Models for assessing biological age. Webinar, 24 September 2020
September 24, 2020
Differences in health status at older ages result from social and biological mechanism responsive to the accumulation of deleterious exposures across the lifespan. These exposures vary across individuals leading to heterogeneity in population health as people age. Chronological age (CA) is a typical indicator used to represent overall risk of morbidity and death such that older age is highly correlated with risk of chronic disease, morbidity, disability and mortality. However, CA is a crude proxy for assessing underlying individuals’ physiological deterioration. An alternative to chronological age is biological age (BA), an indicator that assesses age-related biological change measured through a battery of biological markers of major physiological systems. Several methods have been proposed to estimate BA, most of them rely on linear regression approaches in which chronological age is modeled as a linear function of BA markers. In this presentation we show an alternative way to construct an indicator of BA by improving upon previous literature. First, we generalized previous approaches by considering the link between BA and CA through an structural equation model (SEM). Under this approach BA is a latent construct associated with both each biomarker and CA. Second, we relax the assumption of linearity between CA and BA and assumed that an increase in calendar year (i.e., age) is exponentially related BA. This implies that at older ages, for example, one year increase in age exerts a larger toll than one unit on the underlying biological age of the individual. We apply our proposed method using data from the US National Health and Nutrition Examination Survey 1988-1994 and compare results with three of the most commonly used in the literature (principal components, linear regression and a method proposed by Klemera-Doubal).
Is Associate Professor in the Department of Community Health Sciences at the University of California-Los Angeles (UCLA) and Associate Director of the UCLA California Center for Population Research. Hiram completed his Ph.D and M.A. in Demography at the University of Pennsylvania (USA), a M.S. in Mathematics at Northern Arizona University (USA), and a B.S. in Actuarial sciences at the Universidad Nacional Autonoma de Mexico. Hiram is co-founder, with Alberto Palloni and Guido Pinto, of the Latin American Mortality Database, the largest data repository of mortality from 19 countries in Latin America (including data from around 1850). His main areas of research include the study of health patterns and trends in low- and middle-income countries including issues about compression of morbidity, links between early life experiences and late life outcomes, as well as on biomarker data to study physiological patterns of health and their link with sociodemographic factors.