Influence of parameter values and variances and algorithm architecture in ConsExpo model on modeled exposures

This study evaluated the influence of parameter values and variances and model architecture on modeled exposures, and identified important data gaps that influence lack-of-knowledge-related uncertainty, using Consexpo 4.1 as an illustrative case study. Understanding the influential determinants in exposure estimates enables more informed and appropriate use of this model and the resulting exposure estimates. In exploring the influence of parameter placement in an algorithm and of the values and variances chosen to characterize the parameters within ConsExpo, "sensitive" and "important" parameters were identified: product amount, weight fraction, exposure duration, exposure time, and ventilation rate were deemed "important," or "always sensitive." With this awareness, exposure assessors can strategically focus on acquiring the most robust estimates for these parameters. ConsExpo relies predominantly on three algorithms to assess the default scenarios: inhalation vapors evaporation equation using the Langmuir mass transfer, the dermal instant application with diffusion through the skin, and the oral ingestion by direct uptake algorithm. These algorithms, which do not necessarily render health conservative estimates, account for 87, 89 and 59% of the inhalation, dermal and oral default scenario assessments,respectively, according them greater influence relative to the less frequently used algorithms. Default data provided in ConsExpo may be useful to initiate assessments, but are insufficient for determining exposure acceptability or setting policy, as parameters defined by highly uncertain values produce biased estimates that may not be health conservative. Furthermore, this lack-of-knowledge uncertainty makes the magnitude of this bias uncertain. Significant data gaps persist for product amount, exposure time, and exposure duration. These "important" parameters exert influence in requiring broad values and variances to account for their uncertainty. Prioritizing them for research will not only help fill a large and influential knowledge gap, but also lead to more accurate assessments and thus refine the studies informing policy decisions.