National Academies Press: OpenBook

The Aging Population in the Twenty-First Century: Statistics for Health Policy (1988)

Chapter: 10 Enhancing the Utility of Statistical Systems

« Previous: 9 Health Services Utilization
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 202
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 203
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 204
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 205
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 206
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 207
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 208
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 209
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 210
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 211
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 212
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 213
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 214
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 215
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 216
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 217
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 218
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 219
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 220
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 221
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 222
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 223
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 224
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 225
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 226
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 227
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 228
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 229
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 230
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 231
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 232
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 233
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 234
Suggested Citation:"10 Enhancing the Utility of Statistical Systems." National Research Council. 1988. The Aging Population in the Twenty-First Century: Statistics for Health Policy. Washington, DC: The National Academies Press. doi: 10.17226/737.
×
Page 235

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

lo Enhancing the Utility of Statistical Systems INTRODUCTION The United States has a wide variety of data bases related to the health of the elderly national data bases, state data bases, and some private data bases. The purpose of this chapter is to identify actions that can be taken to improve the utilization of existing data bases for policy analysis for the elderly. It addresses the third charge to the panel "to determine whether changes or refinements are needed in the statistical methodology used in health policy analysis or in the planning and administration of programs for the elderly and to recommend actions or further research." The recommendations in this chapter are much broader than those in earlier chapters because they generally concern all aging-related data bases rather than ad- dressing a specific data system, a characteristic of recommendations in the previous chapters. A few of the previous recommendations are special cases of the broader recommendations in this chapter. This small amount of duplication is intended to help individuals con- cerned with improving data resources for a specific policy area, e.g., long-term care, since it might not otherwise be clear that a general recommendation would be important and relevant to a specific policy area. The first section discusses issues pertaining to national data. This country has a large decentralized data system, at least 19 dif- ferent agencies have responsibility for production and distribution of statistics related to one or more aspects of aging (Wallman, 1985:6~: 202

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS Department of Agriculture Economic Research Service Food and Nutrition Service Department of Commerce Bureau of the Census Department of Health and Human Services Administration on Aging Alcohol, Drug Abuse, and Mental Health Administration Centers for Disease Control Health Care Financing Administration National Center for Health Services Research National Center for Health Statistics National Institute on Aging Office of Human Development Services Office of the Assistant Secretary for Planning and Evaluation Social Security Administration Department of Housing and Urban Development Office of Planning, Development, and Research Department of Labor Bureau of Labor Statistics Employment and Training Administration Office of Pension and Welfare Benefit Programs Department of the Treasury Statistics of Income Division, Internal Revenue Service Veterans Administration 203 Coordination, planning, and priority setting for these multiple data sources are required to make possible the analysis of data from multiple sources a capability frequently required by policy makers. Several issues of coordination and planning are addressed, including content, coverage, data detail, uniform definitions, periodicity of data collection, and accessibility of data. The role of administrative records in policy making and research is also considered, as is the importance of designing surveys and administrative record systems in a manner that permits linkage between data systems. The second section recognizes the importance of states both as sources and users of federal statistics and identifies several actions that could increase the value of federal statistics for use by states. Private sources of health-related information are reviewed briefly, primarily for the convenience of the reader who may wish to use the data but also to round out the panel's coverage of available data

204 AGING POPULATION IN THE TWENTY-FIRST CENTURY to ensure that no recommendations are made that would lead to duplication. NATIONAL DATA The pane! believes that the federal government should place a very high priority on greater coordination and more consistent data policies among programs that produce statistics related to the elderly population. Currently, public policy is being developed in a context of major changes that affect the demand for and delivery of health services, as well as major changes in many other facets of policy that affect the lives of the elderly. Changes are likely to continue in the foreseeable future as budget constraints interact with population shifts and varying public demands. Such an atmosphere emphasizes the essential need to have adequate information to evaluate the effects of policy changes; unless attention is paid to ensure that as changes occur information ~ available to evaluate them, public policy will be made in a vacuum. The issues that face this panel, though focusing on the health of the elderly, by necessity are interdisciplinary in nature. Issues of health cannot be considered without an understanding also of social, demographic, and economic characteristics and dynamics. As a result, although one agency may assume primary responsibility in providing a data series, in most circumstances several agencies will have a vested interest in developing data systems, preparing them for public use, and conducting various analyses. These facets of the need for information on the elderly require a high degree of interagency cooperation and planning. It is with these concerns in mind that the pane! makes its recommendations. Background on Federal Statistical Policy The U.S. federal statistical system has developed as a decentral- ized system: each department and major agency has one or more components that produce data related to the mission of the depart- ment. Responsibilities for the collection, processing, analysis, and dissemination of statistics are shared among program and statistical agencies. In the Department of Health and Human Services, for ex- ample, the National Center for Health Statistics is a major federal statistical agency, but several program agencies such as the Social Security Administration and the Health Care Financing Administra- tion have large statistical program based on administrative records,

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 205 and the National Center for Health Services Research and Health Care Technology Assessment and the National Institutes of Health collect extensive statistical data as part of their research activities. More than 60 years ago, a report prepared for the Congress by the Bureau of Efficiency noted (cited in WalIman, 1985:1) Practically every bureau in Washington collects or disseminates statistics of one kind or another; and there is much confusion in the public mind concerning the work done by the various offices. The statistics collected by the Government relate to nearly every aspect of our economic and social life. Statistics of agriculture begin with the seed and follow through to the marketing of the ripened product. Statistics of manufacturers extend from the mining of crude ore to the production of the manufactured article; those of commerce, from the lighting of rivers and harbors to the consumption of imported commodities; and those of social relationships from a mere enumeration of population to elaborate data regarding the incidence of disease. Decades later, in 1985, the Office of Management and Budget iden- tified more than 70 agencies having outlays of $500,000 or more for statistical activities in any of fiscal years 1984 to 1986 (U.S. Office of Management and Budget, 1985~. Although there have been many debates over the relative merits of maintaining a decentralized system (in contrast to a centralized statistical system such as exists in Canada and a number of European nations), that aspect of the system has not been changed. There are costs resulting from this decentralized structure related to ensuring the development of a comprehensive, nonduplicative, accurate, effi- cient, and accessible body of statistical information for the analysis of social, economic, and environmental issues. Problems in the data system are particularly likely to arise either when agencies share substantive responsibilities or when no single agency can be clearly identified as the most logical one to assume a responsibility. "For example, difficulties in ensuring the availability of a comprehensive and consistent body of data to explore the condition and problems of particular populations may arise when no single agency has lead re- sponsibility for meeting information needs related to that population group" (WalIman, 1985:2~. Increasingly, in public policy forums, issues cut across substan- tive areas. These interrelationships mean that quantitative measures in one area are likely to be of significant use in related areas; data

206 AGING POPULATION IN THE TWENTY-FIRST CENTURY bases need to be combined, Ed consideration of possible relation- ships needs to be built into the processes of creating, expanding, and adjusting any new information systems. As the complexity of the society, the economy, and the envi- ronment increases, the demands for data to better understand rapid changes also increase. Statistical agencies require a Tong planning horizon in order to have the lead time needed to collect and process data relevant to current issues. In the current fiscal environment, fed- eral agencies need to undertake a serious long-term planning process, coordinated with planning of other statistical agencies, to address those areas for which social and policy change can be anticipated and formulated and for which data needs can be identified. Legislative Authority for Statistical Policy Office of Management and Budget Since 1981, the legislative authority for the coordination and oversight of federal statistical programs has stemmed from The Pa- perwork Reduction Act of 1980 (44 U.S.C. S.3501-3520, in particular S.3504 (ad. That act establishes in the Office of Management and Budget (OMB) the Office of Information and Regulatory Affairs (OIRA) and charges the administrator of OIRA with, among other duties, statistical policy and coordination functions. The specific duties detailed in S.3504~) are as follows: The statistical policy and coordination functions of the Di- rector shall include: (1) developing Tong range plans for the improved perfor- mance of Federal statistical activities and programs; (2) coordinating, through the review of budget proposals and as otherwise provided in this section, the functions of the Federal Government with respect to gathering, interpreting, and disseminat- ing statistics and statistical information; (3) developing and implementing Government-wide policies, principles, standards, and guidelines concerning statistical collection procedures and methods, statistical data classifications, and statisti- cal information presentation and dissemination; and (4) evaluating statistical program performance and agency compliance with Government-wide policies, principles, standards, and guidelines.

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 207 The act also assigns to OMB the responsibility to review and ap- prove information collection requests (such as statistical surveys or administrative record systems) proposed by agencies. OIRA reviews such requests to prevent excessive burden on respondents, to pre- vent duplication of information collections among and within agen- cies, and to ensure adequate coverage of important issues (WalIman, 1985~. OIRA currently is not pursuing the goals of this legislation as aggressively as members of the statistical community and of the office itself might like. To meet the statistical policy mandate of the law, a statistical policy office was established in OIRA headed by the chief statistician. OIRA's initial authorization under the Paperwork Reduction Act of 1980 expired in 1983, after which the office and its activities were funded on a year-to-year basis from OMB's general appropriations (Council of Professional Associations on Federal Statistics, 19863. Since 1983, however, that office has had a staff of only six professionals to carry out the required functions for all statistical programs of the government. (By contrast, in previous years, the size of the statistical policy staff in the Office of Management and Budget reached as high as 69 in 1947, when the statistical programs of the federal government were substantially smaller and less complicated.) As a result of staffing and fiscal constraints, the statistical policy office has had to be selective in the specific projects and tasks that it has initiated. The office is not currently focusing attention on statistics on the elderly; "active efforts to coordinate or improve the scope and quality of information on the elderly population are not being undertaken by OMB's Office of Information and Regulatory Affairs" (WaDman, 1985:93. Congress recently passed The Paperwork Reduction Reautho- rization Act of 1986, providing funding for OIRA for fiscal years 1987-1989 and amending the original act to correct problems sur- faced during oversight hearings conducted by the House Committee on Government Operations and the Senate Committee on Govern- mental Affairs. A new requirement of the 1986 act with respect to the statistical policy and coordination functions is that the direc- tor of OMB include in an annual report to Congress (Council of Professional Associations on Federal Statistics, 1986:2~: (A) a description of the specific actions taken, or planned to be taken, to carry out each function;

208 AGING POPULATION IN THE TWENTY-FIRST CENTURY (C) (B) a description of the status of each major statistical program, includ ing information on (1) any improvements in each such program; (2) any program which has been reduced or eliminated; and (3) the budget for each such program for the previous fiscal year and the fiscal year in progress and the budget proposed for the next fiscal year; and a description and summary of the long-range plans currently in effect for the major Federal statistical activities and programs. Congress is expecter] to continue to focus on OMB's implementa- tion of its authorities. Recently the House Committee on Science ant] Technology requester] that the General Accounting Office inves- tigate whether the OMB is "improperly en c] unnecessarily limiting executive branch agencies in collecting, analyzing, ant] disseminating information" (COPAFS, 1986:3~. A`l~ninistration on Aging A more specific legislative mandate for coordinating fecleral sta- tistical programs specifically on the elclerly population is in the Older Americans Act of 1965, as amencleci, which assigns to the Acimin- istration on Aging (AOA) the responsibility to "gather statistics in the fielc] of aging which other fecleral agencies are not collecting, and take whatever action is necessary to achieve coordination of activities carrier] out or assisted by all departments, agencies, and instrumentalities of the Fecleral Government with respect to the col- lection, preparation, ant! dissemination of information relevant to oIcler inclivicluals." The act further assigns to the commissioner of AOA the responsibility to acivise, consult, ant! cooperate with the heat] of each fecleral agency or department proposing or aciminister- ing programs or services substantially related to the purposes of the act en c] requires that each fecleral agency consult ant] cooperate with the commissioner (Waliman, 19853. The AOA has similar constraints to those of OIRA in terms of resources available for these functions (only one incliviclual is as- signec} to work on statistical issues, but the area of responsibility is more narrowly clelimitecI). The agency reviews plans for statistical programs and publications of those few agencies that request such review, but floes not actively attempt to coordinate relater] programs · . ant issues among agencies. With respect to statistics on the aging population, the most notable problem resulting both from the constraints on resources available for statistical policy oversight en c] coordination ant] from

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 209 the manner in which those resources are allocated is the lack of atten- tion to identifying unmet data needs and to ensuring that such needs are met through modification of existing data systems or creation of new ones. ~fonnal Coordination Activities Although only the OMB and the AOA have legislative authority for coordinating federal statistics on aging, other agencies have initi- ated activities to improve interagency coordination. Recently some progress has been made. On May 2, 1986, the National Institute on Aging and the Bureau of the Census cosponsored a Summit on Aging- Related Statistics. The purpose of the summit was to determine how the federal statistical system can provide the data needed to answer policy questions for an aging society in a cost-efficient manner. Par- ticipating agency directors prepared statements of their views on the vital issues regarding the elderly and agreed to establish an Intera- gency Forum on Aging-Related Statistics to encourage cooperation among the federal agencies in developing data on the older population (see Chapter 1 for more discussion of the forum). The agencies were commended for this initiative at the June 3, 1986, Senate hearings on statistical policy for an aging population, held by the Subcommittee on Energy, Nuclear Proliferation, and Government Processes of the U.S. Senate Committee on Governmental Affairs and the Subcom- mittee on Aging of the Committee on I,abor and Human Resources. Congress recently appropriated funds for NIA for a second year of forum activities. Most of the funds are used for contracts with the Census Bureau and the NCHS to provide for intra-agency coordina- tion of forum activities in these two agencies and for an annual report on these activities. A small portion of the funds is retained by NIA to carry out its forum responsibilities. A predecessor to the Forum on Aging-Related Statistics, an informal interagency group with a similar name (Forum for Statistics on Aging), was initiated by the Bureau of the Census in 1985 to exchange Information, to coordinate the collection and tabulation of data on the elderly, and to identify data needs related to the elderly (WalIman, 1985~. The early forum (now The Exchange) was also concerned with policy needs for data on the elderly and programmatic uses to which the data are put, but members were technical, professional agency staff rather than policy level staff. The National Institute on Aging has recently established the

210 AGING POPULATION IN THE TWENTY-FIRST CENTURY Subcommittee on Data Sources within the Interagency Committee on Research on Aging. The subcommittee has slightly different goals from those of the Census forum; in addition to serving as a means of exchanging information, the subcommittee attempts to assess how such sources have been influenced as research priorities have changed, to review data needs from the 1990 Census, and to provide agencies with feedback on concerns and priorities. The subcommittee is more concerned with how data sources relate to research issues than with the broader policy applications of data (WalIman, 1985~. Numerous examples exist of coordination of statistical programs at the initiative of one or two involved agencies. Agencies that are planning or changing a data collection will frequently attempt to no- tify other agencies that are known to have an interest; comment and review are also sought. Such efforts, however, tend to be fragmented and undependable. They are undertaken on an ad hoc basis and do not tend to play a consistent role in identifying duplication, excess burden, or data gaps. COORDINATING, PLANNING, AND SETTING PRIORITIES FOR STATISTICS ON THE ELDERLY In the area of statistics on the elderly population, coordination, oversight, and long-term planning are particularly important. As the number and proportion of elclerly in the population increase, the needs of the population, from both the government and the pri- vate sector, will also increase, and, possibly more importantly, will change. If properly managed, the federal statistical system can pro- vide information required for the society to cope with these changes. The data needs of agencies differ radically, but coordination of sta- tistical programs on aging should be feasible. The pane! commends the initiative taken by the NIA and the Census Bureau that led to the establishment of the Forum on Aging-Related Statistics it is an import ant first step to improving interagency coordination of statistical programs related to aging an activity that should be institutionalized. Recommendation 10.1: The pane! recommends that a mech- anism be provided for discussion and coordination of data needs, standardized definitions and classifications, priority identification, and production of data relating to an aging society.

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 211 The Forum on Aging-Related Statistics was recently established to function in this role. The pane! recognizes that no single con- gressional committee has oversight of the statistical system, but the House Select Committee on Aging and the Senate Subcommittee on Aging have demonstrated strong interest in statistics on the elderly and might assume the oversight function for the activities of the forum. Because little coordination is currently being undertaken, it is not possible to identify the magnitude of the cost effects of its ab- sence. However, it should be clear that there are several areas in which the absence of coordination has direct effects that increase the inefficiencies of statistical programs. Failures of coordination, planning, and setting priorities lead to duplication of efforts, lesser sharing of data bases that might otherwise be of interagency use, un- met data needs, and inappropriate design of data bases for a broad array of data users. These effects alone imply that there are fiscal results of the lack of oversight. In addition, there are potentially serious effects with fiscal ramifications on other program agencies because of the lack of appropriate data for their policy planning and analysis. Resources In testimony before the Joint Economic Committee, U.S. Con- gress, April 17, 1986, James T. Bonnen placed strong emphasis on the deleterious effects of resource constraints on federal statistics (U.S. Congress, Senate, 1986b:304~: We have already mortgaged the future of our capability to sup- port informed decision in federal policy and elsewhere. Real resource constraints have led to major reductions in research activities in many statistical agencies. Personnel ceilings plus budget constraints have led to a failure to hire and hold a new generation of statisticians. Due to assaults on Useless bureau- cratsn and a generally demeaning attitude toward federal em- ployees, low morale and loss of many of the best people to early retirement and the private sector have affected many statistical agencies just as they have the rest of government. The long run has consistently been sacrificed to the short run. Where statis- tical leadership will come from 10 years from now, I shudder to think about. Sacrificing everything to the short run has re- duced our ability to deal with the obsolescence of concepts that comes with change in markets, economic and social structure, institutions and new technologies.

212 AGING POPULATION IN THE TWENTY-FIRST CENTURY This pane] places high priority at this time on the allocation of resources to reverse the trends of the past few years that have re- duced the number of appropriately qualified professionals in existing statistical agencies. Decentralized Responsibility for Statistics on Aging Not only are substantively different data sources located in a large number of separate agencies, but the agencies have interrelated responsibilities. For example, health-related statistics are produced by at least 12 of these agencies. In this decentralized environment, there are at least six issues of coordination and planning related to statistical surveys and administrative data on the elderly population that need to be addressed In an interagency context and by parties with the ability to influence the outcome of contested deliberations. The discussion of these issues here is limited to their relationships to statistical surveys; a subsequent section of the chapter discusses issues pertaining to administrative records. Content User agencies and policy analysts may have need for particu- lar sorts of information that are not available from existing data sources and are not being planned for future data collections. Be- cause primary data users are frequently not in close proximity to data producers, these needs may go unaddressed in the absence of an agency or other party responsible for identifying unmet data needs and determining means to meet them. Specific data gaps have been discussed in earlier chapters: these include longitudinal data on utilization and financing of health care, out-of-pocket expenses for medical and other care, alternative housing facilities, family support systems, types of long-term care, functional abilities, and chronic disease. Coverage Many of the ongoing surveys conducted by the Bureau of the Census are household surveys and do not include in their sampling frames those who do not live in households; they thus exclude the institutionalized population, which constitutes an important segment of the elderly population. For example, the nursing home population, which increases as a proportion of the population with each older age

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 213 group and among women, is not included in many ongoing surveys, nor is the population of veterans in federal hospitals. Although specialized surveys are sometimes conducted to reach these groups that are not otherwise covered, the richness of content of any single survey is of necessity reduced. Data Detail The issue of data detail is, in some cases, related to the coverage issue discussed above. This problem, however, is more broadly de- fined to include all publications and other forms of data dissemination that do not include adequate detail on special populations. As the elderly population increases in size and as a proportion of the total population, it becomes increasingly important to know more about specific groups within that population. For different purposes, addi- tional detail may be needed on the oldest-old, five-year age groups among the elderly, women, minorities, those with health problems, the poor elderly, or other specialized populations. Sometimes sur- veys do not include enough cases of particular subpopulations in their sample, and the issue is one of coverage. Larger overall samples or specialized oversampling may be required. In other cases, informa- tion that is issued in aggregate form may simply not include enough detail in the aggregated groups to identify those subpopulations; the problem in this circumstance is one of data presentation for the needs of data analysts. The rapidly changing distribution of the aged population requires a reexamination of the sampling frames for studies collecting data on which future plans for people of age 65 and over depend. Table 10.1 shows the projected sharp increase in the total elderly population and the substantial increases among the very old over the period 1980 to 2010. Surveys designed in the 1980s for a sample of the entire popula- tion may include too small a number of the age group 65 and over to answer many questions for detailed age subgroups of this population. Numbers in the age group 85 and over may be especially deficient for the description of the characteristics of the oldest-old. The panel urges that the sample design for population surveys be matched to the major policy questions to be addressed. Sample sizes should be large enough to permit accurate estimation for persons in age-specific subgroups of the older population through further disaggregation of data. Subgroups of interest include those near the

214 AGING POPULATION IN THE TWENTY-FIRST CENTURY TABLE 10.1 Size of the Older Population 1980-2010 (in thousands) Age 1980 1990a 2000a 2010a Total25,549 31,697 34,921 39,195 65-698,782 9,996 9,09.6 11,703 70-746,798 8,039 8,581 8,615 75-794,794 6,260 7,295 6,782 80-842,935 4,089 5,023 5,544 85-891,520 2,157 3,025 3,756 90+720 1,156 1,901 2,795 aBureau of the Census Middle Level Projections. Source: Serow and Sly (1985: Tables 1 and 22~. age of eligibility for Medicare and Social Security (both pre- and postretirement groups), and the oldest-old. The pane! recognizes that it may be costly to screen extra cases to identify members of the age group 65 and over and that there are costs attached to increased sample sizes. Nonetheless, we urge that consideration be given to this issue at the survey design stage. The panel is not advancing a specific sample size for the population age 65 and over, but we urge that agencies planning surveys reexamine the question of adequacy of the sample for analysis purposes. We recognize that the adequacy of the sample depends on the nature of the analytic methods being considered. Uniform Definitions, Concepts, and CIassificatione Surveys (and administrative record systems) that collect related information do not always do so using the same definitions and concepts, even though doing so would increase the usefulness of each of the information sources involved. The importance of standard definitions was illustrated in the chapter on long-term care. In the data collection stage, categories used are not always the same; as a result information from different surveys may not be comparable. In later stages of data processing, differences in ciassi- fications may further reduce comparability. Uniformity is required for purpose of compatibility and could, with proper foresight and interest in the planning stages, be achieved. The single most useful standard that agencies could adopt is to

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 215 provide information for standard age categories in publications and public use files. In addition, the pane! urges that agencies include the date of birth in public use files whenever possible. This is particularly important for the elderly, since they frequently report age incorrectly. Recommendation 10.2: The panel recommends that agen- cies that collect and disseminate data on the total population or administrative data by age: (1) use 5-year age groups for the publication of data up to age 90 and larger intervals thereafter, except when limited by privacy or confidential- ity regulations and (2) provide data on geographic areas, income, and other economic characteristics in the greatest level of detail consistent with the protection of confidential- ity. Issues of confidentiality and privacy are frequently invoked as the rationale for limiting the levels of geographic detail or the detailed breakdowns provided on specific data items. Confidentiality restric- tions in these cases can severely limit the utility of the data. It is clear that confidentiality must be respected to safeguard the integrity of the statistical system and to honor commitments to respondents. It is equally clear that confidentiality should not be used as an excuse for withholding disaggregated data. Periodicity The frequency with which surveys are conducted, or periodicity, an well as the scheduling of the data collections is a matter that clearly affects the timeliness of the data and the comparability among data sources. Although a data-producing agency may allocate limited resources by reducing the periodicity of a survey, this may have severe consequences for policy analysts. A recent example of this was the National Nursing Home Survey, which had originally been planned to be conducted every three years; eight years lapsed between the last two such surveys. Little was known in the intervening period about changes in this rapidly increasing subpopulation that consumes an increasing proportion of total medical expenditures a time period during which the major health issue confronting Congress was control of medical expenditures. Moreover, when surveys are conducted less frequently than annually, it is sometimes of concern whether the timing of the data collection itself is synchronized with other surveys

216 AGING POPULATIONIN THE TWENTY-FIRST CENTURY or record systems since policy analysts may need data from multiple sources for the same time period. :Recommendation 10.3: The panel recommends that greater emphasis be placed on establishing and maintaining the pe- riodicity of population surveys. The periodicity should be established with due consideration of seasonality, secular trends, the frequency of changes in policy, the importance of the issues addressed in public policy, and the timing of related data collections. Data Access Although data producers in the federal government have placed greater attention in recent years on producing public use tapes and providing greater access to data, more attention still needs to be paid to this area. Without timely access to data, researchers gain little from the fact that a data base has been collected and processed. The systematic development of a plan to ensure public access to published reports and to microdata files is a necessary part of the development of any data program. The data collection agencies frequently have insufficient research staff to fully utilize data bases, and priority should be placed on providing outside researchers with access When there are issues of confidentiality of data, such as may arise with the development of linked data bases, public access is a more complicated problem. Even so, there are precedents for providing access in innovative ways, such as enabling outside researchers to conduct research within the agency that has produced the files. The Office of Management and Budget on December 12, 1985, issued Circular No. A-130, Management of Federal Information Re- sources. The purpose of this circular is to establish policy for the management of aD federal information resources, providing procedu- rat and analytic guidelines for policy implementation. With respect to data access, the first policy the circular establishes is that agencies shall "create or collect only that information necessary for the proper performance of agency functions and that has practical utility, and only after planning for its processing, transmission, dissemination, use, storage, an] disposition." This circular was issued by the Office of Information and Regulatory Affairs and establishes the interest and authority of that once to review and monitor the dissemination of data.

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 217 The circular indicates that agencies' primary interest is not nec- essarily in disseminating information products and services, but rather that they must attend to the legal requirements to do so, to the relationship between issuing products and services and the agencies' missions, and to public need for the products. In addition, agencies have an obligation to make data available to those who have a reasonable ability to acquire the information, in a manner cost- eflSective for the government, and, whenever nnR~iLl~ to recover the costs of dissemination. . rim , Data access has, in general, been a weak point in the development of federal data bases on health and health policy. This area must be improved if researchers are to be able to conduct studies in a timely manner. The pane! recognizes that the new OMB circular addresses many of these issues, but it also recognizes that the manner in which the circular is implemented remains to be determined. We believe that agencies must adequately plan for public access to their data bases and that a high priority must be placed on timely and complete access. Not only should access be provided to files created for the purpose of providing statistical information, but it should also be Improved for statistical files created from adrn~nistrative record systems. Recommendation 10.4: The pane! recommends that the plans of a federal agency for the creation or continuation of a data system should include a plan for issuance of publica- tions and public use tapes following its creation or updating. The plan should include explicit procedures for disseminat- ing the data and should provide for widespread access and timely, thorough, and accurate data dissemination. In addition to the timely dissemination of public use tapes, provision should be made for archiving such data bases to maximize their utilization by nongovernmental researchers and analysts. Recommendation 10.5: The panel recommends the contin- ued support and development of (a) aging-related archives of research data, such as the National Archive of Comput- erized Data on Aging at the University of Michigan and (b) provision of technical support for users. Confidentiality "The essence of statistical confidentiality is that statistics

218 AGING POPULATION IN THE TWENTY-FIRST CENTURY released by the collecting agency or researcher cannot be related to data provided by or about an identifiable respondent. This principle is considered to be a cornerstone of statistical policy" (President's Reorganization Project for the Federal Statistical System, 1981:200~. High-quality data can be obtained only when the confidentiality of the respondent can be convincingly guaranteed by the data col- lection and processing agencies. In the past, the statistical policy office of OMB has had a long history of placing a high priority on ensuring that statistical data are kept confidential. Some data series are kept confidential until scheduled release times. More relevant for the considerations of this panel, however, is the strict maintenance of confidentiality of microdata records. The Bureau of the Census and the National Center for Health Statistics are the only statistical agencies that now have a statutory basis to ensure the confidentiality of their records. Although other agencies, such as the Bureau of Labor Statistics, have no statutory basis to guarantee the confiden- tiality of their data collections, they do so on their own, and because of their effectiveness in honoring that guarantee they have obtained a high level of voluntary reporting. Another important consideration for the maintenance of the con- fidentiaTity of records, however, arises from the need for data sharing among statistical and administrative agencies. Data sharing can reduce respondent burden, duplication, and the fiscal budgets of sta- t~stical agencies. The types of data sharing for which this is the case include, for example, sharing of sampling frames for specialized sub- populations and sharing of statistical or administrative data bases for statistical or exact matching. Although the context of such sharing should always be to produce statistical, aggregate information Tom combined data bases, when more than one agency is involved, even if both are statistical agencies, problems may arise in the sharing of records. If one agency has an umbrella legislative guarantee of confidentiality and another has no such guarantee, the first is likely to be reluctant to share its data. Agencies without such legislative authority that have nevertheless guaranteed their respondents con- fidentiality may also be reluctant to share data, even for statistical purposes. Efforts to establish a broad cover of confidentiality guarantees for the entire federal statistical system have in the past been met with some skepticism. Many believe that the right of individuals to privacy, to provide what information they wish with full knowledge about how that Formation will be used, should be paramount in

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 219 consideration of these issues. The issue ~ not only that individuals have been promised confidentiality in the statistical sense for the information they provide, but also that they have provided informed consent when they have provided that information. Opponents have expressed concern that widespread data sharing could lead to the creation of large-scale data banks that would contain substantial amounts of information about different facets of individuals' lives or the characteristics of businesses, even though respondents would never have provided all that information to one government agency. Concern for the potential abuse of such a system has weighed more heavily in the minds of some than the conveniences of being able to reduce respondent burden and budgets. A~ninistrative Records There are two major sources for developing information on the elderly surveys of individuals that have the capacity for including characteristics on a personal basis such as health status or demo- graphic characteristics and administrative records that have infor- mation relative to an individual's participation in a government pro- gram in conjunction with certain characteristics of the individual. Surveys are based on well defined sampling frames and are gener- alizable to well-defined population groups, whereas administrative records provide information only for individuals using a service or selected into a program. Administrative records are not a substitute for surveys but a complementary source for some types of statistics on certain subgroups of the population. Administrative records are typically produced by agencies with direct programmatic responsibilities as a by-product of their admin- istrative or regulatory procedures, sometimes for program analysis, but more frequently for monitoring individual cases within specific programs. As a result, the data in these records tend to be limited to data required for administration of the program. Additional so- cioeconomic and demographic background information so frequently essential for analysis is typically lacking. The use of administrative records for statistical purposes has increased substantially over the past decade. Administrative records serve as an attractive source of data for some purposes. Often they are the only available source of specific information needed for pol- icy analysis. In addition, they can frequently provide data for small

220 AGING POPULATION IN THE TWENTY-FIRST CENTURY geographic areas or small subpopulations, while minimizing respon- dent burden and data collection costs. Problems arise in using such records because they frequently cover only limited populations or limited substantive detail, access is inhibited by law or administra- tive procedure, data quality is not always consistent (e.g., financial data may be accurate while health records are incomplete or unreli- able), and records may not be kept for the same units that are needed for statistical analysis. In some cases, however, administrative records may be a superior source of information to respondent-provided information. When data are available on records, they may be more accurate if they are associated with the actual operation of the administrative program for which the records are being kept. In such cases the use of records is desirable, even when linkages must be made to other data sources to complete data sets for statistical analysis. There are costs and operational problems associated with mak- ing administrative records suitable for statistical purposes or with linking them with other statistical data bases to obtain more com- plete information on the units of analysis. However, such costs and the work required to process the records for statistical use may be much less than that associated with new data collections to obtain the same information. Active statistical policy coordination and oversight is needed to enhance the use of administrative records for statistical purposes. Data policies can facilitate the innovative production of data, ensure confidentiality of information on both administrative and statistical record systems, encourage the greater use of administrative records, ensure maximum data quality, ensure that resources are made avail- able for upgrading the quality and content of record systems, and stimulate interagency deliberations as needed. The panel recognizes that there are many problems inherent in increasing the use of administrative records, but we believe that these problems can be addressed and, in many cases, overcome. In cases of statistical uses of administrative records, particular care must be taken to safeguard traditional assurances of confidentiality of statistical data. The statistical agencies, to secure their own data collections, must take special care in how they handle these records and how they share data with program agencies. Recommendation 10.6: The pane} recommends that greater use be made of administrative records as a source of statisti cal information on the elderly. To facilitate this goal, special

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS attention should be paid to designing data tapes, includ- ing public use tapes, from administrative record systems to make them more useful and accessible to policy analysts and researchers to the extent feasible within confidentiality constraints. Data Linkages 221 Related to, but separable from, the issues of administrative records are issues of linkages between data sets. Such linkages can be accomplished either through matches of records on the same individ- uals from different data bases, known as exact matches, or through matches of records on different individuals who are identified as iden- tical or similar in important respects, known as statistical matches. Statistical matching should be used as a last resort, and then used with great caution. Both types of matches serve to provide a broader set of data on individual cases by merging two or more data sets that supply different sets of characteristics. There are specialized problems associated with each type of linkage. In addition to the technical problems, however, there are also broad problems of sta- tistical policy associated with the linking of records from different agencies and of administrative and survey records. The power of linked records lies in the ability to expand avail- able information at the microdata level without undertaking new data collections. The process draws on the strengths of administra- tive records when they are turned to as a source of possibly more accurate data than could be obtained in an interview setting. Link- age of administrative data with survey data can also enrich the administrative data system. The increased use of data linkages can be enhanced by more active coordination among agencies that need such data an well as among those that have the data bases to be matched. The function of coordination and oversight in this case is very similar to the role discussed above for the use of administrative records. It revolves around ensuring confidentiality and quality, identifying appropriate content areas, assisting in interagency deliberations, and reducing respondent burden and fiscal costs. Recommendation 10.7: The pane] recommends greater use of data base linkages as a source of statistical information on the elderly. Unless there are confidentiality limitations, bar riers to linking data systems should be removed. Resources

222 AGING POPULATION IN THE TWENTY-FIRST CENTURY should be made available to systems. . Increase the flexibility of these Enhancement of National Surveys This report includes a number of recommendations to enhance surveys that are relevant to all agencies that collect data about the elderly. In addition, a large number of recommendations have been made to augment specific health surveys. Many surveys other than health surveys are related to the health of the elderly, such as the Retirement History Survey, the American Housing Survey, and the Survey of Income and Program Participation. We have singled out the last survey for discussion since it has the potential for providing longitudinal data on the health of the elderly and for relating health to income. Surrey of Income and Program Participation The Survey of Income and Program Participation, first adminis- tered in October 1983, is a nationally representative household sur- vey that provides detailed longitudinal information on income, living arrangements, disability, assets and liabilities, government transfer program eligibility and participation, pension coverage, taxes, and many other characteristics of individuals. Each February a new pane] of about 15,000 households enters the survey. Members of each pane} are interviewed every 4 months over a period of 32 months; the lon- gitudinal design of the survey, which provides 8 interviews for each household, allows for the study of changes in characteristics. In ad- dition to the core of labor force and income questions, fixed topical modules are administered periodically during the survey. SIPP can be especially useful in analyses relating economic status to health questions. In the first interview of the 1984 SIPP, there were over 11,000 sample cases of people age 55 and over; their distribution by age group is shown in Table 10.2. The sample is too small for statistically reliable analyses by age groups beyond age 75 and over. It would be desirable to augment the SIPP sample of the population over age 65 periodically, e.g., every five years. It would not be necessary to oversample the population age 65 and over in all panels. Oversampling could be facilitated by using Social Security or Medicare records to identify additional persons age 75 and older.

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS TABLE 10.2 First Interview of the 1984 Surrey of Income and Program Participation (unweighted counts) Age Total 55-64 65-74 75-84 85+ 5,116 3,880 1,856 446 Recommendation 10.8: The panel recommends that the Survey of Income and Program Participation sample be aug- mented at the beginning of each five-year interval to provide suitably precise estimates of characteristics for groups ages 75-84 and 85 and over. 223 One of the topical modules used in SIPP is on health and disabil- ity. This module includes extensive questions on health status, need for help in activities of daily living, person providing assistance, type of disability and disability-caused limitations on kind and amount of work. Other modules cover topics such as education and work history, pension plan coverage, retirement plans and expectations, marital history, fertility history, taxes, asset holdings, and employee benefits (the last module has little relevance for the elderly). The utility of the data collected in SIPP might be increased by target- ing specific modules to appropriate subgroups of the population. It would be possible to obtain longitudinal data on health and disabil- ity of the elderly by administering the health and disability module a second time to those over age 65 instead of the employee benefits module. This substitution would require somewhat more complex procedures for interviewers, but it should be feasible. Recommendation 10.9: The pane! recommends that, in ad- ministering the Survey of Income and Program Participa- tion, the Bureau of the Census replace the topical module on employee benefits with the health and disability module for one wave for those of age 65 and over so the latter module is used twice, ideally at a two-year interval. The Bureau is exploring the possibility of matching administra- tive records from the Social Security Administration with the SIPP records to provide additional information on the elderly. Records

224 AGING POPULATION IN THE TWENTY-FIRST CENTURY being considered for matching include summary earnings records, benefit records, and Medicare records. FEDERAL-STATE DATA ISSUES Background Since the beginning of American federalism both the develop- ment and the administration of social policy has depended on the joint efforts of the federal and state governments. Paralleling the cooperative federal nature of policy implementation itself, the collec- tion and preparation of the data necessary to monitor and evaluate social policies and programs also depend on a variety of federal-state cooperative activities. The states, as well as the several agencies of the federal gov- ernment, are both producers and consumers of data. While states may be the greatest consumers of their own, largely administrative, data, the importance of the crossover linkages between federal and state agencies cannot be overemphasized. That is especially true of the area of national vital statistics, in which the states are the key suppliers of what become national data files. Simultaneously, the bloc grant context of New Federalism has placed significantly increased importance upon a variety of state- government-directed intrastate resource allocation, planning, and monitoring activities. That is, the implementation of federal (as well as state) programs is increasingly undertaken through the allocation of resources by state governments to targeted populations identified through county and other substate units. For example, through the Older Americans Act, state govern- ments receive federal funds in support of a broad range of nutrition and social services. Federal law requires that the state governments allocate these funds through an intrastate funding formula to plan- ning and service areas across the state in such a way that funds are targeted to those elderly in "greatest economic or social need." Since these planning and service areas are aggregations of counties, age- specific county-level population data required for assessing need are a crucial element of this planning and resource allocation process. Consequently, the federal supply of national population statis- tics, disaggregated to county and other substate levels, represents an increasingly important and irreplaceable- source of data. It is

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 225 against this backdrop of the necessary cooperative relationship be- tween federal and state data sources and data needs that the pane] directs attention to the following six issues. Vital Statistics and Data on Mortality Vital statistics representing data on births, deaths, marriages, divorces, and other demographic indicators, constitute a critical na- tional resource, yet represent primarily a data collection activity on the part of state governments. As part of the Vital Statistics Cooperative Program, states collect their vital statistics data and subsequently make them part of the national data system through contracts with the National Center on Health Statistics. Under this program, the NCHS, in conjunction with appropriate state agencies, develops definitions and standards for state data collection and also provides technical assistance to the state agencies. Of special interest to an aging society is detailed information concerning the distribution of causes of death. While "old age" may have been an acceptable designation of the cause of death in an earlier era, both the technical capacity and the greater need for more detailed mortality statistics suggests the importance of more focused analysis of death certificates and the reporting of mortality statistics. The analysis of mortality within an aging population represents, among other factors, an important dimension of the epidemiology of the health of that population. A recent NCHS study, for example, reported that nearly 75 percent of the death certificates studied had more than one condition listed and 15 percent had more than four conditions listed (National Center for Health Statistics, 1984a). Up to 20 conditions reported on the death certificate are coded. With greater detail in the reporting and enumeration of deaths, it will be possible to better identify patterns of morbidity and mortality by region, state, and substate areas, as well as by age, sex, and race. The identification of such patterns are important for both scientific analyses, and policy and program decisions. The NCHS has released tapes of multiple causes of death for the years 1968-1983 to encourage their use by analysts and researchers. The Multiple Causes of Death Data Files are good sources of data for research purposes, but they are used by only a few researchers. Recommendation 10.10: The panel commends the National Center for Health Statistics for developing the Multiple Causes of Death Data Files and for preparing public use

226 AGING POPULATION IN THE TWENTY-FIRST CENTURY tapes of those files. The panel recommends that the Center develop an outreach program to promote their use. The death certificate contains an item on whether an autopsy was performed, but because of the legal requirements for prompt filing of death certificates, autopsy information even for the small percentage (13 percent) of the deaths autopsied is not always used in the preparation of the cause of death statement. The panel urges that NCHS and state vital statistics offices cooperate in efforts to improve the quality and completeness of the medical certification through better training of physicians, and through query programs, with emphasis on augmenting the certification by appending autopsy information when available. Disability Statistics While most older persons live in their own homes and are not constrained in most of their daily activities by disability, knowledge of the distribution of disability of an aging population is especially important. A broad range of both state and federal health and social service programs is dependent on knowledge of such disability statistics for the allocation of resources and the monitoring of the programs. While the 1980 census included a disability item, it was notably limited for use with respect to the older population because its context was that of employment-connected disability. A number of federal agency surveys include specific measures of health and disability, yet most such surveys provide only national estimates or state estimates without statistical detail at the substate level. By contrast, health policy planning and scientific analyses undertaken for an aging society require detailed age and age-sex distributions of disability. Since a variety of health and social service programs respond to the varying needs of different disabilities, neither age data per se nor data on institutionalization are an acceptable substitute for precise disability data per se. Furthermore, as noted previously, given that state agencies have increased programs that provide social services to the elderly and disabled, state and substate disability data are increasingly important tools for planning, evaluation, and analysis. The pane! recognizes that in the continuing planning for the 1990 census, the director of the Bureau of the Census is giving emphasis to the development and testing of items that will provide detailed !

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 227 measures of functional disability sensitive to the needs of an aging population. Recommendation 10.11: The pane! recommends that age- sensitive disability items be included in the 1990 census and that estimates for states and major civil subdivisions, by age, be tabulated and published. Age Detail The traditional social definition of "old" as age 65 tied to the long-standing age of full benefits eligibility under Social Security- appears to influence the reporting of many state and federal age distributions. Both demographically and in terms of policy, it is evi- dent that the population is aging, that there are increasing numbers of persons living to older ages, and consequently there is a funda- mental need to change the norms of statistical data reporting for a broad range of variables and indicators. And as noted, state planning requires increasingly sophisticated age data, including more detailed description of the population over age 65. It is imperative that the major statistical agencies of the federal government, not just the specialized age-related or health-related agencies, routinely report detailed age data beyond age 65, so that the requisite planning and analytic data sets will then become more broadly available. For pur- poses such as assessing the potential impact of changes in the age for eligibility for Medicare or Social Security, distributions by single year of age are needed. Recommendation 10.12: The pane} recommends that in publishing decennial census data for geographic subareas of the country, the Bureau of the Census provide age, age-sex, and age-sex-race distributions beyond age 65. Whenever feasible, distributions by single year of age should be made available. At a minimum, such age detail should be made available at the county level by 5-year age intervals through age 85. Furthermore, within the normal constraints of the protection of individual privacy, the Bureau of the Census should make a broad range of population data available with age detail beyond age 65 for states, counties, and other sub- state areas.

'228 AGING POPULATION IN THE TWENTY-FIRST CENTURY Dissemination of Age-detailed Substate Data Throughout this discussion two themes have been emphasized: the need for data to be made available at the substate level and the need for greater age detail in population and vital statistics data. One valuable source of such data is the State Data Center in each state, whose responsibility is to process census data tapes for a variety of data users. (Parallel organizations, such as the Association of State and Territorial Health Officers, also serve as information sources in this regard.) State Data Centers exhibit substantial variation in the kinds of data and data services they supply, some having developed into national wholesalers of census and market research data, while others having defined a fairly narrow set of technical services. A further limitation of State Data Centers can be seen from the perspective of the health and social policy analyst who requires detailed age data from across the country. As one might expect, for example, even when the State Data Centers do prepare intercensal substate estimates of the size and status of the older population and not all states d~they use a variety of different methodologies in the computation of those estimates. A complementary source of age-detailed substate data is reflected in another existing data dissemination model: the County and City Data Book represents an especially rich collection of substate data that has been produced in printed form by the Census Bureau for many years. Its value is enhanced by the fact that it includes, in addi- tion to population census data: a selection of vital statistics; housing data; education and crime statistics; data on poverty, employment, and unemployment; statistics on doctors, nurses, dentists, hospitals, and nursing homes; and substate information on Social Security and Medicare, including dollars distributed and persons involved. The sheer variety of the variables included provides both state and local officials as well as researchers with a unique capacity to understand and explore the population and environment of the American people at the local level, where an increasing amount of planning and service delivery are taking place. The value of the County and City Data Book as a state and substate planning and analytic too! was enhanced when the Census Bureau made the collection available in computer-readable formats. In the context of the microcomputer revolution of the past decade, it

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 229 is especially noteworthy that in 1984 the Census Bureau began dis- tributing this data base on microcomputer diskette, thereby making it more accessible to a much broader array of state planning agencies. Given the general value of the collection of data in the County and City Data Book, its utility for an aging population is unfortunately limited. The old age variable itself is limited to "percent 65+," and none of the other important indicators of the health and wealth of the population is presented for different age categories. Given the availability of such age detail throughout the other parts of the census data system, in addition to the widespread utility of the city- county data base and the demonstrated ease of access to it via its recent diskette dissemination format, this data base should be further developed for use in the context of an aging population. Recommendation 10.13: The panel recommends that the director of the Census Bureau develop plans to produce an enhanced, age-detailed County and City Data Book data set. Consistent with standards for protecting the privacy of individuals in small statistical areas, county and city data representing age detail, by five-year age categories through age 85 whenever possible, should be assembled for the range of population, health, poverty, employment, race and eth- nicity, and other variables currently included in the County and City Data Book. It is recommended that such a data set be assembled and made available using the 1990 census and related data collections and distributed in both printed and machine-readable formats (computer tape and micro- computer diskette formats) on a timely basis. Timely Substate Data Health planners in state, county, and local areas require infor- mation on the size and distribution of the elderly population in order to budget for community health services, hospital beds, nurs- ing homes, and other related services. Decennial figures from the Census of Population and Housing are soon outdated. The sample sizes of the Current Population Survey or the Survey of Income and Program Participation are too small to provide small-area estimates of characteristics of the elderly population. In 1976, Congress authorized a m~-decade census in the year 1985 and every 10 years thereafter by amending Title 13 of the United States Code-Census (U.S. Congress, Senate, 1976~. No census was

230 AGING POPULATIONIN THE TWENTY-FIRST CENTURY taken in 1985; the earliest possible date for a quinquennial census is now 1995. With the rapid growth and internal migration of the population, health planners need reliable and timely estimates of the population by age, race, sex, and income level and in geographic detail to the minor division level. The mid-decade census would fill this need for all jurisdictions. The pane! recognizes that the cost of the mid-decade census requires justification far beyond that which can be supplied by health planners but views this as an important need that should be taken into consideration. Social Service Statistics Since 1965 the federal and state governments have supported an array of social services, including nutrition programs, for the elderly through the Older Americans Act. From the beginning, but especially since 1973, these programs have reflected the principles of federalism, in that national funding, guidelines, and priorities have been implemented by State Units on Aging and a within-state network of Area Agencies on Aging. Data collections that represent services provided, clients served, and dollars allocated and spent therefore reflect both the national guidelines and priorities and the state and local implementation of these national mandates. Despite the fact that these nutrition and social service programs reflect their state en c] local implementations, the statistical data that describe the programs represent important national data, since nutrition and social programs are significant factors in maintaining the health and independence of the older population. Since 1978 the National Association of State Units on Aging (NASUA) and the National Association of Area Agencies on Aging (N4A), professional associations representing the administrators of the state and area agencies, have been supported by the Administration on Aging to collect budgetary, client, and service delivery data reflecting the sev- eral programs of the Older Americans Act. These data are collected on a continuing basis; periodic and annual updates are crucial given the changing nature of both the older population and the service programs that serve them. In many cases the program data can be matched with demo- graphic and other national data, since the planning areas served by the Area Agencies on Aging are aggregations of counties. This merging of program and demographic data sets offers a valuable op- portunity to monitor and assess changes in the aging population and .,

. ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 231 the effectiveness of social programs in helping to maintain the health of the aging population. While the pane! does not make a specific recommendation in this regard, the importance of maintaining this unique data collection is acknowledged. Private Data Bases Very few private data bases in the health area are publicly avail- able, and there is no central repository enumerating all private data bases. Data bases are therefore examined for the industries consid- ered most pertinent to health policy for an aging population, namely hospitals, nursing homes, HMOs, hospices, private insurance compa- nies, and professional associations. Hospital Data Within the hospital industry, their trade association, the Ameri- can Hospital Association (AMA) collects data on a regular basis. Four primary instruments are utilized at various intervals: the Annual Sur- vey of Hospitals, the National Hospital Panel Survey, the Survey of Medical Rehabilitation Hospitals and Units, and the Survey of Med- ical Staff Organization. The data collected by these surveys focus primarily on facility characteristics (numbers of beds, medical stab organization, character of overall facilities), health resources (types of services within facility, types of resources connected to facility, quality and training of medical/health staff), financial data, and services utilization. No new surveys are being planned, but as the industry and relevant technology evolve, the surveys are modified to meet data needs. The AHA makes all of its data bases publicly available in either tape or hard copy, at a charge and within the limits of confidentiality. Nursing Home Data Data collection for the nursing home industry is less well orga- nized. One of the primary obstacles to comprehensive data collection within this sector is the lack of standardized definitions for what ac- tually constitutes a nursing home and what actual services fall within a particular colloquial definition. Given the broad facility licensure authority within each state, it is not feasible to collect comparable data across the industry.

232 AGING POPULATION IN THE TWENTY-FIRST CENTURY Currently, at least one trade association, the American Health Care Association (AHCA) is collecting data from its membership. The data are limited to numbers of facilities within a region, oc- cupancy rate, ongoing growth of industry, and financial expendi- ture/reimbursement information. Currently, these data are available only to association members and have been used to support arti- cles in trade publications and lobbying efforts. The AHCA plans to expand its data collection efforts to include services utilization and opinions on current issues and has indicated that the data could be made available within the applicable strictures of confidentiality. Health Maintenance Organization Data A different arena, one that has undergone substantial growth and change, is the health maintenance organization industry. Particularly because of the uneven growth in enrollment, services, and utilization trends, this industry also lacks a set of standardized definitions with which to conform any potential data base. Because of the unique reimbursement system between HCFA and HMOs, it has become nearly impossible to elicit cooperation from HMO s vis-a-vis services rendered. This is clue partly to the unique record-keeping systems in HMOs, and in part to the proprietary nature of the information. Notwithstanding these obstacles, the Group Health Association of America (GHAN), a trade association for HMOs, has undertaken a major project to develop and implement a uniform data collection system for HMOs. To date, modules of tables based on items with standard definitions have been developed and are being pretested for statistics on enrollment and marketing, health services delivery, finances, and health care purchases. Completion and implementation of this project could take several years. Given that GHAA's member- ship organizations represent a large majority of all HMO enrollment, that the system design group is composed of a broad spectrum of HMOs nationwide, and that GHAA plans to make the data base available to the public within strict confidentiality guidelines, the project has the potential of generating a useful data base. Hospice Data An industry that deals with the aging population as a part of its clientele is hospice. Because of its relative youth as an industry, very little has been clone to develop a publicly available private data base.

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 233 The National Hospice Organization publishes an annual membership directory and the Hospice Association of America is developing a membership survey to ascertain information from providers, includ- ing field of service, types of personnel, type of facility, and whether associated with another health care unit, numbers of patients served, and financing/reimbursement data. The data base from this survey is still being developed; data will not be available before summer 1988. Commercial Health Insurance Data The Health Insurance Association of America, whose member- ship represents 95 percent of the commercial health insurance busies ness, currently is involved in three major data collection efforts. The Survey of Hospital Semi-Private Room Charges is currently available to the public at reproduction cost. It contains average room cost for responding hospitals by zip code. The Prevailing Health Care Charges System is available to members and certain research asso- ciations only. The data are tabulated by zip code and are available at reproduction cost. The SOUTCe Book of Health Insurance Data (a compilation of Health Insurance Association of America data and data collected by government and other organizations) is an annual publication available to the public without charge. It seems reasonable to ask why there are no statistical data produced as by-products from the large number of health insurance providers, HMOs, etc., just as there are from Medicare and the Social Security system. Two elements that are fundamental elements of a statistical system to assess utilization of medical care are lacking in the data collected by insurance companies: (1) a population base, in which every insured individual is separately identified the population data provide the denominator for a ratio and (2) an individual record of use of service and its nature and costs or charges or both, which can be used individually or combined for subgroups of the population (e.g., age groups or race-ethnic groups) as the numerator of the ratio. The rate structure for private health insurance sold to, or pro- vided by large companies is usually the same for every employee with a family and is varied only for single persons. Age of employees, number of children and their ages, dual coverage of husband and wife when both are employed are all unknown factors in the base

234 AGING POPULATIONIN THE TWENTY-FIRST CENTURY population. Utilization is reported by the identification number of the employee regardless of the family member involved. The data on individuals in the population base and the indi- vidual's record of use of services could be designed into the system when it is created, but the insurance company needs only to deter- mine rates that cover utilization costs. The insurance company has no need for analysis of these costs. If the insured company finds it desirable to know what it is paying for in some detail, the company must be willing to pay the costs of a more elaborate system of data collection. Essentially, the same problems prevent the HMOs from pro- ducing utilization data. What data have been published both from insurance carriers and HMOs have been developed through special studies, designed to answer specific research problems. In the future, utilization information for Medicare may also de- pend entirely on special studies or surveys. The Medicare system was established in 1966 with a statistical system that identified the insured individual and each service that was billed through the sys- tem. Medicare data will no longer be complete, however, with the growth of HMOs, the enactment of the law requiring first coverage of employed persons 65-70 by the employer, and the prospect of a possible voucher system. Professional Association Data Three major professional associations, the American Medical Association (AMA), the American Dental Association (ADA), and the American Nurses Association (ANA), have data bases relating to characteristics of their membership and/or of the profession they represent. The AMA maintains two data bases: the Physician Mas- terfile and the Socioeconomic Characteristics of Medical Practice. The Physician Masterfile contains information about all physicians, including non-AMA members, such as medical school attended and dates, year graduated, residency, board certification, licensure data, membership in professional societies, specialty (including geriatrics), and professional activities. The Socioeconomic Characteristics of Medical Practice file, published annually, contains information on the practice characteristics of a sample of active nonfederal office and hospital-based physicians drawn from the Masterfile, for exam- ple, patient care activities by specialty, professional expenses, fee schedules, and number of hours per week devoted to patient care.

ENHANCING THE UTILITY OF STATISTICAL SYSTEMS 235 The ADA conducts a similar membership survey. Its survey, however, contains information only on members of the ADA. This survey includes basic demographic, biographical, and practice char- acteristics of its members, such as schools attended and dates, year of graduation from dental school, location and type of practice (gener- alist or specialists, number of hours of practice per week. In addition, the ADA conducts sample surveys of about 5 percent of its members and nonmembers on practice characteristics, which include types of procedures used, fees charged, and age of patients treated. The ANA conducts an annual survey of its members, which includes demographic, educational, employment, and worksite data. The major clinical, teaching, or practice area is specified, as is the job title, the specialty, the type of education received, and employment status (full or part time), and whether the individual is employed as a nurse. Data on the place or field of employment are also collected and include hospital, nursing home, school, HMO, community or home, occupational, and office nurse. No information on salaries is kept in this data base. These association files provide little data related to services avail- able for the elderly. The AMA Masterfile identifies physicians spe- ciaTizing in geriatrics; the ADA sample survey contains an item for age of patients; and the ANA specifies nursing home as a place of employment. Since the elderly use all types of services of physicians, dentists, and nurses, these files can provide information on the avail- ability of professional services to the population of all ages, including the elderly, but not on the elderly as a group with special needs. Summary It should be remembered that the aforementioned data bases are not an exhaustive list. Rather they are illustrative of the variety of sources providing information, or working to provide information, to the public. Universities with gerontological centers are one example. In addition to the sources described there are a variety of private associations doing health-related research requiring that data bases be created for each reject. Quasi-academic settings are also sources of useful health policy data, for example, the Urban Institute, the Brookings Institution, and the Institute of Medicine.

Next: 11 Statistical Methodology for Health Policy Analysis »
The Aging Population in the Twenty-First Century: Statistics for Health Policy Get This Book
×
Buy Paperback | $60.00
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

It is not news that each of us grows old. What is relatively new, however, is that the average age of the American population is increasing. More and better information is required to assess, plan for, and meet the needs of a graying population. The Aging Population in the Twenty-First Century examines social, economic, and demographic changes among the aged, as well as many health-related topics: health promotion and disease prevention; quality of life; health care system financing and use; and the quality of care—especially long-term care. Recommendations for increasing and improving the data available—as well as for ensuring timely access to them—are also included.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!