Delineating the Possible Mechanisms Underlying Longitudinal Associations in Observational Studies on Aging

• Activity Number: 187
• Type: Invited
• Date/Time: Tuesday, August 5, 2003 : 8:30 AM to 10:20 AM
• Sponsor: WNAR
• Author(s): Karen J. Bandeen-Roche*+ and Qian-Li Xue and Luigi Ferrucci and Linda Fried
• Companies: Johns Hopkins University and The Johns Hopkins Center on Aging and Health and National Institute on Aging and Johns Hopkins University
• Keywords: causal effect ; feedback ; gerontology ; latent variable ; measurement

Abstract

Over recent decades, many cohort studies have investigated the effect of disease occurrence and physiologic change on the course of late life disability. These studies present numerous challenges to establishing causal relationships, e.g., isolating true effects from potential biases not eliminated in design. This paper proposes analytical methods related to 3 challenges that particularly threaten gerontological studies: key health constructs cannot be directly measured, but are inferred from multiple surrogate indicators; bidirectional effects are common, e.g., chronic diseases are thought to cause disabilities, but subsequent deconditioning may worsen disease burden; repeated measures are initiated late in life, so that heterogeneity in "initial" status may be great. Strategies are discussed for integrating the proposed methods with well-discussed approaches for reducing bias in observational studies. Using data from the Women's Health and Aging Study, we apply the proposed methods to analyze a scientific paradigm of common shared mechanisms leading to disability in late life. Our work may improve the adjudication and analysis of causal mechanisms in cohort studies on aging.