Models of the interaction of mortality and the evolution of risk factor distribution: a general stochastic process formulation.

Generally analyses of longitudinal studies of chronic disease risks do not directly model the change with time of risk factor values and the interactions of those changes with risk levels. Failure to account for such process characteristics can lead to incorrect inferences about the specific effects of risk factors on mortality, the inability to accurately forecast the future risk of the cohort, and inaccurate statements about the effects of specific risk factor interventions on mortality. We present a model which does describe such a process and show how it can be estimated from longitudinal studies. We also illustrate the effects of certain risk factor process features on the evolution of disease risk data from males in the Framingham, Massachusetts study.

Duke Scholars

Author Patrick J. Eric Stallard Social Science Research Institute

Author Kenneth G. Manton Demographic Studies