Declines have been documented in the functional status and mortality of the U.S. elderly population in the 1982 to 1994 National Long Term Care Surveys (NLTCS). These are likely due to a mix of demographic socio- economic, and biotechnical changes occurring in the U.S. over time and in different birth cohorts. Important among these are increased understanding of how aging and chronic disease operate in elderly cohorts, and over time, due both to different early life experiences, and health care access over time, and to advances in biomedical technology and the ability to treat chronic diseases at later stages than previously possible. Such changes are, in part, stimulated by basic research showing physical changes associated with aging are more """"""""plastic"""""""" than previously thought. In the proposed research we will model changes in the mortality and the health of the U.S. elderly population as a function of health trends, and advances in biomedical technology and research, as mediated through Medicare which pays for most health services used by U.S. elderly beneficiaries. This will require methodological innovations in population based multivariate models describing the effects of disease dependency, and in assessing cost consequences of diseases in the current and future U.S. elderly populations. Maximum likelihood procedures will be used to combine demographic, social, economic, actuarial, and health survey data to estimate parameters of stochastic population processes. The data used includes the longitudinally linked NLTCS of the U.S. population aged 65+, linked to Medicare mortality and service use records. Ancillary data sources on age specific population disease incidence and prevalence will also be used as well as data on the likely effects of new treatments on the U.S. elderly population. Several other surveys will be used to improve specific expenditure estimates (e.g., three National Nursing Home Surveys (1977-1996), two National Health Expenditure Panels (1987, 1996), and Medicare administrative data.
Showing the most recent 10 out of 24 publications
