FORECASTING LIFE EXPECTANCY AND ACTIVE LIFE EXPECTANCY

点击翻译按钮获取中文摘要

Demographic trends over the past century have taken developed nations through a series of epidemiologic transitions, characterized principally by increasing longevity and changes in patterns of cause-specific mortality. Recently we have entered into a new era in our epidemiologic transition--one characterized by the postponement or delay of major degenerative disease (e.g., heart disease, stroke and some cancers). As a consequence of these trends, policy makers have sought estimates of the size and health status of future cohorts of the elderly. However, current methods for forecasting life expectancy and the healthfulness of tomorrow's elderly require improvement, given statistical difficulties, debate over the validity and breadth of their assumptions an gaps in the data to which they are applied. NORC and the University of Chicago Department of Medicine propose a series of studies medical demography designed to produce improved methods for forecasting life expectancy and active life expectancy. Our approach is unique in at least two respects: first, we will explicitly apply clinical and epidemiology knowledge about disease prevention and treatment as our basis for estimating future trends affecting life expectancy and active life expectancy. Linking clinical information directly to forecasting has not yet been done but the person who has paid this question most attention, Jacob A. Brody, will chair an advisory panel overseeing our efforts, and our principal investigator also brings the expertise of clinical geriatrics to the proposed research. Second, we seek to make explicit and precise the degree of uncertainty associated with our forecasts. We will statistically estimate the magnitudes of major components of error in forecast--including the reliability of experts' opinions as reflected in forecasts--and study the propagation of these error components through the forecasting method in order to estimate the probable levels of error in the forecasts. We will summarize these results in the form of intervals ("confidence intervals") with associated probabilities that actual future population characteristics will lie within these intervals. Taken together, these features will enhance the utility of our forecasts for policy formulation and will also enable us to highlight areas for future research in medical demography.