Toward a Health Statistics System for the 21st Century: Summary of a Workshop (2001)

William Kalsbeek, Member, Committee on National Statistics, and Edward Perrin, Workshop Chair

It is clear from the presented papers and ensuing discussion at this workshop that the nations health statistics system, while reasonably comprehensive in content, is unable to fully meet the nations current and anticipated health information needs. It is also apparent from the workshop discussions that existing health information initiatives elsewhere in the world may provide a springboard of ideas for improvement.

Comments from workshop participants suggest that dealing with the current system’s shortcomings will require more major redirection than minor incremental changes in the systems organization and function. Indeed, the workshop discussions point to a search for innovative ways to gather health data and to transform it into useful health information, not just shoring up existing components of the system. Participant comments also seem to endorse the encouraging trend toward interagency cooperation that has expanded the utility of several health data systems in the past 10 years.

Many important shortcomings of the current health data system emerged in the course of the workshop. Some of the more provocative shortcomings mentioned by attendees (named in parentheses) were the following:

• Content overlap among existing data generating sources (Norwood, Rice)

• Uneven access to proprietary data generated by the private sector of the health system (Kosecoff)

• Limited standardization in reported data items among data sources (Kosecoff, Rice, Wolfson)

• Failure to broadly disseminate health information to public health professionals (Kronick)

• Limits in the amount of data sharing among federal statistical agencies as a way to broaden the utility of the data they generate individually (Rice)

• The absence of broadly focused coordination and vision in the gathering and dissemination of health data (Norwood, Rice, Wolfson)

• An insufficient capacity to generate needed information at the state and local levels, as well as for other small but important subgroups of the U.S. population, such as young children (Klerman, Rice, Sondik, Wolfson)

• Limited ability to longitudinally link disease prevention and treatment to health outcomes in the general population (Kosecoff, Kronick, Rice, Sondik, Wolfson)

• Several notable content gaps in the present health statistics system including:

• health system performance (Kronick)

• health-related quality of life in general (Kronick), and in children in particular (Klerman)

• indicators of functional capacity and other indicators of “positive health” in children (Klerman)

• health assessment for children who are disabled, institutionalized, or possessing other special problems (Klerman)

Papers and discussion from this workshop seem to suggest that some form of integrated planning is needed to improve the usefulness of the current decentralized system. There are at least two possible initiatives that could address important gaps in the nation’s health statistics system. One is the creation of a separate, broadly focused, and semi-independent health information planning board with the responsibility and authority to provide overall focus, direction, and coordination to the national health data system. Patterned after the Australian Institute of Health and Welfare, the Canadian Institute for Health Information, and comparable panels in other countries, such a board might serve a central role in forging an ongoing partnership of health providers and insurers, as well as public and private

users and producers of health data. The board’s main function could be to orchestrate and broker the production and dissemination of population-based health data in the United States. Some of its major specific aims might therefore be to:

1. identify and organize information needs at all levels of the health care system;

2. establish a working collective partnership with all producer agencies that contribute to the health data system;

3. negotiate a clear set of information gathering roles for each member of the partnership;

4. develop and implement a long-term plan to promote the collection of standardized, high-quality health data;

5. develop and implement innovative ways to gather needed health data;

6. promote the privacy and confidentiality of all publicly acquired data, and explore ways to defeat accompanying threats;

7. support the development of new methods to effectively use collected data to answer important health-related questions being asked by the health community at all levels;

8. determine who decides the content and standards for health indicators for national health and health systems performance comparisons;

9. promote access to all existing health data by public and private researchers and policy makers; and

10. explore new outlets and uses for health information based on current information technology.

At first glance, some of the goals of this proposed planning board may seem to overlap with the goals of two existing advisory bodies, namely, the U.S. National Committee on Vital and Health Statistics (NCVHS) and the Department of Health and Human Services (DHHS) Data Council. It is important to note that there are clear distinctions in the roles. NCVHS serves as the statutory public advisory body to the Secretary of Health and Human Services in the area of health data and statistics. In that capacity, NCVHS provides advice and assistance to DHHS and serves as a forum for interaction with interested private sector groups on a variety of key health data issues. NCVHS is composed of individuals from the private sector. The DHHS Data Council coordinates all health and nonhealth data collection and analysis activities of DHHS, including an integrated data col-

lection strategy, coordination of health data standards and health information and privacy policy activities. The Data Council consists of all assistant secretary and agency administrator-level DHHS officials who have a direct reporting relationship to the secretary. The proposed planning board differs from these two bodies in many ways. First, the planning board would be “semi-independent,” meaning it would be a partially public body, not solely an internal federal body. Second, the planning board would be composed of people from both outside and inside the federal government, but not federal employees who could be affected by the recommendations of the board. Finally, the planning board would ideally operate at an integrated national system level, and thus could identify, suggest, and coordinate plans for all data producers, and not be limited to impacting the programs and plans of DHHS.

Another new initiative suggested by this list of shortcomings of the current system is the development of a comprehensive national longitudinal health study of significant magnitude to better understand biological, behavioral, treatment, and policy effects on whole-life health outcomes. This might be accomplished largely by applying existing health data gathering apparatuses to a general population sample based on a rolling cohort of births. A “birth” in this context might be interpreted initially as participation in a population-based household sample, and then subsequently to natural birth and immigration. The sample of “births” could be followed continuously until “death” occurs naturally or by migration to another country. After gathering baseline data this rolling cohort sample could be followed up to obtain a full range of needed data tied to health-related attitudes, behavior, and provider care. Besides continuously monitoring the health experience of cohort members and their families, linkable health services data could also be continuously gathered from their health insurers and providers. An example of this sort of comprehensive information system is the one currently under development in the United Kingdom (National Health Service, 1998).

An underlying purpose of an ambitious longitudinal sample-derived portrait of the American public would be to expand the budding synergy among existing health surveys into a broad and significantly more important source of health information. Each member of this reoriented confederation of the nations most important health surveys would become an important component of a more comprehensive design paradigm, thereby creating a larger whole whose information benefits are significantly more than the sum utility of these surveys by themselves. A key to the success of

such a longitudinal birth cohort study would be, of course, the creation of a reasonably nonthreatening division of information gathering roles among those agencies with a vested interest in these roles.

As with any longitudinal study that contains data linkages, careful attention must be paid to the issues of data privacy and confidentiality. The protection of the health data of study participants must be a priority, but will be a challenge nonetheless. In addition to the obvious needs of privacy legislation and technological protection solutions, efforts must be made to demonstrate to the public the benefits to be gained (or the information opportunities to be missed) from this study and to assure participants of the privacy of the data. Other countries that have already faced these obstacles and implemented solutions can again serve as models for our nations system. For example, as mentioned above, the United Kingdom’s National Health Service (NHS) outlined a strategy for their health system that includes lifelong electronic health records for every person in the country and round-the-clock on-line access to patient records and information about best clinical practices for all NHS clinicians (National Health Service, 1998). Similarly, a goal of the Canadian Advisory Council on Health Infostructure is to create information resources on factors affecting the health of Canadians and to improve privacy protection within the health sector (Advisory Council on Health Infostructure, 1999). The privacy and confidentiality of health data are very important and are not just a statistical issue. Efforts would need to be undertaken by all levels of government and society to ensure that the challenge of privacy protection is overcome without compromising the knowledge to be gained from a national longitudinal health study.

General responsibility for developing survey instrumentation, data analysis, and information dissemination for surveys addressing each information focus would be the responsibility of an agency partner; developing the design and gathering the data in this longitudinal study would fall under the purview of the health information planning board. By dividing responsibility this way, an agency with an information focus need not be consumed by data production, but rather can focus its efforts on data measurement and information generation and use. Beyond the health of particular relevance to their focus, each agency would benefit by having ready access to lifetime person-level data that would enable them to explain health outcomes or to evaluate efforts designed to improve these outcomes.

Challenging but solvable problems would be faced in fielding and producing findings from a national longitudinal study. At the planning stage

the content of existing health surveys would need to be reevaluated and perhaps prioritized around established information needs. Some survey measures would need to be developed or improved to meet new or emerging information needs (e.g., quality of life, gauging the quality of provider care). Creative technological solutions would be needed to preserve the privacy and confidentiality of all gathered data. To minimize followup nonresponse and the risk of a “Hawthorne effect” on the respondents’ health experiences, new ways to easily, efficiently, and innocuously extract data would need to be developed. To avoid measurement softness in portions of the respondent data record, uniform standards and definitions would need to be established and enforced for all study questionnaires. The sample would need to be large enough to facilitate the production of acceptable estimates for states and other “small areas,” including specific race/ethnicity groups of interest. Finally, new or improved statistical methods to produce small-area estimates would be needed.

Another possibility to consider when thinking of the existing national surveys is a redesign and standardization that would allow for state and local estimates. Such a redesign would benefit states and permit flexibility in data collection and use. Furthermore, statewide data collection infrastructures are evolving, and as standardized sources of non-inpatient data become more readily available, the ability to link and aggregate these data into multi-state and national statistics increases, thus suggesting future reevaluation of the current national survey systems.

Although not explicitly included on the formal agenda of the workshop for discussion, there are valid historical and political reasons why the current system evolved as a set of independent “silos” spanning a large number of federal agencies. The many bureaucratic and “turf” issues and constraints that make it difficult to achieve the goals of an efficient data system were also not discussed. A health information planning board and a comprehensive longitudinal study would require a “surrendering of turf” by our major health agencies along with significant changes to remove practical and political barriers before they could be implemented. Those complex issues merit further work on how to motivate the desired organizational and individual behavior, perhaps drawing on expertise in organizational structure, human behavior, and societal and economic incentives. Without the insight of such work, a history of failed reorganization and cooperation attempts will continue.

Continuing growth in health care costs, in the complexity of the health care delivery process, and in the public’s focus on matters of health, makes

having timely, relevant, high-quality health data an essential part of the nations future information landscape. In addition to raising many key issues we must face, this workshop points to an even greater need to search for and critically evaluate new ways to assemble and use health information. In failing to do so, we are reminded of Goethe’s words that “nothing is more terrible than ignorance in action.”

Advisory Council on Health Infostructure 1999 Canada Health Infoway: Paths to Better Health. Final Report. Health Canada Publications, Ottawa. Available: <http://www.hc-sc.gc.ca/ohih-bsi> [July 12, 2001].

National Health Service 1998 Information for Health 1998–2005: An Information Strategy for the Modern NHS. National Health Service Executive, United Kingdom. Available: <http://www.nhsia.nhs.uk> [July 12, 2001].