Within the context of the EU’s REACH regulation, several occupational exposure models are recommended for chemical exposure assessment. These models require exposure parameters to be entered from a few to many, depending on the level of models’ complexity.
As of today, the performance of these models is not fully evaluated for a wide range of exposure conditions because of the lack of adequate exposure data. In this study, we developed and tested a new model for airborne exposure, named “TREXMO Plus (+)”, that uses three REACH models, i.e. ART, Stoffenmanager (version 4.0), and ECETOC TRAv3, as the independent predictors. This model considers that the performance of the different exposure models may vary over different exposure conditions (such as for lower and higher vapor pressure values).
TREXMO+, therefore, is conceptualized by applying weights developed from the training data sets into the independent predictors of three models to derive a refined exposure estimate. The exposure data, which counted 1058 exposure measurements, was split in two sets, where 80% was used to develop the model and the remaining 20% to test its performance.
Compared with the three REACH models individually, TREXMO+ was the least biased model and the most accurate. It was found that, on average, the TREXMO+ estimates differ by a factor between 2 and 3 from the measurements. The model was significantly correlated against the measurements; explained best the variance (R-squared, 0.44-0.71). Although more data is required to test further this model, the concept of TREXMO+ is expected to provide better estimates in comparison with the use of the exposure models individually.
Keywords: exposure factors, exposure models, industrial hygiene, occupational, risk assessment, statistical methods
Contribution to PEROSH Research Conference 2019.
