Statistical Analyses for Assessing Space-Time Exposure Data

Johns Hopkins University, Baltimore MD

Investigators

Linked publications & trials

Abstract

Project Summary Technological advancements in exposure assessment, a necessary component of intervention, control, and compliance, have recently increased the accuracy, reliability, and affordability of portable, direct-reading monitors. These monitors can rapidly assess worker exposures to occupational hazards. By coupling the estimated exposure with a known location, an industrial hygienist has the ability to connect exposures to specific sources. Contour plots of the hazard concentration over space, known as concentration maps, have recently been used to assess the spatial variability of hazards. Concentration maps have the potential to be powerful because they are easily comprehensible for managers, exposed employees, and occupational health scientists to locate areas of concern. Reducing or eliminating exposures in these areas will improve worker health. While we believe there is great potential for direct-reading instruments to aid in the identification and mitigation of workplace exposure hazards, it can be dangerous to apply such a methodology without understanding the uncertainties associated with this new form of exposure assessment. To date, no statistical framework has been applied to these maps. The goal of this project is to evaluate several statistical approaches for the analysis of workplace exposure data collected with direct-reading instruments. There are three specific aims for this project. First, we will employ a spatial statistical mapping method to build reliable mapped exposure estimates and confidence intervals using a previously collected exposure dataset. Second, we will collect a comprehensive dataset of hazard concentration as a function of time and space for noise level and aerosol concentration exposures. Third, we will generate a reference concentration map from the high temporal and spatial resolution dataset, against which we may evaluate simpler statistical models.