Let me begin by telling you about the Public Health Service Data Policy Committee and what we're about. Roz Lasker, our deputy assistant secretary for policy, is heading up this committee. He viewed it as an opportunity to put the Public Health Service at the negotiating table as decisions were being made about what the data elements in health care reform were going to look like. He viewed another area of responsibility as doing that which is within its power to streamline state reporting requirements for various programs. We've recently heard from a variety of states that have begun the move toward more rational data systems, more of an indicators benchmark model.

We are all moving to LANS, and improving communication between states and their local health departments. There are beginning to be systems in which all parties use the same data elements and share data about patients--allowing for the tracking of people through the system. The question is: What will further these efforts? Why don't we begin this session with some questions from the audience.

AUDIENCE MEMBER: I'm from Idaho, and we fight with Mississippi for the lowest physician-population ratio. What we haven't been able to document is the adverse consequences of this. We're struggling with how to accurately measure health status. Mortality is a long-term measurement for any intervention that you take. Outside of greater access to prenatal care, the measurements that we're using now are 20, 30 years down the road in terms of heart disease, cancer, etc. So we're looking for something that's more morbidity-based, where we can see change in a shorter period of time.

We don't have hospital discharge. And a lot of the national indicators that we've discussed today don't really translate down to the state level because the sample is too small.

DR. MILLMAN: I'm not going to give you a ready solution to your problem. But one of the things we tried to do with this paper we've just written is take ambulatory care sensitive hospital admissions from the Medicare data and look at rates of admissions for these conditions across the 600 health care service areas.

These health care service areas are market areas that were designed by the National Center for Health Statistics using Medicare data, in terms of where Medicare patients went for hospitalization. What's nice about these market areas is that they're available on the area resource file. So you can link information about physician supply with the Medicare data.

We were interested in seeing if there are differences in an insured Medicare population for ambulatory care sensitive conditions across the 600 HCSA's. And if there are differences, are these differences related to the physician supply or some other characteristic of the area. Essentially, what it showed is that there are considerable variations in the ambulatory care sensitive conditions, even for an insured population. Physician supply, except in the most underserved areas, explains almost nothing about these variations.

The high income areas have a much richer physician supply than the low income areas. That's why you have to have very powerful statistics to separate out physician distribution from other demographic characteristics. But interestingly, the demographic characteristics of income and rate explain 40% of the variation in admissions. And we've probably reached some threshold in most areas of the minimal number of physicians required to not have ambulatory care sensitive admissions.

Now, I want to add a caution here. We haven't done a lot of validation of ambulatory care sensitive conditions, because we don't know exactly what they measure. In principle, they measure delays in care.

And so our next round of research is to take a look at the outpatient side and see whether there is a difference among patients in their outpatient experience that would explain hospital admissions or mortality.

But I think, what all of this points out--justifies our community health center program--is that you need delivery systems structured to deal with the differences in sociodemographic characteristics of the people within the systems. If you really want to see a health status difference, it's not enough just to have the resources available. The resources have to be tailored to the problems you're going to encounter. And I think those problems have a lot to do with the kinds of enabling services that are available. You have to overcome a lot. You have to overcome transportation barriers. You have to overcome psychological barriers. Unless you have delivery systems that are tailored with the appropriate kinds of outreach and patient education, you're probably not going to have an impact on health status.

So, one of my answers to your question is that I doubt if we're really ever going to be able to match numeric resources to access or morbidity problems in the way that we would like--in a simple way.

When we move to people joining plans and being distributed that way, we can begin to compare how well these plans do because we'll have some characteristics about their structure. Until then, I think it's going to be very difficult for us to tie resources directly to indicators of problems in the system.

Let me answer another part of your question, about the appropriate physician-population ratio. With the HCSA designations, it is tradition to use the ratio of one physician to a 3500 population base. There are thoughts that the measure should be one to 1800 or one to 1500. When I was in England recently, they were using 1800.

We have recently done some case studies of HMO physician planning. This project looks at both staff model HMO and IPA's; their ratio of primary care to specialty physicians; their overall number of physicians; the data they use to do this kind of planning.

We looked at 23 HMO's; 11 of them have a heavy Medicaid population. So we are also asking the question about what difference does the type of population enrolled make.

And let me tell you, the HMO's are not very sophisticated in their physician planning efforts. Most of it is very seat of the pants and has to do with market targeting considerations. We did see signs of some movement toward the development of encounter data so they could match patient's needs more closely with physician resources.

We were also interested in to what extent they're using nurse practitioners and physicians' assistants. And this was heavily dependent on two things. The overriding factor was how the physicians in a particular plan felt about nurse practitioners and physicians' assistants--whether they were convinced that there were productivity and economy gains with using them.

The second reason, we found in some places that it's very hard to recruit nurse practitioners and physicians' assistants.

In general, all but two of the HMO's had no difficulty recruiting primary care physicians. One was a community health center-based HMO, in a very bad inner city area; the other was a very small rural HMO. All of which adds up to the fact that in the private sector, in a very small, controlled environment, they're not any better at figuring out their provider needs. It's still very subjective.

AUDIENCE MEMBER: Have you ever seen anything about whether physicians practice different medicine with their fee-for-service patients and managed care patients?

DR. MILLMAN: At least in the Washington market, there's such hot competition among the HMO's. Everybody is trying to cut costs, but everybody is also trying not to look like they're being more conservative or denying care. I think in very mature markets everybody is beginning to crunch down. And then there is a range of experience in other markets, and they're watching each other very carefully. And, of course, everybody would love standards so that they could point to the guidelines. In theory, you would like to see these feeding in, so that decisions aren't being made just purely idiosyncratically.

One of the most amazing things to me about the experience of putting together indicators is how little we can say. You can't get even a small group of physicians to agree that under this particular set of circumstances a physician should do X and then you won't see a bad outcome in the future.

AUDIENCE MEMBER: Perhaps the appropriate measure is in the other direction--the excess. If you have an inappropriate number of primary care physicians, you might have an inappropriate number of other things. In other words, that's your small area variation on the high end.

I think everybody has been looking for the relationship at the wrong end. They have been looking for low physician-population ratios and problems associated with that when, in fact, it is probably when there is excess physicians that there are untoward events like excess surgery and excess hospitalizations. That's the physician effect.

Whether health effects exist at the other end of the physician-population ratio, you can monitor forever, and I don't think you will ever find anything.

DR. MILLMAN: I think we'll learn more from the medical outcome studies that are being done. We need to rethink the issue of overutilization now. When you use the difference in physician visits between insured and uninsured patients, you find the insured have about half as many. The argument often made is that there is really overutilization among the insured population.

AUDIENCE MEMBER: Our issue has been how are we going to get more primary care physicians into the state. Do we build our educational capacity to put out more physicians? Do we develop a loan repayment program to entice physicians? I think what we've tried to do, unsuccessfully, is to give the legislature some idea of the resources that would need to be allocated in order to adequately serve the population.

DR. MILLMAN: To me, this problem is a health planning problem--a problem of supply and demand. And we have to figure out a way of providing an appropriate level of services to the population, whether it's through circuit riding, or combinations of nurse practitioners and physicians' assistants. And, if you have to pay more in the way of salary, it's a much better investment than putting it into the education system and hoping you'll get something back down the road.

It's got to be some sort of structural change in the delivery system that assures that providers are not out there on their own practicing without any professional interaction and that they have colleagues and they have relief.

My answer is a labor economist's answer. First of all, you have to pay people what it takes. Secondly, you have to find the aspects of the job or site that will make it attractive to providers.

The real measure of need for a National Health Service Corps, ultimately, is you can't hire the providers to meet the need in particular areas. The market isn't working and this, then, is the rationale for government intervention. If everybody were enrolled in a health plan, and particular plans had problems recruiting physicians, there would be rationale for a National Health Service Corps. The problem is we're in this mixed system of fee-for-service and managed care. And I think it's going to be really difficult over the next few years for the states to make a case for needing the NHSC. It's going to depend on what data you can collect, and what you know about the local area.

What frustrates me is that people fixate on the HPSA's and the MUA's. We're not getting a lot of community health centers out there and, despite the fact that we're growing, the National Health Service Corps is still a very small number compared to the designated areas. The places with the most serious problems are going to continue to get the resources.

It becomes important then to make the case to a state legislature and because of their national uniformity, people have gravitated to the MUA's and HPSA's, even though nobody likes them, as a framework. What we need to do is come up with some new frameworks that can be applied and tailored to state and local conditions.

AUDIENCE MEMBER: Do you know of any work that's been done using a household survey that asks people about their perceptions and their access, and compares people's perceptions to standards that exist?

DR. MILLMAN: When these things have been asked on the big surveys, like the Health Review Survey and National Medical Interview Survey, you get very small numbers of people saying there is a problem. When you ask why people don't have a usual source of care--and you take away the people who don't want a usual source of care, or who have just moved--the number is frustratingly small, about two or three percent. The problem is that it is very hard on surveys to get people to think in our terms of access as a concept. I told a study this morning about the new access supplement to HIS that was field tested by looking at individuals who had gone to a free clinic in Rockville, and had been denied care because it was already over capacity. These individuals were then interviewed, given the access supplement, and many of them said they didn't have an access problem.

Again, I think I'll go back to what I was saying when I was discussing the study with Medicare. Ultimately, I think what we are looking for are differences in health status, and differences in the use of services that we think people ought to have based on their age and circumstances.

And if that's not happening, then these populations need a tailored health service delivery intervention that deals with the fact that people often don't recognize when they need medical care, or what kinds of chronic disease followup they should have, or what types of screening, preventive services, and health education are necessary.

Our aim is to have an equity in health status across the country. This requires that you invest in these very tailored interventions that take into account the whole range of psychological, cultural, and financial problems that exist.

The Health Security Act had a provision for $300 million a year to provide enabling services for communities. It was recognized by those on the task force with experience in delivering care to populations that things like transportation and translation were very important. We believe at HRSA that our programs provide these kinds of services. The grant money partly pays for uncompensated care and it also partly pays for these extra services that our populations need.

But nobody really knows how much of these enabling services we provide. I have headed a work group over the past six months that has looked at what we provide and how we should define these services.

We took the HRSA programs--all the way from the Ryan White HIV programs to the Maternal Child Health programs for special needs kids to community health centers to HMO's--and it became clear to us that different populations required different packages of enabling services.

Initially, when costing it out, we were trying to separate out case management from health education and translation services. But, the providers who work with these populations really see all these things as mutually reinforcing. Let me give you an example from the lower middle class. What Kaiser in Los Angeles told me as I was trying to figure out what it is we normally do for nonvulnerable populations that's becoming more popular to provide outreach for asthma and diabetes. It involves identifying the population, making sure they come in, providing health education services, and dovetailing this with case management to assure that they are actually following through at home. In addition, if they can't get in, you need to provide the transportation for them to do so. These things all work together.

One of the things pointed out to us by grantees in the homeless program is that homeless individuals can't just be given a bus token for transportation. Getting them to come in for care requires teaching them how to use the transportation system and making sure they get to the appointment. It's a very heavy investment.

So when you take a special population perspective and you look at the range of services and how those services need to be tailored, the design for the delivery system becomes clearer. And I think with HMO's the good thing coming out of HEDIS is that in order to get your indicators looking good you're going to have to initiate proactive efforts to make sure that your asthma populations and your diabetes populations are getting the services they need early enough. The HMO's are starting to invest in figuring out how to do this efficiently and effectively.

It's one thing to do this for a middle class population; it's quite another thing to do it for a standard Medicaid population. As one HMO doctor commented, when he told his patients to refrigerate a drug they didn't necessarily have refrigerators at home. And some of them don't have telephones. There are now programs to provide telephones to high-risk pregnant women who don't have telephones at home.

I think this is the intersection of public health and personal care. I think some of these programs will work more efficiently as shared resources at the community level. It's conceivable to me that the health departments will either assure that these services are structured a particular way or manage the shared resources.

AUDIENCE MEMBER: Are you suggesting that HRSA may, in the short or long term, be thinking about asking states to start planning in terms of service delivery systems for special populations when we can't even do it effectively for the mainstream population?

DR. MILLMAN: No. We are going to our providers who are getting our grant funds to see what services they are providing. And so we said to them, "How much do you spend on transportation and case management and services like that?" What we were trying to do is understand the broad variation of services grantees are spending money on, and what contributes to this variation.

I think we have traditionally provided money on a population basis. Our grant programs imply populations. Our Ryan White program works with different segments of the HIV population. There are some broad statements about how the grant money should be used. At the provider level, we're asking, "How are you actually spending your money?" Not everybody knows, because it's all folded together as part of their delivery systems.

AUDIENCE MEMBER: We've tried to put a lot more emphasis on integrating systems, as opposed to separating them into little categorical niches. I would hate to see us try to develop multiple, separate delivery systems.

DR. MILLMAN: I wasn't suggesting that. What I was trying to say is that when you look at what it takes to treat special populations, it's a lot more than just straightforward medical care. And the problem is in the past providers have had to integrate a bunch of separate funding streams with separate reporting requirements. The goal is to ultimately get away from those things that inhibit better integration. This all exists because there are interest groups who convince Congress to spend money, and Congress wants to see how that money is spent.

AUDIENCE MEMBER: Can you address the use of emergency rooms by various populations?

DR. MILLMAN: I was trying to determine how to make that an indicator. But, we have a hard time distinguishing between a non-urgent visit and a legitimate emergency visit.

It's a goal of Medicaid managed care to shift episodic care away from the emergency room to a primary care setting. But in practice it's hard to know prospectively whether a visit is a true emergency. And we have lots of doctors in the private sector sending their patients to the emergency room. So it's very hard to identify what is an access problem. I think it's worth a further look.

AUDIENCE MEMBER: Do you think there would be a utility to data-poor states diverting analysts' time to understand the Medicare claims files? And how much state health policy application would there be there?

DR. MILLMAN: It's a good model in the sense that it's close to the encounter data of the future. Our view was: let's look at an insured population and try to determine if there are access problems for this population, and also explore whether we can use the Medicare patient experience as a substitute for claims data for the rest of the population.

I can only speak from what we've done thus far. Remember, we're talking nationally, and the workable claims history file was the 5% file. I don't know how useful it will be at the local level. It's a tremendous amount of data crunching, and you have to know what we're doing. I'm hoping that ultimately we can take this experience, document it, and share it with others. But at the moment, I wouldn't suggest states try to do this on their own.

AUDIENCE MEMBER: You alluded earlier to the concept of cultural barriers. Could you expand on that a little more in terms of indicators and access issues.

DR. MILLMAN: Yes, we've essentially stopped using the race/ethnicity variable as a substitute for the cultural barriers. For example, when you look at the Hispanic women who have actually been to the doctor but have not had a Pap smear or a mammogram, you see a big difference between them and other segments of the population. You can then begin to ask, "Is something cultural going on here?"

We don't yet know how to ask on surveys a good set of questions that identifies individuals as "at risk" for a cultural problem. What's happening in California with Medi-Cal is that they are asking for bids from contractors and the contractors are supposed to indicate in their proposal how they intend to assure cultural competencies.

We were asked to support a small effort with two HMO's that have high concentrations of Hispanics and Asians--to try to work with them to develop a screening tool.

There are two issues here; one is how do we identify the ethnic minorities in a given plan? The experience with social service agencies is that everybody has different standards and methods for collecting that information. And then secondly, what sorts of tools do you need? A survey instrument will be tested out in these two HMO's to see if it can identify people who have cultural barriers.

One particular enabling service is interpretation. And most people who have started programs in this area believe that it should be broader than just providing the translating itself. It requires having somebody available who understands cultural nuances. During the translation, these can be conveyed to the clinician. Some will take it even further. Kaiser, in Los Angeles, hired a Hispanic diabetes educator who is familiar with cultural dietary issues. We haven't been very good about measuring culture as an access barrier. We think it's an access barrier, but it's hard to separate out from income and insurance status.