Far from all exposure situations require obtaining quantitative measurement data to reach a decision. But when measurements are required, the considerable day-to-day variations observed in exposure levels render their interpretation challenging. How can an industrial hygienist convince their boss that six measurements all below the occupational exposure limit probably correspond to a potentially harmful situation?

Despite the availability for several decades of a robust theoretical background for this type of analysis—mostly based on the lognormal model and often termed “IH statistics”—few practical tools have been available to support practitioners performing the rather involved calculations needed.

IHSTAT is one of the few available and free IH statistics tools and is arguably the best known in the IH community. It was made available first through the AIHA monograph “A Strategy for Assessing and Managing Occupational Exposures,” which was published in 1998 and was updated with a multilingual capability setup by volunteer Daniel Drolet in 2008.

Industrial hygiene statistics have recently evolved into a new era thanks to Bayesian statistics, which allow practitioners to obtain answers that are more direct and easier to communicate (for example, there is an 84 percent probability that this similar exposure group, or SEG, is overexposed—no need to explain upper tolerance limits anymore). State-of-the-art processing of non-detects is another notable advantage. Data interpretation tools using Bayesian statistics started to appear around 2010. The first was the IH Data Analyst program from Paul Hewett, famous for its “BDA charts” that allowed practitioners to estimate from data a probability associated with each of AIHA’s exposure control categories. Expostats from the University of Montreal appeared in 2014 in the form of web-based and standalone tools, adding the evaluation of exposure heterogeneity within a SEG and analysis of exposure determinants in addition to single SEG analysis.

Our team at the University of Montreal modified the well-known IHSTAT application with Daniel’s help by replacing the embedded equations with a Bayesian engine similar to the one at expostats.ca, allowing the most recent technical developments in Bayesian statistics while maintaining the simple interface favored by many practitioners. Interested parties should note that the Bayesian engine is open source and available to anyone who wishes to create their own tool.

The new tool, called IHSTAT_Bayes, can be found on the AIHA risk assessment tool webpage. The installation of IHSTAT_Bayes is a bit more involved than IHSTAT because of the embedded engine, but then it’s the same familiar user interface. Users input the OEL and measurement values (which can include values below the limit of quantitation) and are presented with traditional distribution parameters (geometric mean and geometric standard deviation), decision statistics (exceedance fraction, 95th percentile or any other percentile, arithmetic mean), and associated confidence limits (70 percent or 95 percent upper limits), enhanced by the addition of the probabilities associated with AIHA’s exposure control categories. The multilingual capability has also been preserved with eight languages included in the recently released version 1.01.

Through the creation of IHSTAT_Bayes, we hope to improve the availability of simple-to-use yet state-of-the-art IH statistics calculators and ultimately better protect workers.