Am J Epidemiol 2002; 156:488-489.
Copyright © 2002 by the
Johns Hopkins Bloomberg School of Public Health
BOOK REVIEWS |
Risk-Benefit Analysis. Second Edition
National Cancer Institute Bethesda, MD 20892
Edited by Richard Wilson and Edmund A. C. Crouch
ISBN 0-674-00529-5, Harvard University Press, Cambridge, Massachusetts (Telephone: 800-405-1619, Fax: 800-406-9145, World Wide Web: http://www.hup.harvard.edu), 2001, 400 pp., $25.00
In the past year, public and official perceptions of risk have changed dramatically. We are all now acutely aware of the possibility of events that once seemed extremely unlikely—multiple hijackings of US commercial aircraft and biologic attacks spread through the mail. Efforts to respond to these newly perceived threats over the past year have illustrated how risk management priorities evolve with public concerns and perspectives on risk. Thus, the timing for a book on evaluation of risks and the oddities of risk management policies could not be better.
The authors of Risk-Benefit Analysis (1), Richard Wilson and Edmund A. C. Crouch, are both Oxbridge-trained, Harvard-affiliated physicists active in environmental policy and risk assessment. Risk analysis is a practical, applied science, with the aim of providing specific answers to sometimes politically contested questions rather than generating hypotheses or uncovering fundamental laws of nature. Thus, the extensive experience of the two authors as environmental risk analysts, litigation consultants, expert witnesses, and activists for "sound science" puts them in good stead to discuss the messy, sometimes paradoxical practices of risk analysis and risk management. This volume is the second edition of a 1982 book, but it bears little resemblance to the original.
Two main themes emerge from the book: 1) the importance of objective, quantitative measurement of risk, along with explicit description of uncertainty and assumptions, for sound decision making, and 2) observations that public risk perceptions and risk management choices often fail to correlate with objective, quantitative rankings of risks. Through a broad survey of the risk literature and numerous risk policy anecdotes, the authors effectively argue both of these points. However, the tension between these two themes—the ideal of objective, quantitative risk assessment and the reality of public perceptions and risk politics—runs throughout the work.
The first half of the book covers basic definitions and describes methods for calculating risk and uncertainty. Few formulas are used, and the discussion is readable and accessible. Risk is defined as the probability of the occurrence of an event multiplied (though it may not be mathematically multiplicative) by the severity of the event. The authors then distinguish between two classes of risks—"historical" and "new." Historical risks are those for which we have a great deal of data (or at least the ability to collect data if we want to), such as motor vehicle accidents, influenza epidemics, and cigarette smoking. Methods for studying these risks are "simple, direct, and obvious," the authors say, though some epidemiologists might disagree. In contrast, new risks are those for which adequate frequency data are difficult to obtain. These include improbable, catastrophic events, such as nuclear reactor meltdowns, and diseases that are weakly associated with multiple causes, such as cancers caused by low-level chemical exposures.
New risks raise novel challenges for risk assessors, since available data must be used to make inferences about risks that are difficult to observe. Wilson and Crouch include the "ambiguities" of epidemiology among these challenges, though their terse pronouncements could benefit from more in-depth discussion. When information about exposures and potential confounders is incomplete, they state, "epidemiological studies then provide a statistical ‘association’... rather than a clear-cut causal link" (1, p. 47). The authors also advocate, with little explanation, imposing a relative risk threshold of 2 before claiming the existence of a causal association.
In the second half of the book, the authors focus on risk perception and risk management. They discuss psychological studies of risk perception, explain how people rank risks on factors other than their frequency, and reveal several common misconceptions about risks, under the guise that the risk manager must understand how the public thinks about risk. The chapter on risk management surveys a variety of approaches, from product bans to cost-risk-benefit analysis to the use of the best available control technologies. A theme running throughout the chapter is that many risk management strategies harbor inconsistencies, and "it is the duty of a risk analyst to point out inconsistencies so that if they are retained they are retained deliberately" (1, p. 171).
One such apparent inconsistency is the regulation of chloroform versus trichloroethylene in drinking water. Animal bioassays suggest that chloroform is more highly carcinogenic, yet regulatory limits for trichloroethylene are more stringent. Why? Chloroform is intentionally added to water supplies to kill bacteria, while trichloroethylene is present only as a result of accidental pollution, and this difference has significance for risk management decisions. Clearly, risk management involves more than mere quantification of risk. However, the authors insist that quantification of both risks and benefits, imperfect as it is, will make explicit the nonscientific values and assumptions that often underlie risk management choices. John Graham, author of the book’s preface, would probably agree. As administrator of the Office of Information and Regulatory Affairs in the US Office of Management and Budget, Graham has promised to increase the explicit use of cost-benefit analysis in reviews of federal regulatory policies.
To aid readers in putting risks into perspective, the authors have devoted the last 70 pages of the book to tables and graphs comparing a variety of risks and their costs. Chronic worriers may be interested in knowing that the risk of being killed by a falling aircraft is about half that of being killed by a tropical hurricane. Quantification has limits, however, which the authors do not always acknowledge. For example, they fault advocates of the Precautionary Principle for failing to assign a number representing the degree to which we value prudent action. While the Principle itself is vaguely stated and vulnerable to abuse (it simply requires that we act to prevent harm even when evidence is limited), it is not clear how assigning a numerical value to our desire for caution will lead to better risk management decisions.
Despite these difficulties, quantitative risk assessment clearly has an essential role in public policy. This provocative and readable volume provides an excellent introduction to the subject.
REFERENCES
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||