[This Transcript is Unedited]
The Wyndham City Center Hotel
Potomac Room
1143 New Hampshire Avenue, NW
Washington, DC 20037
Participant List
DR. MAYS: The first thing we should start with are some introductions because we have some new individuals who have joined us from some of the agencies. I have no problems with people, this is a big room, with people sitting at the table. You can feel free to sit with us. You don't have to sit back there unless you want to kind of be in the corner. Otherwise, come to the table and join us.
Let me just ask while we're doing the introductions if our two presenters in the very beginning need to set up any AV? Do you all need to get anything set up for your presentation? You're okay? Okay. All right.
Let's start with introductions. This is the Populations Subcommittee. Hopefully, everyone will have an agenda. They have gotten one. If not please just raise your hand and Gracie will make sure you get one, but let's start with introduction. I thought you were reaching for the mike. I was going to say let's start with Barbara because she's reaching for the mike.
DR. STARFIELD: I'm Barbara Starfield. I'm from Johns Hopkins University and I'm a member of the committee.
MS. COOPER: Hi, I'm Leslie Cooper and I'm with NIDA.
DR. MAYS: Thank you. She's a staff person who's new. Thank you.
MS. KEIM: Hi, I'm Sarah Keim. I'm giving a talk this afternoon.
DR. MAYS: You should say who you're with.
MS. KEIM: I'm with NIH.
MS. HEURTIN-ROBERTS: I'm Suzanne Heurtin-Roberts. I'm with the National Cancer Institute.
MR. PEARCY: I'm Jeff Pearcy from the National Center for Health Statistics.
MS. CRUTE: I'm Cheri Crute. I'm the writer for the committee.
MS. QUEEN: Susan Queen, staff.
MR. HITCHCOCK: Dale Hitchcock, staff.
DR. FRIEDMAN: Dan Friedman. I'm a member of the committee.
MS. JACKSON: Debbie Jackson, NCHS staff.
MS. GREENE-KAHN: Good afternoon. Kelly Greene-Kahn, National Institute on Alcohol Abuse and Alcoholism, MIH.
MS. JONES: Katherine Jones, National Center for Health Statistics.
MR. ETTINGER: Stan Ettinger, AHRQ.
MS. LUMLINE: Emma Lumline, AARP.
MR. SMITH: I'm Phil Smith just sitting in for Ms. Edna Pisano who represents the Indian Health Service.
DR. MAYS: Welcome to everyone. They told me I should probably do one thing is to check on time constraints. I'm told that some may have planes and some may need to get back to their institutes and we may just need to adjust the schedule.
Vic and Jeff, are you okay in terms of time? Great. Then why don't we start in the order we actually have in the agenda and this is a presentation that on one of our subcommittee conference calls we talked about. And that is, let me just bring you up to date a little bit and I don't want to steal any of their thunder, but some of us, Suzanne, myself, and I'm not sure who else was here was actually at the presentation.
It's probably been last September. Yes, I think it first occurred in the summer of last September in which NCHS had pulled together a group of people to talk about the issue of the measurement of health disparities. They had examples and we were there to talk about it.
One of the recommendations that we made during that meeting is that they think about some kind of explanation, guidebook, handbook, et cetera that could discuss this issue further. This is being pulled together through Ken Keppell's group.
Ken wasn't able to join us today, but there is a subcommittee that developed as a function of that initial meeting. Jeff has been one of the major players in it. Vic has also been one of the major players in it. I participate on an "as can" basis and I don't know if Suzanne, you're continuing on an "as can" basis.
But we're reaching a point where several drafts have been circulated now about the handbook. I thought the committee might want to hear a presentation about it and to be able to opine about it. So, without further ado, why don't we start with Jeff. Jeff, introduce yourself just so they know a little bit more about you.
MR. PEARCY: I'm Jeff Pearcy with NCHS. I've been with NCHS for three years. I've been working with Ken Keppell for two years now. Before that, I was with the Maryland Department of Health and Mental Hygiene in the Office of Analysis and I don't remember what. They change offices a lot there.
Before that, in a previous life, I worked with Forester College and I developed multi-factor classifications for forest ecosystems. It's not readily apparent how that applies to public health, but it actually turns out that a lot of techniques in data reduction and the interrelationship of environmental factors with biological factors also apply in public health.
DR. MAYS: Welcome.
MR. PEARCY: All right. Let's get right into it then. I know we're pressed for time and so I will move along. I will begin with not so much a discussion of the handbook, but some of the ideas that we think are important in considering the topic of disparity and how they can be applied to the idea of a handbook.
And then I will lead into some of the main ideas we've incorporated in the handbook and then Vic will talk further about some of those ideas and develop some scenarios that may be important and play out further down the road in measuring disparity.
The first question that comes to mind in a number of contexts, and this has always been confusing to me, is what do we mean by disparity. It's not always apparent what that means. In the context of public health, disparity means more than what I was taught, that it takes on additional meaning other than what we're familiar with in other biological and physical sciences.
In the context of public health, disparity is not just how big a difference between things, but it also includes the idea of who's disadvantaged, what is inequitable, and what is unjust. But still, I think it's important to bear in mind that we keep the idea and concepts of inequality and inequity separate in an operational sense.
A number of people also feel strongly. I have an abstract from a quote from Laurie Carr-Hill of the UK Health Equity Network and others that argue that it's important to distinguish between equality which is a factual matter and equity or inequity which is an ethical, judgmental matter.
Inequality asks the question: How different are we? We can measure this in a number of ways and it can be dealt with in an objective way. In contrast, inequity requires a decision about how much a difference should be reduced, whether that is in terms of what another group wants it to be like or an absolute term of what level of outcome should be obtained.
That should be inequality. How different are we? To follow my argument that disparity in context of public health can be decomposed into inequality and inequity, we can further extend the model and this follows on the work of Margaret Whitehead, Tim Worth, and others that inequality can be discussed in terms of what is avoidable and what is not avoidable.
If differences are unavoidable, then we should not consider the differences inequitable or unjust. On the other hand, if it can be determined that differences are avoidable, then the question of inequity can by addressed by deciding whether or not differences are acceptable or not.
We may think of disparity in terms of a decision tree in which case first you determine whether two or more groups are different and how different they are. It's also important to determine or to ask the question of how big of a difference is a disparity in terms of just equality. We have a lot of tools for deciding or for measuring how different things are, absolute difference, percent difference, ratio.
The most appropriate to use may not be apparent and some of these questions we touch on in the handbook. The question of how big the difference is and whether that difference rises to the level of disparity is not apparent and we don't really have guidelines on that matter.
We need to keep in the back of our mind, is the difference significant or is the difference important? Assume we have an inequality that's big. We can ask a series of questions about whether the inequality is unavoidable or potentially avoidable.
To work through this stage, we could utilize a set of health determinants. If we determine that the inequality is avoidable or potentially avoidable, then we should get to the ethical, judgmental question of whether it's inequitable and unjust.
I borrowed these from Tim Worth and Margaret Whitehead and others that there are a number of determinants of health that are unavoidable such as genetics. For an example, I recently had a conversation with a person from the National Cancer Institute who was wrestling with the problem of disparity in melanoma, and they were asking the question of are white non-Hispanics much more greatly predisposed to melanoma than other groups, so consequently are the differences among these groups to be considered disparity or is the inequality unavoidable due to genetic factors.
Similarly, unhealthy behavior choices made in diet or activity may lead to adverse health outcome, but from a public policy perspective it may be considered or prove to be unavoidable. An example might be participating in extreme sports and I'll touch shortly on the lack of free choice later.
Whitehead also notes a transient health advantage as being an unavoidable determinant of disparity. For example, if one group takes up a healthier behavior before or in greater relative numbers than another group such as bicycle helmet use, a disparity may develop, but it can be considered unavoidable as long as the other group was not disadvantaged in its means to catch up.
That is, if they had the economic means and the opportunity to purchase appropriate safety equipment. On the other hand, determinants of inequality or disparity are avoidable potentially if they involve restrictive lifestyle choices.
A good example that I'm familiar with is the lack of availability of a healthy diet for groups that have low incomes or that live in areas with restricted availability of nutritious and fresh fruits or when products such as tobacco or alcohol are targeted at specific groups.
These are avoidable I consider determinants of inequality. Also, group is restricted by income or some other reason to living in unsafe housing or to working in more dangerous occupations, we should consider this to be unavoidable.
Differentials in access to care are clearly avoidable. It is rarely a matter of choice to seek our less or inadequate care. Natural selection refers to the type of situation characterized by loss of job or reduction of income followed by reduced access to care and possibly illness which, in turn, may affect the ability to be able to work and to afford care.
Conceptually, I think in terms of disparity in two ways: inequality in which it's descriptive or monitoring of differences among groups and in terms of inequality in which we have a goal or a target towards which groups should improve.
A disparity model based on inequality is very straightforward and is implemented regularly. However, a
disparity model based on inequality is complex and requires a series of objective evaluations for which there may not always be data as in the determinants of health differentials.
It also requires a subjective judgement of what is unfair and unjust. Alternatively, a simplified inequity model may be used that assumes all health differentials are avoidable and that all are unjust. That's a little bit of background to consider in thinking about disparity.
More directly, to what we've been working with in the handbook are some ideas of what to measure. Clearly, change over time, how do the differences, the inequalities among a set of groups change from 1990-1998? Or, from a program point of view, how has the rate of one particularly group changed over time.
Another way to look at disparity is across indicators, across a number of mortality indicators. I have a brief example of that later, but it may also be of interest to look across indicators spanning mortality and morbidity access to care.
For example, looking at disparity in stroke and then looking again at the same groups at disparity in hypertension and then looking at the disparity in the access to care as a data source possibly from the Health Interview Survey, persons having a regular source of medical care and so on.
We also may look at how disparity varies across the United States and its successively finer level of geographic scale, region, states, medical service areas, and so on. Finally, it may be useful to consider disparity for a specific health indicator across different types of populations, that is, populations grouped according to race, ethnicity, income, education or something else.
It may be useful to understand the difference among those groups. In our work, we looked at a great number of criteria for selecting statistics and evaluating their use. It turns out that a few of them are most important in discriminating and distinguishing the most common or most likely to be used statistics.
Among them, the number of groups whether we're going to measure two groups or several groups. To what are we going to compare groups? Are we going to compare groups to a total or overall population rate, an average of the group rates, the best group rate or some target rate such as might be set in Healthy People 2010 or the HHS Disparity Initiative.
We've also found it important to distinguish between absolute and relative statistics. I'll just go over a couple of quick examples and then turn it over to Vic. I'll move on to the effective reference because the choice of whether to make comparisons between two groups or among groups has been most problematical and for myself, I'm primarily interested in measuring across all groups.
I recognize that some people are most interested in comparing a target population with some other rate. The effect of effects of this example of what you get in terms of an estimate of disparity. In this case, this is all cancer mortality, deaths/100,000 in 1999.
I've got five groups in the box above and along the left-hand side are different referent points, the total population death rate, the average, the best, and the target as set in Healthy People 2010. On the right is an estimate of disparity among all groups using our so-called index disparity which is simply the mean deviation of the groups from the total relative to the total expressed as a percent.
You can see that as you go from the total, the average, to the best, the disparity, the estimate of the magnitude of disparity increases, that it does not increase for the target is a function of the target setting method. If as in the case HHS, the target, if it had been set as better than the best, the estimate of disparity would be 50-60 percent.
This has a practical implication in two ways. One, the first two total and average, you're not making a judgment about what groups should attain. You're just looking at how different they are. When you use the best of the target as a reference, you're saying that this is what the other groups should be able to attain, this is how they should be. As a result, the estimate of disparity is larger.
This is a recap of that, the total or average. You're talking about inequality and you end up with lower estimates of disparity. At best or target, you're talking about inequality and you get higher estimates of disparity and I just want to restate that when you're talking about equity, you generally do not know what the unavoidable portion of the inequality is.
This is just an example of an absolute type of comparison. You talk about the number of groups, the point of reference and whether you're making and absolute or relative comparison. This shows the death rates for a number of cancer and site-specific cancers in the left-hand columns and the mean deviation of those from the total rate where there is specific causes of cancer as well as the range across those groups.
It's apparent from this that the higher the death rate, the greater the mean deviation, the greater the range and so if you're talking about disparity in absolute terms for mortality here, the only conclusion you can reach is that disparity is higher for death rates when death rates are greater or causes of mortality in which death rates are low, you have low disparity.
This is the same analysis using irrelative statistics across all groups. The index of disparity again, and you can see that the estimate of disparity, in this case, across groups, across indicators is not correlated with the magnitude of the death rate. It's independent.
It facilitates comparisons across indicators in a better way. If you look at them together, it just seems apparent to me that in comparisons across indicators that the use of absolute measures is not appropriate and then more relative measure of disparity, it was more appropriate.
We're actually getting into the conclusion. The absolute or relative, absolute shows the magnitude of the difference and is more appropriate in describing burden of disease whereas a relative comparison is more appropriate for making comparisons across health indicators.
Disparity, again, I want to emphasize the need for considering the context of inequality or inequity, the idea should be born in mind that it would be ideal to parse the avoidable and unavoidable, but this is difficult. You more accurately arrive at the idea of inequity, but still recognizing that differing philosophies, goals, and objectives will guide the choice of statistics.
The time line, Ken is hoping to have this finished by the end of the year. This meeting along with other, data user conference, the Healthy People Steering Committee Subcommittee are all going to be providing further input on our efforts. I'd just like to point out that the handbook work group includes those at the state and federal level.
Vic is on the work group committee, Susan Queen is on that in the federal, and we have some people, Tom Lavist from Hopkins, John Lynch from the University of Michigan.
DR. STARFIELD: What is PDG?
MR. PEARCY: Initially, I thought about talking about the differing philosophies, I forget what the D is for, goals. Philosophy something and goals.
DR. SHOENBACK: Data?
MR. PEARCY: Could be and that's all I have in the way of presentation.
DR. MAYS: Great. Thank you. Vic, could you take a moment also to just introduce yourself to the group? Some of us know you very well, but let's share that knowledge with everyone.
DR. SHOENBACH: Well, my name is Vic Shoenbach. I'm a faculty member in the Department of Epidemiology at the University of North Carolina at Chapel Hill where I've been for all of my professional career I guess in epidemiology, including my doctoral training.
I've worked with a variety of research areas, most recently HIV, STDs, and prostate cancer. I've done a fair amount of teaching. I have an evolving textbook on the world wide web. I'll show you the Web site for that if anybody is interested. It's great for people abroad who don't have the money to buy the published ones, and I've served on the American College of Epidemiology Committee on Minority Affairs with Vickie Mays. She is now chairing it. Is that enough?
DR. MAYS: Yes.
DR. SHOENBACH: First of all, thank you very much for inviting me here and thanks to Ken Keppell and the committee that he has been working with because that's where most of my material comes from and certainly my ideas.
I also want to thank Audrey Berwell who actually created the institute that I've been involved with annually now for a number of years that has helped me to learn a lot about this area and certainly to think about it. I also want to mention that this is a work in progress at least that what I'm going to be presenting.
I'm going to skip those topics because those are things that I think Jeff has talked about pretty well. I want to just skip to the point about the quantifying health disparities is a complex topic, especially when we start to analyze it in detail.
As the saying goes, the devil is in the details. The goal of eliminating health disparities can be stated simply and in general we know what is meant. In addition, many disparities are large and there is question about which direction our society needs to go.
There are many subtleties that arise in comparisons that are less dramatic, but nevertheless important. I want to underline the concept that these complexities are really not a reason or basis for holding up, waiting, or trying to figure out what to do in the large sense. The disparities are plenty big and we can tell what to do.
The first point to note is that different measures serve different objectives. One objective might be for setting goals and monitoring progress towards goals. Another could be evaluation of interventions. Another might be ideology.
A second area worth attention relates to the concept itself. Exactly what do we mean by disparities and the magnitude of disparity and Jeff has talked about those in a philosophical perspective, but there are also some other interpretational aspects that I'll be touching on.
A third area involves mathematical considerations and the availability and the nature of data with which to assess and monitor health disparities. Although measurement generally connotes mathematics and statistics, many of the issues that arise in qualification are, in fact, conceptual and political.
One basic question in the area of health disparity measurement is what groups qualify for comparisons of health disparities. Historically, the movement to eliminate health disparities can be traced to periodic reports about the health of African Americans. It goes back probably farther than this, but certainly at least to W E B Dubois classic work at the beginning of the 20th century to Gunner Merdol's AN AMERICAN DILEMMA published in the middle of the 20th century to the report of the secretary's taskforce on Black and minority healthcare report in 1985.
The focus of these studies was the health of African Americans and the gap between their health statistics and those for whites. One question is then, which groups should be compared in assessing health disparities.
Recognizing the changing demography of the US, the Heckler report devoted substantial attention to the health of Hispanics, American Indians, Naive Americans, Asian Americans and Pacific Islanders though there were major gaps in the data and there still are. Historically, in recent terms, the issue of health disparities centers around the health of the five racial ethnic groups that were defined by the Office of Management in the budget prior to the recent expansion.
However, all of the racial/ethnic groups in the earlier OMB classification or in the expanded one are heterogeneous. Latinos have numerous countries of origins that differ greatly in culture, lifestyle, and political economy. There are over 500 recognized tribes of American Indians with different cultures, languages, histories.
The people who we classify as Asian American and Pacific Islander represent about half of the world's population, speak multiple languages, have different religions and cultures, have vastly different histories and have very different socio-economic resources in the US.
Their health statistics are similarly diverse. For instance, Chinese Americans have the lowest rate of teen pregnancy, about 1 percent, and Laotian Americans have the highest, about 19 percent. And yet, we put these peoples all together and call them Asian American/Pacific Islanders.
When African Americans may have been born in Africa or in the West Indies or the US and African Americans who grew up in the southern US, may well have different health risks from African Americans who grew up in the north or in the west.
How many groups should be compared? If we further subdivide, that will influence measures of disparity. When the group is subdivided into heterogenous subgroups, then the disparity is between these subgroups emerge into consideration where before they were submerged.
In addition, the size of an overall measure of disparity will generally depend upon the number of groups being considered. If we divide a group into multiple groups, for example, we separate Asian Americans from Pacific Islander Americans, that will certainly increase a number of measures of disparity. It may actually decrease others.
Jeff and Ken Keppler are in a federal institution. They have responsibilities to actually do things and come up with answers, but I'm in a university and I have the opportunity to just think about concepts. The rest of my talk is going to be about concepts and issues.
One of them is, for which factors should we adjust the comparisons that we're making? There are certainly concerns about health disparities related to gender, rural residence, region of the country, immigrant status, length of time in the US, income and education.
These dimensions overlap, but are not co-extensive with ethnic classifications. In addition, disparities are strong related to many of these dimensions, yet most indicators also differ across ethnic even within the above dimensions. The need exists, or the concern exists that we may need to adjust or standardize the comparisons as we always do when we're making comparisons to death rates and we want to age standardize for example.
A fundamental question in making comparisons across groups is what other factors to control for. We routinely standardize comparisons of vital statistics to take into account difference in the age structures of populations and we often carry out separate comparisons for females and males.
It is also common to attempt to adjust ethnic comparisons for differences in socio-economic status. The conceptual basis for these practices bears examination. Consider first adjustment for socio-economic variables.
When an ideologic analysis is being carried out, for example, to assess the extent to which racial, ethnic differences and research occurrence are explained by socio-economic differences, the purposes of adjustment is reasonably clear. There are major limitations in our ability to make an adequate adjustment because of limitations in available data and the crudeness of socio-economic status measures.
When we translate this practice into analyses for descriptive and policy objectives, the rationale for adjustment is less clear. Adjustment for socio-economic status may not be indicated because of the implicit assumptions made about differences in socio-economic resources which are strongly related to health status and also to ethnic group.
When we adjust for socio-economic differentials, we reduce the reserve disparity among ethnic groups. Does the goal of eliminating disparity require only that the smaller disparity be eliminated leaving the substantial disparity that arises from differences in socio-economic resources.
If that were the case, then the movement to eliminate health disparities is headed for a major disappointment. The striking socio-economic differentials that characterize comparisons of ethnic groups to a great extent drive the health disparities that require elimination.
Even if increasing socio-economic resources for health disparities populations is regarded as outside the realm of public health, which is a somewhat questionable proposition nevertheless endorsed by many. The strength of the relationship between socio-economic resources and health status is not fixed by nature.
When we adjust for a factor, it takes it off the agenda of the discussion. Does that adjusted measure provide a more meaningful indicator of he extent of disparity? Not necessarily. For example, access to health care is much greater for people with greater socio-economic resources.
Since these people are likely have jobs that provide health industry, and more able to afford health care if they lack insurance. This differential in access contributes to difference in health outcomes and constitute disparity among ethnic groups. Yet, the link between family economic resources and access to healthcare arises from the need to purchase to healthcare and the absence of national health insurance.
Universal health insurance is a key strategy to reduced health disparities and would weaken the relationship between socio-economic differentials and healthcare and health status. Such a system would markedly reduce health disparities across socio-economic groups and, therefore, across ethnic groups.
But the effect on ethnic health disparities adjusted for socio-economic differentials may be smaller. The impact of adjustment needs to be considered in relation to the objectives of the measure of disparity. Similar issues arise regarding adjustment for other factors such as rural residence, family size, length of time in the US, and even lifestyle.
Successful efforts to reduce health disparities by providing additional healthcare resources to rural areas may not be fully reflected in disparity measures that have been adjusted for rural/urban residents if a larger proportion of minority groups lived in rural areas.
Even the universal practice of adjusting for age has implications that should be born in mind. Major health disparity groups, American Indians, African Americans, and Latinos are younger than the white population and in the total population. Age standardization to the US population can change the size of disparities that are strongly related to age.
However, unadjusted or crude comparisons are problematic to interpret as well. Here are several recurrent themes that have come in preparation of the handbook, but since Jeff has referred to all of them, I'm going to pass over them, but I do want to say something more about the question of absolute versus relative measures.
They are intuitive and we use them routinely in epidemiology certainly and in other fields, but in thinking about why we use relative measures or absolute measures when we have the choice, as we sometimes do, I think that inherent in the use of a relative disparity measure to measure change is the idea that a reduction in the more favorable disease rate indicates a change in scale.
A life saved is still a life saved which is an absolute perspective, but if the more favorable rate, for example, in the white population has dropped, then our expectation of the level of disease also changes so that the same number of lives becomes more important.
Is the situation similar to a comparison between a high mortality area where high death rates are expected and a low mortality area where high death rates are unexpected, a life in each area has the same value, but do we tend to regard them differently?
If overall infant mortality has fallen to 7/1000, then a rate of 10/1000 may be regarded as a greater disparity than a rate of 18 when the overall level is 14. So if you're comparing 7 to 10, that difference may seem to be greater than 10 to 14 even if the difference is actually smaller between 7 and 10 which is only 3.
Similarly, a gap of 200/100000 in one region for a cause of death may seem greater or less than a disparity against a background of 400/100000 than in one with a background of 100/1000000. Resolving these questions may require research on personal utility.
The concept of utility recognizes that people have different preferences which may need to be measured. Perception of risk is not straightforward. The is a large literature by people like Paul Slovak which explains the perception of risk is not straightforward and perception of utility may be not either.
What I wanted to do was to illustrate this with a couple of scenarios to the extent that we have time to talk about them. I'll put them up and we can talk about them at this time which is will give you an idea of the issues that I think need to be thought about and the kinds of research that we may need to do in order to reach some consensus on measures of disparity.
Here's a hypothetical scenario. Let's suppose that there are several alternate programs to reduce infant mortality that are under consideration. They may have different costs and that would be taken into account, but let's not worry about the cost for the moment.
Which program which reduce the disparity the most and which program represents the best choice overall? Here on the slide instead of labeling them with particular race/ethnicities, I called them higher and lower. This is the group with the higher mortality rate and this is the group with the lower rate.
The current situation is that there is a rate of 12/1000 for the higher rate and 6/1000 for the lower rate. The ratio is 2 to 1. One program might be expected to bring the higher rate down to 9 and the lower rate to 4.5 which would preserve the ratio as 2 to 1 and it would reduce the higher rate by 3/1000 which would translate into a certain number of infants.
Another program might be able to reduce this rate to 8, but it would actually lead to an increase in this rate from 6 to 8 perhaps by shifting of resources. There we have a ratio of 1 to 1 and that's presumably no disparity another is a greater reduction in the high rate from 12 to 8, reduction of 4.
Another possibility might be a program like that that would bring the 12 down to 8 and the lower group rate would also come down to 4. Here we have the same absolute reduction in the higher rate, but we have the 2 to 1 ratio which some people would say is the same disparity that we had in the current situation, i.e., no reduction in disparities.
Here is a situation where we still bring 12 to 8, but now we bring the 6 to 2. That's a ratio of 4 to 1, a much greater disparity in the relative sense and still it's brought the higher rate down by 4. That still doesn't feel good.
What about the situation at the bottom where we bring the group that has 12 down to 5 which is a reduction of 7/1000, the biggest reduction of all the alternatives, but it also brings the lower group rate down to 1 and the disparity is now 5 to 1 in relative terms?
If we were trying to evaluate the different proposals, which one would we say would produce the least disparity or the best improvement overall for the health status of the country taking account the goal of eliminating health disparities?
We can take the differences between these different values, here the difference is 4, here it's 6, here it's 4 and here's it's 6 and here it's 4.5. So we can choose on that basis or we can choose on the ratio. The ratio is often the one that people look at and Jeff was explaining why that might be.
I've heard Bill Jenkins, for example, say that in the last 50 years, the infant mortality rates for African Americans have come down substantially as those for whites, but the ratio is still 2 to 1 which would suggest it's not a reduction in disparity. Yet, when we think about the actual number of lives saved, you can have scenarios where there's a bigger absolute reduction than there is a relative reduction.
Here is another alternative. Several programs, for example, to reduce the prevalence of poorly controlled diabetes. So which program would reduce disparity the most and which program represents the best policy choice overall? Here I have multiple groups and I've given them names. These are all made up data.
The first row in red is the current situation I'm assuming and I'm assuming that the green row is the goal or the target. This should be a 6. That's a typo. Here is a program that doesn't look very impressive at all. It barely reduces this at all or this group or this group.
It makes a slight reduction in the whites and Asians, and Pacific Islanders. This program does a bit better for AMERICAN Indians and Latinos, but not at all for African Americans and none for whites and Asians and Pacific Islanders.
Here is another scenario like that, but the whites come down a bit and here's one where the Asians and Pacific Islanders come down a bit and here's one where we have a smaller reduction for American Indians and Latinos and no reduction for African Americans. I had some more alternatives, but my slide didn't hold them as they cut them off without thinking too much about it.
Without looking at these to try to figure out which we would like because they're actually more similar than I intended to show the ones are, this is the kind of decision that will have to be made at some point conceivably. If it's not a choice among different programs, it will be in evaluating which state did better in reducing health disparity or on which indicators are we being more successful in reducing health disparities or which institute should get more money because they have a bigger health disparities problem.
The decisions will be made on some level and they'll be made on the basis of numbers something like this. I guess my conclusion is that measuring disparity involves conceptual as well as statistical issues, that issues in measurement should not become a reason for delaying action, and that focus groups with stakeholders may help to establish consensus measures. Thank you very much for this opportunity to talk about these issues.
DR. MAYS: Thank you. Great. What I would like to do with our time is open it up for discussion, Let's try and do that within the realm of about 10-15 minutes and then I'm going to bring us back to a couple of questions that we talked about that I'll put back on the table for the presenters to answer and that really has to do with the philosophy, what the intentions are, whether or not in the guidance in this particular document the clarity of these issues are going to be there and whether the is going to be a prevailing philosophy that NCHS will actually recommend.
First, let's just start with questions and comments about what was presented. Then I'm going to bring it back o to the presenters to ask them to comment on those two statements I made.
MS. HEURTIN-ROBERTS: I think maybe my questions or comments or philosophical. You may not have to bring things back very far. I'm a little uncomfortable with the distinction made between inequality which is represented as a factual matter and inequity which is considered an ethical judgement. I'm not sure that inequality which we heard from Vic, is necessarily all that factual because what goes into the question of statistics and how we determine our statistics and choose our statistics are conceptual matters that are philosophically based. I don't know that the question of equality, the discussion of what's an avoidable voice and unavoidable disparity, when you say something is unavoidable, that implies that you fully understand the causes and reasons for a certain outcome.
I don't think we ever really can say that. I don't philosophically whether we ever reach that point, but in practice, I doubt that we reach that point very often. I think the question, as Jeff mentioned, is the question of parsing the avoidable versus the unavoidable is definitely problematic and I think probably that's something that really needs to be highlighted.
I'm not sure we can ever say that something is unavoidable. I also think that often what we consider unavoidable is not necessarily question of the reality of a situation. Historically, we've seen that what we consider unavoidable is perhaps really a question of will to bring something about or will to change something at a policy level. We need to be very careful before we can make clear statements about something being avoidable or unavoidable.
DR. MAYS: Do we have any other comments in the same range and then we'll bundle them together and let them comment? Any others around the same area?
DR. NEWACHECK: I have more of a background question. In terms of the handbook that's going to come out, can you tell us a little bit more context about it's usages? Is this going to be something that will hopefully or be planned to be used as a department-wide guidance for measuring disparities or is it just for NCHS or what is the intended audience for it?
MR. PEARCY: The intended audience, we wrestled with that a great deal and we generally came to the conclusion and I think Vic was in on this as well, that it would be most useful as a guide, as a workbook, as an aid in measuring disparity for people primarily at the state and local level rather than as an internal document. That's part of the reason we work hard to bring in people from at least the state level as well as from state academia.
DR. NEWACHECK: Is there a draft available for us as members of the committee to look at?
MR. PEARCY: There is a very rough draft. There is excessively finer versions. The first section is in pretty good shape. The second section, we're still wrestling with and then there will be a third section that will deal with specific measurement issues and how to calculate standard errors and what have you for the various one. It's not quite there yet.
DR. MAYS: Would you like me to request on behalf of it. I have it because I sit on the committee, but I will request to distribute it for comments. I think that's what they really need at this point are comments. I'll do that.
MR. PEARCY: For comments, that would be most welcome. I was misunderstanding you in terms of whether it's ready to be for use and in that respect, the answer is no. Comments certainly, yes.
DR. MAYS: For comments, I think we'll get it. That's why I added that little bit.
MS. COOPER: One of the concerns that I have is the whole issue in terms of health disparities and how do we actually operationalize it. We all have our own sense of knowing what we mean by health disparity. I know within NIH, we struggle with NIDA coming up with the definition because the definition was originally provided using medical model and I wasn't behavioral oriented.
As we come up with this handbook, is there going to be some operational definition of health disparities which is one of the problems we ran into with the February hearing, that is, people talked about health disparities. It was really all over the place. That would really be very, very helpful.
Everyone is working on the issue of health disparities within their own agency, but as we begin to have more studies come out and we would like to be able to look at the data or cross study and see where we're going and where is the next step.
DR. MAYS: Do you actually have a copy on a slide or do you have it? There is a definition. It really opens up with a definition of how disparity is being defined for the document. Do either of you have it?
DR. SHOENBACH: I have it with me. I think it's actually the NIH definition or at least we proposed using the NIH one. I'm not sure if that's what. I don't know that it will satisfy you. I'll read the definition of disparity. "Disparity is the quantity that separates groups in a population on indicators of health measured in terms of race, percents, means or other quantitative measures." I don't think it's going to satisfy.
DR. MAYS: That's why I wanted you to read it. I don't remember it exactly, but I thought you should here.
DR. SHOENBACH: Now, there is the NIH definition from the strategic plan. I could read that also. That's in the preamble. "Health disparities are defined as differences in the incidence, prevalence, mortality and burden of diseases and other adverse health conditions that exist among specific population groups within the United States." Again, I don't think that's going to provide total guidance.
MS. COOPER: The concern is that as this document or handbook comes out and people use it as a working tool, is it going to confuse people is it actually going to prove more clarity in the field and will there be some type of recommendation that as we try to measure health disparities, will there be a fourth set of questions that we'd like for studies to consider using so that at least in some area, we will be able to do some type o comparison?
DR. MAYS: Let me just ask, what I want to do is take these question and ask you will answer them or take them to the committee and get and answer back. The issue about avoidable and unavoidable, I think it's a very significant issue.
I think that as those who do research 100 percent with what was said which is that we hand' often a other search to know that with scientific certainly what is avoidable and unavoidable.
Particularly with the human genome project where it is, it also raises questions. That's a question for that group to really answer. The other is in the discussion of disparity, will the definition of the disparity be such that the few will benefit from the clarity for all or will it serve to narrow us to a statistically driven, not conceptually or problem driven orientation looking at the health disparities.
I would love to have these question answered and the other is can you predict what the impact of the release of this booklet will be. I'm feeling really bad and sat ion the room thought it was the greatest idea since sliced white bread and now I'm like wondering. What the impact will be to turning people to a formulas oppose to having states at their own level, consider disparity and come up with working does that fit the actual problem or conceptualization that they're working with. I'm going to take questions.
DR. FRIEDMAN: First of all, let me say that I really appreciate the work that NCHS has put into this and I also appreciate the extent to which you tried to bring some conceptual clarification through the issues and having said that, I do think that there is an issue that you put your finger on which is partially a conceptual issue, it's partially a policy issue, it's partially a value issue which is, in fact, are we really interested at a practical level in disparity per se or is disparity, in fact, not really what we're interested in?
Are we interested in reducing the gap between eh most disadvantages groups? I agree with Vickie completely. There is a potential danger f of the statistical measure and then undoing policy. I can't speak for most states, cities and towns, for the areas that I'm operationally responsible for.
My interest conceptional might be in disparity, but may practice interest is not in disparity, but in those health areas and those groups where there is difference between the most disadvantaged which generally white non-Hispanics.
What we're really looking at is a series of rate ratios and that's what drives policy. In fact, that is not necessarily a bad thing. In short practically speaking, the issue that you're going to be confronted with the he handbook is whether or not it's going to resonate.
I realize you've been trying to get feedback from states, but whether or not it's going to resonate with the states in any practical terms or whether or not it's not and particularly whether it's going to resonate with those states that have been particularly committed to both analytic and policy work in this area.
MR. PEARCY: May I just make a few comments that kind of go to a number of these things?
DR. MAYS: Are you in the same mode, Barbara?
DR. STARFIELD: I don't know.
MR. PEARCY: Just briefly, I recognize the problem in parsing out avoidable and unavoidable. I agree with Vic that we really don't need to solve that right now to move ahead with measuring disparity and working to eliminate it.
I brought these points up because I'm knit-picky. I think they're important to keep in mind and as a part of a research agenda down the road, we can improve our measurements. As far as being statistics driven or conceptual driven or goal driven, in the workbook, I didn't talk about this very much, the burden of disease and improving the outcomes of the most disadvantages, but it is covered extensively in the workbook.
It's just that for myself personally, I am not comfortable making decisions about how things should be. That's one of the reasons that I went the length to make the distinctions between some of these concepts. They are important things to consider in the long run, but should not hold us back at this time.
Also, the question was raised about how are we going to operationalize disparity? I think also it's reflected in the idea of how big a difference is disparity? I think certainly we can say that we need to improve outcomes without having an absolute definition or an answer to 1- deaths per 100,000 is close enough. That's not disparity. But it is, I think, and idea that has to be borne in mind down the road at some point.
DR. STARFIELD: It turns out it was related. Even though she's a guru, I think Margaret Whitehead did us a great disservice by putting in this avoidable stuff. We in the international society defined inequity differently. We have defined it as systematic and potentially remedial.
That puts the focus on thinking about what you can do about it not whether it's avoidable or not, but whether you can do something and remediate it. I would urge you to try to shift your focus from the negative to the positive. It's really very helpful in terms of elucidating what the influences on health that create these differences are. You can refer to this as the standard definition on our Web site.
DR. MAYS: Give them the Web site address.
DR. STARFIELD: WWW.ISEQH.org for International Society Equity in Health.
DR. SHOENBACH: You can find a link to it on our minority health project Web site which is at the bottom here. If there are any other Web sites that you'd like to have available on that, I'll put them on.
MS. HEURTIN-ROBERTS: I appreciate all the work you've done. It's wonderful and important work, but we can't be monsters. I can easily the head of some state public health department saying this is an unavoidable disparity. We don't have enough money so we're going to look at this and just turn a blind eye to this unavoidable issue.
What's avoidable and unavoidable changes over time with political will, political atmosphere, social atmosphere, social concerns as well. The other thing is that the ability to measure difference among groups is very useful.
To have this misused as a single measure of health disparities, what could happen is you could easily see health disparities being reduced, have your index getting smaller, but having the health of the population going down, conceivably among some of all groups.
It's just a small part of the picture and how this statistic is used in the context of other information and how it may be used to justify one policy over another is a big question. I think that this is something that really needs to be thought through carefully before you turn it out there to the state and other public health departments.
DR. MAYS: I know that they were aiming to actually have this finished. If I remember one of the e-mails by APHA which is November 2002. I guess on behalf of what's being said, I'm hoping that what you will do is actually take this back and raise it. If I'm on one of the conference calls, I'm happy to also help with that.
There are two things: One is, it is very powerful for this to be released by NCHS because you're looked upon as the analytic arm to some extent. Even if it's just an exercise to some extent where you say it's offered like guidance, at a state or even county level, you put it in the hands of someone and the question might be we can do this as opposed to we will think about this.
I think there is really great work in raising issues, but it may be that there needs to be one more chapter to be thought about and that is something about the philosophy, some of to stuff that Vic is presenting, some of the stuff in which it is clear about the implications of choosing A or B, but in language where it is really about the broader thinking about this level.
It's not choosing A or B in terms of the statistics, but in terms of a long term impact, but are there any suggestions or recommendations that anybody wants to offer before we closer this. We are going to move to Sarah Keim.
DR. FRIEDMAN: Not having seen the draft, what I'm saying may already be in there, but it does strike me that what I'm hearing and what I've said reflects a desire for not a choice among measures, but essentially a sweep of measures that taken together can illustrate a set of health problems.
I would suggest that it be considered approached like that and consider having a final chapter which is essentially a set of example figures, tables that could present what that suite of measures could look like for an example state or county.
Some of it might just be simple rate ratios using white and non=Hispanics as comparison groups. Some might be the index of disparity and so on, but something like that which would be, in fact, an illustrative how-to guide for how to think about the issues and present the data at a very practical level. It could be very valuable and also would be more likely to be adopted than saying you can choose A, B, or C and the implications of that choice is different for each.
MS. COOPER: And probably also some type of statement indicating the potential limitation and we only know at this point in time. We leave it open as if the document is still available even beyond the time of this publication so we let people know that all the answers are not there yet and this is continuing to evolve and maybe something from the agencies that are really doing some of the things that may be able to be contacted and use the information.
DR. SHOENBACH: I agree with that very strongly. This is an area where we're trying to leap ahead and we're ignoring the fact that really there has been fairly little methodological attention to this. There are serious questions about how one should measure disparity if it's going to drive policy and that deserves to be taken a look at not just by a committee of volunteers, but by people who in different places with different agendas who are exploring it and publishing and getting back and forth to meetings and so on.
It's definitely a work in progress and I've been arguing that it shouldn't try to be prescriptive because I don't really think we are in the position to be prescriptive knowingly. I'll make some summary comments in response to the questions and let you move on. Is that okay?
The question of what the impact of this will be and if you take these rate ratios, for example, and use them to make policy decisions. My feeling is that rate ratios don't drive policy and that they shouldn't. Choices depend upon many considerations as you know.
But even in relationship to the data, no measure can possibly accomplish that job by itself and maybe only a few can. For example, with some times, you would choose absolute difference. Where are the most lives being lost? You might pick the most burdensome condition. This condition is absolutely terrible. Alzheimer's disease people suffer for years and years and years.
That might drive more than the actual number of cases or the age at which it occurs. This one if occurring to people in prime of life, 20-30 and that is more important even if the numbers may be lower. Here is something that we actually know how to prevent, whereas this other thing we don't actually know how to prevent.
Prostate cancer has this very big difference, but we actually don't know what to do about it. One might pass up the largest in order to take on something that we do know something about. Whether we have the ability to prevent the problem or fix it with our current political institutions.
Healthcare insurance is something that absolutely would make a difference and we can't seen to get there from here or political will. I really think that they way these things need to be decided is by democracy. The groups so affected need to be at the tables saying what they want.
I don't think there's any statistician that can this is what you machete need to fix because this what the numbers come out. The numbers are certainly a part of it, but the groups that have lobbied and brought about this resolution, this goal are the ones that should be driving what they want, what they feel is most important. That would be my summary comment.
MR. PEARCY: I don't really have anything more to add. Perhaps except to go to Dan's comment and to reiterate Vic's comment that in the workbook, we have not in any way conceived of making a recommendation of this statistic or that statistic.
There is a number of views of health disparities and philosophies, goals, and objectives. We're trying to provide the most useful set of statistics along with some discussion of their limitation and application in addressing health disparities in various respects. I don't think we've gone to detail conceptually as Vic has raised. That certainly needs to be considered, but in terms of the statistical mechanics, it's very broad-based.
DR. MAYS: I want to thank you both for your time. Your presentations were very enlightening and despite everything that was said, please take away that we greatly appreciate the dialogue that the work is starting. It matches great with what this committee is doing. We really appreciate the opportunity to dialogue with you.
We also appreciate the opportunity that you'll let us comment on the actual draft. If you send it to Gracie and to Susan, they'll make sure that we get it distributed to the subcommittee for comment and give us a date by which you need it back and we will try our best.
Thank you both for your time. We appreciate your being here and you're welcome to stay as we finish the meeting or go to the airport whichever. Thank you. The next person who has graciously given us time and stayed beyond that time is Sarah Keim.
Again, on our phone calls one of the things that we talked about is a new study that NICHD along with several other partners are wanting to launch. It would be a longitudinal study and they have a process in place by which they have work groups.
One of the work groups is on disparities and that's part of why we thought at this point in time it would be very useful to learn about the study and know that there is a disparities work group and that they are currently looking for ways to measure, questions, thinking about disparity.
This just fits very well and given the interest o this subcommittee, we are happy that you were able to take some time to bring us up to speed on what's going on. Welcome.
MS. KEIM: Thank you. Did everyone get a handout? I'm Sarah Keim from the National Institute of Child Health and Human Development at NIH which is the home for the National Children's study even though it is very much an inter-agency partnership.
I'm the study coordinator for the study. Unlike most of the people who have graciously put in time working on the study, this is my full-time job. There are only three of us for whom this is our full-time job.
We're growing very quickly as the planning phase moves forward and we get closer and closer to enrolling our first pregnant women in 2005. I'm going to cruise through quite a few slides and I don't mean to gloss anything over, but if you have questions, feel free to interrupt or ask later or send e-mail.
It's a very complicated project and trying to focus on disparity, but also to give you a general overview of the project. Fortunately, the presentations right before me have given an excellent foundation to the difficulties in measuring disparity. It's an enormous and daunting task.
I'm going to go quickly through the first ones. In epidemiology, it's very hard to collect data that's both useful and accurate on disparity because of the classification issues, methods, and definitions. We're going to do the best we can. Projects like the one we heard about today and others will help inform us to do a better job and look at those things.
Looking at racial and ethnic groupings, it also has economic status as defining characteristics of disparity and how those have been associated with each other, poor health outcomes, higher exposures from the environment, and trying to get a grasp of all of these things and how we can measure them in a longitudinal study like the National Children's study.
Think about whether the disparities are cultural, genetic, in terms of exposure, proxies for racial and ethnic differences, socio-economic status? Are all of these things interrelated? Trying to sort out confounders and what are independent and dependent variables of each other.
It's a massive web of complexity. Often epidemiologist rely heavily on socio-economic statutes. Even that, how do you define that and what differences in socio-economic status are associate with and including environment issues which is something that's very core to this study.
Even in terms of developmental stage and how child development experts would think about disparities. Children experience critical windows in development and vulnerabilities via those windows to environmental factors are a major concern in child development and how disparities interact with exposures in those critical windows is a major issue for child development and overall health.
A little background about the National Children's study. It's a longitudinal cohort study proposed by the President's Taskforce on Health and Safety Risks for Children which was established in 1998 under the Clinton Administration via executive order.
The taskforce has since been renewed by the Bush Administration and is co-chaired by the administrator of the EPA and the Secretary of HHS. It's a very active taskforce. In thinking about the task to better analyze risks posed to children' health and safety, they determined that it was an impossible task without much more research and formally proposed a study like this to better be able to address those risks to children.
Another critical milestone in the development of the study was the authorization of the study through the Children's Health Act 2000 which is a very complex piece of legislation. One section pointed out the need to conduct a national longitudinal study of environmental influences, including physical, chemical, biological and psycho-social in children's health and development.
It specifically mention the need to address health disparities in the legislation. It' something we've taken very seriously and are looking for ways to adopt the most innovative technologies in measuring certain chemical exposures, for example. We want to adopt some of the more innovative ways to measure disparities and better be able to address them.
We think this study can be a very powerful tool to do that where other smaller studies have not been able to do that because of size or scope. This is very much an inter-agency project with several components of HHS involved and the NIH. CDC particularly and the new Birth Defects Center and the US EPAS Office of Research and Development and representation from the Office of the Surgeon General.
I mentioned the President's taskforce, but in January 2000, the question came up is something like this even possible. It's always been a dream in the epidemiology community and is it doable to do a national study like this to address the health and safety environmental risks.
In January 2000, we held a first consultation where the directors of major cohort studies came together and discussed feasibility. For example, we had representation from the Women's Health Initiative, the Bogloosa Heart Study, the Framingham study and so on. We had a larger consultation in December 2000 to start focusing on specific issues.
Here are our basic study concepts forming the framework for the study that it be a high-quality longitudinal study of children and their families in their environment, national in scope and what exactly that means is a very difficult thing to define, that environment be defined very broadly to include the chemical, physical, behavioral, social, and cultural environments so that it's not just a study of toxicants, pesticides and so on, that it be of sufficient size to study the common range of environmental exposures and less common outcomes.
Outcomes occurring on the level of incidence of a sample size of 100,000, that it take a look at the interactions of the environment and genetic expression, incorporate state-of-the-art technology in terms of tracking measurement, data collection, storage and analysis, involve a consortium of multiple agencies and take advantage of public/private partnerships, be a national resource for further studies so that the data be made available widely and very early on so that as much analysis as possible can be done.
The scientific rationale for a study like this rests on the concept that children have an increased vulnerability to environmental exposures because of their behavior, size, the critical windows of development and history has shown that exposures to some agents have caused serious development effects such as the experience with lead and alcohol.
We know there are current exposures of very high frequency that we don't quite understand their full impact, that existing studies have addressed many issues in children's health, but size and scope often limits what they can do and questions they can answer.
A study is needed to identify the effects of these exposures or to assure safety of many of them. We want to be able to tell not only what is harmful to children, but what is harmless what is helpful. We want to be able to learn a lot more about healthy development, not just abnormal.
It be of a longitudinal design to be able to infer causality and be able to look at multiple exposures and multiple outcomes. You can read the next two slides. They're pretty completely laid out. I'll skip over those for purposes of time.
Some considerations about sampling. We know the sample needs to be generalizable to the US population, but needs to focus on the issues facing some certain study subpopulations including women of childbearing age being able to look at very early pregnancy and fertility effects of certain exposures and that we be able to address the health considerations of certain high-risk populations such as agricultural communities, communities living in heavily industrial areas, communities that face a serious economic disadvantage for example.
Currently, we're weighing potential core hypotheses and we do know that in examining possibilities, we need to apply certain criteria. We know that this study will not be driven by one hypotheses. It will be multiple hypotheses, but it does need to be hypotheses driven, that public health significance is key to determining what the core hypotheses are and how you define that as also very complex and tricky.
We have folks from NCHS and the rest of CDC who can help us with that. The hypotheses demonstrate the need for longitudinal assessment. We want to be able to take advantage of the longitudinal design, that it also require a study of this size and we're talking 100,000 children and want to take full advantage of having that many children and be able to look at subtle differences between subpopulations and outcomes that are less common.
The hypotheses have a theoretical rationale that makes sense and that it be feasible measurable, that it's something that's possible to be done. here are some sample hypotheses related to health disparities that we've been discussing lately.
A lot of these are originating in the health disparities and environmental justice working group which has been instrumental in articulating some of these issues in a way that we can measure for the study. The first one related to in utero and early childhood exposure to diesel exhaust and how that alters immune function in children and adults.
Psycho social stress is associated with adverse pregnancy outcomes and those outcomes are related to disparities we see. One related to injury and socio-cultural factors that socio-economic disparities in general and we can be very specific about them, but we want to be able to collect data to be very specific and look at how they mediate and modulate environment effects in child health.
We're also embarking on a project to engage communities in the design of the study. As was mentioned, we have several working groups working pulling together the state of science, what will work and won't work, what is something that is critical to be done in s study like this. One of those is the health disparities working group.
We have a federal advisory committee that has had its second meeting now and provides valuable input and judgements on what should and could be done with a study. The study assembly involves all the stakeholders interested in the study and we have meetings about once a year to discuss current progress and bring in input.
This is our organizational chart. It is in the slides and on our Web site. We have not just the health disparities working group that works on these issues, but others related to community outreach, health services and the social environment. This has been focused a lot of developing guidelines for the project and a framework for how to measure disparity and human subjects projections and how we involve our community.
The community outreach working group is a very hands on group working on developing strategies and guidelines to get community input into development of the study and health services as its name implies discusses healthcare access and related issues that influence health outcomes and the social environment working group.
We're planning to have another study assembly meeting in the late Fall and the advisory committee is meeting in September and then in the Fall with that larger meeting. We've been undertaking quite a few pilot studies to address a lot of the questions that remain before we embark on implementing the study and there are many more to be answered.
Here are some of the key issues currently under discussion, some of the very difficult-to-answer issues a lot of which we're working on right now relating to sampling and health disparities is integrated into that whole process. For example, will it be possible to design the study so that those represented? How do we do the sample in a way that's feasible, but in a way that we can collect the kind of data we need and address the issue of health disparities?
Here's our time line you can take a look at. Here's our Web site, National Children's Study. gov. YOU can look at that and get an updates about our progress, and you can e-mail the address listed there to join that or to ask any questions you might have. Thank you.
DR. MAYS: Great. Thank you. What I'd like to do is to open it up for questions or comments in terms of clarity and then what I'll do is address both you and Leslie how we might be helpful. Leslie sits in the health disparities work group. Let us address how can we be helpful to you? Questions, comments, need for clarity.
DR. STARFIELD: I just wonder how much advantage in the tremendous knowledge about longitudinal studies. I have really great doubts about whether this can be generalizable because of the way you're going to go about selecting the sample. You're going to have to put out an RFA and invite people. How can it possibly be generalizable? If it's not, how can you make any conclusions about disparities? You're selecting centers by the excellence of their research proposal. How can that possibly be generalizable?
MS. KEIM: Before we select study centers, we're going to come up with the sample strategy that we want to see implemented. We need to be able to select centers that can build their sampling locally into that overall strategy.
We can't have half the centers in New York City if our strategy tells us we only need 1,000 kids from New York City. Part of the criteria for evaluating the selection of those centers needs to be, how do they fit into the overall sampling strategy? If they don't, that's not a plus for becoming a center.
In terms of the other cohort studies, we have on our working groups, folks who are very involved in those other studies and we've been in very close contact with some of the older studies that can still provide a lot of very good guidance talking to them about best practices, what works, what doesn't work, both scientifically and in terms of process and strategy.
We know there are, however, some big differences between the ability to do a study like this in the United States versus in certain European countries. In some respects, it's easier in those countries to carry out such a national sample. We're trying to take as best advantage of those as we can and through our interactions, through the working groups and other ways.
DR. NEWACHECK: What's the expected retention rate over the 30-year period? You're starting with 100,000. What do you expect to end up with?
MS. KEIM: We're probably going to have to actually enroll more than 100,000 and those calculations haven't been made. We need 100,000 children to answer certain very important questions about health outcomes, some that don't come about until far after birth into the teenage years. We need to be able to backtrack and make calculations based on the numbers we need at various life stages. That decision can't be made until we have the selection. The 100,000 has been the working number.
DR. NEWACHECK: For the starting sample or for the endpoint or midpoint?
MS. KEIM: The number that sounds good and we know it's not going to be a million kids or 10,000 kids, maybe 200,000 kids. It all depends on what stage do we need certain numbers to answer certain questions.
DR. NEWACHECK: I'm wondering that because in the US you lose about 10-15 percent each 10th. That means if you really want to have 100,000 years out or 20 years out, you would have to have 200,000.
MS. KEIM: It's a very big challenge and it's something we're trying to think of strategies to combat that loss to follow-up. We'd like to be able to take advantage of things that have worked very well in certain studies and those are some of the reasons we have a working group like the community outreach working group. We need close interaction with communities.
DR. STARFIELD: I have real doubts about whether this could be a hypothesis generating exercise over 30 years because you've got a cohort and things change over time. So what's true for one cohort may not be true for another cohort. Hypotheses, I realize, they're just examples, are incredibly simplistic given the fact that they don't into account developmental effects and changes within cohorts over time.
I could think of better reasons to do a study like this rather than hypothesis testing. The more important thing to me is to include all the relevant variables and to brainstorm about what are relevant variables rather than to test hypotheses over this period of time.
DR. MAYS: Let me take Suzanne's question also and then you can take them both as a way to wrap up.
MS. HEURTIN-ROBERTS: Could you say more about what it means to draw a nationally representative sample? Representative of what at what time? I think the population is going to be changing over the long time that you're going to be in business. I'll tell you up front that my concern is how do you know, what are you doing to make sure that smaller groups such as Pacific Asian groups or Native American groups don't wash out, that there is enough information and enough data collected on them that it's worth their while to participate?
MS. KEIM: For all of the questions related to sampling, I'm giving our speculation as what probably will work. We have an analysis ongoing right now of all the various sampling strategies that we could undertake and then analyzing each, the pros and cons of each and doing some very thorough analysis of what actually is going to work.
We know, for instance, we can't do a pure random sample of all the kids born in such a such year. We're going to have be center based in order to carry out the research. We can prescribe on those centers criteria or standards for oversampling certain populations that they have in their community, that we know have health outcomes or exposures that are of particular concern.
We know we're going to have to oversample. How you define representative? Everyone has their own definition and we have these competing concerns that we have need to deal with. What comes out at the end is going to be uncertain right now.
We know we need to be able to address health disparities and how that impacts the sampling strategy is a major criterion. In getting to your questions about hypothesis, in order to field a major initiative and get all the support we need to do it, we need to be able to demonstrate that these are the hypotheses that will serve our goals.
At the same time, we know we're going to be collecting a lot of data about a lot of things and a lot of things and questions that we don't even know exist yet, concerns about exposures and things that are going to come up 20-30 years from now.
While it has to be hypothesis driven, we know we're going to be collecting a lot of data in general and we need to preserve the data and, for instance, specimens, for analysis subsequent for questions that come up later and that gets into the national resource idea I mentioned earlier that we want to be able to have this data available to be able to do analysis on questions we haven't even thought of.
MS. COOPER: I wanted to make the point in terms of how others may participate and that the uniqueness of this study in the sense that it involves private partnership so there's representations.
MS. KEIM: One thing we've been doing that is fairly unique for an endeavor like this is our ability to involve experts from outside of the government at this very early stage in actually planning the study and designing the hypotheses and the protocol that will eventually go out.
The federal governmetn has lots of rules about conflict of interest and financial obligations and things like that which we're tip toeing around very carefully, but we feel we've been pretty successful so far and have 275 scientists involved on our working groups right now and putting in input and with a lot of enthusiasm about the study.
We feel that getting the best people involved is going to make this the best effort we can put forth, just having this be a federal effort isn't going to cover all of our bases. If you think you're interested in participating on a working group, putting in some solid time and working on these things in a group, just dash off an e-mail to the e-mail address I have listed up there and our office can communicate back and forth with you about those opportunities.
MS. COOPER: All of the information will be available on the Web so that when the RFPs come out, everyone has equal access?
MS. KEIM: Yes, very much.
DR. MAYS: Thank you. We kept you over your time, but I appreciate your ability to be able to stay and share the information with us. We will also be in touch in the sense that if you go to our Web site, you'll see some of the background information about health disparities that might be useful in terms of questions that people have raised and commented here.
MS. KEIM: We'll be monitoring all of yours too.
DR. MAYS: We'll do the backup for Cheri. Thank you. Okay. On our agenda, several things and let's see how we're going to zip through some and focus on a few. That's going to be our strategy here.
The racial privacy initiative that's on the agenda, I can actually do that quite quickly and I want to thank Audrey Burwell from the Office of Minority Health for sharing this information with us and I also had consultation with a couple of the California groups.
Just so that you know, I guess it got brought and I thought, okay, we should talk about it quickly. It's not something that we need to worry about at this moment in time in the sense that the racial privacy initiative was instigated by Ward Connelly who is our regent at the University of California.
In it, similar to the SP-1 and 2, he is objecting to the collection of information with race attached to it. Several people have raised concerns about what this would do in terms of health statistics. In terms of medical data, there is some guarantee that it may not impact medical data, but in terms of health statistics which is a little different, there are some concerns about it.
But as you can see both in terms of the qualification for the ballot and as I understand it also for reasons of when strategically it is wanted to be on the ballot, I think this is to be delayed. As I understand it, it may not be on the ballot until 2004.
At this point in time, the campaign director is no longer even there. I do think this is something to just think about as you make a case for why health statistics in racial/ethnic groups are important to just be laying the groundwork. It is predicted that this will revisit us in 2004. As we were saying today, 2005 is around the corner so 2004 is probably a short time away. Thank you for that one.
The hearing and we're also going to blend Cheri Crute who is our writer into that when we talk about the hearing. As all of you know, we've been talking about this issue of the measurement of health disparities and racial/ethnic minorities, looking at the adequacy of data to respond to the notion of disparities.
We had the February hearing and part of what we said was wanted to move to the next level of discussing some of the same issues we discussed at the federal leave with the states, but also recognizing that the states have some very different and unique issues which is why I'm very gala that Dan is able to be here with us who will represent the states.
It's a burden he'll be okay about I think. We've had on our conference calls some discussion about what is it that we want to achieve in the state hearings, what are the state issues and who should we be inviting. The sum of our discussion has been one of the things we want to do is to look at what are some of the states that actually have very good data collection and mechanisms in place that they've been able to integrate not only issue of racial/ethnicity, but they're dealing with multiple race and some of the other variables that we say are somewhat emerging to be important in terms of looking at the health disparities of racial/ethnic minorities.
At the same time, we wanted to hear in that time from states for whom this has been somewhat difficult despite the fact that those states may also expericne a disparity. If you'll turn to the tentative agenda that we have for the meeting of July 18, we can walk through this a bit and discuss it.
We also wanted to look at slates. We thought slates was a good example. We wanted to look at parallel surveys. When we had our hearing in February, if you remember, Raynard Kington brought that up. Unfortunately, Raynard is not available on the 18th. We're probably going to have to hear from him at another time.
In talking with some of our state colleagues, there is the issue of the reporting back and forth of, for example, states collecting data in one form, but needing to report to the feds in a different form and the guidance from the feds that go to the states.
There is an interaction of data that goes back and forth and what happens in terms of it being reduced up or reduced down and there may be some concerns about it. If there were any other issues that were characterized in what we wanted to hear from the state hearing, somebody else should probably remind me that those were the ones that I thought were the critical ones.
Dan, was there anything else? We're also including in here the broad overview about some of the vital statistics also. When the states talk, they will also be talking about vital statistics as well as surveillance and it's more broadly in terms of the type of data that they're going to be talking about.
DR. NEWACHECK: (Conversation off-mike.)
DR. MAYS: What it will raise is two issues. One is part of why the date was selected and I thought we polled and I thought Barbara wasn't available. I thought you all were, but may not have been is that we were doing it on the tail end of the data user's meeting. We thought that we would actually have several of the state people there at that meeting. We've been looking at the list. It may not be so impossible that we do or don't.
DR. NEWACHECK: We see names from people from a lot of the states. We don't see any of the names we discussed here in our meetings. I'm thinking that these people are probably analysts. Initially, when we thought about this, we were thinking it was the way that NCHS used to do it when they combined the public health conference on records or whatever it was called with the data user's meeting and we had a larger audience come in and there were eight registrars and so forth and it looks like that's not really the case.
DR. MAYS: We just got the list today that we were looking at and part of what we're trying to determine is whether or not they were already there or if we were talking about a big bill in the sense of many people having to be brought it.
Since we looked at the list and they're not there and it looks like you're not there, we may go back to the drawing board about the date. Let me ask the question because the issue came up about September. We talked about the state hearing and the hearing in which we would look at specific subpopulations that we've had particular difficult with the collection of data and that's Native Americans and the Asian/Pacific Islanders.
The Native Americans are having a meeting right after our meeting in September and it's in Denver. The question would be whether or not the committee would be able to leave our meeting and get to Denver to have a meeting the next day.
You were suggesting it would have to be the next day as to whether or not that would be feasible. It would mean that the hearing is actually in Denver. The Native Americans are having a large meeting there. It would make it very easy for us to hear from them.
The Asian/Pacific Islander group, several of whom I think we would want to hear from are also on the West Coast. We could either choose to stay in Denver or move you yet again to like San Francisco of LA to actually hear from them. The question is:Couldd we, after the September meeting, potentially meet with the Native Americans; is that a possibility?
DR. STARFIELD: What date?
DR. MAYS: I think it's the 27th.
DR. STARFIELD: I think the full committee is 25, 26.
DR. MAYS: I think we were talking 27th.
DR. STARFIELD: I know I can't make it.
MS. COLTIN: I could do the 26th.
DR. MAYS: I don't want us meeting Saturday. No need to have precedent set or anything like that. We all want to go home by Friday.
MS. COLTIN: It means spending Saturday flying anyway.
DR. MAYS: Oh, sorry. Some of us can get home by that night for a change. We actually have a little more time then because if that many of you aren't going to be there and we don't have the advantage we thought we would have at the data user's meeting, I guess it's the one where NCHS is doing its presentation. I don't think as many people are coming in as has been in the past.
Let us sit and look at a date. Can I just bring up some logistics around what the issues are and to figure out how you'd like to do it? The issues that we wanted to continue to have hearings on is the Native American, Asian/Pacific Islanders, language but we couldn't do language in a short verse, but really it's about half day I would say and then the state issues.
What do you see as your pleasure in terms of whether we can combine some of this? I don't think you want just hearing after hearing. We can't even find the time to do them, but let's talk about being able to combine some things if possible. Suggestions.
DR. NEWACHECK: It seems like if we're going to do the Denver meeting, we would meet half day.
DR. MAYS: Would you like the other half of the day to be the API because we're on the West Coast. Major offices are going to be in LA, San Francisco and here in DC. It would be between those places? The state does need to be separate and we were already jamming it into a day.
I'm wondering if we can do language relative to a separate time or to try and do language built into November 19 and 20 to either stay? I don't know if we're ending at 1:00 or 3:00. If we end at 1:00, we could ask for the rest of the day to do it.
If we're ending at 3:00, then we have a problem. We'd have to end up starting another day. What do you think you want? I shouldn't ask at 5:00 about anything about travel. Let's see if we can do it this way. As a West Coast traveller, we're very cognizant about this. If we don't get out by a certain time, we're really hurt.
Why don't I try and see what the schedule is going to be for November and see if even what we could do is a block of two hours for the subcommittee and then two hours afterwards. If they're ending at 1:00 and then we could take another two hours, we can do it and we're done even though it's not like everybody is there together rather than having a whole separate trip. That would be my preference.
Does that sound doable? Remember, I have to be on that phone call to make sure when we do the schedule for November I can see if we can do it.
MS. COOPER: Just one question. I know my agency has very limited money so travel.
MR. HITCHCOCK: NIH has limited money.
DR. MAYS: We feel that way on research grants too.
MS. COOPER: I was just thinking that when special groups have meetings that are already set up, do you ever send representatives from the committee or if there is a committee member who is going to be attending the meeting, to try to identify what are some of the issues of concern so you can bring it back to the larger committee?
I was just thinking of that because I know that the API's for the first time are going to have a large HIV meeting in California in September. I don't know if the data coincide with the meeting you're planning on holding or somebody else is going to be there. There may be a larger audience that might be available and you might be able to do some brainstorming and try to identify some of the needs of the committee and what their needs are.
DR. MAYS: That is in September. I got some information about. In essence, I'm not sure we send people to something, but if somebody is already going to be there, it might be whether or not we can ask questions or something. It may be informal, but on a formal level, we're FACA committee and things have to be formal for us.
MS. COOPER: I'm not going to be there, but I just wanted to get that out because there might be a possibility of getting feedback from some of the special ones.
DR. MAYS: We can see at the gathering what the outcomes of the meeting is in terms of issues for them.
MR. HITCHCOCK: Can I make a suggestion of an item as a distraction and we probably can't do much about it and that's the state data reporting to federal agency policies. There's one federal policy around race and that's the OMB policy. The Congress gets involved in it to a much greater extent than you might in determining what agencies have to report.
We've had many committees looking at that over the years I've worked at HHS. Nobody has really moved it schematically. They're pushing it forward. This committee can get involved directly on this without really rethinking of centering on some aspect of it. It would not be a good use of our time.
DR. MAYS: This was raised by Dan or Bruce in terms of the concerns about the need to report data in ways in which they wanted at a more broad level and then the reporting requests that go back to the feds are such that you have to scrunch it down.
DR. FRIEDMAN: We collected at a real level in terms of ethnicity and national origin and language. We collect it differently than we report it because there's a conflict between the census/NCHS standard and the revised federal standard.
DR. MAYS: That was what you were getting at.
DR. FRIEDMAN: That's an issue and there's a second issue which is we don't have a consistent implementation policy from federal agencies on what's acceptable and what's not.
DR. MAYS: There's not much we can do about that?
DR. FRIEDMAN: Dale knows more about it than I. I'm not convinced. What I've seen is a lot of confusion and sloppiness and lack of coordination.
MR. HITCHCOCK: We may be saying the same thing. I'm just saying I don't think this is the right group to really look at that.
DR. MAYS: It's a broad issue that doesn't apply specifically to the group we're looking at. It's out of POP. You have to take it to another group then. Can you look at the outline and just get some quick feedback before we bring the meeting to a close? We need to have 21st Century meet after this.
DR. FRIEDMAN: To me, it looks really good. My only question was North Carolina. I was just wondering what North Carolina was doing there?
DR. MAYS: In the second group, it was really to try and get a sense of what are some of their needs and procedures and policies. North Carolina was suggested as one that had little resources, but seems to be able to do well. They're in the group of those that were looking at policies and procedures and their needs and they've managed in some ways with not a lot to do well. We think that some of it is the contribution of the university.
DR. FRIEDMAN: I can't speak to how they deal with racial/ethnicity data, but generally it's regarded as a state that does extremely well and is relatively well-researched as well in this particular area. That was the reason. They would contribute to a hearing. I was just surprised.
DR. MAYS: I see what you're saying. That was just a suggesting that was made on-line and these are things that we need to check out. Anything else that you see as a little out of whack?
DR. NEWACHECK: (Off-mike.)
DR. MAYS: I think there was a person identified. Steve Blumberg. I didn't have his name at the time we did this. What we will do is look at a different date. We can generate the questions and have the notes that we've talked about. It's really taking off the same questions we asked at the federal level with some more specific ones. I will get some guidance from Bruce about some of the vital statistics issues.
I'm glad that Kathy distributed hers. They're very well done and a good model for me to see how to do them. I appreciate having them. I had hoped to be finished by September and you all tell me I wouldn't. I've just given up. Now, 2005 seems around the corner. That's an idea. Who's going to go to APHA in Philadelphia? It's over on the 13th. Somebody asked me to do something after that. I think it may start like the 9th.
MS. HEURTIN-ROBERTS: It actually starts on the 10th.
DR. MAYS: It's over on the 13th. It's November 10-13. Remember we have a committee meeting November 19 and 20. The 13th is what, usually they're over on a Wednesday. It would mean doing it like Thursday. No, we could do it during.
Our biggest thing would be getting space. APHA takes up everything. I will discuss this and see what's possible, but actually that is a good suggestion. We could look at time and see what we could do. Cheri, we have probably just a few minutes left. I thought that would be good at this point is to have you introduce yourself and for us to just comment on where we're headed in terms of the hearings and what some of the materials are that we'll need to pull together.
The committee had expressed that it would be useful to us and now we can even do it better is that as we proceed in the hearings, it's important for us to know what we've done already, to go to our archives and see what recommendations we have and what testimony we have and to look at that as a blueprint so that as we conduct new hearings, that we don't cover the same ground.
That's probably going to be one of your first tasks that now we don't have to rush and get to July 18th. It might be useful for people to comment and particular those who are left to comment a little bit about some of the earlier products that you're aware of that we should look at.
MS. CRUTE: My name is Cheri Crute and I've been writing about health and medicine for about 15 years. I have interviewed many of your colleagues. I am one of those people who is an end user of things like the Haines and Framingham work study.
This is an interesting process for me to be in the beginning of the process and understand how those numbers come together. My purpose is to objectively and clearly represent all of the things that you say in this room and to put an emphasis on the subjects that are most important to you.
All I ask for in return is minimal or at least clear explanations of said jargon and clarity and letting me know if something is critically important to you or if you need to get it through me through Dr. Mays.
DR. MAYS: Anything in particular that you want her to focus on first in terms of pulling from the archives. Debbie did a great job of giving us an overview of the things that are there, but I don't know if there is any particular report, particularly if it's something that we won't recognize right away as being in POPs.
MS. COLTIN: Look at that both for a model for a report as well as the fact that there are within that report some important comments and testimony that we heard.
DR. MAYS: Was there something in quality too that you want?
DR. NEWACHECK: I agree with Kathy about that model. It's a similar process.
DR. MAYS: Anything else. It's been a long day and we still have another hour to go.
MS. COLTIN: If it comes up, you'll see it in the Medicaid report, but one of the things that we've found has been universally appealing in the reports that we had it is to pull out comments from people who testified that really highlight a point or an area where we want to make a recommendation. It adds life to a report that can be otherwise somewhat dry.
DR. MAYS: They have a history that I can't even begin to touch. It's important that as we go through this since I'm just beginning this process that you feel free to call on them also.
MS. CRUTE: There is actually some of that in the draft of the 21st Century report. It's the first thing that caught my eye in terms of some of the points.
DR. MAYS: I want to thank everybody for staying around and your great questions. I feel like we've done a lot in this short amount of time today in terms of our piece of contribution for making the world a better place for at least today. Thank you.
(The meeting was adjourned at 5:14pm.)