Student Research: Ravi Sanga

Effects of Uncertainties on Exposure Estimates to Methylmercury: A Monte Carlo Analysis of Biomarkers of Exposure vs. Predictive Dietary Estimation
Faculty Advisor: 


Presented is a general model for differentiating between population heterogeneity and state of knowledge uncertainty in methylmercury (MeHg) exposure assessments. Using data from fish consuming populations in Bangladesh, Brazil, Sweden and the United Kingdom, we compare dietary methylmercury exposures of predictive model estimates against those derived from biomarkers. Parameter uncertainty disaggregation was performed by two-dimensional Monte Carlo analysis. By disaggregating the uncertainty into components (i.e., population heterogeneity, measurement error, recall error and sampling error) detailed information was obtained regarding the contribution of each component to the overall exposure estimate uncertainty. Diet:hair and diet:blood MeHg exposure ratios were assessed across populations to develop distributions useful for conduction probabilistic assessments of MeHg exposure between predictive modeling, hair Hg, and blood Hg analyses. Fifth and 95th percentile estimates of median population exposure are presented. For example, results from the United Kingdom population show that the hair biomarker model results in an 86% higher variance about a central mean estimate relative to a predictive dietary model, and the blood biomarker model results in a 50% higher variance about a central estimate relative to a predictive dietary model. When comparing the dietary recall, hair biomarker, and blood biomarker models to an assumed bias free duplicate diet exposure model, the blood biomarker was seen to have the least bias amongst all models for the UK population. Such analyses, used here to evaluate MeHg exposure biomarkers, can be used to refine predictive exposure models, to improve information used in site management and remediation decision making, and to identify sources of uncertainty in risk estimates.