Efficacy of predictive modeling as a scientific criterion in dermal hazard identification for assignment of skin notations

Skin notations (SNs) represent a hazard characterization tool for alerting workers of health hazards associated with dermal contact with chemicals. This study evaluated the efficacy of a predictive model utilized by the National Institute for Occupational Safety and Health to identify dermal hazards based on potential of systemic absorption compared to hazard assignments based on dermal lethal dose 50% or logarithm of octanol–water partition coefficient. A total of 480 chemicals assigned an SN from at least one of seven institutes were selected and partitioned into seven hazard categories by frequency of SN assignment to provide a basis of evaluation for the predictivity of the examined criteria. We find that all three properties serve as a qualitative indicator in support of a dichotomous decision on dermal hazard; the predictive modeling was identified from a multiple regression analysis as the most significant indicator. The model generated estimates that corresponded to anticipated hazard potentials, suggesting a role of the model to further serve as a hazard-ranking tool. The hazard-ranking capability of the model was consistent with the scheme of acute toxicity classification in the Globally Harmonized System of Classification and Labeling of Chemicals.

► We compared scientific data for efficacy in assignment of skin notations.
► Dermal LD50, log KOW, and predictive modeling supported a dichotomous decision.
► Predictive modeling was the most influential in association with notations.
► Model output was consistent with GHS classification based on dermal toxicity.
► Predictive modeling may be considered for application in dermal hazard ranking.