Final Program - Society for Risk Analysis

verbal decision analysis to classify different combinations of influencing factors intodifferent levels of the operator’s abilities to successfully perform the required action.The specific feature of the suggested approach is that it presents the opportunity ofusing the decision support system to logically analyze errors and their underlying factorsin the process of data collection, instead of following the traditional approachof drawing conclusions from the investigation reports. The results allow to furtherimprove the model of SAFE (System for Analyzing and Forecasting Errors) for accident/incidentinvestigation in aviation.W1-D.1 Yemshanov D, Koch F, Lyons B, Ducey M, Koehler K; dyemshan@nrcan.gc.caNatural Resources Canada, Canadian Forest Service, North Carolina State University, Universityof New Hampshire, Canadian Food Inspection AgencyASSESSING RISK OF INVASIVE SPECIES UNDER SEVERE UNCER-TAINTY: A DOMINANCE-BASED APPROACHA critical issue for modern risk assessment of invasive species is severe uncertainty.The uncertainty associated with spread and establishment of invasive organismsinfluences how regulatory decision-makers respond to expanding incursions.The omission of uncertainty in estimating long-distance spread leads to very low establishmentrates (as is typical for rare events) and may communicate overconfidencewhen used to support decisions regarding managing the expansion of pest population.In this paper, we present a model-based approach to invasive species risk assessmentthat combines two potentially conflicting analytical tasks: (1) estimating thelikelihood of a new organism being established at a given locale and (2) quantifyingthe uncertainty of that prediction. Our methodology focuses on the long-distancespread of invasive pest species. We use the stochastic simulation model to generatedistributions of plausible invasion outcomes for a species of interest over a particularspatial domain. We then apply second-degree stochastic dominance (SSD) criteria toorder all geographic locations within that domain by their distributions of plausibleoutcomes to an ordinal risk rank. The SSD rule implies that decision-makers are riskaverse and would choose a course of action based on the minimum tolerable distributionof invasion risks. We apply the modeling approach to evaluate the pathways ofspread of the emerald ash borer, a major pest of ash trees in North America. Resultsshow the tradeoffs between the pest`s estimated establishment rates at remote locationsand the amount of uncertainty in those model-based forecasts. The integratedrisk ranks delineate major ``crossroads`` where the movement of the pest with commercialtransportation is most likely to occur. Overall, the new approach generatesmore realistic predictions of long-distance spread than models that do not accountfor uncertainty and can help design more effective pest surveillance and regulatoryprograms.198P.53 Yoon M, Yang Y, Tan Y-M, Clewell HJ; myoon@thehamner.orgThe Hamner Institutes for Health SciencesDEVELOPMENT OF A HUMAN PBPK MODEL FOR CARBARYL US-ING AN IN VITRO TO IN VIVO EXTRAPOLATION APPROACH ANDITS APPLICATION IN REVERSE DOSIMETRYThere is increasing interest in applying in vitro data and pharmacokinetic (PK)modeling to predict in vivo kinetics in order to improve risk assessments of chemicalsin humans. Physiologically based pharmacokinetic (PBPK) modeling providesan effective integration tool to quantitatively extrapolate in vitro data to in vivo. Thegoal of this study was to demonstrate an in vitro to in vivo extrapolation (IVIVE)approach for developing a PBPK/PD model for carbaryl. A parallelogram approachwas used in which successful application of the IVIVE approach in rats served asthe basis for application in the human. Carbaryl is a widely used pesticide with acutetoxic effects due to its interaction with cholinesterase (ChE) in the brain. Carbarylmetabolism was determined using human hepatocytes. Interactions between carbaryland ChE were determined in red blood cells in human blood, as a surrogate for brainChE, the actual target for carbaryl toxicity. These in vitro PK and PD data were extrapolatedto in vivo using biologically relevant scaling factors to describe the dispositionand ChE inhibition dynamics of carbaryl in humans. The Markov Chain MonteCarlo technique was used to integrate the in vitro data and known information on humaninterindividual variability of enzyme expression instead of simply extrapolatingthe average values from the experiments. The variability in model predicted carbarylexposure in the target tissue thus reflected the impact of human variability in metabolism,information that is not readily available from individual studies. Reverse dosimetrywas conducted using the PBPK model to estimate carbaryl exposures consistentwith human biomonitoring data by integrating carbaryl exposure patterns simulatedwith the CARES software and Monte Carlo analysis. This modeling approach canserve as a template for developing models for other environmental chemicals using invitro data to support risk assessments.P.110 Youngblood SA, Chaney PL, Weaver GS; susan.youngblood@auburn.eduAuburn UniversityTHE SOUTHERN STATES OUTBREAK SEQUENCE: SURVEYINGSURVIVORS AND INFORMING RISK MANAGEMENTRisk management depends on fully understanding constraints—for instance,communication problems and social elements such as perception (e.g., Covello, 1991;Trumbo, 2000)—and is incomplete when it does not address these issues. Post-disasterquestionnaires are a critical means of helping the profession better understandconstraints by examining how people prepare for risks, what resources they hadavailable during a real situation, how they actually reacted, and why they made those