Exposure of the American People to Iodine-131 from Nevada Nuclear-Bomb Tests: Review of the National Cancer Institute Report and Public Health Implications (1999)

The U.S. Department of Health and Human Services (DHHS) asked the Institute of Medicine and the National Research Council to provide information that would help the department educate and inform the public and health professionals about the health implications of exposure to I-131 from the nuclear-weapons testing program in Nevada in the 1950s and 1960s. Chapter 4 discussed approaches to patient and clinician education and included sample information statements that might serve as a starting point for DHHS work to develop, test, and disseminate this kind of information. This chapter provides a broader framework for DHHS to consider as it develops its information strategy.

Communicating effectively with the general public about the health risks posed by exposure to I-131 fallout from aboveground nuclear tests some 40 years ago presents a difficult challenge to DHHS for several reasons:

• The aboveground nuclear tests in Nevada were purposive, man-made phenomena that left behind toxic residue. Since the tests ended, governments and residents of areas adjacent to the test sites have engaged in intermittent, often acrimonious, debate about possible health effects and about the release of information about the tests. The legacy is a government with a record of poor credibility as an information provider, and a subset of the population convinced that the health consequences of the tests are significant and severe.

• The best scientific estimates of exposure to radioactive iodine and of developing radiation-related thyroid cancer or other thyroid problems are burdened by significant uncertainties. These uncertainties must be explicitly considered in any effort to estimate the likelihood that a specific individual will be diagnosed with thyroid cancer or that a particular diagnosed cancer actually stems from radioactive fallout.

• Although a diagnosis of thyroid cancer must be treated seriously, it is an uncommon cancer and is rarely life threatening. Most individuals exposed to I-131 fallout from the Nevada weapons tests—even those few whose doses approached or exceeded 1 gray (Gy, 100 rad) are unlikely to develop exposure-related thyroid problems. Communicating small probabilities and changes in small probabilities may be difficult.

• A widespread attempt to alert the public to the possible health consequences of exposure to I-131 fallout and to promote systematic screening for thyroid cancer might not lead to entirely benign outcomes. No evidence shows that screening for thyroid cancer is effective in improving survival for this highly survivable disease. Screening tests are imperfect and can cause harm in the form of unnecessary surgeries and other procedures, anxiety, insurability problems, and other problems. The quality of the tests also depends on the skill of the screeners, which can vary widely.

Despite the complexity of the topic, the uncertainty of estimates of exposure and of probabilities of developing cancer, and public questions about government credibility, DHHS must devise ways to communicate accurately, credibly, and effectively about its 1997 (NCI 1997a) report (taking into account sound criticisms of its methods, conclusions, and presentation). Media and other attention to the report and to the fallout issue more generally will undoubtedly attract interest and concern and lead some people to want to learn more about their own potential chance of developing health problems. The committee believes that DHHS must accept responsibility for helping people understand the possible relevance of the NCI report to their own circumstances but recognizes that the limits of available data and methods will make this difficult.

The following sections of this chapter discuss characteristics and principles of risk communication that emphasize how people construct their own judgments of risk; the importance of source credibility in those judgments; the probability that the audience for risk information will be heterogeneous rather than homogeneous, thus creating the need for a variety of different information efforts; and the need to promote involvement rather than exclusion of the public in the risk communication process. The chapter also explores some specific communication strategies aimed at both groups and individuals. The resources and effort expended on risk communication should be proportional to the potential for harm and the likelihood that the risk communication will be successful.

As is the case with most risks, communicating effectively possible health hazards stemming from exposure to I-131 fallout from the Nevada nuclear-weapons tests will be complicated. Most risk communication efforts fail because communicators believe the process is relatively simple: ''educate the public" (Liu

and Smith 1990; Rogers and Storey 1987; Hyman and Sheatsley 1974). Informing the public is a useful goal, but there are contingencies that, if not accommodated, can frustrate even the most earnest effort.

The success of any message will depend on considering a variety of factors in designing an information campaign. Several of the important contingent conditions are listed here, both to illustrate what is available in the literature and to establish an interpretive framework that can inform DHHS public-education efforts. Success may be very generally defined as communication that is viewed as credible by intended audiences, that allows these audiences to accurately interpret factual information (e.g., probability statements), and that does not provoke unintended reactions (e.g., serenity when concern is warranted or vice versa).

One mistake made by risk communicators is to assume that the audience is an empty vessel waiting to be filled with the communicator's interpretations of reality. The assumption of audience passivity is widespread, and it has found expression in such theoretical positions as the following:

• The powerful media hypothesis, sometimes called the hypodermic model, predicts that people generally are extremely vulnerable to media influence. It is the basis of most propaganda efforts and served as a catalyst for much of the communication research during World War II. Today it is resurrected whenever a new communication channel opens. The cyclical phenomenon is seen by looking back in time at public and policy reactions to the movies, comic books, radio, television, and now the Internet (Lowery and DeFleur 1995; Reeves and Hawkins 1986; Wartella and Reeves 1985).

• The third-person effect hypothesis states that when individuals encounter information that is potentially alarming, they suspect the information will strongly influence everyone but themselves. We believe ourselves capable of evaluating messages, but we do not ascribe the same skill to our neighbors (Perloff 1993; Davison 1983).

Audiences are far from passive and failure to recognize this can sabotage effective communication. People are more likely to bring interpretations to a message than they are to glean interpretations from a message (Zaller 1992; Derwin 1981). For example, Stevenson and Greene (1980) studied individuals' judgments about possible bias in news stories about presidential candidates and found that "people see as biased news information which is discrepant with the cognitions they already hold about the situation described in the news story" (p. 119). Put another way, a judgment of bias told the researchers more about the views of the individual doing the judging than it did about the content of a news story.

Communicators must learn that the audience participates in the construction of meaning. People bring complex sets of cognitions to any issue, including risks. If they attend to the new information, they will likely incorporate it in some way into their understanding of those issues. But that effort can change the meaning of the message more often than it will change the understanding of the issue at hand.

The idea that audiences will be active in constructing understanding of an issue presents a challenge to information campaigners who wish to influence attitudes and behavior. With respect to I-131 exposure, DHHS must remain sensitive to this audience role and, like all information providers, it must make strategic decisions about how much of an impact it wishes to have and with whom. There are ways to work productively with active groups; some of them are discussed below.

One task for DHHS is to learn more about how members of the general public and more interested groups and advocates think about exposure to I-131 and the risk of developing thyroid cancer and to identify how they filter information through different experiences and conceptual frameworks. Ongoing evaluation of its communication efforts should allow DHHS both to learn more about public thinking and to assess how well its communication program is doing.

As noted above, the resources and effort expended on risk communication should be proportional to the concern about the risk as well as to the likelihood that the risk communication will be successful. It might be appropriate in some instances, therefore, to use less resource-intensive risk communication strategies. Whatever strategy is pursued, it must be remembered that people who encounter DHHS information about exposure to I-131 will predictably incorporate the messages in ways that are consistent with their personal understanding of the Nevada tests and their fallout. If the DHHS message is inconsistent with that understanding, the message will suffer.

There is no general audience for risk communication. Rather, any group of individuals will probably display a wide, often contradictory, array of meanings regarding a particular issue. A host of variables account for this variance including that people will differ in their knowledge about the risk at hand, in their motivation to learn more, in their capacity to learn more, and in the complexity of the belief systems they have constructed to make sense of threats generally.

When there is a broad audience, it is difficult for communicators to judge what individuals already know about a particular hazard and its potential consequences. Studies of risk perception find a strong correlation between the salience

of a risk and the amount of knowledge. During the 1980s, for instance, AIDS often topped the list of important problems for the American people, and respondents in surveys generally demonstrated reasonably high levels of knowledge about the risk (Fisher and Fisher 1992; Becker and Joseph 1988). Similarly, the summer drought of 1988 put global warming on the personal agendas of many individuals, likely leading to increased knowledge about that phenomenon and its implications (Trumbo 1995). It would seem, then, that for any risk, some number of individuals will have learned a great deal. Those persons also will tend to have established strong beliefs about that risk.

Simply knowing little about a risk does not ensure that someone will work hard to learn more. People vary dramatically in their motivation to learn, but one strong incentive to learn is personal experience with a problem. Knowing someone with AIDS, for example, has long been a good predictor of information-seeking about that disease (Becker and Joseph 1988). The Centers for Disease Control and Prevention's Cancer Information Service offers another example. It experiences an upsurge in the number of telephone calls soon after news about a particular cancer is disseminated in the mass media (Freimuth 1998).

Sometimes, however, even highly salient conditions will not produce an effort to learn. We all make choices about what we do, and one option is to do nothing. For example, Wynne (1991) was at first surprised to find that apprentice workers at the Sellafield nuclear power plant in Great Britain knew little about basic radioactive processes and felt little need to know more. He and his colleagues subsequently learned that those workers had put their cognitive energies into learning organizational procedures, not science; instead, they placed their trust in the institution to protect them from harm. Dependency on their jobs may also have encouraged disregard.

At times, even though motivation exists, an individual cannot summon the resources to learn. Socioeconomic status confers on some individuals an information-rich environment and on others an information-poor one. Sociologists (Tichenor and others 1980) hypothesize that structural differences can create a knowledge gap even in a society brimming with information. Given a societal need to know and the routine availability of information, they argue that people with greater resources will become increasingly knowledgeable while those with few or no resources will show little appreciable change for the better. Over time, a knowledge gap forms and widens.

Wynne (1991) reinforces this notion of structural limits to learning in his research on public understanding of a variety of risky situations in Great Britain. Specifically, "an important discovery from our research has been the enormous

amount of sheer effort needed for members of the public to monitor sources of scientific information, judge between them, keep up with shifting scientific understandings, distinguish consensus from isolated scientific opinion, and decide how expert knowledge needs qualifying for use in their particular situation" (Wynne 1991, p. 117).

Confronted with a possible risk, people vary widely in the sophistication of their cognitive processes for making sense of risk. As in many other situations, people tend to employ judgmental shortcuts and personal theories of how the world works to decide what to make of a risk (Freudenberg 1992; Gilovich 1991; Kahneman and Tversky 1979). Those theories can be sophisticated or they can be naive. For example, some people have notions of causality that make chance an acceptable explanation for a cancer cluster; for others, assumptions that all effects have causes will preclude chance as a factor. As noted in Chapter 4, even basic literacy and numeracy must be recognized as challenges.

Creating appropriate risk messages requires recognition of the heterogeneous nature of the audience and of the importance of risk salience, information resources, and other factors that inform people's responses to information. Information providers can profitably use survey research, focus groups, and strategic informants (including citizen advisory groups made up of representatives of a variety of concerned groups) to shape their understanding of these variables. Although the participation of concerned groups is important, representatives of the general public should also be included to ensure that the larger community feels neither ambiguous nor hostile to the communications and that its perceptions of risks related to I-131 exposure are understood.

Because people may vary in language and numerical skills or comfort levels, DHHS may want to develop materials that vary in scope and depth or that unfold in ways that allow people access to information at the level that is comfortable for them. On the Internet, for example, documents that are relatively short and simple can include links to more detailed information. Unfortunately, as discussed below, much remains to be learned about the Internet as a medium of communication. Also, those who might benefit most from this feature of the Internet may be among the income and educational groups that have the least access to the Internet. This is one reason why the Internet should not be the major communication tool.

Most message designers try to change beliefs and behavior. The strongest effect of a message, however, can be to reinforce the status quo.

A typical example occurred some years ago when the government of Taiwan engaged in a national risk communication campaign to persuade residents that the construction of its fourth nuclear power plant was both safe and in the public's best interest. Underlying the campaign was the assumption that, as people learned more about nuclear power and the country's energy plans, they would develop beliefs in keeping with the campaign's goals. The idea backfired: Although many people believed they had learned more about the risks of nuclear power, they used that knowledge to strengthen their existing views. Residents who supported nuclear power before the campaign tended by its end to be equally or more enthusiastic about the technology; those who were skeptical about nuclear power at the start ended more strongly against the proposal (Liu and Smith 1990).

Of course, this reinforcing pattern works only when a person brings a belief system to the issue. When a risk is not well known or understood, information campaigns can have a substantial influence on the construction of knowledge and belief systems, and sometimes that effect will be consonant with the goals of the information provider. Even information touting the benefits of a technology, substance, or process over its risks can produce avoidance or hostility (Lopes 1987; Kahneman and Tversky 1979). Sociologist Allan Mazur (Mazur 1981) has examined this process and finds, for example, that "balanced" media accounts—which discuss the benefits and the harms of a process or technology—produce predominately negative reactions among people who are unfamiliar with the process or technology in question. And the more coverage provided by the media, the more negative public attitudes become (Mazur 1981).

Traditional informational messages about the health effects of aboveground testing will serve to reinforce existing beliefs, whatever they are, among individuals who already have complex and enduring beliefs about those risks. Effective communication with these groups may require more intensive work such as community meetings, establishment of advisory groups, and working with other information sources that may be more credible to those with strong existing beliefs. Other people will perceive they know little and, if they are motivated to learn, will be much more open to DHHS's attempts to frame the issue for them. News releases, Internet communications, and other more traditional approaches may work well with this audience.

One finding from the literature is that individual risk judgments are formed by many factors. This contradicts the common supposition that an informed risk judgment is one that achieves a tight fit between the likelihood of coming to harm and individual willingness to take preventive action. As many researchers have demonstrated, the possibility of harm does contribute judgment, but its relative

weight depends on a host of other factors, among them assessments of a risk's controllability, its magnitude, the extent to which a hazard might harm future generations, whether or not it is voluntarily assumed, and its ability to harm non-human organisms (Slovic 1992; Slovic 1987; Johnson and Tversky 1984).

Some risks seem inherently unnerving; others do not. This fact has led scholars to argue on behalf of a cultural basis for perceptual differences. That is, a particular interpretive framework can be nurtured among groups of people such that the framework defines some risks as more terrifying and less palatable than others regardless of the risks' other attributes (Douglas and Wildavsky 1982). Kasperson and colleagues (1992) articulate a process, called the social amplification of risk, to explain this. In their view, "events pertaining to hazards interact with psychological, social, institutional, and cultural processes in ways that can heighten or attenuate perceptions of risk and shape risk behavior" (pp. 157-158).

Public perceptions of nuclear power likely have been shaped by this complex amplification process. For example Slovic and colleagues find the term "nuclear" is a strongly aversive term for Americans, evoking feelings of peril and ugliness (Slovic and others 1991). Weart (1988) contends that these nuclear fears are rooted in our social and cultural consciousness, and Peters and Slovic (1991) offer evidence that attitudes toward the world and its social organization (called "world views" by scholars who study the phenomenon) serve as "orienting dispositions" that guide individuals' responses to all things nuclear. Regardless of cause, Americans' affective response to nuclear technology is so intensely negative, say these scholars, that it overwhelms any positive affective response. The "nuclear connection'' may have a strong negative impact on individuals' thinking about the health risks of I-131 independent of health issues themselves.

Deciding what to emphasize in messages about risk depends on intelligent assessments of what dimensions of a risk matter to the audience. The aboveground nuclear-weapons tests will provoke cultural reactions as well as disparate individual responses. DHHS may need to focus on more than I-131 exposure and probabilities of developing thyroid cancer and to acknowledge that radiation is not like most risks because it is not observable and exposure is not voluntary or often even known by those exposed.

As individuals, when we think about risks, we make important distinctions between our personal risk of coming to harm and the likelihood that other people will be harmed. Specifically, we view ourselves as less vulnerable; everyone else is at greater risk than are we.

The distinction between "me" and "them" pervades life—not just risk estimates. So it makes sense to explore the possibility that we all use information

differently to inform our understanding of "me" and "them." Such a distinction is particularly important to the success of information campaigns, where the goal is typically influencing individual perceptions of personal situation, often to the extent of inducing a change in behavior.

Numerous studies find that we seem to differentiate between channels of information that we find appropriate to informing our understanding of "them" and those acceptable for an understanding of "me." Specifically, we value mediated information channels—newspapers, television, magazines, radio—as sources of information about other people while we resist the relevance of the information channel to our own understanding of our situations. Indeed, we rely heavily on interpersonal channels—people we trust—when we need to make personal choices.

This tendency is called the impersonal impact hypothesis (Mutz 1992; Tyler and Cook 1984) and it seems to be pervasive. Illustrative of findings in this arena are those that examine influences on individual perceptions of victimization by crime. Such studies typically find that people use media accounts of crime to inform their understanding of the prevalence and threat of crime in society generally but that they do not interpret those stories as telling them anything about their likelihood of personally being victimized (Tyler 1984; Tyler 1980).

Similarly, we prefer to use media accounts of risk to inform our general, societal understanding of a risk, but rarely do we employ these accounts to inform our personal risk situations. Put another way, the primary effect of media accounts of risk seems to be knowledge gain (see, for example, Schooler and others 1998). But when it comes to making a judgment about personal risk, media accounts are insufficient sources of information. Instead, we need to talk to someone (Dunwoody and Neuwirth 1991).

It is important to distinguish between individual personal risk judgments and perceptions of the risk to others; some channels of information will be more appropriate for one than for the other. If DHHS is trying to communicate information about the general risk stemming from the Nevada tests, mediated channels are fine. There is some evidence that individuals who attend to those messages will learn from them. But once people begin to seek information to inform their personal situations, other channels, interpersonal ones, must some into play.

People make the acceptability of information a function of the perceived trustworthiness of the source. Simply put, untrustworthy sources will not be believed, regardless of the quality of evidence they present about an issue.

Freudenburg (1992;1988) attributes this heavy reliance on trustworthiness as a heuristic device to "recreancy" and sees it as a normal by-product of an increasingly

complex society. Formally defined, a recreant is someone who fails to do his or her duty or to be faithful to his or her obligations. We have become heavily dependent on specialists to make society run, argues Freudenburg. Because the typical citizen cannot evaluate the detailed work of specialists, he or she must resort to making judgments about the ability of the specialists to do their duties carefully (Freudenberg 1992; 1988). And all it takes to generate a diagnosis of untrustworthiness within a complex system, says Freudenburg, is one mistake, one display of carelessness or ignorance. Because the typical American cannot know a system intimately enough to interpret errors as more or less serious, he or she instead defines any error as proof of recreancy, and credibility suffers. Freudenburg argues that a public perception of recreancy explains more of the variability in public concern with such hazards as nuclear power and nuclear waste than do such characteristics as sex and socioeconomic status.

Other scholars, among them Kasperson and others (1992) and Peters and others (1997), offer sets of underlying dimensions for trust judgments that include perceptions of a source's competence, objectivity, fairness, and compassion.

Typically, we assume that trustworthiness judgments are directed toward the individuals or organizations who are sources of information. That might not always be so. People sometimes rely on judgments of the credibility of information channels rather than of information sources. The choice can depend on the complexity of the belief system that a person brings to the message.

If someone knows little about a risk it suggests that the risk is not highly salient (at least at the moment). When someone has less at stake from an information perspective then he or she will invest less effort in making a credibility judgment about available information. One way to invest less effort is to employ judgmental shortcuts that focus on the credibility and trustworthiness of information channels, not information sources. That is, rather than enter into the labor-intensive task of picking apart a message to judge the credibility of a particular source, an individual will make a broader judgment about the credibility of the channel. In this context, a source is the originator of information and a channel conveys the information: A newspaper story (a channel) carries information from a government official (a source).

Like all heuristic devices, this default to channel is efficient. Most of our information about risks comes to us through some mediated channel. And a channel judgment often is grounded in culturally negotiated notions of channel quality; for example, most people will regard NBC television as a more credible channel than the National Enquirer.

People who have enduring belief systems will already have defined the risk in question as highly salient, will probably be willing to engage in more systematic information processing when they see messages about it, and thus will make much finer credibility distinctions among the sources than solely among the channels. They will not simply judge a risk on the basis of the channel itself but will instead attempt to learn who is speaking with authority and to decide whether the authority is warranted.

Information sources bring their own cultural baggage along. Some recent research indicates that general cultural expectations of sources influence our perceptions of credibility. For example, Peters and others (1997) found that the mix of underlying predictors of trust and credibility varied by type of source: The two strongest predictors of the credibility of government sources were the perceived commitment to a goal (such as public health) and knowledge and expertise regarding the risk. The two strongest predictors of trust in industry sources were the perceptions that the industry was concerned about the risk and that the industry sources tried to provide full information quickly.

Evaluating the role of trust and credibility is made even more complex because of the possibility that credibility judgments are at least partially contingent on the source's area of expertise. In a survey of perceptions about a variety of risks among residents of 5 European countries, Jungermann and others (1996) found that those surveyed clearly regarded sources that are critical of industry (environmental groups, for example) as the most trustworthy. But when they were asked to match specific types of risk information with the ideal source for that information, respondents indicated that even low-credibility sources could provide some trustworthy information. For example, although they were critical of industry sources, survey respondents regarded industry as the best source of information about the characteristics of their own products. Government sources, another category with credibility problems in this survey, were regarded as the best sources of information about relevant environmental regulations.

If a source or channel is not trusted, it is almost useless as an information provider. And in such long-running and volatile issues as the aboveground Nevada tests, credibility is important in public acceptance of information. In lieu of changing perceptions of credibility—a decades-long task at best—it might be useful to separate types of information such that a specific type of information is matched with an information provider who will be seen as credible in that instance (see Jungermann and others 1996). For example, a federal agency, such as NCI, might be defined by the public as a highly credible source of information about how to calculate the risk of developing thyroid cancer from exposure to I-131. But NCI's expertise might not give it legitimacy as a source of information

about how concerned any individual should be about that risk. By involving members of the public in developing and testing information, DHHS may also identify other, more credible information sources (e.g., some state health departments) and encourage their involvement in providing information.

Risk managers now understand that communicating successfully with a set of constituents means involving those constituents in the process much earlier than was done previously. Risk managers of old viewed communication as a one-way, temporal educational process in which experts pronounced and the audience listened and learned. An unsuccessful risk communication effort was blamed on audience failure to assimilate the information and act appropriately.

That unilateral approach is slowly being replaced by a multilateral one as agencies with a communication mission work to involve the audience in planning and execution. Serving as an important catalyst for this change was the National Research Council report Improving Risk Communication (National Research Council 1989), which declared "risk communication should be a two-way street," between experts and various groups, that should exhibit a spirit of open exchange in a common undertaking rather than a series of "canned" briefings restricted to technical ''nonemotional" issues and an "early and sustained interchange that includes the media and other message intermediaries" (p. 10).

In response, many state and federal agencies are beginning to assemble citizen advisory groups and to meet and seek the counsel of the public in the course of designing risk messages (see, for example, Boiko and others 1996). There are good reasons for this. First, individuals who have developed complex belief systems about a risk situation tend to be highly knowledgeable about the risk and often have developed insight about the risk perceptions of various interested groups. They can thus bring useful information to the table. Second, involving citizens early in a risk communication process enhances interpersonal communication of the type most likely to influence the perceptions of even the most hostile individual. Numerous efforts are under way to engage citizens in interpersonal dialogue about issues to enhance decisionmaking. The most visible of these recent efforts was the 1996 National Issues Convention. Before the 1996 presidential election, a random sample of voting-age adults assembled in Austin, Texas, to discuss issues with policy makers and with one another. The object was to encourage people to deliberate and negotiate with others, and subsequent surveys of the participants indicated not only that they learned a great deal about the issues under discussion, but also that many of them did indeed change their understanding of the issues. An added benefit was an enhanced sense of political efficacy; these persons now believed their opinions mattered (Fishkin 1996).

As judged by the degree of community involvement, perhaps one of the better current risk communication templates is the Hanford Health Information Network,

sponsored by the state health departments of Washington, Oregon, and Idaho. Like the Nevada tests, the federal Hanford Nuclear Reservation is the subject of lengthy and continuing public concern about the health implications of radioactive contamination. However, the Hanford situation is sufficiently different from the Nevada Test Site (e.g., range of hazardous materials and exposure routes involved, time period, geographic area involved), that the various elements of this model and the kinds of information provided would need careful examination.

If DHHS deems this particular risk worthy of a sustained campaign to inform not just the general public but also various interested groups, it will be important for the department to design information structures that bring the public into the process early. The "stakeholders" are the people who have developed complex belief systems about the risks posed by the Nevada tests or similar events and those who are most at risk, regardless of their a priori knowledge and beliefs.

A simple, albeit ambiguous, answer to this question is "Yes." The difficult part is making educated guesses about how to accomplish the goal. There has been some empirical exploration of how to communicate about risk with nonspecialists—studies of risk comparisons, of explanatory devices, of appropriate information channels—but the work is relatively sparse and one must generalize with care. Equally challenging for risk communicators is the task of selecting the interpretive context within which facts will be presented. That is, should one talk about an individual's immediate or long-term prospects? Should one talk about the risk to the average person or instead emphasize contingent conditions that would make some people much more susceptible than others?

The remainder of this section briefly discusses elements of the interpretive context, offering a list of domains that DHHS should consider as it decides on the content of its messages. The use of specific communication strategies is discussed, including risk comparisons, the question of whether the World Wide Web can contribute, and a means for people to estimate personal risk from exposure to I-131 during the Nevada tests. Finally, a recommendation is made for how to accomplish communication planning by bringing the affected public into the process.

In the course of studying the health risks created by the Nevada atomic weapons tests, this committee has come to several conclusions that need to be reflected in the communication strategy developed by DHHS.

• Being at increased risk of developing thyroid cancer has little influence on individual mortality. This is true for two reasons: Thyroid cancer is rare, and long-term survival after diagnosis is high. No evidence suggests screening would improve these already high rates. It will be important to account for contingent conditions when trying to evaluate an individual's future lifetime risk of being diagnosed with thyroid cancer. Such evaluations should emphasize that increased probability of developing thyroid cancer caused by I-131 exposure is restricted to a relatively narrow birth cohort and that, even for members of that cohort, future risk of developing thyroid cancer is most likely less than a few percent. For the public, explanations of probabilities of developing or dying of thyroid cancer related to I-131 exposure should be provided in both qualitative and quantitative terms as described in Chapter 4 and explanations should include baseline risk (e.g., risk of naturally occurring thyroid cancer).

• Risk estimates will come bearing high levels of uncertainty that will complicate efforts to explain probabilities of developing thyroid cancer related to I-131 exposure. Although NCI (1997a) has estimated as much as a two-thirds increase in risk for developing thyroid cancer for selected birth cohorts (people born between 1948 and 1952, for example), this estimate of increased risk is accompanied by a wide range of uncertainty. Moreover, available epidemiologic evidence so far provides scant support for the higher estimates of excess cancer cases.

As suggested in Chapter 3, if individual exposure to I-131 from fallout can be accurately assessed, then for a diagnosed case of thyroid cancer, the probability that the cancer was caused by fallout can be assessed.

One obvious way to place a risk in context is to compare it with a second, third, or fourth cause of harm. For example, Chapter 3 compares thyroid cancer incidence and mortality with comparable statistics for other cancers.

When scientists and risk managers attempt to put risks in context by making comparisons, their audiences may derive meanings from them that are quite different from what was intended. For example, when one scientist equated the likelihood of coming to harm from eating contaminated fish from the Great Lakes to the risk of getting cancer from exposure to chemicals found in the food, drinking-water, and air of some of America's more polluted cities, his attempts were interpreted as venal efforts to downplay the risks of eating fish (Dunwoody and Peters 1992).

Although comparison is an excellent explanatory technique, it works only if it allows the user to connect appropriate dimensions, such as the quality of life,

associated with different health outcomes. A scientist will offer a risk comparison with the intention that the user will compare the likelihood of one cause or another leading to harm. But a lay user of a comparison might attempt to make comparisons between other dimensions of the risk, for example, whether the risk is voluntary or involuntary. Most people consider multiple dimensions when making risk judgments, and therein lies the problem. In the example above, the scientist intended the comparison to illustrate the magnitude of various risks that are commonly encountered and accepted. But the audience inferred that the scientist was comparing dissimilar values. If a source's credibility is under attack, risk comparisons are problematic. When a source that is deemed untrustworthy offers comparisons, those comparisons can be interpreted as attempts to persuade rather than to inform.

Freudenburg and Rursch (1994), who have conducted one of the few empirical explorations of the value of making risk comparisons, add that the use of comparison can be successful explanatory strategy only when the comparison is by a source that is trusted by the audience. The authors note, "Essentially, all risk comparisons—even those that are widely seen as acceptable risk comparisons—involve the provision of information by proponents and officials. If those proponents and officials are not trusted, then even 'legitimate' risk comparisons can do more to arouse suspicions than to assuage them" (p. 954).

A comparison of estimates of harm works only if the audience interprets the comparison as intended by the communicator. One way to achieve that is to identify risks for comparison that are similar in areas that are important to risk judgment, for example, the extent to which the risks are voluntary and the extent to which they are familiar. Such a strategy could, however, narrow comparisons substantially and make it difficult for people to put a risk in context. For radiation-related risks, identifying even this narrower set of appropriate comparisons is difficult.

Thus, DHHS may want to test how different kinds of comparisons involving radiation-related risk are perceived by different audiences. A later section of this chapter returns to this issue.

Most of us try to explain complicated things by a largely intuitive process. We employ simple words and short sentences. We use active voice verbs, analogy, and metaphor. But sometimes this is not enough. Scholars in science education have become specialists in explaining science to the public; they help educators and communicators develop a keen focus on how audiences interpret information. For example, Rowan (1990) suggests that even the most carefully

selected analogy or definition will fail if it collides with an incompatible belief held by the audience member. Beliefs inconsistent with theory are rife in daily life. If an individual happens to believe that a co-occurrence of apparently highly unlikely incidents—for example, the diagnosis of the same rare illness in three players on a football team—simply cannot occur by chance, then even the most carefully articulated explanation of the role of chance in such an event will not be credible. Rowan (1990) notes that an explainer must first refute the "naive theory," in this case by helping someone understand that the occurrence of a rare illness in three proximate individuals is not nearly as unlikely as originally supposed. Only then can the explainer safely offer the new theory.

In addition to naive theories, audience members can come to a risk message with only the barest numeracy skills, making it difficult for them to interpret quantitative representations of risk. In a study of adult women in New England, for example, Schwartz and others (1997) found that numeracy skills were strongly predictive of the ability of women to accurately interpret information about mammography and risk reduction.

Explaining scientific concepts and processes and representing risk numerically requires an understanding of what the audience knows or believes to be true. If numeracy skills are low or if individuals have accepted competing naive theories the risk communicator must tailor the message accordingly, and consider the possibility and consequences of communication failure.

The World Wide Web's popularity has led many to predict an explosion in public reliance that will eventually exceed that for today's traditional information channels, with the possible exception of television (CommerceNet 1997; Worthlin Worldwide 1996). Indeed, although the Web is a fundamentally new kind of information channel that could be ideal for communicating some information, it is important to be cautious about using it as a primary channel of information for at least two reasons.

First, it is not yet a mainstream channel. Although use of the Web is spreading among the American population faster than did the telephone or television, it is still available mainly to people who have regular access. This excludes a large portion of the public. The emergence of technologies such as the Web TV will increase the rate of use, but for the near future access will be available only to a minority of the U.S. public (Hoffman and Novak 1998). An information manager who is responsible for communicating with a wide audience is not well served by employing only the Web alone.

Second, we are just beginning to explore the role of the Web as a public-information channel. Research on hypermedia sites exists but has looked primarily

at use of such channels for formal education purposes. We have no idea how the Web will be used as an informal channel, but it would be fair to expect that such uses would be very different from those in the classroom. Is the Web a mediated channel or an interpersonal one? If it emphasizes nonlinear communication—by urging the user to dart from one place to the next—what is the nature of the learning that takes place? How do users make reasonable judgments about the trustworthiness of the information presented?

Although research on how people use the Web as an information channel is just beginning, one possibility, admittedly optimistic, is that, for risk communication purposes, the Web combines the best attributes of mediated and interpersonal channels. As a mediated channel, it can reach millions of people inexpensively. And its interactive qualities can encourage users to interpret a site's contents as telling them something about themselves personally—a conclusion usually reserved for interpersonal channels.

If users are willing to extrapolate site content to their personal situations—something most people are extremely unwilling to do with information encountered in the mass media—then the Web could become an important means through which to teach users about a risk and help them make personal risk judgments.

The National Cancer Institute seems to be actively considering this possibility. Its own Web site already contains a section that allows a user, given the relevant background and behavioral information, to calculate his or her own likely dose of I-131 from Nevada Test Site fallout. The section is not very accessible, however, and it could frustrate users more than it helps them.

DHHS should actively and creatively use its Web site but not regard a Web site as its primary communication tool. The Web should supplement other information channels, but it cannot replace them. The committee has found no empirical evidence to support the idea that calculating one's personal risk from information and a formula available on the NCI site will be helpful to members of the general public. But a combination of careful information choices and thoughtful design might make such a Web site worth trying and evaluating.

Many scholars, and an increasing number of risk managers, encourage a differentiation between technical and democratic approaches to risk communication. As described by Rowan (1994), the technical perspective is characterized by "the use of physical and social sciences to describe and predict health effects posed by environmental hazards" (p. 392). Such risk assessments construct a picture of some hazard's health effects and are used to set safety standards. People who subscribe to the technical approach, says Rowan, put their faith in scientific

expertise. Those who subscribe to the democratic approach, on the other hand, are concerned about the political fairness of the assessment or risk. Notes Rowan (1994), "The democratic view is that who decides whether some risk should be incurred and who benefits or is harmed are just as important, if not more, than the severity and likelihood of the risk" (p. 396).

Communication strategies are different for the two approaches. To facilitate a technical view, one would use an expert-to-novice flow of information. The democratic approach requires that all parties affected by a risk are guaranteed participation and power in decisionmaking. Each strategy has limitations. For example, the democratic approach cannot be implemented retrospectively, and thus, in the current instance, this strategy will be heavily burdened by past decisions relative to weapons testing and the resulting exposure of the public to I-131 and to other fallout radionuclides. The democratic approach now can be used only to guarantee that all affected parties can participate and affect future decisionmaking.

The issue of health effects caused by I-131 fallout from the Nevada tests is one for which technical and democratic perspectives are both present and, at times, seem to collide. DHHS can choose to communicate with the public only with the technical approach. But the department must realize that such a choice will be quite unsatisfactory for some subgroups of the public. Communicating effectively with these often highly charged subgroups may well be expensive and time-consuming. If DHHS wishes to promote a public perception—particularly among these subgroups—of the government as trustworthy, it will be a challenging undertaking. To that end, this committee provides a set of recommendations for how to achieve a communications arena in which all parties may participate.

Effective communication by a government agency about public exposure to radiation from federal facilities must account for the public perception of radiation as a health threat and the history of less-than-forthright communication about such exposures. To establish its credibility as a source of information, DHHS will need to engage interested groups—affected members of the public and the health professionals who serve them—as partners in the development and dissemination of information materials. Because interest in the health effects of I-131 exposure may continue for some time, attention should be paid to maintaining and improving—not just initially creating—the infrastructure to communicate information to the interested parties. DHHS should consider the following process of developing and communicating information about the health effects of exposure to fallout from nuclear weapons testing at the Nevada Test Site. DHHS will also need to consider the effectiveness and cost of different approaches as it develops its strategy.

• Development of Public Information on Radiation Health Effects.

To promote the provision of consistent public information about exposure to I-131 from the Nevada weapons tests, DHHS should consider creating a national resource center to help inform people concerned about their individual risk from such exposure. Ideally, the center would be under the guidance of an organizational unit familiar with radiation health effects and with communicating such information to the public, for example, the Centers for Disease Control and Prevention. If such a resource center were created, DHHS should also consider establishing a council to provide guidance on the center's communication methods and materials. The council should consist of individuals selected to represent the concerned public, including the "downwinder" organizations and Native Americans groups; physicians; public-health authorities; and scientists knowledgeable about dose reconstruction, risk assessment, epidemiology, and risk communication. Before they are published, materials should be reviewed by a panel of technical and lay advisors.

Public communication materials should be developed to include at least the following items:

• a description of the NCI (1997a) report and its limitations;

• a description of the report's focus on I-131 rather than on other isotopes;

• a description and explanation of the factors that place an individual at risk, e.g., age at exposure, sex, source and volume of milk consumed;

• a method to help people place their risk in perspective, again with adequate explanation of the uncertainties in individual estimates of risk;

• a description of the possible health effects and their likelihood of occurrence;

• a description of the limitations and risks of harm associated with screening for the health effects of radiation exposure; and

• recommendations about factors people should consider in deciding whether to seek further information or advice.

The addendum to this chapter includes for DHHS consideration an example of a method people could use to assess their exposure to iodine-131 and their risk of developing thyroid cancer and to help them decide whether they wanted to seek further information and advice. The specific county classifications, risk categories, and multipliers provided for in the assessment method would need to be chosen carefully and validated appropriately. The method should provide qualitative as well as quantitative information and should include information on baseline as well as relative risk. The inherent limitations of individual dose estimates should be clearly explained. The general approach illustrated in the addendum would replace the unwieldly method for individual dose assessment currently included on the NCI's Web site for the I-131 report.

• Development of Information for Health Care Providers.

Of equal importance to public information is the development of information for health care professionals. Chapter 4 discussed the kinds of information needed by physicians and presented a sample information sheet that DHHS might use as a starting point. Consistent with the suggestions above regarding public communication strategies, it recommended that professional groups and health care organizations be involved in developing, testing, and disseminating information materials through a variety of channels including professional meetings, journals, education videos, and public and private Internet sites. The information for clinicians described in Chapter 4 builds on the chapter's assessment of the evidence about the benefits and harms of thyroid cancer screening. It recommends against a program of routine screening. If DHHS chooses to disregard this evidence-based assessment and substitute a political judgment, materials provided to clinicians and others should make clear that there is a lack of a scientific justification for routine screening of the general population or potentially exposed population subgroups.

• Distribution of Information.

In developing a method that could be used to distribute information, one approach it might consider is funding the assignment of public-health personnel educated on the topic of Nevada Test Site radiation health effects in some or all public-health-service regions. These personnel would provide information to interested members of the public, as well as to state health agencies, and would respond to questions and concerns. In general, plans for communication to the regional audiences should be developed at the regional level. These plans should be provided to the interested public and to state health agencies for input on their adequacy in meeting local needs. Special attention should be paid to identifying the most effective means of communicating with the interested parties, such as establishing a user-friendly Web site and a toll-free telephone information line staffed by skilled risk communicators who are knowledgeable about the health risks of exposure to radiation from the nuclear weapons tests. At the national level, DHHS should consider sending information to publishers of wellness newsletters and to organizations that represent affected groups.

This chapter has provided a broader framework for DHHS to consider as it develops its information strategy. It notes the challenges that DHHS will face in communicating complicated information about nuclear fallout, radiation exposure, probabilities of developing thyroid cancer, and clinical and public health responses. It also notes the challenge of communicating against a historical backdrop

of controversy over government actions and information. Involving the public in the further development and testing of its information program is one way for DHHS both to better understand public concerns and to develop information that is credible and understandable.

The IOM/NRC committee believes that some Americans will want to learn more about the thyroid cancer risk they individually face. In this context, risk refers to the chance that a specific person will develop thyroid cancer as a result of exposure to I-131 from the Nevada tests. As discussed earlier, the lifetime risk of developing thyroid cancer is, even for most people exposed to I-131, still small, and the disease is rarely life-threatening. About 90 percent of people diagnosed with papillary thyroid cancer are alive 30 years after diagnosis. Any method to estimate I-131 exposure and the chance of developing thyroid cancer should be accompanied by clear information about these points and about the potential benefits and harms of screening for thyroid cancer.

Each person's exposure to I-131 and his or her chance of developing thyroid cancer are usually very difficult to determine as well as difficult to communicate. The method developed by NCI to help people assess their exposure is extremely complicated involving complex calculations for each of the test series and recollections about events that are decades past. It does not provide information about cancer risk.

The IOM committee believes that if DHHS wants to make available a method for assessing exposure and cancer risk, the method should be simple for the average person to use. The method presented here illustrates one approach that could be presented in a brochure or Web site along with the kind of information about thyroid cancer and thyroid cancer screening that is reviewed in this report. This method and related information would need to be tested with potential users.

Though the committee has reasoned that the individual county average dose estimates are imprecise to a degree that they are not useful to assess risk to individuals, there are more readily determined factors about each person that are related to their risk.¹ Individual thyroid cancer risk is determined by several factors beyond simply where a person lived during the time of the Nevada nuclear weapons tests. People exposed to I-131 can be found in almost any part of the United States. Gender, year of birth, some lifestyle characteristics, in particular, the rate of consumption of milk, and the source of their milk have to be considered too. Although age and gender are easy to determine, dietary recall information—for

example, recall of milk consumption some 40 years past—is generally viewed as unreliable.

The method described here is neither fully developed nor tested and this should be understood. Several steps by DHHS would be required as part of its development. Foremost would be to classify the 3,000+ counties into one of 3 groups that would differentiate between concentrations of I-131 in commercial milk available in the county (averaged over the testing period), and into one of 3 groups that would differentiate the counties on the basis of the concentration of I-131 in milk obtained from backyard cows (averaged over the testing period).² These two lists would be necessary companion tables to the method. In addition, supplementary material would also be necessary to provide information to the reader about the nuclear testing program, the years of birth that would result in the highest risk, thyroid cancer risks specific to each sex, radioactive fallout and I-131, natural occurrence of thyroid cancer, the benefits and possible harms of thyroid disease screening, the prognosis of thyroid cancer once it is diagnosed, and information about seeking counseling and guidance from a family physician or other medical specialist.

Again, the intent of the committee is to provide an example of a relatively simple procedure to assist in assessing and communicating an individual person's risk of developing thyroid cancer following exposure to the Nevada tests. In addition, the method should be accompanied by information about thyroid cancer and the potential benefits and harms of thyroid cancer screening similar to that described in Chapter 4. The committee expects that the brochure material, of which this would be a part, could be made available to those persons seeking information from DHHS about their risk related to the Nevada test. The committee does not suggest using the method as a mass screening device.

The following steps present the method as it might be written in a brochure.

Procedure: If you were born after 1910 but before 1960, you might be at some risk of fallout-related thyroid disease. Continue to Step 2. If you were born

at another time, you do not need to continue further. Your risk from exposure to I-131 released by the nuclear weapons tests is near zero.

Explanation: Because age at exposure is the major determinant of risk, the first step is to decide whether there is an appreciable individual risk, based on age at the time of the nuclear testing program. For anyone born after the completion of testing, the risk as a result of the nuclear testing would have been zero,³ that is, there would have been no exposure to the I-131 released from the tests. Equally, for anyone over the age of 20 years in 1951, although there might have been exposure, the risk would have been miniscule and would have been essentially zero for anyone over the age of 40. Thus, there is a potential risk for people born after 1910 and before 1960. Anyone else has zero risk, and need not continue with this procedure.

Procedure: If you drank commercial milk routinely, obtain a factor from Table 1A. If you routinely drank milk from your or your neighbor's backyard cow or goat, obtain a factor from Table 1B.

Explanation: The NCI (1997a) report shows that the most important route of exposure to I-131 was milk consumption. The contamination of milk was a result of the contamination of the pastures where the animals fed. Because milk is not always consumed at or even near where it is produced, the distribution system for commercial milk was one factor that determined the variation in doses across the United States. If milk was supplied by a backyard cow or goat, the main determinant for dose was the amount of fallout where the animal was pastured, presumably near the home of those who consumed the milk. Thus, the second step in the self-assessment is to determine whether the main source of milk was a commercial source (A) or private (B).

Those who consumed mostly cows' milk from a commercial source during the testing refer to Table 1A, which is derived from the NCI data using as a reference group, males aged 5-14 who drank average quantities of milk during the 6 major test series between 1951 and 1957. Commercial milk supplies were not necessarily produced and consumed in the same counties, the geographic distribution of exposure reflects the pattern of distribution.

Milk from a backyard animal was likely to be more contaminated because of the much shorter time between milking and consumption and because of the possibly wider area over which such animals grazed due to the smaller amounts of supplemental (uncontaminated) feed provided. Exposure is related to deposition

in the area. Thus, to determine relative risk according to location, use is made of Table 1B, in which the counties are divided into 3 categories—low, medium, and high deposition. The allocation of counties was determined for males, drinking average quantities of milk from a backyard cow summed over the 6 major test series between 1951 and 1957. The multiplier from Table 1A for the county of main residence during the testing is thus selected as the starting point for the risk self-assessment. Table 1B is analogous to Table 1A but the distribution of counties in the low-, medium-, and high-deposition categories is somewhat different.

Procedure: Obtain 3 factors from Table 2: A factor for age at the time of the tests, a factor for your gender, and a factor for milk consumption rate at the time of the tests.

Explanation: Several factors are important for assessing individual risk. Age is a major determinant both because the young thyroid is more sensitive to the effects of radiation and because the younger the child, the smaller the mass of the

thyroid, which leads to a higher concentration in the gland. Because intake of radioactive iodine is relatively constant with age, radiation doses are usually greater in smaller children. Both factors are considered in determining the multipliers given in the age panel of Table 2. A proportion of exposure came from sources other than milk (often as much as one-third of the total) so a different multiplier should be assigned to the non-milk drinker. Persons who consumed more milk than the average are allocated a multiplier of __. Finally, a gender-specific factor is included because females are more sensitive to cancer induction by a factor of about __.

Procedure: Multiply the 4 factors obtained from Steps 2 and 3 to determine your relative risk value.

Explanation: The product of the 4 factors is a number that indicates the risk of developing thyroid abnormalities (cancer in particular) as a result of exposure to I-131 from the nuclear weapons testing. This value is risk relative to the group to males aged 5-14 and above at the time of the testing. Men in that age group are the ''reference category" because, of all the age and gender combinations, they represent the group with the lowest risk of exposure-induced disease (older men were of minimal, if not zero, risk). The risk of those in the reference category is about ___ on the scale developed here. The risk anyone else might calculate could be as high as ___. Example calculations are provided to assist the reader.

Here are some examples of the way the calculations would work.

• Male, born in 1941, living in county "a" (medium-deposition county) who drank commercial milk at twice or more the average rate in the 1950s:

• Female, born in 1951, living in county "b" (high-deposition county) who drank milk from a backyard goat at the normal rate:

Table 3 provides information for the person making the assessment for understanding the calculated risk value.