Risk assessment confronts uncertainties and variabilities at every step. Typical risk assessments attempt to account for them by choosing extreme values in order to be conservative; but this may lead to compounding levels of conservatism and unreliable results. In such cases, where more extensive analysis is needed, we use probabilistic methods such as Monte Carlo simulations to quantify uncertainties and variabilities and to characterize risk more completely and reliably. Sometimes the uncertainties or variabilities can be adequately summarized using analytical techniques (e.g. using means and standard deviations, together with evidence-based ideas about the probability distributions that apply); but in most cases numerical methods are necessary. In addition to performing probabilistic assessments, we offer consulting services to determine whether such advanced techniques can or should be used, whether others should have used them in specific situations, and to assess the validity and accuracy of such assessments when they have been performed. We have been working in this area since the mid-1970's, and are nationally recognized experts in risk assessment and related fields.
Report to the federal Commission on Risk Assessment and Risk Management
Under contract with the Commission on Risk Assessment and Risk Management, we designed and conducted probabilistic assessments of cancer risks due to drinking water contaminated with part-per-billion concentrations of the chemicals 1,1-dichloroethylene and vinyl chloride, and to occupational exposures to the latter chemical. In particular, we were asked to design and implement methods that could provide regulators and other decision-makers with information that went considerably beyond "upper-bound" point-estimates generated by default methods. We conducted case studies based on holistic uses of the underlying dose-response data generated from rodent bioassays, and on thorough evaluations of the variabilities and uncertainties in key input parameters and models. We also discussed the qualitative distinctions between carcinogens, such as vinyl chloride, known to cause cancer in workers, as opposed to chemicals, such as 1,1-dichloroethylene, known to cause cancer only in laboratory rodents. Explicit recognition of the needs of environmental regulators and others was also central to our work.
PCBs in the Kalamazoo River
For a private company, we performed a probabilistic health risk assessment for people presumed to ingest fish from a river contaminated with polychlorinated biphenyls (PCBs). Our assessment took full account of all the available measurements on the exposed population, and incorporated the variability among members of that population and the uncertainties inherent in these measurements. Rather than evaluate hypothetical populations (as is done in screening-level assessments), we examined the population of anglers who actually ate fish from the Kalamazoo River, a population that had been extensively characterized in a survey designed for that purpose. We also quantified and took account of the uncertainties and variabilities in the dose-response data upon which the carcinogenic potencies of the PCBs are estimated. Our assessment provided risk-estimates applicable to individuals and to various populations, including hunters and gardeners in the surrounding flood plain.
Clostridium perfringens in ready-to-eat meat products
We collaborated with USDA to perform a probabilistic risk assessment for the potential effects of Clostridium perfringens in ready-to-eat and partially cooked meat and poultry products. Consumption of high levels of the vegetative form of some strains of C. perfringens may lead to diarrheal illness. This bacterium is ubiquitous in the environment and grows well on meat and poultry products in the absence of oxygen, particularly at relatively high temperatures. Its control is complicated because in adverse conditions it sporulates, and the spores are resistant to the high temperatures used in cooking meat products; but these high temperatures activate the spores to develop into vegetative cells which may then grow during the cool-down or storage of the products. We evaluated all available literature to derive variability and uncertainty distributions for the growth behavior of this organism, consumer habits of storage (e.g. refrigerator temperatures) and cooking, and the dose-response for diarrheal illness. We then wrote a Monte Carlo simulation program and used these distributions in the program to estimate the number of such illnesses (and the uncertainty in that number) to be expected in the U.S. under current conditions.