You are here:
STATISTICAL METHODOLOGY FOR THE SIMULTANEOUS ANALYSIS OF MULTIPLE TYPES OF OUTCOMES IN NONLINEAR THRESHOLD MODELS.
Citation:
Coffey, T., C. Gennings, AND V. C. MOSER. STATISTICAL METHODOLOGY FOR THE SIMULTANEOUS ANALYSIS OF MULTIPLE TYPES OF OUTCOMES IN NONLINEAR THRESHOLD MODELS. Presented at Society of Toxicology, New Orleans, LA, March 06 - 10, 2005.
Impact/Purpose:
xx
Description:
Multiple outcomes are often measured on each experimental unit in toxicology experiments. These multiple observations typically imply the existence of correlation between endpoints, and a statistical analysis that incorporates it may result in improved inference. When both discrete and continuous endpoints are measured, methodology to appropriately analyze the endpoints is scarce. We propose a new methodology to simultaneously analyze binary, count, and continuous outcomes for nonlinear threshold models that incorporates the intra-subject correlation. This methodology uses quasi-likelihood estimation and a working correlation matrix, and is therefore appropriate when the mean response of each outcome is of primary interest and the correlation between endpoints is of secondary importance. The methodology is illustrated with data collected from a mixture of five organophosphorus pesticides. Five outcomes-gait abnormality, tail pinch response, motor activity, and brain and blood cholinesterase-were analyzed using nonlinear threshold models. Compared to the individual analyses of each endpoint, the simultaneous model that incorporates the correlation between all five endpoints resulted in similar parameter estimates but had increased power to detect differences due to more precise estimates. In particular, the dose-threshold parameters for three endpoints were found to be statistically different from zero in the simultaneous analysis while none was significant in the individual analyses. Supported by T32 ES07334-01A1 (NIEHS, NIH). This is an abstract of a proposed presentation and does not necessarily reflect EPA policy or the official views of the NIEHS, NIH.