Several tools exist to process and interpret occupational hygiene samples, for example estimating mean exposure, the fraction of exposures exceeding an exposure limit. Unfortunately, almost none of them allow the dataset to include non-detects, i.e. samples reported as lower than a limit of quantification (<LOQ). As a consequence of this, practitioners have traditionally replaced non-detects with LOQ/2 or LOQ/√2 when analysing such datasets. This approach, called simple imputation, has been severely criticised in the literature because it causes bias in the estimation, especially when evaluating uncertainty around estimates.
NDexpo is a Web application that implements a rigorous censored data treatment method proposed by Denis Helsel in his monograph: Statistics for Censored Environmental Data Using Minitab and R (CourseSmart). Wiley 2012.
NDexpo implements an approach labelled regression on order statistic (Robust ROS). This approach compares well with other methods such as maximum likelihood, and represents a significant improvement over simple imputation. It is based on fitting detected values to a lognormal distribution, and then predicting a replacement value for each nondetect as a function of the fitted distribution and their rank among the detected values. Each nondetect therefore ends up replaced by a prediction, and, while the predictions do not have any individual meaning, the modified dataset can be processed in traditional software as if there were no nondetect.
Note : NDexpo will not work under certain Internet Explorer settings. We recommend using Firefox or Chrome
Few tools are available to help assess potential risk following dermal penetration of chemicals. The proposed tool’s objective is to help occupational health practitioners perform this assessment through summarizing relevant information from several sources and calculating a hazard index. Information provided by Upercut for over 1 500 chemicals include:
- Standard classifications of systemic percutaneous toxicity (including 2 skin notation systems)
- Toxic potential based on animal toxicity as coded by the General Harmonized System
- Intrinsic potential for transdermal passage
The dermal hazard index calculated by Upercut compares the doses potentially absorbed through skin to a reference dose deemed safe for most workers. The index is accompanied by a fully quantitative uncertainty analysis.
The Upercut tool was designed with the help from a grant by the French Agency for Food, Environmental and Occupational Health & Safety (ANSES) part of it's 2007 request for projects. The research scientists involved are Jérôme Lavoué, Assistant professor at the Environmental and Occupational Health Department of the Université de Montréal / Research scientist at the University of Montreal Hospital Research Center as well as Antoine Milon and David Vernez, Research scientists at the Institut universitaire romand de santé au travail (IST).
Current recommendations for interpreting exposure levels rely heavily on statistical procedures that assume the data come from known distributions (mostly the lognormal and normal distribution).
It is therefore logical to verify that these assumptions are met; this is the main purpose of Distrib.
Note: Distrib is ‘non-detects friendly’. These types of value can be entered and are treated according to the NDexpo procedure.
CANJEM is a job exposure matrix, i.e. for a given occupation and time period, it provides information on the probability, frequency and intensity of exposure from a list of 258 occupational risk factors.