All ETDs from UAB

Advisory Committee Chair

David B Allison

Advisory Committee Members

Charles Katholi

Inmaculada Aban

Nengjun Yi

Kevin Fontaine

Document Type

Dissertation

Date of Award

2013

Degree Name by School

Doctor of Philosophy (PhD) School of Public Health

Abstract

Analyses of the National Health and Nutrition Examination Survey (NHANES) have suggested that the mortality rate (MR) associated with obesity has decreased over calendar time. However, there has been conflicting evidence about the apparent change in the obesity-MR over calendar time. This dissertation investigates some of the empirical and methodological issues involved in assessing the longitudinal change in the obesity-mortality associations. In the first manuscript, Obesity and Mortality: Are the Risks Declining? Evidence from Multiple Prospective Studies in the U.S., a multiple longitudinal design is proposed and implemented to evaluate whether the obesity-mortality associations have declined over calendar time after controlling for age related effect-modification, length of follow-up, stable study level-factors and recency. The U.S. Hispanic population has increased substantially over calendar time. The second manuscript, Does Obesity Associate with Mortality among Hispanic Persons?: Results from the National Health Interview Survey, provides estimates of obesity-mortality associations in the U.S. Hispanic population. The third paper, Plausible Nuisance Contributors to the Appearance of a Decreasing Deleterious Association of Overweight and Obesity with Mortality Rate over Calendar Time is about the role of plausible nuisance contributors in the apparent change in obesity-MR relationship. Simulations were used to evaluate whether changes in BMI distribution over calendar time coupled with the use of broad BMI categories (< 18.5, 18.5 to <25, 25 to <30, 30 to <35 and 35 or greater) could contribute to the apparent change in obesity-MR association. The fourth paper, Comparing Predictive Accuracies of Competing Non-Nested Parametric Survival Models, presents significance tests to evaluate differences in the predictive accuracy of non-nested models based on loss functions (absolute and quadratic). Years of life lost and deaths attributable to obesity, predicted using BMI, influence the attitude of policy makers and health care providers. The proposed tests, for example, can be used in evaluating whether the accuracy in predicting longevity can be improved by departing away from standard BMI categories to a continuous form of BMI.

Included in

Public Health Commons

Share

COinS