Our general aim is to examine determinants of multiply caused mortality in male and female cohorts using both demographic time series of multiple cause mortality data (from national vital statistics sources) and longitudinal population surveys of health, functioning, and mortality in the U.S. elderly and oldest-old populations. In examining sex differentials in mortality, we pay special attention to the oldest-old populations where females predominate. This is appropriate because females have about a 7-year greater life expectancy than males in the U.S. and much of the late age mortality may be due to disease and disability processes initiated post-menopausally, i.e., initiated at a later age than in males. In addition to examining the effects of standard physical risk factors on sex differences in mortality, we will examine the effect of income, education, and other social dimensions. We will be interested in determining what health and functional factors mediate the effect of income, education, and other social conditions on late age mortality. Finally, we will place considerable emphasis on examining the role of chronic functional impairments and disabilities in causing death. In several preliminary studies such functional impairments were a much stronger predictor of female mortality risks than standard risk factors. A class of functional impairments which will adopt special importance in the study is cognitive and related memory and sensory impairments and their sex specific effects on mortality. The study is well suited to examine these effects because a.) where dementia or Alzheimer's disease is reported on the death certificate, it is generally not the underlying cause (we are using multiple cause data), and b.) the National Long Term Care Survey (NLTCS) contains a number of measures describing cognitive, sensory, and behavioral problems for longitudinally followed individuals. The sample of persons with such problems in the NLTCS, due to the large number of oldest-old, is expected to be large enough (e.g., 600 to 700 cases in each sample year) for detailed multivariate analyses -- especially of females who will tend to predominate in the elderly sample. The data to be employed will be U.S. multiple cause of death files for 1968 to 1990 (and later) and the 1982, 1984, 1989, and planned 1993 NLTCS with linked Medicare mortality and service use files. In the 1993 data, there will be a specialized supplement of persons 95 and older on which special studies of age reporting quality will be conducted. Additional use will be made of several other long term longitudinal studies such as the National Health Examination Surveys and their follow-up (i.e., NHANESI and the 10-year follow-up). Analyses will be conducted with multivariate procedures that are designed to simulate the physiological processes underlying health change and mortality in the U.S. elderly male and female populations. It is felt that to separate the effects of multiple physiological and functional measures that interact over time, multivariate time series modeling will be a necessity. As such the analyses will be cohort based with parameterization to detect specific period effects. Statistical tests will be conducted using likelihood ratio principles.
Showing the most recent 10 out of 54 publications
