[This Transcript is Unedited]
Hubert H. Humphrey Building
200 Independence Avenue, SW
Washington, D.C.
Agenda Item: Call to Order and Introductions - Ms. Coltin
MS. COLTIN: Some of us have them from the last meeting, but we have some new members so I wanted to be sure that everyone had the materials who we're going to be working with today. Let's start out by doing introductions around the table and then we'll get started.
My name is Kathryn Coltin and I work for Harvard Pilgrim Health Care in Wellesley, Massachusetts, and I'm the Chair of this Workgroup.
DR. FERRER: I'm Jorge Ferrer, I'm a Medical Officer at the Centers for Medicare and Medicaid Services. I came to CMS after completion of a Medical Formatics Fellowship at Johns Hopkins for Medicine, and I was assistant professor there.
DR. JANES: I'm Gail Janes from CDC, staff to this Workgroup.
MS. BEREK: Judy Berek from CMS.
MS. CARR: Willene(?) Carr, Office of Planning and Evaluation at HRSA.
MR. HUNGATE: Bob Hungate, Physician Patient Partnerships, and my interest, which is directed at co-management of health measured by outcome fundamentally. I also Chair the Group Insurance Commission in Massachusetts which buys health care for all our employees and dependents.
DR. STARFIELD: I'm Barbara Starfield, a member, no I'm not a member of the committee anymore, past member of the Committee, and this is my last meeting. I'm from Johns Hopkins.
DR. HOLMES: I'm Julia Holmes, and I'm from the National Center for Health Statistics, the division of health care statistics, and my primary responsibilities have included working collaboratively with other federal agencies in developing the national health care quality report and the national health care disparities report.
MR. RILEY: Tom Riley.
MS. JACKSON: I'm Debbie Jackson, the National Center for Health Statistics, and staff to the Committee.
MS. KANAAN: I'm Susan Kanaan, writer for the Committee.
DR. FITZMAURICE: Michael Fitzmaurice, Agency for Healthcare Research and Quality.
Agenda Item: Review Draft Recommendations to AHRQ
MS. COLTIN: Alright, well we have two agenda items today that are continuations from our meeting in September. One is that we had discussed offering some recommendations to AHRQ on the National Health Care Quality Report. And we had expected actually to receive from them a copy of the summary of public comments that had been taken on the Report but that never did arrive and so I think we're kind of on our own in terms of forming our recommendations.
We did hold hearings ourselves in July, July 25th in Chicago, and so those of us who were there did hear what some of the major stakeholders in the delivery system had to say about the Report. And so I've taken the liberty of going through and putting together some straw recommendations that I thought we could just talk about.
And in order to be able to do that with any rationale, everyone needs to have this document here, which is the preliminary measure set. And if you haven't seen it before I think the most important thing is to actually go right to the tables and they show I think, the pages are not number unfortunately, but the first table says measurement categories and it shows components of health care quality effectiveness.
What the report is doing is using a framework that was recommended by the IOM where they basically have adopted the foundation for accountabilities framework of consumer needs which has categories, which in this table represents the rows, staying healthy, getting better, living with illness or disability, and end of life care. And there's also a category called basics in this particular table, that category may not fit because it has to do with access, and this is the effectiveness of care table.
The columns are based on the quality chasm categories from the other IOM report, which look at the attributes of safety, timeliness, patient centeredness, effectiveness and so forth. And so what they have in this report is a table for each of those categories.
So looking at the effectiveness table here you see that the areas in which they've identified measures are cancer, chronic kidney disease, diabetes, heart disease, HIV AIDS, maternal and child health, mental health, respiratory diseases, and long term care. And then there are similar tables for safety, timeliness and patient centeredness.
And in each one of the cells they have identified potential measures and data sources and whether in fact the data are only available from a national database, that is could be reported at the national level, or whether there is a compatible state database so that these data and measures could be taken down and reported at the state level. And then whether there are other measures that may have been arrived at through a consensus process, and that may only be available for pockets of the delivery system, but not necessary for entire states, they may be less population based. An example of that might be some of the HEDIS clinical effectiveness measures that are produced by health plans but they're certainly not representatives of the entire population in a market area.
So that's in general how the Report is structured, or how the measures rather are structured. So in terms of just taking a quick glance over these, some of the recommendations that I have drafted for us to consider are based on some of what we heard in July, on some of what I observed in looking at the table, and also Barbara Starfield had shared with me some of her comments as well.
So the first had to do with the content areas within the effectiveness of care category. One of the recommendations was that the Report needs more balance, that the measure set needs more balance, that there are many adult measures and few child measures. That there are many physical health measures and very few emotional health measures or mental health measures.
DR. STARFIELD: What are you reading from?
MS. COLTIN: What am I reading from? These are my straw recommendations, they're just my notes. I'm throwing these out for consideration. And the main comment, the theme, is that it needs more balance, and what I'm then doing is just proposing examples of what I mean by that, the two that I just mentioned are the adult/child and the physical versus emotional health. The other is there are many hospital-based measures, fewer out-patient ambulatory measures, and those that are are focused primarily on prevention as opposed to management of acute or chronic illness.
There are many disease specific measures and very few population specific measures, so there is this one column called maternal and child health which seems to be an attempt to identify some measures that might cut across and be relevant to that population, but most of the others are disease specific. And as we know as people age, the likelihood that they have only one condition diminishes, and there are really no generic population type measures that would cut across conditions. Those would be, as an example, things like rates of follow-up of abnormal lab tests regardless of what your condition is, but are laboratory tests being followed up. So they're more getting at the process of how the delivery system is working, and they could be relevant to more populations because you can combine disease categories and you can combine screening, even if it's follow-up of abnormal screening tests.
So it's a really broad cross-cutting type of measure. The same would be true for things like medication monitoring, so whether what you're taking is Lithium for a mental health problem or an anti-coagulant for HO fibrillation
, you still need to have testing on a periodic basis to make certain that your levels are staying in control. So those are examples of population based measures.
And part of the reason I bring them up is I understand why they're not in the Report, the data that we need to measure these types of things are not widely available, but that's part of what our group is about is trying to say maybe we need to build the infrastructure to be able to report on these population based measures that are more broadly relevant to large population segments. So let me throw that one out there and anybody want to react? Have discussion? Does this make sense?
DR. JANES: I have a question for you Kathy. As you talk about these population based measures, this is a world that I'm less familiar with then obviously the preventative services world. Are there prototypes for some of these measures? You just mentioned two of them and I'm immediately wondering whether there are HEDIS measures or other things that we could point to or would we also have to draft some sort of an approach?
MS. COLTIN: I think that these are areas where a lot of work still needs to be done to develop measures. There are examples of these types of measures that are primarily disease specific, but not good examples of putting them together to build composites. So there's a measure, HEDIS measure being developed, it's not currently a HEDIS measure but it's being proposed as a HEDIS measure, which is looking at the rates of monitoring patients who are taking anti-coagulation. But the concept for that measure could be extended to many other conditions where when someone is taking a particular medication, there's a measurement for how frequently, so just at a minimum they are to be monitored.
DR. JANES: Whatever level is recommended for that particular intervention.
MS. COLTIN: Right, and you can combine then across medications by simply, it's the same metric percent monitored within the time, recommended time interval, so you can then combine. Another example of that would be something like stage of cancer at diagnosis, where instead of just looking at breast cancer you could actually combine and look at across four or five major cancers what is the stage, what percent are diagnosed in late stage or early stage. So in this report they actually have those measures on a disease specific basis but not, for instance, as an overall measure thinking about cancer in general.
So sometimes more detail is helpful and where you can break it out, it's good to do that, if you have large enough numbers and specific enough data systems. So I really don't take issue with the fact that they've broken out the breast cancer and the cervical cancer and so forth, but if in fact you have to leave out large groups of different types of cancers where the numbers are too small, and an aggregate measure would allow you to consider those as well, then I think it's worth pursuing the idea of an aggregate measure.
DR. STARFIELD: Kathy expressed my concerns extremely well. The measures that HEDIS uses are probably half a century old, most of the kind of process disease oriented stuff is nothing new. They have gone a little bit further than that, there are some things in here, the percent of the elderly who receive potentially inappropriate medications, that's patient population oriented rather than disease oriented. And they ask about a usual source of care. But I think it's our role as a Committee is to push the envelope a little bit, so that the next time there's a Quality Report we do things that are increasingly relevant in this age of co-mobidity. Patients don't have diseases, they've got multiple diseases, and the guidelines don't really fit.
Most of what we do in health services, there is one measure that is available that's not here, it's population oriented and it seems to me they should include the ambulatory, hospitalizations for ambulatory very sensitive conditions. That's a pretty good measure of access to primary care and it is available. It doesn't come from clinical records but it comes from stuff that we have.
The other thing that I think that we have instruments for, although it's not really been well, they haven't been really well incorporated into existing surveys, is the quality of primary care. They ask about a usual source of care, but that's not adequate anymore, but the usual source of care really should provide high quality primary care, and there are characteristics in primary care that can be measured. It's not enough to have people report that they have a usual source of care when that usual source of care is an episodically oriented out-patient department, it's just not enough.
And there are systematic differences between people who use episode oriented hospital out-patient departments as their regular source of care and a good source of primary care. We would have to add measures to the existing services to get that, but it's not a big job, after the fact it's been working on measures of a medical home --
MS. COLTON: With HRSA?
DR. STARFIELD: With HRSA, right, and with maternal and child health. So soon we will have those measures, very soon. And we ought to be ready for it when they're there, so I think there's a lot of strong constructive recommendations we could make about improving this whole process.
DR. JANES: What's the timeframe on this?
MS. COLTIN: Well, the deadline for public comment has already passed and AHRQ has basically said if we could get them comments within the next week would be fine so what I'm trying to do is get some agreement on a set of recommendations that we can put together via email and be able to approve.
DR. FITZMAURICE: Kathy? You mentioned earlier that AHRQ was going to get you a summary of the comments. Is there a reason why we didn't do that? Do you know?
MS. COLTIN: I think it just slipped through the cracks and then last week when I requested it Ed was out of the office and he delegated it to someone and I never heard back.
DR. FITZMAURICE: I'll follow-up on that. And Barbara, I wanted to ask you, are there measures of good quality primary care that might not be too difficult to collect?
DR. STARFIELD: Yes, there are full characteristics in primary care. You go there the first time you have a new need or problem, it is person focused care of a time, not disease focused care, so that right away let's out the heart disease specialist as your primary care. Comprehensiveness care, that means the place takes care or the person takes care, all of these that are common in the population except those that are too uncommon to maintain confidence. And coordination is that when people do have to go elsewhere, there is actually a measure of coordination in here right now.
DR. FITZMAURICE: Is this something you would ask the patients? Or you'd ask the doctor, or you'd ask --
DR. STARFIELD: I think you can get it from the facility. I know a lot about this because we developed the instruments at Hopkins, we have a comparable exactly parallel version for the provider as well as for the patient. It's not incorporated in the surveys, but it could be.
DR. FITZMAURICE: I think the direction that things are going and that what Kathy mentioned at the beginning about it has this but it doesn't have that, those are good directions, not a criticism, but this Report will give directions for the future.
MS. COLTIN: And that's the main intent here, is to recognize that in fact AHRQ is limited by what's available today but in terms of really putting together a comprehensive national quality report, these are the kinds of things we'd like to see in it. And some of what we can hopefully try to do in our role is think about how to improve the data infrastructure to be able to enable them to incorporate some of these types of measures in the future.
So I think that's really, our role is to push the envelope, to say we'd like to see things other kinds of things in there. And in order to do that we recognize that that may involve adding additional items to some of the national or state surveys, it may involve pushing the envelope on some of the HIPAA electronic transactions in terms of the types of data that could be included and collected. There are a number of different ways in which some of the kinds of measures that Barbara and I were describing could be implemented. But in some cases right now, the data elements aren't even widely available to certain measures, but could be.
DR. FERRER: I know that this sounds perhaps a little bit futuristic, but are there virtual, often times when we're presenting this kind, these ideas to the Administration, for example, CMS, often times there are, in this age of information technology, can we do a virtual demonstration of what that actually would look like? Because we have these topics come, given people to see exactly what we're getting if we were to increment our clinical data streams to the organizations. Clearly we could do a virtual simulation of this of what we have that we're calling patient centered outcomes now, and this is where we'd like to go.
If anybody in the room has ideas of kind of putting that together, feel free to give me a call because they concretely want to see some, there are models by which you can, and I don't want to say cost estimate, but get a better handle as we increase our sort of data requirements of the agency, how do we incrementally put a price tag so that this is what you get for this kind of information and if so, sort of wrap it up in not just a sort of an absolute model where you have, you'd have to all, because we tend to run to where we need all this world of information now. Incrementally we'd get this and MTR(?) approach, we'd get this and sort of a full scale adoption architecture we'd would get this. So anybody has ideas, my email is on the packet.
DR. FITZMAURICE: By simulation do you mean, let's assume a practice that has a wide variety of patients, what kinds of information would we collect from each patient and then as you aggregate it across different practices, you then are able to estimate the cost of collecting that information? Or if you have to ask the patient a question in order to get it, that may be more expensive than something that summarized what you're already collecting and then you can aggregate that across all primary care physicians and practices in the United States. You can look at the cost of collecting it and make some assessment of the value compared to the cost. Is that the kind of simulation you're talking about?
DR. FERRER: Correct. Also from our standpoint, the data obstruction efforts from a quality care organization standpoint are prohibitively expensive from a manpower and data collecting efforts. As we standardize these data elements, we always want to sort of see well, what's that ratio really look like, and I've been to meetings for years where nobody has really been able to come up with a conception model.
DR. FITZMAURICE: That might help because of measuring primary care physicians don't have enough patients in any one particular category that would be meaningful for them, so there's very little value for them to collect some of these statistics and at a summary level might be very important for a population.
DR. STARFIELD: No your name isn't in the packet. You're going to have to give it to everybody. Some of these measures, you know the ambulatory care, you don't have to collect data from anybody. Basically the data are already there.
MS. COLTIN: Some of them are.
DR. STARFIELD: And the usual source of care, the characteristics of that are easily meth's that could be changed in relatively minimum ways with the stuff that factor's coming up with. So this doesn't require collecting from individual offices, a lot of the stuff we're talking about is really population data and it comes from populations.
MR. HUNGATE: I want to comment on what Michael just said. Let me use a little background first because you folks don't know me. But I worked at Hewlett Packard for 30 some odd years, worked in marketing, I implemented quality improvement in the business that I was running. And the take away that I have from that is that the measures that are good are the measures that the people take themselves for their own self management of what they're doing. And I came into health care with the interest in electronic patient record as the enabler of that kind of thing.
Now the reason I bring this up, I don't know what the discipline is in this group, to think about that in terms of what are we doing that is going to be additive in that way in the reinforcing the work flow, which is then rollable to the population level, if you will, as opposed to which is additional work for somebody else. Which, my experience is doesn't work very well because it's artificial, and it's better done statistically by survey sampling, which is useful for population work but it's not part of improvement. Is my question clear?
MS. COLTIN: Yes, I understand what you're saying, and actually we'll probably get into some of those issues in our next agenda topic when we talk about some of the testimony, we heard about the barriers to measuring quality, and where we need to go from a data systems perspective, because clearly being able to build on data that are collected in the normal course of practice and self generated whatever, is a much more efficient approach.
MR. HUNGATE: It'll work.
MS. COLTIN: I think one of the issues we'll face as we look at that and that did come up in some of the testimony that we heard is that there are some aspects of care that are typically not collected well through medical records or where patients really are the best source of the information. So we need to sort that out and say here are the things that ought to be routinely collected as part of the care process, some kind of an electronic medical record would be the best way to go, but we need to think about how we get from here to there because we've been saying that for 20 years.
MR. HUNGATE: The reason I ask is the same issue you just raised. I started working hard on health policy quality improvement issues 15 years ago figuring that I would make some progress by the time I went on Medicare. It didn't happen. And so I've been trying to think about what I think the barriers are and I think some of the barriers, are at least there isn't a clear visible way of tracking of whether we're making progress, and so that's why I got interested in this activity, to say maybe that, this piece I don't know, this is maybe out of the context of what you're doing, but it seems to me --
MS. COLTIN: Not really, because I think that's one of the goals of the National Quality Report is that it would be produced on a regular basis and could be used if it were comprehensive enough and if it were granular enough that one could see the information at a level where it was relevant.
MR. HUNGATE: I guess what I'm partly arguing is that there should be some kind of an overlaid measurement that says is this collectable in the future information structure and does it become an automatic, it's in the value added to the question.
DR. FERRER: That is sort of my plea that if at CMS, quality improvement organizations which are designed as an effectiveness of the care medium to collect this data has been designed from a claims based financial viewpoint. We can continue to do that but then you can't be coming to these meetings and continue to hear we need more clinical data granular elements to look at effectiveness of care because those things --
MS. COLTIN: I'm just going to cut us off here because we aren't getting, we're actually getting into our next agenda topic, so I'm interested in what you're saying and it's all relevant but I really want to get through these recommendations as opposed to the solutions, what it is we think needs to be done and I think our report is where we're going to be talking about making some recommendations about how some of the kinds of gaps that we're observing in this report actually are reflected in the proposed national quality report, are reflected in a lot of the barriers and obstacles that we heard in the testimony for our own report, where the gaps are and what the data needs are for measuring quality broadly.
So they feed into one another quite nicely, but for now if we're going to make comments to AHRQ we really just need to deal with do we have some consensus. So I guess the first thing was I put on the table a recommendation about striving for greater balance. Are people comfortable with that as being a fair criticism and recommendation? Do I hear any?
DR. STARFIELD: I'm comfortable with it, I think obviously you are going to --
MS. COLTIN: We're going to go beyond that, yes. And we'll talk about the specific examples that I mentioned, adult versus child, disease specific versus population based, and physical health versus emotional health, those were the main areas.
DR. STARFIELD: And you will add the ambulatory care? And the quality primary?
DR. COLTIN: Right, those will be offered as examples of things that can potentially help to achieve greater balance. So does anyone have a problem with making that recommendation? Do people want to speak against it?
DR. HOLMES: No, I think that's very appropriate.
MS. COLTIN: The second recommendation that I want to put out for consideration is that using the IOM framework may limit the incorporation of important relevant content. And in this proposal AHRQ has actually chosen to go beyond the framework that was recommended by IOM to actually include some resource based measures. And I applaud that, I think it helps with the balance issue I think as well, but I think they could have gone further. So in addition to including some efficiency measures, which they have proposed and which are beyond the IOM framework that was offered for the Report, I would consider that they go back and take a second look at the fact framework.
Because they took a category which isn't that framework is actually called changing needs. And they focused on end of life care, which is one changing need. But throughout life we have life stages that each raise their own issues. So for example, the foundation for accountability has developed a measure on promoting healthy development in infants. Healthy developments in infants and toddlers is an important life stage issue. And whether the right things are being done during that period of time is important. Anticipatory guidance with teens and adolescence is a very important area and there are even the youth behavior risk survey and ZACT(?) has a survey and so forth, that measure those sorts of things.
Management of menopause, a life stage issue for women, very important, there is a HEDIS measure around that. So these are the kinds of things that I mean. Healthy aging, maintaining and improving functional status in elders, there's the HOS survey that CMS does, the Health Outcome Survey. So it's seems to me that why narrowly define it as end of life which is the very last life stage, why not think about these other aspects. And again, those are more population based in concept so it goes along with this measurement of population based measures, but not necessarily disease specific, they're something that everyone passed through.
DR. JANES: Kathy, it sounds like most if not all of the measures that you have just mentioned come from fact. Are those in the public record at this point?
MS. COLTIN: Yes, they are and a lot of them are actually being used by the Medicaid and SCHIP program, just like the promoting health in the adolescent survey, and a number of them actually were included in a Marker-Blue-Johnson(?) national survey, so there are some national benchmarks for some of these.
DR. JANES: A survey that's tied to the SCHIP program?
MS. COLTIN: No, just a separate, they were, RWJ National Strategic Indicators project that was done as a survey, so a lot of the children measures, the promoting health development and the adolescent measures, those were implemented in a lot of the state Medicaid and SCHIP programs, the adult measures that they've developed were measured through this RWJ Survey, National Survey. So my sense is look to some of the stuff that's in the privacy sector and think about how it fits.
But my main comment, the measures are kind of an example of the point. The point was really to think about not being constrained by the framework that was offered by IOM and to expand that last category beyond end of life care to use the framework that FACT(?) had actually recommended, which is changing needs, which opens up these other types of measures that could be used that are population based measures. So those particular measures would simply be examples to illustrate how they could do that. Is that a reasonable suggestion? People think that's ok?
Alright, we're doing well here.
DR. FITZMAURICE: Kathy, how would you phrase that? Would you phrase it as expand beyond the IOM framework?
MS. COLTIN: Yes. And I would start out by applauding them for already doing that in the area of efficiency, because I think they did. That was not in the IOM framework and I think it is a good idea to do that and to balance the value equation essentially.
DR. JANES: Isn't the term that you used when you first opened this discussion, a descriptor that you used to refer to all of these, I've since forgotten that word --
MS. COLTIN: When I said they made it more balanced?
DR. JANES: You said they need more balance, I would suggest they add measures. They've already added efficiency, and I can't remember what that was, you don't remember either. It was a nice word.
MS. COLTIN: Resource, I talked about, I'll have to look at --
DR. STARFIELD: While we're talking about frameworks, the original IOM framework did have equity and it's missing from all of this.
MS. COLTIN: I think that's because they're doing the National Health Care Disparities Report, so they've actually broken that axis into a separate report.
DR. HOLMES: The same measures that are in the Quality Report are included in the National Health Care Disparities Report, so they're simply arrayed differently and they focus on the priority populations in terms of the measures.
DR. STARFIELD: That's not the point, the point is to compare those priority population with the whole population.
DR. HOLMES: They're all comparative.
MS. COLTIN: And that one is, by the way, any of the members who would like to comment, the comment period on the National Health Care Disparities Report is still open until December 13th, and the proposed measures are available on the AHRQ web-site, and I did have, I know Debbie had copies made for Vickie to see where the Populations Subcommittee might want to formulate some recommendations, but I think they would probably use ours as a springboard because the measures are the same measures as are in the National Quality Reports, so hopefully what we're doing today will provide an opportunity for them to piggy-back on that.
The next point I had had to do with the fact that many of the key measures are not available at the state level or below the state level.
DR. STARFIELD: Before you get to that, before we get off that, is it the National Health Care Disparities Report right?
MS. COLTIN: Yes.
DR. STARFIELD: It's not a National Health Disparities Report?
MS. COLTIN: Correct.
DR. HOLMES: And it's actually the National Health Care Quality Report and National Health Care Disparities Report, those are the formal names.
DR. STARFIELD: I'm sorry.
MS. COLTIN: No, that's ok. Good clarification, and I actually, that's two points down on my list.
So the next point I had was that a lot of the key measures are not available at the state level, that in order to really be actionable, most of the action on these occurs at the state market or provider organization level, and that even if a problem area is observed at the national level, it will require looking at the state level to decide whether that problem exists before resources or any attention to improvement will get focused on that problem at the state level. It becomes too easy to say that's them not us. And so unless the data are provided at levels that are more actionable, it may go on unattended, the problems that are identified may go unattended.
So I think this is again an area where we're speaking to some of the gaps in the data system where we're talking about the fact that we don't have either sample sizes that support reporting out at the state level on some of the national surveys, or that we have inconsistencies between state data systems that don't allow us to report uniformly across states on some of these measures. So I think those point out areas.
The fact that there are those gaps and that we can't report the data at a level that makes it more actionable and that allows a state legislature to say yes, we've got a problem, we need to do something in our state and let's create this to have large provider organizations within a state come together and say we've got a problem here let's collaborate on trying to fix it. When it's at the national level, the best you can hope is that at the state level they're going to measure to see if they have a problem, but it's going to require, it's going to delay anything occurring, because that measurement's going to have to occur at the state level before any resources are going to end up getting allocated, which is particularly true at a time when health care costs are experiencing double digit increases and states are experiences huge budget deficits.
So nobody's going to throw resources at fixing any of the problems that are only reported at the national level, without first confirming that they have a problem in their area. So I think it's really important if the report's going to have impact to be able to measure more at the state level.
Does anybody have anything that they'd like to add or disagree with around that? Is that a reasonable point?
DR. HOLMES: It's a very important key point.
MS. CARR: Just an observation in terms of how that is presented because you appropriately couch the difficulties around doing that, and also I think that you would not want to sound inconsistent in terms of the recommending more measures, some of which are only national and then saying we need state measures. But perhaps the recommendation can be couched in terms of those that can be addressed in this immediate report, some that are more developmental in nature, short term and long term, might be a more constructive way of presenting observations on the Report.
MS. COLTIN: Which sort of gets to the point you were making about thinking about phasing, having a vision of where we want to be and what might be done within one to three years versus five to ten and so forth. So developmental short term and long term.
DR. JANES: What's happening in terms of the development of provider or practice level --
MS. COLTIN: Well, I think CMS really is taking the lead on that. They have a Doctor Office Quality Project that --
DR. FERRER: Barbara Fleming is leading that.
MS. COLTIN: That is getting started and that may be something that this group would like to hear more about at a future meeting. But there is a ground swell I think beginning for measuring at the individual physician level and being able to measure once and roll up, which is how you get away from the kinds of constraints that we were just discussion. Right now we've got a top down kind of measurement approach and we need huge samples and national surveys to be able to break it down whereas if you measured from the ground up you could measure once and roll the data up to various levels. So I think there is some work going on to try to look at exploring those approaches and it would be good for this group to hear more about those as you go forward. I won't be here but it's a thought that you might want to pursue.
DR. JANES: Well by including that in your recommendations on this point I think it gives you something else to point to that has an enhanced level of grandiosity if you will beyond the national data. I agree that I think we need to keep pounding away at this issue of state data. I must just be getting old but I am somewhat discouraged at the likelihood of seeing a lot of movement on that topic at least in the near future. Certainly the issue has been put before Administration after Administration.
MS. COLTIN: And actually I think this bottom up approach is a better alternative.
DR. JANES: It has strengths and weaknesses but at least we're not just sort of one more time being the drum for state level data and saying there is nothing else. That is another thought.
MS. COLTIN: There is another approach, but, in all fairness, it goes back to the first recommendation that we talked about, about population based measures, because as Barbara pointed out, when you're measuring at the individual physician level, or even sometimes at the group practice level, the numbers of cases you have of particular diseases are not large enough to report at that level, but if you can't create these composite type measures, like follow-up of abnormal lab tests, and you can combine across diseases, you can report at those levels. So it's a way to actually make the suggestions about how data could be reported at more relevant levels that also goes along with an earlier recommendation about using these types of population based and composite --
DR. JANES: And if we are talking about health care rather than health that it makes it even more relevant.
MS. COLTIN: Ok, so are people comfortable with that approach?
MS. MALINA: Cynthia Malina, the CEO of Alternative Link. I'm really pleased to hear the comments about patient centric measures occurring from the ground up. And my only concern is that I hear you talking about physician measures and I don't hear a whole category of care that patients are accessing outside of conventional health care systems, the complementary and alternative medicine areas of care or nursing care. And I'm sure all of you are aware of the David Eisenberg from Harvard study on the 629 million visits to complementary and alternative medicine practitioners compared to 386 million for primary care. So it's a key area that may need some measurement.
MS. COLTIN: That's a good point. I think we had started off actually saying what was going on in that area and I mentioned the Doctor Office Quality Project which is why I focused on physicians, but I think the bottom up approach that we were describing really does need to be broader than simply focusing on care that patients get from physicians. Thank you for that.
So people are comfortable with the need for information at finer levels of the geographic or geopolitical as well even market and provider organization levels.
There were some comments that we heard in July that I think we might want to consider reinforcing. One has to do with presentation of the measures. There really isn't much in the Report that says how the measures will be presented, whether it will simply be the mean for the measure or whether some range or measure variation will also be provided. And we did hear particularly from some representatives of the business community that it would be helpful and important to have measures of variation in addition to measures of the mean. And while that may not be that important to consumers, to relevant to consumers, I think most consumers are getting more sophisticated even political polls of the plus or minus whatever, just presenting it in some way that simply can indicate how much variability there is in these measures would be useful. Do think people think that's a good idea? Is that a recommendation that we'd want to pass on?
DR. EDINGER: Are you thinking more in terms of actual numbers or some explanation of what the limitations are so they can, or maybe both?
MS. COLTIN: I think we'd want to understand something, maybe even the range, that this was the mean but it ranged from here to here, people can understand pretty simply.
DR. FERRER: Just bar a little bit on the caution of people can understand that pretty simply because in Medicare has spent millions of dollars of testing of beneficiaries. For example this Nursing Home Public Reporting Initiative that we just launched and we're going to be moving into home health and hospitals, I'd hate to say it but from a Medicare standpoint, less is best because you understand that and people in this room understand it, but when Medicare preparation begins to navigate the data set on nursing homes, they want very simple information they can quote "use", so there are people, a lot of researchers at CMS have spent a lot of time studying format, the mechanism by which you communicate this information to a sort of end user beneficiary. And I know that Regina McPhillips(?) in HRSA, and I'm sure she could probably get some guidance as to what has "worked" for the Medicare population. That doesn't mean that that should just be the model --
MS. COLTIN: I've been involved with that so I understand what you're saying. But I think the primary audience for this Report is Congress initially and policy makers and then kind of down the line and the consumer version of this may look different and may look much simpler but I think in those versions that would be most likely to be used in either setting policy allocating resources or drawing the attention of the delivery system and the provider community to the need for improvement, that having these kinds of measures --
DR. FERRER: Well if that's the audience then I would support something --
MS. COLTIN: I mean I'm not leaving consumer out I'm just saying that my understanding is that this is a report that was initially requested by Congress and that's the primary audience, but that AHRQ does want to make this comprehensible and get to grab the attention of other audiences as well.
DR. FITZMAURICE: I support that by saying that if you're a member of Congress or if you're staff, you're looking at this and say gee, the numbers look however they are, let me ask my expert. So the expert's going to look at it, I want to go to the appendix to see what the standard deviation, what's the prohibition of variation, what's the range. Even if the consumer doesn't get to see the appendix the expert called upon by staff is going to want to see more about it so there will be a lot of phone calls to CMS.
DR. JANES: I couldn't agree more. I agree with you. I think giving even range or whatever to the consumers in many instances would be pushing the edge of the envelope. But for people who are sitting policy makers and making resource decisions about this, they're going to want to know what this distribution looks like. Is it a very narrow distribution around that mean or median or is it all over the place?
DR. FITZMAURICE: How confident can they be in the quality of the data. That's one of the first things that they'll do.
MS. COLTIN: Right. I agree. So and interestingly, this comment was brought up by one of the business coalitions saying that purchasers in terms of if they're paying the bill, as CMS is for a large segment of the population, they want to see those kinds of measures as well in terms of their own policies around benefits design and so forth. So that was a comment that was made that I thought was worth reinforcing, that would be useful.
Another comment that was made that's a bit more controversial is that this is a report on health care, not health. And that we need to consider issues of attribution when reporting outcomes and so while on the one hand it's important to have information that's broken down to finer levels of granularity, you have to balance that with the risk of potentially attributing an outcome that may be measured inappropriately to given entities. So I'm not sure if we want to weigh in on that one or not, but that was a comment that was made, I think it was brought up by JCHO, particularly as if affected some of the measures of mortality that were not in hospital mortality but that might potentially be attributed but where there were many, many more factors that went into the outcome than simply those that were a function of health care that was received.
I'm less inclined to want to weigh in on that one but I wanted to put it out there because it was an issue that was raised.
DR. FITZMAURICE: So if I understand your point Kathy, it's not how's our health care system doing it's what is the health of the population? What is there about the population that can improve the health care even if it's auto accidents or something else that interferes with being healthy?
MS. COLTIN: I think the point was it's not a report on health, it's a report on health care and yet there are some outcome measures that are the result of more than just health care. And since it is a health care report the implication is that this outcome occurred because of some failing in health care, and that may actually be true but it may not be the only thing that contributed to that outcome.
MR. HUNGATE: I would argue that our viewpoint should be that of how well does this health care report help in reporting on health. Because that's what it seems to me our primary mission is. And if that might then lead to a discussion on a measure that says this might be used for attribution, it should not be because, these are the qualifications on that measure, which is useful, and not doing it is worse than doing it. It seems to me trying to deal with it in some way that gives a viewpoint.
MS. COLTIN: I agree, I think that's a good idea, to suggest that there be some text that provides that perspective.
DR. HOLMES: And I believe that AHRQ, the people working on the National Health Care Quality Report are intending to include that text but I think it would be important for this group to underscore that. Because I think providers are concerned.
MR. HUNGATE: I would add a personal comment that I just debated with Steve Jenks on the issue of release of mortality data and I lost the debate in terms of what's happened because a lot of these things have been hidden rather than made more visible so accountability which might have led to more improvement hasn't occurred. So I would argue that reinforcing that and maybe underlining that risk addressment is needed underneath that making it effective, could highlight the areas of improvement that might make the measure better.
MS. COLTIN: Those were the main recommendations. There was one other recommendation that had less to do with the report itself but more to do with kind of how it was presented or released. Which was there was a suggestion that AHRQ develop one or more companion pieces for different users, and maybe what this is different versions of the report, maybe it's one report with different companion pieces, I don't know. But that there needed to be some sort of a users guide, particularly for consumers around maybe how to interpret the information or linking it with other initiatives. Like AHRQ has a web-site called talkingquality.gov and that might be a vehicle for thinking about this kind of companion report about how to use this kind of information. What could states do in looking at the information?
So that was a suggestion that again was raised by the representative from the business coalitions which I thought was a good idea and might be worth weighing in on. Is that something that people would like to see us comment on? It's a little bit out of our bailiwick in terms of data but it does have to do with actionability of the information, being able to really not just put out a report but help the report to be used.
MS. HOLMES: well, I think that's an important point because the issue is you get a lot of information but what use can this information be put to, how might we interpret it and go on from there, whether in terms of future research, future reports or actionable items, so I think that's a very appropriate recommendation.
MS. COLTIN: Are there others? That was kind of my straw list of recommendations, we've done fairly well by it, but I'm sure it wasn't comprehensive. Are there other things that anyone observed in reviewing this that they think we should comment on?
DR. STARFIELD: What about the whole area of adverse effects of interventions? I'm not talking about errors, I'm basically talking about unanticipated adverse effects as well as anticipated ones. Doesn't get much in here, are you going to include it in your balance things? I don't remember --
MS. COLTIN: I haven't identified that one, I have no objection to doing that if people think that would be a good, another area to raise.
DR. STARFIELD: There's a lot of that, something like 50 percent of people report having an adverse affect in the last year.
MS. COLTIN: I think a lot of what I've seen in terms of the availability of that type of information --
DR. STARFIELD: That's pushing the envelope.
MS. COLTIN: Yes, it is pushing the envelope, but it's been doable in projects, I know a lot of work was done on this at Group Health Cooperative in Washington with linking pharmacy databases and claims in looking at adverse affects after patients were started on particular medications and things like that. So that's one example of an unintended adverse affect. It wasn't an inappropriate prescription it was just reaction.
DR. STARFIELD: Well this is something that you can get from patient surveys but you can also get it from ongoing while learning from --
MR. HUNGATE: Yes, I think this is a good example of something where you could position a major amount of where your data collection is going to be patient centered through the personal health record and their own identification of something happened I didn't expect in describing that.
MR. FERRER: Steve Blackwell at CMS is actually, he's a pharmacist, health services researcher, and an attorney, dangerous combination. Nonetheless, he's doing a pharmaco-economic data set to track claims in pharmaco adverse affects, so that might be somebody, if anybody's interested, may want to follow up with. That's Steve Blackwell.
MS. COLTIN: Ok. Good suggestions. Anything else? I think we're going to need to move along. Alright, so our next agenda item is --
MS. JACKSON: One thing, in the process just to clarify I think what you will do is to present to the full Committee this afternoon --
MS. COLTIN: These recommendations.
MS. JACKSON: Because they're already alerted that something is coming from --
MS. COLTIN: They won't have anything written I don't think unless I can just quickly summarize these --
MS. JACKSON: If you can do that then I'll see that in the full Committee and the process I assume is maybe they can clarify that the Executive can make the final decision that would be a statement from the Committee.
MS. COLTIN: And I think when I present it to the full Committee I won't go into the depth, I won't have the time to go into the depth that we went into today about the examples, it will be more the 20,000 foot level of what our recommendations are and then what we will actually put together in writing we will include some of the examples that we talked about more specifically.
Agenda Item: Review Themes: Date Issues in Measuring Quality
MS. COLTIN: Ok, now going to get back to our report and the last time we met we were working up the document that's Susan Kanaan had put together where, and you should have a copy of this, some of the summaries of the testimony that we heard over the past four, four and a half, five years, we've been doing this a long time. Basically for new members, we started this initiative right after the President's Advisory Commission report on consumer protection and quality in health care was released. And in that report they made several recommendations that were data dependent and so what we intended to do at a lot of the hearings that we had is to get perspectives on some of those areas. So one of them had to do with patient safety and medication errors and so forth, and we had a panel where we took testimony about what's happening in that area, what kinds of data systems exist now, voluntary reporting versus mandatory reporting issues, things like that. We have information about what's happened as well as thinking about what we heard about what needs to happen, what people would like to be able to measure to know whether they're making progress in these areas. So that's one example.
There was a recommendation about the need to measure quality in vulnerable populations, for example, and so now we have coming out the National Health Care Disparities Report which is a big move in that direction, but we're still faced with the issue of the fact that we don't have good data on race ethnicity, we don't have good data on education or income in any of our standard data sets, so that we have to rely on survey based measures for looking at those kinds of factors because those measures that we can derive from a lot of the administrative data sets or even medical records don't always provide that information, we don't always get education and socioeconomic information from medical records even if you were to use them as a source.
So that's an area that the Subcommittee on Populations has been exploring but I think they are looking to us to also weigh in on that and weigh in on the need to be able to collect that type of information more routinely in our data system so that we can have measures for vulnerable populations and be able to actually do a lot more with the National Health Care Disparities Report than we're currently able to do.
So those are just two areas that they recommended, but basically where I'm going with this is that the hearings that we conducted over these couple of years were started primarily to topic areas that had been raised in the recommendations for the Advisory Commission Report and so what we're trying to do now is to go back and see well across all of those areas, were there some common themes that we heard about data barriers, gaps in information, limitations in what we're able to do.
And so what we've done, this is sort of one of those things, like which is the best way to cut the pumpkin. Should we have cut it by what were the topic areas in the Advisory Commission's Report where we'd be really redundant within each area about some of the data issues and gaps, or cut it by the data and then go across how it affects different populations, different types of measures and so forth.
So in this thematic summary that you have here we actually organized it primarily around obstacles related to different sources of information. And this isn't a final decision but it was a way to at least be able to discuss some of the issues that we heard across a wide range of panels in a wide range of topic areas but that had some common themes in that. They all used administrative data one way or another to do different things with it and they all encountered certain problems in doing that. They all used surveys or many of them at least used surveys to measure whether you're talking about the panel mental health services for instances, they use administrative data, they use surveys, they had problems, they've sort of been using surveys, problems they observed in using administrative data. So that's the way this is organized right now is by the data topic.
And the other reason we did that is we felt that there were actually, it was a logical way for a committee like this one to be able to look at these issues because our role is data and how to improve the data systems and to make recommendations about the information infrastructure and the data that are needed, so organizing it in this way seemed to be natural. If we look at administrative data we can tie our recommendations to what the Standards and Security Subcommittee is recommending around administrative data sets and modifications to administrative data sets. We can also build on their recommendations around patient medical records. So the PMRI standards and other recommendations that they make around electronic records.
That's part of the rational for organizing it this way. And what we started to do at our last meeting was to go through and prioritize these and so I don't want us to take a step backward, even though some of you were not here at the last meeting, it would not be efficient for us to go back and revisit all of these. But we had gone through, I can summarize for you kind of where we were and what we thought was really important for us to include and comment on.
In the general notes and questions area, the last three were ones that we felt raised some cross cutting themes that we wanted to be able to comment on and we added one that is not on the list here as a cross cutting theme, was around the business case for quality and documenting the business case for quality because we really felt that convincing people to invest in the data systems that are needed to measure quality was in many ways dependent on making the business case for quality. So that was an addition that we made.
Under the obstacles related to administrative data, in terms of priorities, the ones that we felt were most important to develop recommendations around were code set limitations, obviously we heard a lot about code sets yesterday, that's a hot topic, and it's a topic that's very timely for us to make some recommendations around given all that's going on in that area now.
The second one, privacy concerns. The third, delays in adopting standard identifiers and the impact on being able to link data, so the kind of thing we were just talking about in terms of linking pharmacy and claims, standard identifiers are a very important component to doing that. Or being able to measure from the ground up, standard identifiers for providers become very important in being able to do that.
And then we actually skipped down to the last one on the list about data collection hasn't been streamlined, multiple collections. So a patient can have a visit at a particular medical group and get a survey from the federal group, a survey from the integrated delivery system that that medical group belongs to, a survey from the health plan, and everyone is paying for these surveys, the poor consumer's being bombarded, we're paying three times through premiums or visit co-pays or whatever else with supporting these surveys. And so this was an issue that people felt was important and it kind of speaks to that ground up philosophy. Could we measure one time and then be able to attribute different measures to different levels of the health care system rolling the data up accordingly. So that was another priority area.
And then on the next page, the very last one in that category about ambulatory claims practices, which relates back to the one on payment arrangements also which is on the previous page, those can kind of be combined because the claims practices are really dependent in great part on the payment arrangements.
So at this point those have been identified as priorities. Is there anyone who would like to, who wasn't here, who sees something else on this list that they would like to argue for adding under that topic area as a priority area in which they would like to see us make recommendations?
MR. HUNGATE: The only argument I would make would be the one that I made before, in terms of it having the spaces for self improvement within the process, then it's going to be the most efficient. If it's externally asked for it's not going to be. So that's a qualifier that I think we ought to make.
MS. COLTIN: Well, one way to bring that in and one way to also address what we were discussing in terms of the CMS vision, is this next to the last one which is called limited clinical richness around administrative data. And I think that might be worth adding because it gets to both of those points.
MR. HUNGATE: I would agree with that.
MS. COLTIN: So we'll check that one off as well then. Alright, and then moving along to obstacles related to surveys, we actually felt the first three were the most important. Did anyone want to argue that we should include the fourth one as a priority area for recommendations? It's more, if we address the problems in the other areas this one could go away. So anyway, so people seem to be comfortable.
The next one, the obstacles related to medical records, we combined the first and last under those that we would actually address the cost consistency issue through the notion of the fact they're not electronic and so there is not the standardization that you would like to see. So the first one was a priority area but it also incorporates the last one.
The other two areas were the second one of no standardized core data and definitions, and the one on multiple charts for the same patient, which has to do with again, issues of should there be a single patient control medical record or some way of either accessing a common medical record, so we've actually heard I think as a full Committee and perhaps I may have also heard at the time when Barbara and I were on the Standards and Security about some of the technologies that are available now and are actually being implemented in some places to where the patient kind of holds the key and can give the key to different providers to have access to their medical record.
DR. STARFIELD: This is probably a good time, in the interest of continuity, it's our last chance to call attention to new members to the core data elements, which was the product of this Committee, the whole committee, in 1996, and it's really a key document that we ought to keep referring to. Elements that ought to be collected for, administrative ones, it's mostly clinical things, so keep that up there.
MS. COLTIN: And it is important because it does get to, some of the core data elements are not necessarily things that make sense to put on claims but ought to be routinely part of the medical record collection system.
MR. HUNGATE: Is that Report on the web-site?
MS. COLTIN: It is, yes. I think it's called the Core Data Elements Report. So that was where we had stopped the last time in terms of prioritization and we hadn't really prioritized anything that was in the other obstacles category. So what I'd like to do today is, and what I had done was actually asked the members if they would email me their recommendations for priorities in these other areas. But I only received one email so I think we actually have to spend a little bit of time going through these and trying to get some consensus.
If you could take a moment to look through the category that's labeled other obstacles, we'll do the data issues category as well, and suggest candidates for priorities, what you see under there that you feel we really need to comment on.
DR. STARFIELD: So what we're thinking of should in fact relate to the things we identified in the first hour right?
MS. COLTIN: That would be helpful. I would say, I wouldn't limit ourselves to that but if we did highlight things in the first hour, we want to make sure we don't miss the opportunity to reinforce them here. But we can add to that.
DR. STARFIELD: Well we've got the inadequate focus on ambulatory which is related to the two focused on hospitals. So I would suggest that.
MS. COLTIN: And the separate state reporting systems fits into what we said as well.
DR. STARFIELD: I guess the existing area reporting systems are incompatible and that sort of relates to the adverse affects one, I mean we focused on errors and not sufficient can adverse affects.
MS. COLTIN: Some of these could be collapsed, too. I mean like the separate state reporting systems and the lack of standardization and gaps make it difficult to access vital statistics, that really kind of, those two relate together.
DR. HOLMES: Here you have the need for a business case.
MS. KANAAN: This was sort of the catch all category that didn't, was neither particular to the other two categories.
MS. COLTIN: I think the issue of data available to risk adjustment are primitive gets back to the clinical richness area that we had identified under administrative data sets, too, so that might be worth using as an example.
DR. STARFIELD: This was a comment in 1998. The field has advanced.
MS. COLTIN: But in terms of some of these data sets, it's still true.
DR. EDINGER: What about the issue, in the GAO Report on HCVA's Nursing Home measures, one of the issues they raise was that some of the nursing homes, if the way that they rated the quality in there, that the survey and certification data indicated that some of them had either more or less severe problems, but when necessarily indicated by the measures collected. And you can go into long arguments about why there's a difference, but maybe the issue is not only the patient surveys but the license survey and certification type measures, and the public data surveys, not necessarily giving compatible or comparable results for whatever variety of reasons that that happens.
MS. COLTIN: Well, it also had to do with one of the arguments that I've heard which I heard Lisa Iozonni make as well in the case of hospital measures, is that because we don't have a single longitudinal record, we have all these chopped up records, we don't always know measurements at different points in time. And so whether or not something was present on admission, California does have a field for that in their hospital discharge data so that they can tell you whether this diagnosis was present on admission as opposed to something that occurred during the hospitalization, and should be attributed to the hospital, and that came up with the nursing home like in terms of pressure sores, do you want them to not admit a patient with a pressure sore because it's going to count against them? And yet do you have the longitudinal data to show that this was present on admission and therefore hopefully it was treated and at different points in time it was no longer present. But not having that kind of longitudinal record kind of gets back to the issues of linkages and problems with linkages but it is an issue that sometimes the data may be available but they may be available in different places and you can't link them appropriately.
DR. STARFIELD: I don't know if you mind jumping ahead but related to the issue of risk adjustment is the issue of co-mobidity in the next set or the issue of standard qualifiers for diagnosis and the explicit recommendation. Do you want to check those off now or wait until we get to them?
MS. COLTIN: Let's wait until we get to it, I just want to make sure, I would like to suggest that we add the last one at the bottom of page two, the data collection not integrated into routine work because I think that was another comment that was made earlier about when we were talking about electronic medical records are building these measures and data elements into those systems. Can you do it in a way that it's not add on, that it's the data are extracted from elements that are or should be routinely recorded.
MR. HUNGATE: In a way I think you can link that to quality measurement is unrelated to payment. And the preexisting condition issue also, the mixture of risk adjustment, a payment system unrelated to quality, and data collection I think other than routine is I think a core set of things there.
MS. COLTIN: So these are mostly concepts that I think in many ways we can bring in in our discussion of these other data sources, but they may come up under more than one of those data sources because they will cut across.
So now Barbara let's move on to the next one, the data issues of sources. And the ones you were suggesting --
DR. STARFIELD: That's why it's related to, it's because of co-mobidity in that category. Down near the bottom.
MS. COLTIN: I think I would vote for the lack of integration, I think that gets us back to this idea of separate records, silos, no longitudinal records issue, standard identifiers for linkage, there's a lot of things that go into that. I think the inadequate codes is another one, it gets back to our issues in the earlier one about the code set limitations.
DR. HOLMES: Lack of availability, when you talked about before, that some data simply are not available, they're not gathered, it's not that they're not gathered consistently, they're simply not gathered.
MS. COLTIN: And I think that actually is something that operates at multiple levels, it's not even just the data that's not gathered, but it's the records themselves that often aren't available. If they're not in electronic form, if they're not integrated, I know that there are some measures that have been proposed in thinking about HEDIS measures, where plans said they might have to go to six different offices to review charts because it could be in any one of a number of different medical records, because there isn't a single record. And how do you figure out where to go to find the information of the patients' record.
DR. STARFIELD: That's the lack of integration point.
MR. HUNGATE: Now it's the doctor to date capture electronically. As soon as you have to go to review you've lost the improvement gain.
DR. STARFIELD: Also related to things that we picked out before with the children and adolescence, the ambulatory care, and the state level data, at least those.
DR. HOLMES: I would also add race ethnicity.
DR. STARFIELD: Well if you're going to add that I think you need to add the next one, too.
DR. HOLMES: Right, right. Well, what you're saying is these are all issues that have been identified, but you're saying the ones that are, you're identifying the ones that you think are the most important.
DR. STARFIELD: Mental health was in our list from earlier.
DR. HOLMES: They also relate very nicely back to the recommendations with respect to the National Health Care Quality Report.
MS. KANAAN: A couple of people spoke repeatedly, and I pulled out a list of the sort of synthesizing comments, at the end here, specific recommendations of key strategies for improvement, and a lot of people identified laboratory and radiology results as sort of the one, if I could only have one thing.
DR. STARFIELD: Test results, it's the test results, that's right.
MS. COLTIN: Test results. It's saying if you had them, they in and of themselves can be outcome measures. They can be used to track process and improvements. And that's true, there seem to be consensus. We raised that issue with a lot of people, different people who did quality measurements. If you could have one piece of information that you don't have now, and everyone zeroed in on test results as being the most important piece of information. It's not only important for measuring quality but in terms of targeting improvements, like being able to use it for case mix adjustment but also use it for stratifying and finding high risk populations to target interventions to those particularly populations. So it was brought up in a number of different context.
MR. HUNGATE: It seems to me that the functional status is if you're interested in outcomes, functional status is very important.
DR. STARFIELD: That's interesting because our core data set that remains one of the few still big gaps that we don't know how to measure it.
MS. COLTIN: The other issue under this incompleteness and gaps is that I think it's really important and relates to the area of vulnerable populations is the uninsured, because you're not going to get good claims data in many cases, you can put some of the state hospital data sets, you can identify those, who got free care, was self pay or whatever, but there are some real limitations in the ability to measure what's going on in that population. You often have to rely on surveys and then you're stuck with the limitations of survey data. So I think that's another one that we should include.
DR. EDINGER: Kathy, I'm wondering, on the laboratory and radiology results, were they just thinking in terms of quantitative data, like as we put cardiac output or pulmonary results? Because that typically, the BUN result but they are sort of results that you can quantify.
MS. COLTIN: Let me give you a real concrete example. For the past three or four years, CMS and NCQA(?) have been trying to develop a measure, a quality measure for patients with congestive heart failure. And they needed to, they wanted to look at the use of ACE inhibitors or ARB's, and they needed to limit it to people who had a left ventricular ejection fraction of 40 or less I think it was. So if they had the lab result for the LVEF, they could identify the population who should be on these meds and measure whether they're getting it.
But they don't have the test result so the best they could do is find people who have an LVEF to limit the chart review at least and say ok, but you'd still have to review many more charts than necessary to then find the subset whose measure was 40 or less and then look at whether they had the meds so it's that kind of thing. It's necessary for defining the target population you want to measure, it's necessary for targeting interventions.
I know when we look at interventions in our diabetic population around the d-quip(?) measures we're focusing on people who have measures that are currently above recommended thresholds, those are the ones we want to go after first and say can we get these patients in control and then we can worry about getting the other patients from progressing to out of control. But if you don't have the lab data you have to implement an intervention across a population of in our case 16,000 diabetics when in fact, which you can't afford to implement an elaborate intervention, it's not necessary for half of them. So you really want to say well can we identify that group that really will benefit from this intervention and really needs it and without the laboratory result data that's very hard to do. So again, it comes up in measurement, it comes up in improvement and so forth.
DR. JANES: Kathy, is the problem that, and I think gets back to what Stan was asking, is that when I think about this I tend to think of lab data and admittedly I also think in terms of sort of organized systems of care in which I think well hell that data's available, you can go to your system and pull it up and take a look at it. That's one of the things that is consistently on line.
MS. COLTIN: Where is it available? It's not available to a health plan that only has administrative --
DR. JANES: Well that's what I'm asking, is it that these numbers are not consistently across nuclear medicine and radiology that are simply not in an electronic form or is it that they're not being picked up in the billing streams --
MS. COLTIN: I think a big part of it is that they're not in electronic form. Therefore they can't be shared in electronic form. Even if the information is only going to the ordering provider and I think there are real privacy issues about why that is the right way to go and it only gets passed maybe to the health plan on the particular select circumstances for measurement purposes, but it can't be passed along easily if it's not in electronic form.
DR. JANES: So it does live in a more basic level than just a question of not being picked up in billing streams, it simply is not consistently available in electronic form.
MS. COLTIN: Yes, it's an upstream problem, it's that a lot of doctors first of all don't have the capability to accept it in electronic form even if the lab can give it to them that way. So labs are increasingly able to provide the information in electronic form but if a physician doesn't have a system that they can use to store the information and that's where it's easily retrieved and can be passed along, that's a problem.
So the 837 claim form, for instance, actually has a field for a laboratory test code and a test result. They're both situational data elements which means they'd only be filled in under certain circumstances. But one can contractually define those circumstances and say ok, for quality purposes our plan is doing this. Any time you bill us in a patient with this diagnoses for this test we want to get the lab results, which means you don't get it for everybody, you don't get it for every condition for every test, you get it for very targeted situations.
Even if it were handled in terms of a claims attachment request for a laboratory result where they're developing a lab result claim attachment, the provider would have to have it in electronic form to make that a viable option for measurement. If you're getting back a whole lot of pieces of paper on 6,000 people, it's still not going to be an efficient way to operate.
MR. HUNGATE: Given your description, why not push for a demonstration project with community health centers to give them the capability to get in an electronic form? In other words to say let's go the route which then can build up which solves the health plans problem of getting it.
MS. COLTIN: I think what's happening is that we've got different parts of the health care system moving at different rates in terms of adopting electronic solutions and for those who are going directly to an electronic medical record well that solves this problem because the data can be easily integrated int. There are others who are doing it incrementally, so they're beginning to build patient registries or clinical data repositories. They may only be collecting the lab data right now. But I don't want to preclude any progress but I think we want to identify different ways that this could be accomplished and hopefully accomplished in a way that ultimately would lead into an integrated medical record. But in the interim the --
MR. HUNGATE: But I would argue for this full piece that deals with congestive heart failure of the patients that you've described in a setting for they're predictably going to be as a way to improve the care which will coincidentally then improve the measurement.
DR. EDINGER: I think in the Secretary's Report there are recommendations and there was a piece in there on community health centers and --
MS. COLTIN: I think actually some of the more innovative quality improvement work is actually being done in some of the neighborhood health centers, so it is a reasonable starting point for some of this.
MR. HUNGATE: Is it appropriate for us to try to reinforce that sort of a --
MS. COLTIN: I think we can offer suggestions for how it might be done in the report. We can make recommendations that it should happen kind of at the macro level and then how it could begin to happen, this would be an example to start with a community health center and think about doing collecting this kind of information.
DR. CARR: I would just comment that our community health centers at HRSA have an evolving project around sentinel health centers which are selected health centers for which we are trying to build the capacity to participate in research and data collection efforts. This sounds very consistent with that thrust that's already underway.
DR. STARFIELD: I'm glad you mentioned because we've been involved in that at Hopkins Center. Also including primary care measures is a good laboratory looking at the use of new measures for quality.
MS. COLTIN: Let's just make sure that, I think we've covered pretty much all the key issues under data issues. We've checked off so many of them --
DR. STARFIELD: I think that whole, probably all of them should be at this point.
DR. HOMES: You want to include all of them Barbara? Is that what you said?
DR. STARFIELD: I think so. I mean severity is one of the issues.
MS. COLTIN: I think that on the last page there's explicit recommendations. Most of them actually relate back to things we already discussed, the accurate and complete lab data, the standardization of code sets, identifiers, I'm not really seeing anything in here that we haven't covered under other, maybe the last one.
DR. STARFIELD: The standard qualifiers for diagnosis, because that's related to the risk adjustment, it's also related to the present on admission requirement.
MS. KANAAN: So we want to star that one? We can put that someplace else.
MS. COLTIN: It related to an issue that Barbara and I have talked about over the years.
DR. STARFIELD: Eight years.
MS. COLTIN: That has to do with sometimes inaccurate diagnostic information in administrative data sets because physicians don't have and other providers as well the opportunity to indicate whether this is a presumptive diagnosis, a rule out diagnosis, or a confirmed diagnosis and even though the coding rules say only code to the known level of specificity, that's not what providers do so, right they don't get paid, so what ends up happening is you'll see a diagnosis of diabetes and a test, a glucose test to determine if the patient has diabetes, but you've got the diagnosis that says they have diabetes, so you're thinking this is a diabetic when in fact the test was done and it came back negative and you don't have test results, which is the other problem, so you can't say rule out anybody who came back normal.
So it's an issue if in fact there were some sort of a modifier that allowed you to say this is a presumptive or rule out diagnosis versus this is a confirmed diagnosis, that would be a helpful thing. I think the big issue there is that's a recommendation that may be a very useful recommendation for the PMRI and electronic medical records unlikely to, we've fought for eight years around the administrative side, it isn't going to happen, I mean it's not part of --
DR. STARFIELD: Well, it ought to be in data issues actually.
MS. COLTIN: Yes, it's a data issue but I think our recommendation for how to address it would probably be through the electronic medical record being able to distinguish those kinds of differences, the status of the diagnosis.
MR. HUNGATE: Well the absence of that status indicator limits the use of the data to the one doing the work and so collecting it administratively for somebody else doesn't help you, but if you could do it that way it could.
DR. FERRER: That's actually a segue to what I was going to mention. Often times we hear little discussion as to providing the basis of and other code information back to the clinician and that inherent value, in academic centers often times, but that's just a very self selected, people who are very motivated to use information to gauge sort of the population based on the data. Should we also be recommending or perhaps viewing that, we should also be analytically providing that information back to the clinician for their improvement of their particularly patient care. And that's something that is talked about but is it recommended, if so, I don't know what the answer to that would be.
MR. HUNGATE: It is, as Kathryn said, part of the electronic medical record. Unless it's there on the machine as you're managing the patient it doesn't seem to me that it's helpful. To report back it after the fact doesn't help very much.
MS. COLTIN: It helps but it's inefficient and far less effective as well. So I'm reaching back 20 years now when I did research on the difference between concurrent and retrospective reminders in electronic medical records systems because you can have both. Retrospective reminders are still helpful, here's your patient on whom you ordered this test, they haven't shown up to have it done, or here's your list of patients who had an abnormal that hasn't been followed up, but the concurrent reminder is that if the patients in front you and by the way this patient needs a mammogram. The action can happen immediately, it's far more effective than getting a retrospective list that says two years have gone and these patients don't have mammograms. We have to reach out to the patient, some percent will respond, some won't. So yes, there are reminders and reminders and they work differentially in terms of effectiveness.
MR. HUNGATE: In a data set you've got to have it available for the retrospective before you can do the other.
MS. COLTIN: The last one, I'd actually like to make a pitch on the recommendation that we heard from SAMSA about integrating behavioral health into general health. There's really very poor coordination I think right now between the mental health provider community and the primary care physicians and others and I can't tell you how many times I've heard pediatricians lament the fact that they don't know a patient of there's has been given a medication by a child psychiatrist or whatever, that that communication isn't happening. It leads to errors in terms of prescribing medications that may interact or how the patient might be managed and so forth.
I think it feeds into what we were talking about the need for an integrated kind of longitudinal record but it also is something that it's a highly sensitive area because of the privacy and so we're not talking about the notes but we are talking about the diagnoses and the major treatments being shared among those that are caring for, providing primary care to these patients, not just those that are dealing with the emotional side of their well being. So I think it would be useful to make a recommendation in that area and to recognize some of the privacy concerns around them. Are people comfortable with that?
DR. JANES: I assume you're talking about the example you were using calls for category of mental health, but I assume you're talking about the sole issue of referrals and treatment by specialists outside of the primary care realm, that you're not limiting --
MS. COLTIN: Well, this particular issue here, the last one on the last page, says integrate behavioral health into general health. That meant that this was a particular problem area of behavior. Behavioral being both mental health and substance abuse issues.
DR. STARFIELD: How about saying mental and behavioral health?
MS. COLTIN: Well, the term behavioral health is used to include mental health and substance abuse, it's shorthand for the two combined. And in health plans when they talk about the benefit, it's the behavioral health benefit, which covers both of those.
MS. KANAAN: Well, I only know the English language.
MS. COLTIN: We'll put in English in the report. We'll make it clear in the report that we mean mental health and substance abuse.
DR. JANES: Kathy, do you want to limit it to just this section, because it does, I think, it plays out in other issues?
MS. COLTIN: Yes, but I think we have covered it in the other issues where we've talked about integration and so forth. This is highlighting an area that's a specific problem and often a problem because of the privacy issues as opposed to the other issues which pertain in many other areas of coordination of integration.
DR. ORTIZ: I'd like to make a comment, I came in late so I'm not sure, you're talking about behavioral health and maybe you covered this, I don't see clear that that's one, most people don't actually assume behavioral health means mental health, I think it really does need to be distinguished separately. And number two, behavioral health, you're talking about mental and substance abuse but it a lot of time behavioral health actually looks at things like sexual risk behavior, smoking, dietary, things that you do in terms of riding your bike, driving your car, all that kind of stuff, so unless that was covered earlier, to me that's behavioral health, it really encompasses all those pieces of the puzzle. And that's something that's not typically covered in a primary care setting or a clinical setting but should be, and so I think that's something that we need to address.
MS. COLTIN: This particular suggestion is not dealing with health risk behaviors. It is talking specifically about mental health and substance abuse care and so I think it sounds like we should not use the term behavioral health because it's widely interpreted and the recommendation here was specifically from Eric Oppelwood(?) at SAMSA about mental health and substance abuse services being included. I think we actually have addressed some of the other features around completeness of data.
DR. ORTIZ: We should just call it was it is and be very explicit in the terminology.
MS. COLTIN: Yes, it's sounding that way, it's very clear that there isn't a common interpretation of that term and we should be explicit.
MS. KANAAN: Do you want to add a point about risk behaviors?
MS. COLTIN: Do we have that covered in any of the others? Because if we don't we probably should. I think it probably belongs under the gaps, the long list of incompleteness gaps, on page three, so risk behaviors.
DR. JANES: Since public health is out there pushing aggressively on particularly these issues of trying to get primary care docs to counsel on a lot of behavioral health issues, but particularly in areas where it has been shown to impact subsequent behavior, then maybe you'd want to tie that into then some recommendations about being able to assess, the extent to which you would think.
MS. COLTIN: Yes, you're right, and actually, that's usually my primary example under inadequate codes but it probably warrants being broken out in that CPT, for example, but the list would be true for almost any code in the system. A weak area has to do with cognitive tests and cognitive therapies. There are wonderful codes involving blood or tissue or anything, but when you're talking about three questions to determine if someone's depressed, there isn't a code for it, and yet it's a standardized kind of screening tool for depression. And if a provider administers it there's no code for coding that. And that's true of the standardized instruments for alcohol problems and other areas that warrant I think their own code for testing.
But because it's a cognitive test as opposed to a physical specimen test it gets lumped into this big catch-all single CPT code about all these screenings, and you really have no clue what was done, whether they asked about do you smoke, or whether they asked about questions that have to o with drinking behaviors, or they asked about sexual behaviors, whether they asked about depression, you cannot tell. And yet, if somebody had a cardiac cath(?), I believe there are 28 different codes for it, so it's a real inconsistency in the coding system and I would bring it up I think under the inadequate codes but it's a particularly good example to bring up.
The other is the cognitive therapies so it's counseling for example, anticipatory guidance. It's another one where there are these catch-all codes as opposed to specific codes for counseling about what, smoking cessation, it's very different.
MR. HUNGATE: Do these code gaps all tend to fit the general thought of primary care?
MS. COLTIN: Yes, they tend to be primarily primary care oriented but not solely primary care oriented, because one would hope that any specialist is asking about smoking or some of these other behaviors so not just the primary care physician. But I think it's an orientation in the profession right now we're in clearly in the coding world to kind of lump these all together and it makes it very difficult to measure some very important behaviors and elements of care.
MR. HUNGATE: Or related specifically is this whole area of shared decision making where there's a lot of materials developed and should be utilized in those setting and there's not code for that either, so there's no incentive to doing it.
MS. COLTIN: The other thing is they're not well recorded in medical records either. So this is an area in terms of PMRI, thinking about standards for electronic medical records, we've tried to get this kind of information from Charberviews(?) and it's just not there reliably. So if the provider is screening the patient for depression and they screen negative, there's no mention of it. If they screen positive, there maybe then a mention that the patient's depressed, but you don't know who got screened. And likewise if they provide some counseling about smoking cessation, hardly ever do you see it recorded in the medical record. Big gaps there so I think there are issues in terms of the CPT coding and the payment policies that are paying for these kinds of things that lead to not having in the codes, that's part of why we don't have the CPT codes, but it also is an issue on the electronic medical records side as well.
DR. ORTIZ: That's one of the things though that are important I think that we're going to deal with behavioral issues and health risks and things like that, one you've got the inadequate coding issues but then there's also things you have to think about, one you just brought up, that a lot of these things that we do in terms, first of all, health risk behaviors are a huge issue, in fact a lot of the stuff that occurs in health care could be prevented just by people being more responsible about their health behaviors. So that being said though one, often times even when you do counsel, it's not captured so that's a big issue.
But the other thing is I think we have to be careful as far as taking an evidence based approach because a lot of things we say we should be doing because it's a good thing to do but it turns out in fact counseling for a lot of these things like weight loss is terrible and really is very ineffective, so I think as we decide some of these things we can't be led necessarily just because health risk behaviors are a big problem we should be advising them. We really need to look at well what really does make a difference and use those in terms of what we're going to be measuring because otherwise you end up putting a big burden on a lot of people for things that don't make any difference, don't improve outcomes.
And especially given the fact that nowadays we keep asking more and more and more things to be done by clinicians, we really have to start balancing what really makes a difference versus what's doesn't because you can't do a thousand things at once and now we also have to start looking at other groups in the health care system doing this thing, kind of stuff, instead of it always being put in the clinician, where the nurses should be doing it, where the patient should be doing at home, interacting through computer, there are a lot of other ways of doing this beside the clinician which is the way we traditionally think because it's the way we've evolved in medicine. So I just kind of want to warn people about that.
MR. HUNGATE: In order to establish the effectiveness don't you need to code no matter where it's done so that you can link?
DR. ORTIZ: You definitely need some kind of a coding to capture it.
MS. COLTIN: This is one area where the problem of payment policies can actually become a plus because most companies are not going to pay for something that hasn't been proven to be effective and so if they're not going to give you a code unless you're going to pay then at least we're in that area, which may be one of the few, the payment policy can be supportive in that particular issue.
Alright, we need to think about next steps on this. What we're going to do is cull this list down and then what I'd like to do if people are comfortable with this is take an approach similar to what we did this morning with the National Health Care Quality Report and that is put some straw proposals for recommendations together under these. Groups the ones that make sense to group and make recommendations and then get those out to people for comment electronically. I'm going to be going off the Committee although they asked me to stay involved with seeing this Report through to completion and we are hoping to have a draft report to bring to the February meeting.
So I think we will try as much as we can to formulate these, an agreed upon set of draft recommendations so that we can bring a draft report to the February meeting, but we'll have to do it using email. So for those of you whose names may not appear on the roster, can you just make sure that I have your email addresses so that I can be sure, the staff in particular, I mean the members I'm fine with I have everybody's, but I want to be sure I have everyone's email so that I can get this out to people. And Susan and I will work together to collapse these into some meaningful structure and to put together some straw recommendations for consideration. Does that sound all right with people as a way to go given the time limitations? Alright, great.
PARTICIPANT: When do you expect to have a draft?
MS. COLTIN: Let's try to get something out to you by maybe the second or third week or in December. Does that sound reasonable?
DR. STARFIELD: It's not very far away.
MS. COLTIN: Well, this is going to be draft with straw proposals. It's going to be a little rough, so we're going to want your help in crafting these recommendations.
DR. STARFIELD: When do you need comments back? By the end of December?
MS. COLTIN: I'd say right after New Year's. This isn't going to be a draft report, this is going to be just these priority areas with draft recommendations. Once we have that part settled we will put these into a draft text for the report and we're hoping to get that out to you maybe by the middle of January. Does that sound about right? But if I can get your comments back so we can get these priority areas and recommendations clear then we can fold them into the broader report and I believe you have a condensed outline of what the report will be covering.
The one other area, once we have these major priority areas and recommendations drafted is to be sure that we've done a good scan of both the public and private sector around progress, things that are going on that we've either heard about in our testimony or that we're aware about and want to acknowledge. If the IOM did a big report on something, we don't want to put out a report that appears ignorant to that fact. So one section of the report does involve this sort of what's going on and in particular about what's going on within the government.
For those of you are liaisons for various agencies, as you see the areas in which we're making recommendations, if you're aware of things that are going on, things that you are doing, if you could let us know that, send us back that kind of information, because for one section of the report we want to talk about progress and what's been going on over the period of time since the President's Advisory Commission report was published.
PARTICIPANT: Including emerging things because when we talk about opportunities for implementation that will include sort of identifying things that are sort of just on the horizon, either the Committee's own PMRI recommendations or other things.
MS. COLTIN: So if you're aware of things both in the public sector and the private sector, I'm aware of a lot in the private sector but there's stuff I'm not aware of so the more of you that know of things that are going on if you can let us know so we can try to incorporate that, but in particular I think we want to make sure we cover what the federal government is doing in terms of leadership in these areas as well.
That make sense? Ok, thank you all.
DR. FITZMAURICE: Kathy, you have a copy of the public comments.
MS. COLTIN: Oh, wonderful.
DR. FITZMAURICE: I have several copies here and there are more being printed off right now.
MS. COLTIN: Thank you so much. So we can pass those out. These are a summary of the public comments that were received on the National Health Care Quality Report.
[Whereupon, at 11:00 a.m., the meeting was adjourned.]