Evaluation of the exposure prediction component of Control of Substances Hazardous to Health Essentials

• Published in: Journal of occupational and environmental hygiene Vol. 17; no. 2-3; pp. 97 - 108
• Main Authors: Kimbrough, Leshan J., Oestenstad, R. Kent, Beasley, T. Mark
• Format: Journal Article
• Language: English
• Published: England Taylor & Francis 03-03-2020; Taylor & Francis LLC
• Subjects: Control banding; Controlled Substances - adverse effects; Controlled Substances - analysis; COSHH Essentials model; Exposure; external validation; Hazard assessment; Hazardous materials; Health hazards; Humans; Inhalation Exposure - statistics & numerical data; Linear Models; Occupational Exposure - statistics & numerical data; Occupational safety; qualitative risk assessment; Vapors; Volatility; qualitative risk assessment; external validation; COSHH Essentials model; Control banding
• ISSN: 1545-9624; 1545-9632
• DOI: 10.1080/15459624.2020.1717501

The exposure prediction component of the Control of Substances Hazardous to Health (COSHH) Essentials model (paper version) was evaluated using field measurements from National Institute of Occupational Safety and Health (NIOSH) Health Hazard Evaluation (HHE) reports. Overall 757 measured exposures for 94 similar exposure groups (SEGs) were compared with the COSHH Essentials predicted exposure range (PER). The SEGs were stratified based on the magnitude of measured exposures (high, medium, or low) and physical state of the substance (vapor or particulate). The majority of measured exposures observed involved low-level exposure to vapors; thus, overall findings from the current study are limited to low-level vapor exposure scenarios. Overall, the exposure prediction component of COSHH Essentials vastly overestimated low-level exposures to vapors. This study went beyond the scope of previous studies and investigated which model components led to the overestimation. It was concluded that COSHH Essential's tendency to overestimate was due to multiple complex interactions among model components. Overall, the magnitude of overestimation seems to increase exponentially as values for predictor variables increase. This is likely due to the log-based scale used by the model to allocate concentration ranges. In addition, the current banding scheme used to allocate volatility appears to play a role in the overestimation of low-level exposures to vapors.