New Longitudinal Methods for Trauma Research

Boston University Medical Campus, Boston MA

Investigators

Linked publications & trials

Abstract

[unreadable] DESCRIPTION (provided by applicant): The primary assumption guiding this proposal is that contemporary longitudinal methodologies largely advanced and applied within the realm of life-span human developmental and aging research can be--and should be-implemented in the study of mental health outcomes subsequent to exposure to disasters and other highly stressful life events. As with developmental and aging research, which uses time since birth as a fundamental component of the methodology, research on the sequelae to traumatic events has a starting date, the point of exposure, from which the manifestation of an individual's symptoms can be mapped and profiled as a function of time. The broad, overriding goal of this project is to adapt, build, refine, and test state-of-the-art statistical models of individual growth/decline for the prospective study of mental health following trauma exposure and to disseminate strategies for design, model specification, and analysis to the community of trauma researchers. The heart of this proposal is the reconceptualization of a variable from a static representation of an individual's position on an attribute of interest at some designated timepoint(s) to a dynamic representation of the individual's progression of symptom expression from time of exposure. Specific Aim 1 is to adapt and refine longitudinal dynamic change score models of individual growth/decline using data reflecting repeated assessments of continuous trauma-related outcomes; a latent difference score approach will be used. Specific Aim 2 is to extend the latent difference score operationalization of dynamic change to novel approaches for analyzing longitudinal data reflecting dichotomous or ordered polytomous diagnostic status. Specific Aim 3 is to frame latent difference scores within the context of multilevel regression analyses for situations where post-exposure times of assessment vary widely over cases. Specific Aim 4 is to integrate practical solutions to the problem of incomplete data in longitudinal research on trauma outcomes. A number of high-quality longitudinal databases, appropriately reflecting variety in terms of gender, age, race/ethnicity, and trauma types, are available for secondary analyses to demonstrate the viability of the new methods. Models will be built and tested using variations of structural equation modeling and adaptations of multilevel regression procedures. Results will be translated and disseminated in a form suitable for use by mainstream trauma researchers. [unreadable] [unreadable]