Skip to main content

Über dieses Buch

Models to forecast changes in mortality, morbidity, and disability in elderly populations are essential to national and state policies for health and welfare programs. This volume presents a wide-ranging survey of the forecasting of health of elderly populations, including the modelling of the incidence of chronic diseases in the elderly, the differing perspectives of actuarial and health care statistics, and an assessment of the impact of new technologies on the elderly population. Amongst the topics covered are - uncertainties in projections from census and social security data and actuarial approaches to forecasting - plausible ranges for population growth using biol ogical models and epidemiological time series data - the financing of long term care programs - the effects of major disabling diseases on health expenditures - forecasting cancer risks and risk factors As a result, this wide-ranging volume will become an indispensable reference for all those whose research touches on these topics.





Chapter 1. The Scientific and Policy Needs for Improved Health Forecasting Models for Elderly Populations

Models for forecasting changes in mortality, morbidity, and disability in elderly populations are essential to national and state policies and health and social programs. The rapid growth of the elderly and oldest-old populations have implications for the size and long-term fiscal soundness of programs, such as U.S. Social Security and Medicare. Less well understood are qualitative health and functional changes of future elderly populations and how changes affect federal and state health policy, public and private health-care providers, and private acute and long-term care (LTC) insurance.
Kenneth G. Manton, Burton H. Singer, Richard M. Suzman

Methodological Issues


Chapter 2. Actuarial and Demographic Forecasting Methods

The purpose of this chapter is to present common methods of forecasting mortality used by actuaries in the American insurance industry. In a loose sense, the term forecasting refers to two different tasks, both of interest to the actuary. Of primary interest to actuaries are forecasts of the future mortality experience of groups. Individuals of a group are assumed to be subject to mortality according to some known set of probabilities. A second type of forecasting involves assessment of future patterns of the mortality process.
H. Dennis Tolley, James C. Hickman, Edward A. Lew

Chapter 3. Demographic Change in the United States, 1970 – 2050

Population projections demonstrate the implications of sets of fertility, mortality, and migration rates, combined with an initial population age structure, for future population size and composition. Age-specific mortality rates are applied to project the living population forward in time; age-specific fertility rates are applied to project births. Immigrants and emigrants are typically added at the last stage. Projections make endogenous one of the most important determinants of demographic change, a population’s age structure. Populations with larger proportions over age 50, for example, have higher death rates and lower birth rates and growth rates, ceteris paribus.
Samuel H. Preston

Chapter 4. Health Forecasting and Models of Aging

In two prior chapters, legal, financial, and regulatory constraints on actuarial forecasts were discussed (Chapter 2), and the implications of such constraints for projections by the Social Security Administration (SSA) and the U.S. Census Bureau were reviewed (Chapter 3). This chapter discusses statistical and mathematical forecasting techniques that may improve health forecasts by (1) using a broad range of data types, (2) appropriately combining data from multiple sources, and (3) improving the biological realism of forecasts. Identification of methodological issues in health forecasting that need further research is also covered. This need is, in part, evident from past problems with federal projections (e.g., U.S. Senate, 1983) and, in part, is due to the rapid growth of scientific insights on the physiological mechanisms underlying aging which has changed our understanding of the substance of forecasts (e.g., changing concepts of senescence, chronic disease, disability) and their relation to mortality (e.g., Kristal and Yu, 1992; Mooradian and Wong, 1991a,b). Forecasting models, whatever mathematical techniques are used, must take advantage of this new knowledge to be credible. This requires advances in several areas.
Kenneth G. Manton

Forecasting Techniques for Specific Diseases


Chapter 5. Cancer Forecasting: Cohort Models of Disease Progression and Mortality

There is an extensive literature on cancer mortality trends and forecasts based on them (e.g., Davis and Hoel, 1990). A common feature of most forecasts is that they deal with trends in site-specific cancer mortality and/or incidence rates but ignore progression from the well-state, to tumor initiation, to detectable disease, to death. Analysis of mortality trends may be useful for determining how much progress the Naional Cancer Institute (NCI) has made toward its 1986 long-range goal of reducing cancer death rates (Bailar, 1990). More refined assessments, whether of public health interventions (e.g., advertising against smoking), screening programs, or new treatments such as chemotherapy for early breast cancer (Early Breast Cancer Trialists’ Collaborative Group, 1988; 1992 a,b) requires use of individual data and stochastic process models. In particular, biologically based models of cancer progression in birth cohorts are important (e.g., Cuzick, p. 85, and Hoel, p. 86 in Davis and Hoel, 1990) for analyzing trends, producing forecasts, and assessing interventions. Nevertheless, the use of cohort models is an exception (Manton and Stallard, 1982, 1988, 1992), not the rule, in forecasting.
Kenneth G. Manton, Burton H. Singer, Eric Stallard

Chapter 6. The Effects of Risk Factors on Male and Female Cardiovascular Risks in Middle and Late Age

Epidemiological studies have identified a number of risk factors for chronic circulatory disease. Though these risk factors (e.g., serum cholesterol, systolic and diastolic blood pressure, smoking) have significant associations with heart and other chronic disease risks in multiple longitudinal studies, there is little evaluation of the lifetime effect of controlling those factors. When these risk factors have been evaluated in forecasts, the effect on population risks was often less than in longitudinal studies because the models employed were not biologically realistic (e.g., risk factors were assumed to operate independently and cohort dynamics were not represented; e.g., Weinstein et al., 1987). The results in this chapter are based on a multivariate model (presented in Chapter 4) of the effect of risk factors on (1) the risk of disease at specific ages; (2) the interaction of causes of death (i.e., the effect of dependent competing risks); and (3) the effect of selective mortality on the age-specific, relative survival of at-risk populations. Estimates of the effect of intervening in risk factors produced from these models should be more realistic because “diffusion” is represented; that is, “uncertainty” about individual risk factor values measured at any single time is described (Strachan and Rose, 1991).
Kenneth G. Manton, John E. Dowd, Eric Stallard

Chapter 7. Frailty and Forecasts of Active Life Expectancy in the United States

The concept of active life expectancy (ALE) becomes ever more important as it becomes possible to extend the life of impaired, elderly persons. It suggests that, as we endeavor to increase the length of time an individual lives, we be aware of the quality of life gained. To this end, a number of different methods, using different types of data, have been used. One procedure, which is due to Sullivan (1971), combines health survey data with period life tables to provide prevalence based ALE estimates. This has been used to analyze health priorities in, for example, Canada (e.g., Wilkins and Adams, 1983), Japan (Nihon University, 1982), Britain (Bebbington, 1988), and France (Robine et al., 1989; Robine and Ritchie, 1991). Calculations have been performed for the United States using both the Health Interview Survey (e.g., Crimmins et al., 1989) and the National Long Term Care Survey (NLTCS) (e.g., Manton and Stallard, 1991).
Kenneth G. Manton, Eric Stallard, Korbin Liu

Chapter 8. Risk Factors Affecting Multiple-Disease Efficacy and Effectiveness of Intervention Programs

An important issue in public health planning is the assessment of efforts to reduce the mortality, disability, and morbidity associated with chronic diseases through the modification and control of life-style, occupational, and environmental risk factors. Evaluation of the effects of a health intervention for chronic diseases, however, is complicated for two reasons. First, the simultaneous action of the health intervention with other factors such as changes in population composition, in the distribution of different risk factors, and so forth, may mask the health effects of the intervention. For example, an intervention may serve only to partially offset the negative effects of, for example, an aging population. Without proper adjustment in the analysis for these simultaneous effects, the intervention may be interpreted to actually have had negative effects on health. Second, the positive effects of the intervention may take many years before they become manifest. One reason for this lag is that the intervention must first diffuse through the population. As documented in both sociological and anthropological studies, the diffusion process may be slow, with the desired changes in population health behavior not visible for many years. This lag period is further increased because, even after the contents of the intervention program have been diffused, the resulting behavioral changes do not immediately improve the individual’s health. For example, a smoking cessation program may take years before significant numbers of individuals quit. Then, after quitting, there is still a period of time before the physiological benefits produced by smoking cessation become manifest.
H. Dennis Tolley, Kenneth G. Manton, J. Richard Bumgarner

Effects of Interventions on Health Costs


Chapter 9. Estimates and Projections of Dementia-Related Service Expenditures

The increased prevalence of dementia, and its primary cause, Alzheimer’s disease and related disorders (ADRD), in the U.S. population has resulted from increased survival to the advanced ages and the increasing size of elderly and oldest-old cohorts (e.g., Evans et al., 1989; Hay and Ernst, 1987; Office of Technology Assessment [OTA] 1987). These trends will continue and, without advances in treatment and prevention, dementia and ADRD will increase in prevalence.
Kenneth G. Manton, Larry S. Corder, Robert Clark

Chapter 10. A Forecasting Model for the Assessment of Medical Technologies: End-Stage Renal Disease

Medical technologies possess life-cycles like those of other technologies. That is, they pass through phases of development, diffusion, established use, and obsolescence. The primary criterion for the adoption and abandonment of medical practices has been their impact on the health and well-being of the patient. However, it is also recognized that differences among competing technologies in their use of human and financial resources must be considered.
Henry Krakauer

Chapter 11. Projections of the Aged Supplementary Security Income Population: The Implications of Uncertainty

The Supplemental Security Income (SSI) program for aged, blind, and disabled persons was enacted by amendment to the Social Security Act in 1972. The act combined a number of income security programs for these groups and set nationwide standards of eligibility to receive income and Medicaid benefits under the new programs (certain exceptions to the national uniformity rule exist in Medicaid “209B” States). Designed to provide some measure of economic security to those persons who would not otherwise fully qualify for other Federal transfer payments, the program has grown rapidly since its inception and might be expected to continue to do so. When payments to individuals began in 1974 approximately 3.1 million persons received benefits. Fifteen years later, approximately 4.6 million persons received benefits. For the aged population receiving benefits, there has been a decline since SSI began. If historic data are examined, the rate of participation in SSI-“type” programs declined from 217 per thousand in 1940 to 66 per thousand in 1988—a 70% drop (Committee on Ways and Means, 1990; Social Security Administration, 1983).
Larry S. Corder, Lisa M. LaVange, Fred A. Bryan

Chapter 12. Evaluation of Long-Term Care: Estimation of Health Transitions in Frail Populations

A problem in developing long-term care (LTC) insurance and service systems is the complexity of the relation between the personal and health characteristics determining LTC needs, and the observed use of services. Factors related to the need for LTC are medical conditions, behavioral problems, loss of function, lack of informal care, and loss of spouse, income, and education. Each may be represented by several measures; for example, medical status may be represented by diagnoses; functional ability by scores on activities of daily living (ADL; Katz and Akpom, 1976), or instrumental activities of daily living (IADL; Lawton and Brody, 1969); informal care may be represented by family relations—their duration, level, and kind. If the population were cross-classified on all these measures, the number of cells would be large and, in any practically sized sample, the average number of observations in a cell would be small.
H. Dennis Tolley, James C. Vertrees, Kenneth G. Manton

Chapter 13. Financing and Use of Long-Term Care for the Elderly

Perhaps no other part of the health-care system generates as much frustration as does the organization and financing of long-term care (LTC) for the elderly. The disabled elderly and their families confront a fragmented delivery and financing system, a relative lack of noninstitutional services, long waiting lists for institutional placement, mediocre-quality care, and financing hardship (Vladeck, 1980). Public financing, primarily through Medicaid, is perceived as both overly expensive and inadequate.
Joshua M. Wiener, Raymond J. Hanley, Denise A. Spence, Sheila E. Murray

Longitudinal Research: Current Status and Future Outlook


Chapter 14. Molecular Biological Approaches to Understanding Aging and Senescence

Human longevity depends strongly on health, but there is considerable uncertainty about the potential increase in human longevity that could result from diverse curative and preventive interventions (Manton et al., 1992). The basis for some of the most profound intervention strategies in the future lies in the realm of rapidly developing molecular biological technologies. The purpose of this chapter is to review some molecular biological approaches that may be useful in studying the aging process and intervening to delay or prevent age-dependent degenerative processes that lead to disease.
Huber R. Warner

Chapter 15. Biomedical Research and Changing Concepts of Disease and Aging: Implications for Long-Term Health Forecasts for Elderly Populations

In prior chapters, we reviewed models for forecasting the health of the U.S. elderly and oldest-old population. In Chapter 4 we suggest that a major stimulus to the development of forecasting models is increased scientific knowledge about human senescence and chronic disease. As knowledge increases, the boundaries of senescence and disease become less distinct (Mooradian and Wong, 1991a,b). Models of senescence have evolved from a unidimensional, genetically fixed failure process, to multidimensional stochastic processes where polygenic determination of multiple pathologies and losses of function interact with environment (Finch, 1991; Manton, Stallard, et al., 1992). Forecasting models must keep pace with, and anticipate the future direction of, this evolution—with the complication that, even using multiple data sets, processes will be only partially observed. This chapter reviews and illustrates how biomedical research can be used to construct models, and substantive scenarios, to forecast the effects of future health interventions.
Kenneth G. Manton


Weitere Informationen