Godfrey Woelk of the University of Zimbabwe served as discussant.

On May 5, international experts discussed these complexities at "Making Disparities Count: From Government Statistics Systems to Action," at the Harvard Conference Center in Boston. The event was the final installment of the "Health Disparities & the Body Politic" series, organized by the HSPH Working Group on Health Disparities. Major co-sponsors were the David Rockefeller Center for Latin American Studies and Harvard Center for Society and Health. For a full listing of other co-sponsors or to watch webcasts of all three symposia, visit http://www.hsph.harvard.edu/disparities.

Today, disparities within and between some countries appear to be increasing. "If we could mobilize that information and present it both to the leaders and to the public in various countries, we believe there are basic tendencies in the best instincts of people to try to worry about the disparities and do something [to address them]," explained HSPH Dean Barry Bloom. "So, this in a sense is our effort to create a call to action."

Nancy Krieger

The other side of the sword is when data is not collected at all, which can hide that a problem exists, Krieger explained. "Whether and how health disparities are documented is thus a key sign of the priorities of those in power," she said.

The U.K. Experience

The first scheduled speaker, John Fox, director of statistics for the Department of Health in the United Kingdom, was unable to attend due to a family emergency. With Fox's approval, Krieger walked the audience through his PowerPoint presentation.

The U.K. has a long tradition of considering social inequalities in health, dating back more than 100 years. The country also has produced two of the best-known reports on health inequalities, the Black and Acheson Reports.

In power since 1997, the Labour Government has developed a platform to address health inequities, issuing about one major report each year, with a theme of monitoring indicators to track progress. Importantly, they have linked policy proposals to the budget of the Chancellor of the Exchequer, ensuring that worthy mandates can be funded.

The overall aims and objectives of the government's platform revolve around several ideas. Subjects, such as education, that can cross government agencies have been identified in intersectoral work. Health programs on specific issues, such as cancer, have been established. Target populations for intervention have been identified, and the need for making health data and policies available at local levels has been recognized.

The government has aimed to develop action plans for reducing health inequalities based on four primary ideas: supporting families, mothers, and children; engaging communities and individuals; preventing illness and providing effective treatment and care; and addressing the underlying determinants of health principles.

"What they are showing is that there needs to be targets and also that progress needs to be measured en route to getting to those targets," said Krieger.

Among the measures being taken by the government nationally or locally are the following: Impact assessments, which gauge the impact of policies on population health, are being emphasized; The Decennial National Report will soon contain data on many health outcomes, not just mortality; For the first time, several health questions have been added to the U.K. census; Electronic medical records are being established; Funding has been allocated to improve neighborhood statistics and to give people access to their neighborhood data, since local data are needed to inform local policies.

A health poverty index, capturing data from individual to societal levels and from the root causes of health inequalities to the health situation itself, has also been produced. Such an index can be used to compare the health of a particular area to the health of the population as a whole.

"The key point that [Dr. Fox] emphasized to me was that they have taken the question of data infrastructure very, very seriously and are really putting resources into it," said Krieger, "because they do want to have policies that are guided by evidence. They do want the evidence very much to influence what the policies are, and it's that tradeoff back and forth between them that they are working on."

The French Experience

French studies on health inequalities are numerous, but they rarely focus on ethnic criteria, explained François Héran, director of the Institut National d'Etudes Démographiques in Paris. Instead, they emphasize what Héran described as indicators of precariousness: poverty, insecurity, and vulnerability.

François Héran

The second kind, statistical files, such as sample surveys, are created for the sake of knowledge and not to directly intervene in someone's life. Generally speaking, sensitive questions may be asked in a public questionnaire provided that two criteria are fulfilled: are the questions relevant and proportionate to the objectives of study and have they been authorized by written consent.

Information on race and ethnicity is available-or at least can be inferred-through a number of surveys in which written consent is not required. For example, a family survey associated with the French census in 1999 carried questions on former citizenship, country of birth of participant, country of birth of both parents, languages spoken by the parents, and date of entry into France.

Yet, to some degree, the availability of such information is not as important to French researchers as other kinds of data, such as socioeconomic status. The lack of emphasis revolves around several issues. Universal health care coverage and other social policies apply to the entire French population, as well as to foreigners with permits-there is even a program to aid homeless people and illegal foreigners. There are no ethnic lobbies that demand being counted to obtain official recognition. There is concern that singling out groups may cause stigmatization. For example, when the results of an HIV/AIDS survey were released, some newspapers exploited the fact that more than half of infected immigrants came from sub-Saharan Africa, said Héran.

The researchers themselves have voiced concerns. One of the biggest is that statisticians wish to remain independent from the limitations that would be imposed on them by a central, official classification system.

The overall question, then, is whether ethnicity is the right variable to emphasize. Most researchers in France would say that categories in general, including those about ethnicity, have less explanatory power than investigating trajectories of precariousness, said Héran. This broader approach helps researchers recognize that foreigners and immigrants tend to accumulate disadvantages, while considering other factors that may lead to health disparities.

The Brazilian Experience

There are two main sources of health data in Brazil, explained associate professor Eduardo Mota, chief of health statistics at the Instituto de Saude Coletiva in Bahia, Brazil. The first is the National Health System, which provides health care for about 70 percent of the population. The remainder engage in a private system. The data include information on births, deaths, medical care, and epidemiologic surveillance.

The second source includes population sample health surveys and the census. Those provide data more frequently used to analyze health disparities.

	Eduardo Mota and Vickie Mays

To address some of these issues, the Brazilian government launched a pilot project in 2000, issuing 13 million people a national health card, which provides access to services as well as a sign of societal inclusion. The card carries allows easy access to outpatient care data, and carries data on along with age, sex, race (skin color)/ethnic group, place of birth, place of residence, education, and occupation.
The next challenge is to reorganize health services so that they can use the newly available information, said Mota, to make people use the health card and health professionals to accept changes. This challenge becomes more important as Brazil moves from a centralized health care system to a decentralized one, in which local communities expect to play a larger role in the health of their people.

The U.S. Experience

UCLA Professor Vickie Mays has spent the last four years chairing a Subcommittee on Populations for the National Committee on Vital and Health Statistics (NCVHS) in the U.S. The NCVHS is a federal advisory committee that reports to the director of the Department of Health and Human Services.

Race/ethnicity is a social-not biological-construct, Mays pointed out. The terms are used in public health to compare the health of groups of people and to serve as proxies for social/psychological issues. Also needed are data on socioeconomic position, which is not always available in U.S. health data sets.

The two most critical questions right now about the direction of health disparities research in the U.S. are:

• How should health disparities be measured?

• How is progress determined?

Underlying both questions are ethical/policy viewpoints that need to be addressed. Mays believes that lessons need to be taught in the country's programs of epidemiology and behavioral sciences that statistical decisionmaking carries ethical considerations. Each decision has implications for certain people's lives.

Other challenges exist. Despite federal guidelines, there remains confusion about what constitutes race/ethnicity and nationality, particularly when it comes to federal versus state data collection. States tend to require more detailed information, which can reveal important information. Linking these kinds of data between state and federal governments will be crucial in understanding patterns of heath disparities.

Data classification is another issue, particularly when considering the increasing numbers of children or young adults who can identify with more than one racial or ethnic group. This issue extends to more than checking off a box on a survey. While a care provider may judge a patient as African American based on skin color, the patient may actually consider her American Indian ancestry as pertinent as her African ancestry to her family history of health risks. Classifications based on how one is treated, due to racial discrimination, are not necessarily equivalent to those concerning geographical and cultural ancestry, and both may be related, albeit in different ways, to health risks.

In addition to state and federal data collection efforts, the private sector is rapidly emerging as a credible source of information on race and ethnicity in relation to health. Aetna Health Insurance recently found that nearly 80 percent of new applicants volunteered information about their race, something that Mays says indicates the public's willingness to have such data collected to improve health, as long as confidentiality is protected.

Godfrey Woelk, associate professor in the Department of Community Medicine, College of Health Sciences, at the University of Zimbabwe, served as the discussant for the event. He observed that: "We're concerned with addressing differences in health status that are unnecessary, avoidable, and unfair. Clearly, I think all of us would agree, [that] equity in health and health care is a particularly important topic because unfairness or discrimination-disadvantage-is hazardous to one's health. The challenge however is to make these disparities visible to the general population, to decision makers as a whole, in order to generate action." He added that identification of both advantaged and disadvantaged groups is necessary and requires improved statistical methods.

Following the main speakers, a panel convened that included: Howard Koh, Associate Dean for Public Health Practice, HSPH, and former Massachusetts Commissioner of Public Health; Evelynn Hammonds, Professor of the History of Science and African and African American Studies, Harvard University; and David Rehkopf, HSPH Doctoral Candidate. Mary Waters, Professor of Sociology, Harvard University, was unable to attend, but did provide a question posed to the panel through Robert Blendon, HSPH Professor of Health Policy and Management, who moderated the panel discussion.

Conversation about Data Collection Continues over Informal Breakfast

At an informal breakfast in the Kresge cafeteria the day after a May 5 symposium on health disparities, HSPH Associate Professor Nancy Krieger summed up the overriding concern of participants: "It's amazing how much we consistently underestimate how bad things really are."

Krieger anchored the early morning event, which gave HSPH students a chance to discuss issues with the symposium's speakers: Eduardo Mota, chief of health statistics, Instituto de Saúde Coletiva, Brazil; François Héran, director, Institut National d'Etudes Démographiques, France; Vickie Mays, professor of clinical psychology, UCLA; and Godfrey Woelk, associate professor, University of Zimbabwe. The fifth invited participant, John Fox of the Health and Social Care Information Centre, UK, was unable to attend due to a family emergency. Also present was Lisa Berkman, chair of the Department of Society, Human Development, and Health, who helped organize the health disparities symposia series.

Although they work in different nations, the speakers reported challenges in getting accurate information that is necessary to develop and evaluate health care policy and to monitor health disparities. They also expressed concern about coordinating data retrieved from different sources.

Woelk of Zimbabwe said that census information is incomplete in his country. Not all births are recorded, and the same might be true for 20 to 30 percent of deaths. Zimbabwe's socioeconomic data is also spotty. Often, data sets are not linked because different ministries oversee different populations.

"It's difficult to link or even get access to this data because access is controlled," he said. "That's partly because you want to have control and partly [to maintain] confidentiality."

Mays of UCLA said that there are challenges in accessing and coordinating survey data in the U.S. In particular, noted both Mays and Héran of France, there is great difficulty in accessing data from or about prison populations. In the U.S., the reason stems from differences in jurisdiction over prisons, jails, juvenile detention centers, and youth diversion programs.

"The ability to get information varies, ability to collect data varies, and jurisdiction for institutional settings varies tremendously," Mays said. What you end up with is a data hodgepodge, and that leads to confusion and inaccuracy.

"We are missing a big number of people, especially [in data related to] infectious disease and HIV/AIDS," Mays said. "We end up missing [data for] those individuals who are most likely to be in need of disparity-related services."

Mota talked about his work in Brazil, where census figures set the health capitation funding doled out equally among the states.

"We need to recognize that it's not reasonable to put the same amount of federal money in [a state that needs little] as in [a state that needs more]," said Mota. "That's the great challenge of the health information system."

Even in France, which guarantees health care to all residents, disparities in services can result from poor data collection and coordination.

"It's possible to have access to medical care without papers [in France]," Héran said. "You only need a declaration and signature for a number of services of care." Yet, it's tough to register illegal migrant workers, who often have no permanent address or who are afraid of the government.

Most participants agreed that, even when data is collected, it can be biased by terms, categories, and subsets of categories chosen for surveys. Mota asked, for example, Is poverty the same in one neighborhood as another? Who should decide: the collector or the respondent?

"We may live in the same neighborhood and have characteristics of being poor," Mays said, "but there are differences between us."

Krieger noted the many common themes running through participants' reports: the need for local data, who gets a say in what these data are, and how are the data presented and interpreted.

"The way [data] translate in different countries depends on their histories," Krieger said, noting that which social categories are used in each country's health statistics reflects their economic and political priorities and context.

In helping to plan to the symposia series, Berkman had thought "different countries would illuminate different issues." Instead, she found "how they are similar, and different only by a hair's breadth."

--PHC