Bayesian Hierarchical Structure for Quantifying Population Variability to Inform Probabilistic Health Risk Assessments

Capturing and quantifying the aggregate effects of multi-source factors affecting miners' health and well-being: Construction of Bayesian belief networks

Factors originating at the organizational, work, and individual levels are closely interrelated and intricately intertwined, affecting health rates. There was limited research on the interdependence and aggregate effects between multi-source factors and occupational health and well-being (OHW). It is challenging to achieve management goals. Therefore, considering cross-level factors and across the "work environment-stress-exposure-OHW" chain, individual vulnerability was considered. A Fuzzy Bayesian Belief Network (FBBN) driven by both domain knowledge and data was constructed to carve out the logic between multi-source factors and OHW. Workers from four coal mines were surveyed twice in 6 months. 714 valid samples were included in the analysis. The interdependencies among multi-source factors were identified by the Interpretive Structure Modeling method and the visual probability estimation was achieved based on FBBN. It revealed that the work and the organizational level were the root factors. Eight factors involved in work stress were mainly mediating, and actual exposure and individual vulnerability were direct factors. Pathway interventions and joint interventions were proposed. The prediction ability and scheme feasibility of FBBN were verified. The approach developed allows robust assessments of aggregate effects and obtains multi-source factor importance. This study provides vital insights and evaluation tools for understanding workplace stress and OHW management.

Using Prior Toxicological Data to Support Dose-Response Assessment─Identifying Plausible Prior Distributions for Dichotomous Dose-Response Models

The benchmark dose (BMD) methodology has significantly advanced the practice of dose-response analysis and created substantial opportunities to enhance the plausibility of BMD estimation by synthesizing dose-response information from different sources. Particularly, integrating existing toxicological information via prior distribution in a Bayesian framework is a promising but not well-studied strategy. The study objective is to identify a plausible way to incorporate toxicological information through informative prior to support BMD estimation using dichotomous data. There are four steps in this study: determine appropriate types of distribution for parameters in common dose-response models, estimate the parameters of the determined distributions, investigate the impact of alternative strategies of prior implementation, and derive endpoint-specific priors to examine how prior-eliciting data affect priors and BMD estimates. A plausible distribution was estimated for each parameter in the common dichotomous dose-response models using a general database. Alternative strategies for implementing informative prior have a limited impact on BMD estimation, but using informative prior can significantly reduce uncertainty in BMD estimation. Endpoint-specific informative priors are substantially different from the general one, highlighting the necessity for guidance on prior elicitation. The study developed a practical way to employ informative prior and laid a foundation for advanced Bayesian BMD modeling.

Bayesian benchmark dose analysis for inorganic arsenic in drinking water associated with bladder and lung cancer using epidemiological data

Abundant epidemiological evidence has shown that there is a strong causal relationship between long-term exposure to inorganic arsenic (iAs) through drinking water and a few types of cancer (e.g., lung and bladder cancer). Traditionally, a linear low-dose extrapolation assumption was applied in risk assessment for iAs which resulted in a relatively conservative cancer risk estimate. Growing biological evidence suggests that the mode of action of iAs-induced cancer follows a threshold process (e.g., sufficient concentration of trivalent arsenic is required to disrupt normal cellular function). In this study, we applied the benchmark dose (BMD) methodology to model the relationship between the relative risk of bladder and lung cancer and the iAs concentration in drinking water using the high-quality epidemiological data reported in recently published papers, with a special focus on the low exposure range (i.e., <150 μg/L). Because of its biological plausibility and statistical flexibility, the Hill model has been chosen to model the data under a Bayesian framework. A Bayesian hierarchal model together with a bootstrap method for exposure estimation were applied to quantify uncertainty from various sources, including the within-study, between-study, and exposure uncertainties. Dose-response assessment results obtained from a number of alternative model structures and methods consistently demonstrate a threshold type dose-response curve with a threshold in the range between 40-60 μg/L of iAs concentration in drinking water. The BMD for iAs in drinking water associated with 0.1 % increase in relative risk of bladder cancer is 42.2 μg/L (BMDL 39.2 μg/L); for 0.05 % increase, the BMD is 41.6 μg/L (BMDL 38.6 μg/L). For lung cancer, the two counterpart BMD estimates are 57.0 μg/L (BMDL 43.6 μg/L) and 55.7 μg/L (BMDL 42.5 μg/L) for 0.1 % and 0.05 % increase, respectively. These analyses provide additional statistical support for a non-linear dose response for cancer risk from inorganic arsenic which may have important policy implications.