From the Editors

Do risk analyses of controversial issues tend more to unite or to divide those who care about the results? The answer doubtless varies from case to case, but one reasonable hope for risk analysts is that risk-analytic thinking can help to clarify not only key facts but also key uncertainties and values, all of which may be important in formulating wise public policies. This month's Current Topic essay, by Society for Risk Analysis President Pamela Williams, presents follow-up interviews with some of the keynote speakers from last December's Society for Risk Analysis (SRA) Annual Meeting. The speakers addressed hydraulic fracking and legalized marijuana use as examples of politically divisive issues to which risk analysis can potentially contribute useful facts, data, and analysis. The follow-up interviews contain additional thoughts and insights on how risk analysis and policy making intersect as we collectively work through such issues. How sensible is it for a risk model to assume that excess lifetime risk of an adverse health effect is proportional to lifetime cumulative exposure to a substance that causes it? One obvious objection is that such a model does not consider how the exposure is distributed over time, and yet this is typically crucial to determining lifetime excess risk. A person who takes up intense smoking at age 80 may be much less likely to die from lung cancer than someone who smokes the same cumulative amount starting at age 40, simply because there is less life span left for the 80-year old to develop lung cancer. Bofetta et al. offer a perspective on the current state of air pollution health effects risk analysis that emphasizes the need for greater biological realism. They recommend allowing for realistic latency periods, ensuring that current excess risk rates do not depend on future exposures, not attributing to air pollution exposures effects that are due to cigarette smoking or other local confounders, and not attributing to current ambient levels of pollution effects that were caused by higher historical levels of exposure (which tend to be correlated with present, lower, levels). Considering fine particulate matter (PM2.5) and heart disease as an example, they conclude that many influential models and analyses have probably led to exaggerated estimates of the health risks that are associated with—but not necessarily caused by—current and recent past exposures to air pollution. Mansfield et al. accept past estimates of PM2.5 health effects, and investigate how different patterns of urban development affect PM2.5 concentrations and projected health effects. They compare the estimated exposure and health risk impacts of compact development and sprawling development scenarios in a case study for Raleigh, North Carolina and conclude that compact development slightly decreases estimated regional average annual PM2.5 concentrations while increasing by 39% estimated PM2.5-attributable mortality rates, i.e., mortality associated with (though not necessarily caused by) PM2.5 exposure. The explanation is that compact development increases local variations in PM2.5 concentrations and the severity of local hotspots of relatively high concentrations, even while reducing regional averages. This work highlights the importance of considering spatial variability in the effects of efforts to reduce air pollution concentrations. Boelter et al. model cumulative inhalation exposures of drywall workers (including specialist and generalist professionals and do-it-yourselfers) to dusts and chrysotile asbestos fibers, taking into account random variations in exposure intensities over time. They conclude that eight-hour time-weighted average (TWA) concentrations of respirable dust and chrysotile fibers are highest for specialists but are low enough (below historical, but not current, recommended threshold limit values) so that few or no mesothelioma or excess lung cancer deaths should be expected from these exposures. Nonetheless, use of available controls to reduce inhalation of dust is prudent. Gutting et al. advance the state of the art for risk assessment of inhaled anthrax spores by developing species-specific factors for converting inhaled dose to deposited dose. They find that rabbits appear to be more sensitive than the guinea pig, with nonhuman primates in between. How people perceive and respond to flood hazards, and to both direct experience with floods and warnings about flood hazards, are topics of long-standing interest in Risk Analysis. Atreya and Ferreira use a quasi-experimental approach to examine the reductions in prices for properties in inundated compared to noninundated properties in the same floodplain following the 1994 “flood of the century” in Albany, Georgia. They applied a hedonic model of price determinants to estimate the effects of the experience of being flooded as distinct from the effects of information about flooding. They find that being flooded led to a discounting of properties that faded over time following the flood, whereas simply being in the floodplain did not have the same effect. Direct experience has a greater impact than mere information. Koks et al. model the dynamics of economic loss and recovery following floods in the Rotterdam region, using input-output modeling and a Cobb-Douglas economic production function. They quantify both direct and indirect economic losses and find that direct losses usually outweigh indirect economic losses, but the reverse is true for low-probability high-consequence floods. Reader et al. note that research on how high-risk industries foster and manage cultures of safety has largely been conducted in the context of Western countries. They show that the construct of safety culture for air traffic management can be measured reliably across national boundaries for 17 European countries, although operational and management staff often have distinct perceptions of safety culture, suggesting opportunities for improved communication and alignment. Moreover, reliable psychometric scores on different aspects of safety culture are significantly associated with national culture scores for collectivism, power distance, avoidance of uncertainty, masculinity, and long-term orientation. The practical implication of this work is that safety culture for air traffic (and probably other organizations and industries) is a measurable construct, but that the safety culture of an organization or industry reflects the larger national culture(s) within which it operates. How these insights can be extended to understand what kinds of nations and industries are most and least able to sustain operation of high-risk, high-reward facilities (e.g., nuclear power plants) appears to be a fascinating and worthwhile area for further research. Which affects attitudes toward debated hazards, such as genetically modified organisms (GMOs), more: information about risks, or information about benefits? Zhu and Xie use three successive waves of surveys to discover that, as expected, knowledge of benefits affects attitudes toward GMOs positively and knowledge of risk affects attitudes negatively. Risk information has a larger and more enduring impact on attitude change, especially for relatively knowledgeable recipients. Of particular interest to modelers, the authors fit a structural equation model (SEM) of causal pathways to the data to clarify the directions and magnitudes of the causal relations among knowledge, information, and attitude variables. Jacob and Schiffino review the concept of public “risk policies” for reducing threats to people or the environment, as they have been implemented in the United States and described in 21 articles published between 2000 and 2010. They note that such policies typically address recent threats—especially perceived threats to the environment, but also to human health—that are considered known to the authorities, who usually rely on selected experts (and, less often, on public participation) to reduce conflict. Risk policies are implemented mainly by executive authorities at all levels of government, as well as by affected members of the private sector, and receive relatively little judicial review. At a more tactical level, Tran et al. present a dietary exposure screening tool useful for rapidly determining whether food consumption patterns and concentrations of metals in foods jointly create a risk that is not clearly too small to be of no concern. Hazard characterization, dietary exposure assessment, and risk characterization modules populated with data from NHANES and other publicly available data sources allow the screening tool to quickly assess potential public health risk when metals are detected in the food chain. Khazraee et al. examine the application of a generalized linear modeling statistical framework using the hyper-Poisson distribution (which allows data that are under- or overdispersed compared to the equal mean and variance of a Poisson model) to better understand car accident and other vehicle crash data. They show using data from Toronto, Canada and from Korea that the hyper-Poisson model compares favorably to previous models in its ability to describe real data sets, and that it has the significant advantage for risk managers that the effect of each of many risk factors on average accident rates can easily be quantified and interpreted using this model. Price and MacNicoll take a fresh look at the problem of how to allocate the total risk of a disease among multiple risk factors that combine to cause it—a problem that arises in risk management, litigation, and compensation decisions. They show that an approach based on allocating interaction terms to risk factors using equal weights leads to the same allocation of excess relative or attributable risks as an axiomatic approach based on similar problems in game theory for allocating the product jointly produced by multiple cooperating players among them. Thus, simple formulas for allocating interaction terms can produce the same answer as previously developed approaches that rely on risk accounting axioms to justify a unique risk allocation formula. Spackova and Straub critically assess the meaning and use of benefit-cost ratio (BCR) and marginal cost (MC) criteria for selecting among costly alternatives for reducing risks. They note that the BCR is often not well defined, and that the MC criterion can be used to produce optimal decisions in a hierarchical decision process where budget constraints are set centrally but specific allocations among costly alternatives are decided locally. Equating MC across subsystems coordinates the distributed, hierarchical decisions and achieves Pareto-efficient allocations.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.