Global effects estimation for multidimensional outcomes

Many public health studies focus on multifaceted outcomes that are not easily measured with a single variable. Examples include cognitive function, quality of life and general health. To fully explore such an outcome, researchers typically collect information on multiple clinical endpoints. The resulting measurements constitute multidimensional outcomes data.;This dissertation is in part motivated by two real-world examples. The first, an investigation of lower extremity trauma, quantifies health through physical and psychosocial function as well as pain. The second, an study of Alzheimer's disease, measures quality of life via cognitive and physical function and the time up to each subject's institutionalization.;An object of great interest is the overall---or global---effect of a covariate, such as a treatment intervention, on the multidimensional outcome. Quantifying such an effect can be difficult because different clinical outcomes are usually measured on different scales; the problem is enhanced by the fact that multiple measurements on a given subject are often correlated. Additional complications arise for longitudinal or censored data.;This dissertation presents a modeling scheme permitting estimation of global covariate effects when multiple endpoints are examined in concert. The method overcomes certain limitations of preexisting approaches to analysis of multivariate outcomes data. The assumption that the effect is homogeneous across outcome measures is testable, and may be relaxed.;The global treatment effect is conceptualized as a change in the distribution functions of the outcome variables. It may thus be interpreted as a connection between outcome distribution quantiles for the treatment and control groups. This allows the presentation of a global effect as a scalar quantity applicable to all outcomes simultaneously, easing interpretation of results.;We begin by formulating a model estimating global effects for multiple continuous outcomes. We then propose an alternative scheme for analysis of multiple discrete or mixed discrete and continuous measures. We evaluate the performance of the global effect estimator through simulation.;To illustrate application, we present data analytic results for the two motivating examples mentioned above. In doing so, we apply the global effects method to quantify the impact of an intervention on the health of individuals following lower extremity trauma as well as that of a detrimental exposure on the health of sufferers of Alzheimer's disease.