problem with this whole business is this word attachment just conjures different images in everybody's head, and none of them -- you can't bet on any of them being right. It just isn't necessarily what you think it is.
I think it is easiest to think of them in categories, and there is a set of them that really correspond fairly closely to sort of standard and -- and if you will forgive me -- dumb forms; I mean, where someone made up a form and there are words on them and there is a field and something goes in there.
In some cases, it is pretty well defined what goes in there because it comes from a long HCFA elaboration, and in some cases it really is hard to tell, absolutely, reading it by itself, what is supposed to go in there.
So, that is one category, and those are sort of the most concrete ones. The ambulance form is an example and it has things on it like -- well, I can tell you more what they are getting at is things about how far they drove and how far they drove to a second hospital, and if they passed by one they should have gone to, and why did they pass it by, because you are supposed to go to the closest hospital, and those kinds of questions.
Some of them are fairly specific multiple choice answers, and some of them are free text. You just sort of say what you think.
There is another category that I have a more difficult time characterizing, or I will, but it is more or less an identifiable element from the chart. I say more or less deliberately.
So, there are things like the last creatinine value, or the last hemoglobin result, or the discharge summary.
Now, there is a tangle, having said that, because the attachment by itself doesn't specify which one of those. There is an implicit connection about which one they want, but usually it is the one related to the claim.
You can't just do an automated thing like give me the last hemoglobin within the date range of the claim. Well, you could for that one probably, but there are other ones like give me the pathology report.
Well, they don't say which pathology reports. Pathology reports usually aren't named formally as the kind of specimen they are looking at. That is buried within the report, so especially thinking about computers.
The implication is there is a charge code that says something. Maybe it says liver biopsy. Then you would probably want to look for the pathology report of the liver biopsy. But there is a challenge there to making those links in an automated fashion. The claim attachments has wrestled with many of these things.
Let me get to the third and fourth category. There is a third category that I think of as more of a retrieval, in the sense that it will say, give me all of some broadly defined thing.
The biggest one is like give me the chart, give me all the parts of the chart. Another one, there is something in the ESRD, the end stage renal disease protocol that I am still not quite sure what it meant. I think it means, give me some lab results you think I want, or give me all of them. It also specifically says, give me the one before that. So, it wants the last one and it wants the previous one.
So, it is clear that it wants the last one and the previous one. It is not clear that it wants them all. I think there are only four of them. So, you have to pick which ones you actually stick on those slots.
Then there are other ones like that, such as give me the detailed billing. I think of it as a statement, you are pulling out all of it and they are not saying which ones they want. They are just saying give me all that sort of thing or give me the orders. So, you give them all the records that have orders in it.
The second and third one are highly automatable, at least in theory. There are those kinds of data sitting around in computer systems somewhere.
The fourth category is what I think of as fill in the blank questions. In fact, as you read them on hard copy attachments, they are actually just random questions, really free form questions.
I think there really is a structure that you could limit down the scope. It will be like, why did you order X of these procedures within Y months, as maybe a general form, where they find that you have done three bypass surgeries in six months and they think that is suspicious.
The other one that they will often think is suspicious is why did this patient get these three procedures four days after they died. Well, there is something to that, you know, that it might be suspicious. Typically what it is, is a date transcription or some kind of screw-up in the dates that are put in.
There is a series like that, and you see the same pattern. They are not always phrased the same way, but they usually have one or two fill in the blanks that you vary and they come out the same.
Finally -- well, not finally -- there really is kind of a running off tangle of attachments that mostly the group doesn't want to worry about and thinks they should be handled a different way, or many of which should be handled a different way.
Resolution of multiple payers payment, that is an attachment. That is not using the right word -- coordination of benefits. In truth, it is forced on the provider to do that coordination, when the payers have all the computers and in theory could do all that coordination. There is some movement afoot to try to solve that through some more direct communication.
There is a whole bunch of things that are like, give me a copy of. It ranges from license plate, to social security card, to they just really want the detail. Those aren't really a high pressing thing right now, and we will just kind of hide those.
You name it, you will find some other things in there probably. I like to say about the medical record, that the medical record room is a great magnet for anything with a patient's name on it.
We once saw a McDonald's hamburger wrapper which had a patient address on it, and it got into the charts, stapled in there beautifully, because it had the name on it.
Attachments are just about anything else. Literally, it is anything you can staple to something. Plus, now they are talking about images and the rest.
MR. MOORE: Where are you now?
DR. MC DONALD: I just want to make sure that people understand, this isn't the medical record. It isn't the this, it isn't the that. It is this big protean vague set of things which is in some ways hopeless because the providers don't want to have to deal with any of them and the payers don't want to change anything they are doing.
The combination to me is, okay, we are finished. There is nothing we can do. But HIPAA will prevail.
Getting back to the specifics, really we have to deal somewhat differently with the different categories. Let me tackle the forms one, sort of in two ways, the way I would like to handle it and the way the insurance payers would like to handle this.
MR. MOORE: Who are we doing this for?
DR. MC DONALD: For both. I look at it as a symmetrical view. There is a sender and a receiver and we are trying to please them both. We don't want this to be an automatic machine that chases the provider all day begging him to spend all day writing things down and doing automation. I think it was called administrative simplification.
I know you guys have a bigger clock, but I am on the committee. So, try to keep remembering the poor provider.
More or less, what we define is, we define an attachment and give it a number. I am not going to burden it with any particular brand of numbers.
What is important is that these numbers be done with care. I mean, the names associated have a formality to them, so that people know what the heck they are, because the labels on the forms, you cannot guess necessarily.
It will be something like date. Date of which. Just as an independent reader looking at the form, I cannot guess which three or four options a lot of these things are. There is often a book behind it. We can do better with names.
So, we will have an attachment and give it a number. This will be like 678. This attachment, then, will have other properties such as a name, that would be somewhat descriptive, maybe some commentary, and then a list of the pieces of the attachment and the type of the attachment.
So, this is a question. Then what you will have is other attachment all of which, in one model -- would be the numbering scheme and the individual questions would be numbered.
We probably have who wanted it, some way to track it, where these came from and if you want to change it, who you have to argue with about it, or something rational.
So, the questions, what I like to think of these as -- let's say the question is hemoglobin. You want to know the last hemoglobin result.
Then I would think of this as identifying a result in hemoglobin. There happens to be a LOINC code for hemoglobin, but that is not the important point.
It would come along with a value, a date and a unit and maybe some other stuff.
Now, in the historical attachments, they are very irregular about how they deal with these different pieces, because they are just forms someone makes up and says, send it to me.
Sometimes you will see body weight pounds. It is supposed to be body weight in pounds. That is the name of the variable or that is the name of the question. These are in real attachments from different parts of the same world, the same uniform world. In other places, body weight kilogram. I think the ambulance has one and the ESRD has the other. Another one has body weight, no unit, and then there is another question, body weight, units, that you put together.
Now, this is maybe not necessarily gold, but my thought on all this is to put body weight is body weight, and you just send the stuff you need for people to figure it out.
So, the net context, if this was body weight, you would have a field. This thing always has a place for units, but you would specify in this question which of those pieces can be sent, in case people are typing them in. So, you could at least accommodate the current kind of messiness.
The more general structure is this result, the measurement or result. Then when you want all of them, the same thing. You have a way to specify in this question, I want all of them, not just the last one. And the same thing, you just give more of these.
Now, there is another model percolating in this committee right now and that is that we basically just strictly imitated these questions for skill positions, and we won't get anything better out of it, but we would be able to do things just exactly as we do now except that, I thought that wasn't that good, and that is why we are doing all this.
There is a little bit of discussion -- maybe a lot of discussion -- about whether we can transmutate from this kind of chaotic forms based world to a more general results based world.
Now, there are still some other types -- the list of drug orders and all -- that doesn't fit this model. So, we have another type to deal with that model.
Then we have still other types that are just new types of questions, like how many miles was the patient driven in the ambulance. Those would new invented ones.
What they are not trying to do, when they send us a copy of the emergency report, no one on any of these committees is trying to dissect that into its full detailed complexity.
They are simply saying, copy the bugger. So, if you have got a xerox machine you copy it. If you happen to have dictated it, you copy it. There is no intent to try to structure it, except for these existing forms types that do exist. They are part of the HCFA billing stuff.
The ESRD one, which is the one that fits best in this model, says that they want the creatinine, they want the hemoglobin, they want the BON, and they want dates for each of those and they want the erythropoietin dose. That is all obviously to kind of calculate whether they are using this expensive stuff right, because there is a theory that you don't keep giving hemoglobin if the hematocrit is above 35.
Then there are these fields that aren't quite decoded but people are going back to their payers and sorting this out.
There are these fields that say lab result one, lab result two, lab result three, and previous lab result one, previous lab result two or three. Actually, they don't say three or four, but somewhere there is a note that says you only have room for four of these.
I don't know whether that is give me all of those or give me the ones you think I am going to like.
DR. COHN: So, Clem, what is going on right now. Let's try to synthesize what you are saying and bring it up a level of abstraction. You are currently discussing how you are actually going to structure the internal parts of the standard.
DR. MC DONALD: We are trying to describe how you would at least model, in a flow sheet, or a spread sheet or a data base the pieces you would need to describe this.
Then there are some judgmental factors that some force is pushed on them, that when you say it is this way, then this implies certain things.
I should also say that there is a really hard focus on, I think, only four attachments right now. It includes ambulance, this next thing is what they call outpatient therapy, physical therapy, occupational therapy.
MS. DE LA VARIS: Outpatient rehab, I think is what it is called. And end stage renal disease is one.
DR. MC DONALD: I have got them to add on and maybe we will deal with discharge summaries. Those are pretty simple things. We can have a name here that identifies that piece of the chart that you can actually pull out.
Now, there are a lot of technical issues related to, you get this request and how do you know which things you want.
DR. COHN: Right, and I don't think we need to deal with that part right now. So, basically they are down to four pieces?
DR. MC DONALD: Four or five attachment categories.
DR. COHN: And they are in the process of writing the internal part of the standard right now.
DR. MC DONALD: Yes.
MR. MOORE: What portion of the attachment requirements does that represent; 10 percent, 50 percent, 40 percent?
DR. MC DONALD: Essentially, this is a place where we are chasing denominators, because nobody knows, literally, what are all the requests from everywhere.
This does not include Medicaid state requirements, which opens up another universe, literally. Now you have got all this kind of stuff all over again.
I think the theory is, we have got to make a beginning and then maybe the pattern will get easier and one can do this faster.
It is challenging and tricky. Even the ambulance attachment, which if you look at a form you can spot it as an ambulance attachment. But they all have different nuances.
What is it going to take -- we would like to say -- I think the core of the standardization is to take the elements that are now required as distinct elements and make a catalogue of those things.
Then you say, this is the catalogue you have to use, and you have to say the question the same way for this kind of information. I think we can get there.
MR. BLAIR: Can I echo the previous question, though? I think there is some difficulty gaining realistic expectations on the part of a lot of us, in terms of what we will be able -- how much of a dent, the standards that we specify a year from now are going to make in the overall volume of claim attachments.
DR. MC DONALD: You have got to ask that question a couple of different ways. Most office practitioners, as we heard from the AMA last week, do not have automated machinery.
I think that five percent of the small practices, less than three or four. Twenty percent of 20 in a group or larger, and then it got bigger as you got up to the larger groups.
Hospitals, it may be 40 to 50 percent will have opportunities to send some of this stuff electronically.
So, having defined this rule, all you can do in a first epoch is really make it easy for those who are already prepared to do this.
It will take them a year, probably, or six months, for the vendors to line up and deal with all these little details.
Then when you are talking about volumes from claims, the volumes are largest from the smallest, but the dollar volumes aren't.
I would like to talk in dollar volumes, and I would guess that we could probably help 10 to 20 percent in a two-year time frame.
MR. BLAIR: Is that with the four out of 17?
DR. MC DONALD: Yes, but those numbers are os meaningless, because one of those 17 says, all results. They all have different nuances. There are really separate kinds of issues in each of them.
So, there is a whole lot of lumping that has gone into the counting of the 17.
MS. FYFFE: Could someone refresh my memory about how the rule is eventually going to be implemented? If we define, if we very closely and very strictly define the particular attachments, when can that definition be changed?
What if, heaven forbid, someone should game the system and HCFA should discover a new type of fraud -- we will have an exception; you have answered my question.
One of the reasons that this is so important is to prevent public and private payers from losing money to fraud.
So, we have to have the flexibility to be able to change this.
DR. MC DONALD: Here is the challenge. Let me first answer your question. My guess -- but I am not policy maker. I have no power in any of this. My bet is that all they will be able to practically do is find some of them and then a pattern for how to do it in the time frame we are talking about.
I think the flexibility that you are talking about has to be allowed the process so that it can learn its way.
The other side, I view this simplification as what we want to do is avoid having practices having to hire armies of people to review charts and to interpret and put into computers and the same on the other side, armies of people to read these copies that get sent in.
In this context, the only way that can be done is through computers that talk to each other. In that sense, the computer that would send you the answer has to be prepared to have that stored.
I can't see how you would make up questions, just totally new ones. Now you could say, parameterize them differently and I want to have all the results for the last six months, because we could define this in the beginning with parameters that could be adjusted with pragmatics.
I am having trouble figuring out how you are going to say, tell me the color of all your patients' eyes because you are going to find a way to catch them gaming, if you haven't told them ahead of time you have got to record the color of the patients' eyes. This is where this linkage is between the limits of the vocabulary.
One of the other categories of this that I call retrieval, one can make it reasonably flexible. You want to have all of something within a certain period of time based on a certain -- I think both sides have to accommodate, because it is too much work right now.
It is not like people die if you don't catch fraud today.
MS. FYFFE: I don't know.
DR. MC DONALD: Those of us who have been on the payer side of the Medicaid fiasco in most states where it was installed, where they do all this random stuff on the machine, everybody has gotten caught on that.
We didn't die. We complained a lot because it is money but it is not the same as patients' lives.
DR. STARFIELD: The main purpose is maybe fraud and abuse but I think patients have a lot to lose from this whole business of claim attachments, a huge amount. A lot of medical records are being sent because providers don't know what else to do, and because they are being requested.
A lot of the information is being used to deny patients benefits, because you find something six months ago in the medical record that makes this particular visit, this particular claim just not relevant because it is a pre-existing conditions.
I don't really think that we have given enough attention as a committee to the whole issue of privacy and confidentiality vis-a-vis information.
DR. MC DONALD: We have not talked about doing anything in any of the discussions ever -- we don't plan -- we have never seen it on a list, any of the mental health records. I think that would be a good thing to keep off the list for a long time.
MR. MOORE: I would just like to make one point of clarification; that in the Medicare program we have like 12 million admissions. I don't think we denied one tenth of one percent of the admissions that we had in the Medicare program, or from anything.
In the PROS now, we don't look at case review. We look at patterns of care. We get back to the providers to try to work with them.
We changed the whole focus on the inpatient side. One of the things we need to do on the ambulatory side is to look at this in maybe the same way.
Kathleen is right. As payers we have a fiduciary responsibility, just like you all have a health care delivery responsibility.
If we don't do that well, some of us get fired, or should be fired. I think it is because of the fraud and abuse and the distrust between the payers and providers that these attachments sprang up.
DR. MC DONALD: Could I just say something? We had a really good meeting and I guess I am basically born and bred a provider.
Working with the attachment group, who are all payers, they are a good lot. There are no people out there trying to -- everybody gets the idea that the enemy is on the other side.
At one of the meetings someone pointed out, if we separate out the pre and post approval or the medical necessity from the fraud and abuse, it does make the world get a little bit easier.
The fraud and abuse often is going to require a visit. You have got to go in and audit and look under carpets. There are things that require a physical presence.
So, some of the requirements that get very hard on attachments, some of those come out of this fraud and abuse, which should be maybe thought of as still manual. You are going to go out and look at things if you have to anyway.
MR. MOORE: And a lot of times when it gets right down to the claim and you go to the IG and they go to the Department of Justice and there is a lawsuit, just what is on the face of the claim is not sufficient evidence when you go to court in order to have that adjudicated.
So, you have to have this other documentation in order to build a case.
DR. MC DONALD: That is something that probably has to be manual for a long time because it requires physical -- people sign documents and you talk to the Secretary and you take positions.
DR. COHN: In the interests of not going to midnight, let's see if we can raise this up a couple of levels. First of all, the time check for everyone, the original agenda said 5:30. Are most people okay with that? I don't see how we can figure out what the next steps will be.
MS. GREENBERG: We will just go to 5:30 and have to have a different plan. That is my concern about what I said about 5:30. We will have to have a different plan about how to revise that document. If you are willing to stay from 5:30 to 6:00 to work on the document, fine.
DR. STARFIELD: We don't have any directives from John, unless you do.
DR. COHN: You are right. We don't have any direction, but we agreed at the last meeting of the claim attachments that we would take some time now to figure out what sort of next steps we needed to have. We are currently without any next steps.
DR. MC DONALD: Could I give you some other problems and that might stimulate some --
DR. COHN: Either suggestions or next steps.
DR. MC DONALD: I want to paint an even broader picture. How it is currently planned is that the request for attachment would come as a 277, for those who know the jargon.
That really then gives you -- this is the bill I am talking about, this is the patient that I am talking about, that kind of information.
That chunk, that leader information, would come back on a 275 along with this attachment stuff.
Now, the attachment stuff can be sent in a couple of ways, and this is still under heavy discussion. There is a minimal ability to put a few variables into the X12 leader, just all by itself.
There is also something called a BIN segment, which in 275 is mostly designed to be just a truck that carries BIN segments.
So, the BIN segment would be an HL7 listing of the results or the orders of whatever, and that is where some of the arguments come.
There are those who say, we just don't want to screw with HL7 -- this is both on the payer side -- and then some of the payers say we don't want to screw with two ways of doing it and the HL7 is the raw data and we would rather have the raw data. That is not settled yet but hopefully we will settle that, but you want to adjudicate that.
The advantage of sending it all as X12 is that the shipment, the trucking stuff is done throughout all these parties and payers.
By and large, this committee is reasonably happy with that harmonized compromise. That is one other problem.
I guess the other thing is whether one would go with a proposal that says -- and I think Steve would really like to have a concrete proposal that says, for these three attachments, this is how it is done and this is kind of a model, how it will evolve the other attachments.
There is the other approach, to just say very god and motherhood things. We will do attachments where -- sort of like the security proposal, only more general, or it may be somewhat specific. Those are choices, too.
I think that until you get a concrete proposal, I think you are going to have a tough time deciding.
DR. COHN: This is not the occasion to decide. This is the occasion to decide what our next steps need to be to decide. Do you have any comments? I know you have been very quiet.
MS. DE LA VARIS: Maybe this will help clarify where we stand at this point. We are currently going through these different spread sheets line item by line item.
We have met at HL7, we have met at X12. We have been mapping the elements on the spread sheets to both HL7 and X12 as well as to LOINC codes.
That has helped to develop the pattern that Clem was referring to for the spread sheets. We are meeting next week in Baltimore for three days -- Monday, Tuesday and Wednesday -- to continue working.
We are planning to try to start testing the different spread sheets with the providers that we have lined up to do so very soon. In the spring, we will do that.
This summer, we will hopefully spend going over the results and the feedback that we get back from the test of providers, and then developing the notice of proposed rule making that hopefully will be put out some time this summer. That is kind of, in a nutshell, where we are now.
DR. COHN: Do people have comments about that plan?
MS. GREENBERG: Could I ask one question? We were talking about 12 or 14 attachments at one point and then Clem mentioned four or five. How many are going into the proof of concept?
MS. DE LA VARIS: We would like to do the 17 for the proof of concept, hopefully, but we will see what we get done.
MR. MOORE: Clem was talking about categories. You can group some of those into --
MS. DE LA VARIS: Right.
MS. JANES: Clem, to the 17 map cleanly into your four?
DR. MC DONALD: No, my four are a subset of the 17. Even some of those could be counted differently. You could get four up to nine probably.
MS. DE LA VARIS: There is a lot of overlap. Like for example, patient name. That is something that is going to be carried pretty much everywhere.
MR. BLAIR: Somebody joined the conversation, but I didn't know who. Gail, but there was somebody who was part of the committee with Clem.
MS. GREENBERG: Phyllis DeLaVaris.
MR. BLAIR: Okay, Phyllis. This is addressed to both of you. Help me understand this. I got the impression, when I sat through that one day at the HL7 health claim attachment sig -- and this impression may not be right -- I got the impression that with the information that is available in coded form within a provider setting, that those four cases were the ones that had the greatest promise where we could wind up having some kind of codified information.
The rest of them, for the most part, were either why questions and you had to send textual information -- is that correct?
DR. MC DONALD: No, that is beyond the 17. Most of the 17 have some formality to them, either an existing form or some data base that is part of HCFA's requirement for a certain kind of billing, or people can kind of recognize them.
Like the one, the ER, we call it the TBNA, or the ER visit sheet, now they are saying, just send us the whole thing, whatever it is.
If you did have, like the ER docs are doing all the dictation on line, that would be a way that you could send that.
Now, it brings up another issue, and that is how do we send fax documents in this mechanism. In fact, there are a lot of places that are scanning the paper documents and they would have the opportunity to send them electronically.
That has just raised its head in the last couple of weeks.
MR. BLAIR: I am still confused with your answer.
DR. MC DONALD: Of the 17, most of them, if you have energy, looked like something you could do with them and some places would have some of that data stored computer-wise.
We chose the four because either they are good examples or they were more likely.
MR. BLAIR: Of the four that you picked, I know there is some codified data. But within the four that you picked, does that also include responding to questions where it is also saying, send me the progress notes?
DR. MC DONALD: One of the categories, what we call in quotes one attachment, what we actually said was admit note, discharge note and op note. If you got those right --
MS. DE LA VARIS: It all falls under notes and reports, the OR report or the op note.
DR. COHN: I am going to break in and make a comment here, and I think maybe I am moving into the area of what do we do next, which is really the key point here.
We have heard -- I have actually been hearing a couple of things. One is that everybody is a little confused about really what is going on. Dr. McDonald and Phyllis are obviously very clear. But generally, everyone else is a little confused.
I mean, I know personally that I feel that somehow somebody has stuck the gun to my head to make me move much quicker than I would ordinarily on this.
I have personally, as I have thought about this thing, thought that it makes a tremendous amount of sense to do a proof of concept process. I personally fully support that.
What I do have a problem with has to do with finishing the proof of concept up in late spring or summer, writing an RPM over the summer, putting it out on the street, and then sometime in January 1 or whatever of next year have it become the national standard way that all of this will be sent out.
I personally think that the committee has a responsibility, when we dealt with other standards, we took public testimony.
We actually had something that we could show to vendors, providers, payers and others, that we could then ask them for their input about.
We don't have any of that here. What we have is something where we have a process -- and I fully support the process -- but we are somehow expecting that the country, on the basis of that process, will be able to approve, in this relatively short time frame, the results, without asking them.
I guess as I listen to all this, there is nothing against claim attachment. I guess in terms of next steps, I think that there needs to be some comment made by the subcommittee or maybe by the committee to move things along, perhaps in a slightly more reasoned way.
MR. MOORE: I don't think that you have any less on this than you had on the 837 that was presented at the table.
The 837 was so long and had so many data elements, I don't know that anyone who testified to the committee could tell you everything that was in it. The same thing with the 275. It has been out there for a long time and everyone has a different view, as Clem said. It depends on whether you are HCFA or Medicaid or a proprietary payer or some other.
What we are trying to do is pull all those thoughts together here, and you have got different people from the community trying to pull that together. It is not just a HCFA person.
DR. MC DONALD: Could I comment? I have some of the same anxiety that you have. I mean, I am basically always a little bit anxious, so that may not mean a lot.
Basically there is not an arguable document fully specified in front of us. It is still sort of -- frankly, there is disagreement on this committee about how this stuff will work still.
Having said that, I don't see -- the testimony is not what you need more of. You just need to let the committee come up with something, either shoot it, bury it, or else say, oh, that is not so bad.
Right now you just don't have it yet. It hasn't been born yet. It is cooking and it is close to birth.
DR. COHN: Let's hear from some other members of the group. I sort of went out of my role here by making a straw man, but go ahead.
DR. FRAWLEY: I feel very much the same way, in terms of the fact that we really don't have something that we can put our hands around.
I mean, I think having the proof of concept project going forward will be helpful. I am still struggling with some of the next steps in terms of, you know, our role and also, again, for industry in terms of being able to react to something.
That is where I am -- we do the proof of concept and then there is going to be an NPRM written, and I am not sure on what; the results of the project, pretty much?
DR. MC DONALD: May I suggest, I think the key thing is, when we put together this implementation package, and it is going to be for four -- help me; I am not sure, but I think it would have to be for only a limited number of these because there is no way in the time frame to specify this whole thing.
You know, if you go out and ask people what they want, I will bet you a good bottle of wine, you go ask the providers, we don't want that, we don't want any attachments. They said it very clearly last week.
The payers that we have talked to -- the payers are the most active in this. They say, oh, shoot, we don't want to change anything. They said it right out loud. We are not going to change. We want to just get our same silly forms in. Well, your forms are different from the other guy's forms. Yes, but we can't change our forms.
We really have this conundrum. If we are going to try to come to a common middle ground, there is going to have to be some give and take. If you come to people and say do you want it, they will all say no.
You wouldn't have to put anything on the table. They are going to say no ahead of time. But the law says they have to do this.
DR. FRAWLEY: The provisions of HIPAA say that this committee should be making a recommendation to the Secretary, and that the Secretary has to adopt a standard next February; correct?
My concern is, what are we going to recommend to the Secretary, and in turn, what is the industry supposed to be adopting. This is where I kind of see this from the proof of concept project going out.
DR. COHN: Dr. Braithwaite, maybe you have some insight here.
DR. BRAITHWAITE: I would just bring up two points. One is that standard setting is a never-ending process. We should not go into this with the expectation that the standard we adopt in final form a year from now is the end all and be all for attachment standards.
A useful function for this committee might be to advise the Secretary on what would be a reasonable first bite at a standard for claims attachments, given that the expectation is that if it works, they will come.
Then we can, over the years, gradually evolve that standard into something that takes on more like an 80 percent rule for attachments, and maybe even further as time evolves.
People haven't had any experience to speak of with electronic attachments, whereas we have had experience to speak of with respect to claims for 10 years or more, and some of these other administrative transactions.
There has been some experience out there. Some consensus has developed within the industry, although that has been focused a lot in the last year.
I would suspect that if the committee could recommend to the Secretary something along the lines of the kinds of things that Clem is talking about, we are going to have to hear from some more people before we can write that recommendation specifically.
But if we can say, here is a subset that is thought out well enough that we can get our minds around this as a starting point, let's do that with the full expectation that this is not the standard for all attachments. It is the current state of attachment standards, and move on from there.
DR. COHN: Thoughts or comments, please?
MR. FITZMAURICE: I would entirely agree with what Bill said. I have a caution, though. I wouldn't want to take away the pressure from the standards developing community and the payers to say, well, not as much is expected of you; therefore, slow down.
I want to keep the same sense of urgency of the law, to keep the ball moving ahead. I would say we are not a plane here, starting to approach take-off speed saying, do we abort or do we not abort. We are not even in the waiting line.
We have a balsa airplane that we are going to try to throw out and see if it flies. If it flies, then we will quickly build a large one and then we will quickly get on the waiting line and then we will get on the runway, for at least one component or for four components.
I want to keep that model building going, but for us to make a decision right now and say it is go or no go, we don't have a plane on the runway yet.
We shouldn't worry about that, as long as we see very reasonable and very fast progress being made, relative to all the other standards in the world that I have seen, with attachments.
It is a very hard job, and I would just like to keep the pressure on them to do good work, which they have, and recognize that it is going to take some time. Let's not tell them that.
DR. COHN: Other comments from the group?
MS. FYFFE: Yes, I have a question. There have obviously been a lot of man hours put into the proof of concept and that is only for maybe 25 percent of all the 17 attachments. What happens after the first four?
Is there a structure in place, over the next several years, for example, to start tackling the others and to do monitoring of the proof of concept to see if it works?
DR. MC DONALD: Firstly, they have actually wrestled with more than four. You know, let's some of these really, really, really done.
I think your point is well taken. I think there needs to be an organization or some entity or structure that can carry this forward.
The current sentiment is that you use the standards, like HL7 or X12, as the railroad train. But I think there is a sentiment to keep the codes outside of that process, so that they can be rapidly adopted, modified, changed, and do that in some collective that is really -- UUNC has been mentioned -- I don't even know all the letters.
MS. DE LA VARIS: NUCC and WEEDI.
DR. MC DONALD: They might accommodate it or maybe some government agent will accommodate it, or maybe who knows. There is -- you are absolutely right and this is probably something the committee should probably wrestle with, is some recommendations about the process.
If the skeleton works -- and I am not going to be on that until we get agreement and we do a little bit more work -- but if the skeleton work -- and the future work will mostly be trying to get a process so that you really know what these things are when you make up names for them -- I would hope it would be better than some of the processes for making up names in other fields.
I hope we have some formality in the naming so that people can read them and distinguish them.
MS. FYFFE: When I think of the other standards, the transaction sets, for example, there has been a decade's worth of work on the 875, et cetera.
I think of the employer identification number. That is an existing structure, the EI. When I think of the provider ID, hopefully there is a continuing structure for that.
When it comes to claim attachments, the very ugly thing that it is, I mean, we have to be very serious about who is going to do that over time.
MS. DE LA VARIS: That would actually be a very helpful recommendation for the committee to make to the group, would be for follow up and maintenance and so forth, who would be --
MS. FYFFE: It is not an insignificant amount of work.
MS. DE LA VARIS: No, it is not.
DR. COHN: In terms of the proof of concept project, when will it be either complete or done enough that we would have a clue about whether or not it flies?
DR. STARFIELD: When are you testing now?
MS. DE LA VARIS: We are hopefully going to have something in testing this spring. I am tentatively going to say -- I mean, I hate to be held firm to anything without being able to talk to Steve Barr first -- but maybe summertime we could give you some initial idea as to how our data is coming back.
DR. MC DONALD: Don't worry about that, because that will only be about three days before you write the NPRM.
DR. COHN: As I said, I feel funny because obviously I have a position on this, which is let's slow down and make sure we do it right, rather than just do it.
What information does the -- I have heard from some people the question of will this fly and knowing even what to recommend. What role does the subcommittee and work group what to have in relationship to that?
We have a number of options. We can just not make any comment, and decide it really isn't any of our business. These are possible options.
We can, at a June meeting, say that if it meets certain requirements, it is fine to go. We can decide to either get more information from the work group or from others, and then make a determination at that point.
There are probably other options that people can think of but how should we proceed on from here? We don't have any hearings scheduled on this topic, at least as I look at the schedule. We don't have any time until our meeting in June.
DR. MC DONALD: Can I make a suggestion? That would be that when a document that describes this richly and clearly -- and hopefully that is going to be available in six weeks or so -- is available, there would be an opportunity for the subcommittee to digest it and discuss it and critique it. That may stimulate additional kind of insights or goals.
MS. JANES: Is that likely, Phyllis? Will we have that opportunity?
DR. STARFIELD: That is in fact what we decided to do at the February meeting, is that in fact, Clem is going to be working on some things and then we are going to discuss it at the June meeting.
DR. MC DONALD: I don't want to make people nervous, including the committee. The committee is marching along and doing the X12 part. Then the naming of these entities -- I have volunteered to do a lot of it, and I don't know how big that is.
These are just the things that mostly are existing fields on forms somewhere. Then there is the business about all the detail which I didn't even get into, about how do you really communicate in an electronic way, that when you are sending this message with the 275, which really doesn't have much structure, it is really just a box car, how do you signal the receiving system. What actually do you pull out of their system.
I mentioned the one issue about the pathology one. So, what we will come up with is operating principles, that there will be a time window specified and you will get all the surgical pathology results, because there is no way to tell, through the semantics of that message for sure, which particular code or result you want.
So, things like that we are working through. Some of these things may end up being ugly. Now the question is -- two questions, because this is the only way to do this. This thing has to know this, this has to send this, and somehow you have to communicate those two things.
Maybe people will adopt the slower or faster in terms of actually sending this stuff. I would be optimistic that we do well on discharge summaries, op notes, those things fairly quickly.
They are fairly simple. They can be adapted more quickly than some of the other ones. Now it turns out that there is a surgery system that has the codes for the surgery attachments in it.
One of the proof of concept sites is going to be actually, that system is going to use -- it is a set of tools from a place called -- it is a long story, but anyway, they can mix HL7 and X12 and generate these things.
The interesting thing, even this discussion is changing history. Most of the translators now are developing the ability and selling -- at least claiming the ability to do HL7 and X12 simultaneously.
So, there is sort of a push. The dynamic is having an effect already, but we just have to be flexible, I think, and write these things in a way to be flexible to respond to reality.
DR. COHN: What is the interest of the subcommittee in terms of how to proceed?
DR. FRAWLEY: I think having the document to review would be a good first step. I think at this point there is not much we can do until that takes place.
It would probably be helpful at our June meeting to get an update from the proof of concept team in terms of how their work is going.
I guess it would not be until the September meeting, in terms of the time line they are envisioning, that we could maybe give real guidance to the team as they begin to develop the notice of proposed rule making.
I just don't see that there is much more that we can do at this point, unless I am missing something here. I mean, I think we would like to be more helpful but at this point I think it is kind of like there is a time line laid out for things to happen. Some of those things need to take place before we can really react, or offer guidance or input.
I guess report and see how the project is going and then maybe next fall we may need to go into more activity at that point. I don't see that we can do much more than that.
DR. COHN: Other comments?
MR. MOORE: I was just trying to respond. Your schedule, Kathleen, was really based on the full committee. This kind of effort could be done at the subcommittee level as far as getting it to a final point to make a recommendation to the full committee.
What we need is getting the document to you, having it reviewed, having the changes made that you can do, without having to have a meeting, but then having as much of the work done in preparation for such a meeting, so that we can see where we are.
Some of these things could be done through conference calls to share ideas, once you get something to work with.
DR. MC DONALD: I think that is all correct. I am really going to say something that is sort of tangential or separate, but it relates to the process of continuing this, and there are two parts to it.
One of them is deciding what should you ask on an attachment. Should you get agreement that all the players will ask it the same way.
The other one is something I am having trouble describing. It is this formality of how do you actually model the name.
The only place that it really goes on now -- that I know of -- SNOMED does some of that, NLM does some of that and there is the HL Sig thing with this drug thing doing some of that.
I can't get anybody to buy into this, but somehow we need this formality if we are going to have this work over the long haul.
When you just make up names just willy nilly, you can't tell after you get the first thousand of them, whether the 1,001 of them is already in there as the 850th, if there isn't some formality.
I am only suggesting if there is some way, instead of doing this on the strictly pure billing side, which is how I view the UUNC and the various NC things, if there is some way to get a marriage of some of these other operatives to bring that influence in without stalling it, I don't know how to do that but I would like to at least open that up for thought.
DR. COHN: I think you are bringing up an important issue, which is the issue of the overall maintenance and management of not only this one but a number of the other standards that have been identified.
DR. MC DONALD: The goal of the standards group should be get out of the business of having to do a lot of diddling and turn it into tables. The tables are what are managed.
This is a more conceptual thing. This isn't a history. There is no way in the clinical data to make up a new message for every kind of thing you send.
It might be easier if we can make these buckets correct the first time. They wouldn't need a whole lot of maintenance, but we would use maintenance as the semantics, the contents and the names of things, and the answer list and those kinds of things, which is something that you know something about.
I just would really like to sort of mix the cultures. The other thing -- this is really a long hope -- is that the tragedy of this whole thing of payers and providers is that they are like enemies.
One guy does something and says, do it, and the other guy says no, or finds some counter force, like I will just send you the whole damned chart and let you drown in charts.
If they both knew about what each other had, it could be made easier for both. You know, instead of saying, I want this thing answered exactly like this, well, what do they have already there that we could pull and we could do a decision on it.
The old story was you wanted to see do they have hypercholesterolemia. You make them answer this question. This wasn't a payer, but there were other contexts.
You send the cholesterol and they have that and it lasts forever, whereas every time you change the rules for what is hypercholesterolemia, your data goes bad.
If we could get this marriage, in a way that they both understand each other's models, I don't know how to get there, but that is the other reason, if we had the right kind of committees or meetings or groups.
DR. COHN: I certainly wouldn't try to answer that question. I would just leave that on the table as something that we have to deal with.
I guess what I am hearing from a number of people is the idea of obviously getting the document, is it the choice of the work group to have conference calls or hold the discussions in June.
DR. STARFIELD: I don't know that we can answer that now. I think we have to wait now for John Lumpkin in the discussion.
I think it is likely that we are going to do it. When we are going to do it, I am not sure we need to decide right now. It has to be before the June meeting for sure.
MS. GREENBERG: Well, wait. We are planning a meeting in Chicago on May 4 and 5 on the unique identifier. What about the middle of April?
DR. MC DONALD: Eight weeks would be better. You are saying do it at that meeting?
MS. GREENBERG: There is a possibility that we could carve out a few hours in the middle of that. It is a time that -- this is kind of bigger than a subcommittee meeting.
DR. STARFIELD: That is a question. Is that meeting more than a subcommittee?
MS. GREENBERG: I know both of the other subcommittees wanted to be involved. How many people actually go, potentially the entire committee could, but I think that is unlikely.
DR. MC DONALD: There is a risk, too, that if we do this all on our own and then drop it on the committee at large, this is messy stuff. It wouldn't be bad if they were exposed to it earlier and at least did some questions.
DR. STARFIELD: Yes, but we are exposing them only to the unique identifier at that meeting. You are thinking about having a separate work group or subcommittee meeting?
MS. GREENBERG: It could be an hour afterwards or something like that. We haven't set the agenda yet. We have two days set aside, a day and a half on the unique identifier? It is just that it is a time when people are actually in a meeting.
DR. COHN: I think what I am hearing is a number of different options. One which I think sounds pretty good, what I am hearing is that our next steps is obviously getting that document, reviewing it, we will either arrange a couple hours at that May meeting, whoever is able to be there, and/or do a conference call. It may require a combination of both.
We will basically, I mean, assess the situation as we go between now and June and see what sort of recommendations we can make and whether we are able to make recommendations in June or September. Does that summarize sort of the agreement that we have reached?
MR. MOORE: Nothing says that you have to agree with the work that comes out. Like, when we gave you the unique identifier for individuals, you all said you didn't like any of those options and you all went your own way.
MS. GREENBERG: And we followed.
DR. COHN: With that, noting that it is 5:25 or 2:25 California time --
DR. STARFIELD: We are adjourned for today.
(Whereupon, at 5:27 p.m., the subcommittee meeting was adjourned.)