A large part of industrial hygienists’ work consists of measuring workers’ occupational exposure levels. The task of interpreting measured levels relative to occupational exposure limits (OELs) remains a considerable challenge owing to the high variability in exposure. Despite a consensus-based theoretical framework described in the literature and recommended by the leading occupational and health (OHS) institutions, the best practices in measurement and interpretation strategies have not been widely adopted by industrial hygiene practitioners. In fact, these strategies involve statistical concepts not generally covered in the usual training programs and require computations that are unfeasible with currently available tools such as calculators and spreadsheet programs.
The aim of the WebExpo project was to improve current practices in the interpretation of occupational exposure levels by creating a library of algorithmic solutions for the most frequently asked questions concerning industrial hygiene risk assessment.
Most of these questions require estimating the parameters of one or more statistical distributions, and WebExpo used Bayesian statistics to perform the required tasks. Bayesian methods were chosen because they provide direct probabilistic inferences (e.g. What are the risks of the workers being overly exposed?), which facilitates the communication of risk. They also address methodological questions that are rarely taken into account, particularly regarding non-detect values.
The WebExpo project produced a library that makes it possible to estimate the parameters of normal or lognormal probability distributions, and to perform analysis of variance, while potentially integrating external information such as expert judgment into the analysis. The library comes with functional prototypes and allows for the generation of a comprehensive toolkit for the industrial hygiene community designed to improve interpretation of occupational exposure levels, with added flexibility for users to create or adapt their own software rather than having to use a new one.
› Read the News Release