Occupational Co-exposures to Multiple Chemical Agents from Workplace Measurements by the US Occupational Safety and Health Administration

Abstract

Objectives The occupational environment represents an important source of exposures to multiplehazards for workers’ health. Although it is recognized that mixtures of agents may have differenteffects on health compared to their individual effects, studies generally focus on the assessment ofindividual exposures. Our objective was to identify occupational co-exposures occurring in the United States using the multi-industry occupational exposure databank of the Occupational Safety and Health Administration (OSHA). Methods Using OSHA’s Integrated Management Information System (IMIS), measurement data from workplace inspections occurring from 1979 to 2015 were examined. We defined a workplace situation (WS) by grouping measurements that occurred within a company, within the same occupation (i.e. job title) within 1 year. All agents present in each WS were listed and the resulting databank was analyzed with the Spectrosome approach, a methodology inspired by network science, to determine global patterns of co-exposures. The presence of an agent in a WS was defined either as detected, or measured above 20% of a relevant occupational exposure limit (OEL). Results Among the 334 648 detected exposure measurements of 105 distinct agents collected from 14 513 US companies, we identified 125 551 WSs, with 31% involving co-exposure. Fifty-eight agents were detected with others in >50% of WSs, 29 with a proportion >80%. Two clusters were highlighted, one for solvents and one for metals. Toluene, xylene, acetone, hexone, 2-butanone, and N-butyl acetate formed the basis of the solvent cluster. The main agents of the metal cluster were zinc, iron, lead, copper, manganese, nickel, cadmium, and chromium. 68 556 WS were included in the analyses based on levels of exposure above 20% of their OEL, with 12.4% of co-exposure. In this analysis, while the metal cluster remained, only the combinations of toluene with xylene or 2-butanone were frequently observed among solvents. An online web application allows the examination of industry specific patterns. Conclusions We identified frequent co-exposure situations in the IMIS databank. Using the spectrome approach, we revealed global combination patterns and the agents most often implicated. Future work should endeavor to explore the toxicological effects of prevalent combinations of exposures on workers’ health to prioritize research and prevention efforts.

Application of Pattern Mining Methods to Assess Exposures to Multiple Airborne Chemical Agents in Two Large Occupational Exposure Databases from France

“…However, it is also possible that they are not relevant if the interval between exposure measurements is long (e.g., several years), during which work processes or practices could have evolved, such as by substituting one product by another. In order to account for this temporal dimension, previous analyses of coexposures in OEDBs have used time windows of one month [10], one year [9], and almost 20 years [2]. The latter study focused solely on carcinogens that may have a long latency period before the onset of disease, compared to the other two studies that also included agents with acute and non-carcinogenic chronic health effects.…”

“…Another challenge to the identification of coexposure scenarios in occupational databases lies in the definition of a minimum threshold to define a meaningful exposure level to an agent. For example, some of the analyses in Bosson-Rieutort et al [9] considered only measurements with a concentration greater than 20% of an OEL. At the other end of the spectrum, samples with concentrations below the limit of quantification (LOQ) could be informative as to the agents that may be present in the workplace (even at very low levels) based on the preliminary survey by the industrial hygienists.…”

“…However, very few studies have leveraged the data contained in these OEDBs to gain insight into prevalent patterns of coexposure across a wide range of sectors within the population. For example, Bosson-Rieutort et al [9] assessed clusters of agents present in IMIS; while Clerc et al [10] used Bayesian network models to identify mixtures of agents associated with poorly controlled work situations in COLCHIC and SCOLA.…”

“…The application of these methods to data stored in OEDBs could be valuable in identifying frequent coexposure patterns occurring in the working population for prevention purposes. To our knowledge, the only previous application was in the development of "spectrosomes" providing graphical depictions of coexposures between agents in the IMIS database [9].…”

MiXie, an Online Tool for Better Health Assessment of Workers Exposed to Multiple Chemicals

“…In the USA, multiple exposure to chemicals was demonstrated in 31% of 125,551 working situations identified at 14,513 companies. Two clusters were highlighted, one for solvents (toluene, xylene, acetone, hexone, 2-butanone, and N-butyl acetate) and the other for metals (zinc, iron, lead, copper, manganese, nickel, cadmium, and chromium) [ 6 ]. In Québec, there is little data on the frequency of multiple exposures or their main characteristics.…”