In January 1997, a National Institutes of Health Consensus Development Conference [2] concluded that "the available data do not warrant a single recommendation for mammography for all women in their forties. Each woman should decide for herself whether to undergo mammography." Two months later, the American Cancer Society, referring to the same evidence, took a different stand with their recommendation that all women in their 40s undergo annual mammography [3]. Shortly after that, the National Cancer Advisory Board's recommendation for performing mammography every 1 to 2 years was quickly adopted by the National Cancer Institute [4].

The debate, heated at times, has perhaps never been more contentious than in the past few months. With emotional outbursts at public meetings and in the press, positions seem to have hardened, and the discourse that might find common ground is not occurring.

This controversy provides an opportunity to gain important insights into a policy problem in which there is, at substantial economic cost, a large benefit for a small number of persons and a small amount of harm for a larger number. Although some may have tired of this controversy, we believe that it should be pursued for two reasons. First, contentiousness may be getting in the way of rational decision making, at both the individual and policy levels. Second, there may be a reasonable middle ground. The purpose of this commentary is to identify the sources of disagreement and contention so that discussion and decision making can move ahead constructively.

Anatomy of a Decision: Two Steps

Medical decision making, whether for individual persons or for policymakers, involves two major steps, as described by Eddy [5]. The first step involves "facts" and the analysis of evidence. For mammography screening, the facts include a description of the options (screening or no screening), the possible outcomes of those options (cases of cancer diagnosed, lives extended, the effort of screening and workups, the effect of false-positive results, and costs), and the chances that any of these outcomes may occur. Step 1 should be relatively straightforward because, as Eddy notes, "assuming there is some evidence to evaluate, it should be possible to get reasonable, open-minded people to agree on the results of this step" [5]. The second step is more problematic because it concerns not facts but rather each decision maker's informed personal judgments about the importance of the outcomes and risks of each option: that is, values and utilities. The concept of "informed" decisions is important because uninformed decisions are more likely to be bad decisions and to lead to disappointment.

Sources of Disagreement: What the Debate Is about

The disagreement in this controversy is less about facts than it is about judgments. Thus, most participants in the debate, but not all, agree that the mortality benefit over 10 to 12 years is about 16% to 18%, or about 1 or 2 lives extended per 1000 women 40 to 50 years of age who are screened annually for 10 years [6]. This agreement should not be surprising because screening mammography has been studied for more than 30 years. However, perceptions of this mortality reduction and of judgments about the balance between benefits and possible harms and effort differ substantially.

Description and Perception of Risk

Description of Risk

A central challenge in the mammography debate concerns how to describe benefit (reduction of risk) and harm (additional risk) "fairly" and completely so that individual persons as well as policymakers are "well informed" and can go to the next step of factoring in their own subjective values and judgments. The dramatic effect of how "facts" are "framed" or presented to a decision maker has been well studied in the cognitive psychology literature [7-10]. To take advantage of this effect, persons on different sides of the mammography debate have presented risk in very different ways, much like businesses trying to market a product. Thus, it sounds more impressive to state that mammography reduces the relative risk for death from breast cancer by 16% to 18% than to state that the absolute risk for death is reduced by 1 to 2 lives per 1000 women screened annually for 10 years [11]; however, both of these descriptive statements are correct. A different approach to stating the benefit would be to describe the "number needed to screen" over 10 years to prevent one death from cancer (about 500 to 1000) or the number of side effects incurred to prevent one death from cancer [12]. Another approach would be to state that the "average" amount of life gained by mammography screening for women in their 40s is only 3 days (Berry D. Personal communication), a number that sounds "small." In contrast, a statement that about 800 women in their 40s in the United States would have their lives extended each year by universal screening sounds "large." The point is that the different descriptions of the same facts may sound different. How to select a "fair" method with which to describe the facts is still uncertain. We need both additional research and open discussion about the ground rules for and principles behind "informing" people.

Risk Perception and Misperception

Regardless of the method used to describe the facts about risk reduction, an entirely separate set of features may affect a person's perception of that risk. Thus, some features of everyday life tend to carry both the message that breast cancer is a very common problem and the message that screening is effective in extending lives. Indeed, many women, especially those younger than 50 years of age, greatly overestimate their own risk for breast cancer and similarly overestimate the risk reduction possible from screening mammography [13-15].

Why might such overestimates occur? The field of cognitive psychology suggests that the "availability heuristic" may make persons perceive risks as higher than they actually are [16]. Nearly everyone knows someone who has received a diagnosis of breast cancer. It is a diagnosis that, when it occurs, is widely discussed and makes a lasting impression on anyone who hears of it. The "availability" of this mental impression has been shown to lead to an exaggerated sense of risk.

Similarly, much everyday experience may suggest that mammography screening is more effective than it actually is in reducing breast cancer mortality. Breast cancer has never been a universally fatal disease, and most women in whom breast cancer is diagnosed actually die of something else. This was true even before widespread screening [17]. However, as more and more women are given a diagnosis by screening, the natural perception of each "survivor" is that screening "saved" her life. Although clinical trials show that screening in women 40 to 50 years of age sometimes does extend lives (because breast cancer mortality is reduced by 16% to 18% in this group), most survivors actually have a type of breast cancer that could have been treated as effectively when found later as a lump or that might never have caused problems. The problem is that for each survivor, there is no way to know whether screening caused survival, and it is only natural to want to believe that an unpleasant medical intervention has helped. Thus, the number of women who perceive benefit from screening is substantially greater than the actual number who receive benefit, and the impression of benefit is exaggerated.

Given our problems in fairly communicating risk and the intuitive belief of benefit among survivors of breast cancer, it is not hard to understand how these perceptions can add fuel to the fire of disagreement.

Disagreement about Effects of Screening on Quality of Life

Although reduction of breast cancer-related mortality rates is the primary outcome in clinical trials, "quality of life" outcomes may be important as benefits or harms in decisions about screening. There is disagreement about both the non-mortality-related benefits and the harms of screening.

Non-Mortality-Related Benefits of Screening

Potential non-mortality-related benefits of screening include less intensive therapy (for example, lumpectomy rather than mastectomy) because disease detected earlier may require less aggressive treatment. However, although lumpectomy rates are increasing, mastectomy rates are not declining proportionately [18]. Furthermore, the decision to treat with lumpectomy does not depend greatly on finding cancer at a prepalpable stage, and factors other than tumor size enter into the treatment decision. It is not clear that more screening per se will substantially decrease mastectomy rates, although these rates may decline for other reasons.

"Peace of mind," another potential benefit of screening that is reported by many women after a negative result on a mammogram, is also in dispute. Statistically, at least, the reassurance that a woman should derive from a negative result on a screening mammogram is the difference between the probability of cancer before the mammogram and the probability of cancer after a negative result on the mammogram. For a woman 45 years of age, this is a probability difference of about 0.0012 (1.6 in 1000 before mammography – 0.4 in 1000 after a negative result on a mammogram), or an absolute change of a little more than 0.1%. The degree of reassurance that many women report from a negative result on a mammogram appears to be out of proportion to this calculation. Indeed, one wonders whether many women might derive more peace of mind simply from a better understanding of their actual risk.

Thus, the non-mortality-related benefits of screening look larger to some persons than to others.

Harms of Screening

Screening may also decrease quality of life, at least to some degree, for many women. Mammography has a specificity of about 95% (or a 5% false-positive rate); it thus incurs about 30 false-positive results for every true-positive result among women in their 40s. Over 10 years of screening, as many as 30% of all women would have at least one false-positive results [19]. Widely acknowledged anecdotal evidence suggests that many women have marked anxiety from the time that they are told of a positive result on mammography until further testing finds no cancer. In addition, women experience substantial worries that affect their mood or daily activities as long as 3 months after being told that they do not have cancer [20]. The trade-off is between the important benefit of screening for the few women whose lives are extended and the lesser (but real) harm incurred among the many women with false-positive results.

In addition, anecdotal evidence suggests that some women who later develop breast cancer feel "betrayed" if an earlier negative result on a mammogram was falsely reassuring and that women who did not have mammography before being found to have breast cancer may have "regret," even though there is no way to know whether earlier mammography would have detected the disease and extended their lives.

The nature and magnitude of these harms are not precisely known, nor is it clear how they should affect the ultimate decision about whether to have mammography screening. Some persons see these non-mortality-related harms as important and potentially large, whereas others see them as exaggerated and easily eliminated by appropriate counseling. In any case, these potential harms cannot and should not be ignored.

Disagreement about Roles of Physicians and Patients in Decision Making

When confronted with the trade-off of benefit for a small number of women versus harm of uncertain magnitude for a larger number, the 1997 Consensus Development Conference suggested an approach that others have termed shared decision making [21]. In part because of the format of its report, the Consensus panel did not explain the approach in detail, and many persons who heard the report were understandably skeptical. It sounded as if health care professionals were simply passing a very difficult decision on to their patients, who would be even less equipped to make an informed decision.

Although physicians and patients have always shared decision making to some extent, the degree to which physicians' decision-making styles include patient participation varies greatly [22, 23]. One style, with a high degree of patient participation, has been termed shared decision making; this process is becoming more formalized and elaborated for select problems. In this process, the physician first presents evidence about choices and their outcomes (both benefits and harms). Next, the patient's values about the outcomes of each option are elicited and clarified. Finally, patient and physician agree on what option best meets the patient's values. This process, although it has not yet been fully developed and evaluated, is intended to occur in a highly explicit, consensual, give-and-take manner so that the clinician provides medical expertise and helps the patient understand the options and their consequences (corresponding to decision-making step 1) and the patient provides values for each potential consequence (step 2). Because input from both clinician and patient is critical, the decision is "shared." This formal approach has been applied to several specific medical problems [21, 24-26] and is especially useful when "the superiority of one option over another is uncertain and depends on patient preferences" [27]. As one part of this process, individual persons decide how involved they want to be in decision making [23].

There are several variations on this theme of formal shared decision making [28]. In situations in which the health benefits clearly outweigh the potential harms, the physician may play a greater role, encouraging the patient to make a particular decision. In other situations, such as that of screening mammography for women in their 40s, in which the balance between benefit and harm is open to interpretation, the physician tries not to substitute his or her values for those of the patient.

Despite the growing body of literature describing the benefits of shared decision making, some remain skeptical about its feasibility in the hectic clinical setting. Research is ongoing about methods to be used during this process (for example, use of printed educational materials, story boards, a "balance sheet," or computer programs to implement this approach in medical practice [21, 26, 29]). Because the approach has not been fully evaluated, it is understandable that many persons were skeptical about the call of the Consensus Development Conference for shared decision making. At the same time, it is difficult to understand how the concept of sharing medical decisions can be seen as anything but a core principle of good medical practice.

Disagreement about How To Factor Cost into Policy Decisions

Consideration of cost-effectiveness provokes disagreement at the policy level because different decision makers have inherently conflicting perspectives. From a societal perspective, cost-effectiveness analysis provides a way to compare different policy choices in the allocation of scarce resources, but it is generally not a major concern from the perspective of an individual person.

The recent policy debate about mammography for women in their 40s is remarkable for the almost total absence of discussion of cost, given that cost in 1997 is a major consideration in almost every other aspect of health care. Thus, cost was ruled out as a consideration at the beginning of the Consensus Development Conference and was never mentioned in the statements of either the American Cancer Society or the National Cancer Advisory Board. Although all three statements advocated third-party coverage of mammography for women in their 40s, there was no supporting economic analysis or comparison with other competing health care activities.

Because mammography for women in their 40s is technically less effective and its costs are, if anything, higher (because of the consequent need for more frequent screening and the greater numbers of false-positive results requiring work-up), mammography must be less cost-effective for women in their 40s than for women in their 50s and 60s. Screening thus becomes a very expensive strategy for controlling breast cancer for women in their 40s.

Although factors other than cost-effectiveness clearly enter into policy decisions about screening, cost-effectiveness should still be explicitly considered. Indeed, when the results of a cost-effectiveness analysis appear to be dramatically at odds with a policy decision, that fact itself should provide strong motivation to identify the other factors that are driving, and perhaps distorting, the decision.

Special Features Distorting Discussion in This Debate

In addition to the features that can distort perception of risk, other features of this particular controversy may affect decision making and exacerbate disagreement at "step 2," both at the individual and policy levels.

Single-Issue Advocacy

Because breast cancer is such a frightening disease and many persons understandably feel strongly about it, several single-issue "advocacy" groups have successfully influenced policy on this topic. Although such groups have a legitimate voice in the debate, they cannot claim to speak for all women in the United States. An advocacy group necessarily looks at policymaking from a limited perspective and does not consider a broad range of policy choices and implications. Much of the policy debate on this topic appears to be driven by posturing and considerations of political gain more than by thoughtful deliberation of the issues [30, 31]. An important challenge in policymaking is how to accomplish step 2 in a rational and balanced manner. Reducing the influence of single-issue groups through alternate educational programs might be useful in reducing the contentiousness of the debate.

Gender Politics

It is fascinating to think about why this policy problem, in contrast to some others, has become so heated. We must wonder whether some kind of "gender politics," both in the U.S. Congress and nationwide, is being played out in the making of cancer screening policy. The U.S. senate's resolution on this topic was led by female senators [32], whereas prostate cancer is a "man's cancer" with its own advocates and celebrity spokesmen [33]. Yet, perhaps tellingly, the magnitude of benefit from screening for these two types of cancer is either small (for breast cancer for women in their 40s) or not yet proven at all (for prostate cancer) [34]. In contrast, the demonstrated efficacy of screening for colon cancer is greater (a relative mortality reduction of 15% to 33% for fecal occult blood test screening and 60% for endoscopic screening [35]) and is, by several calculations, very cost-effective at about $20 000 to $40 000 per year of life saved [36, 37]. Yet the colon is "gender-neutral," and there are few advocates for colon cancer screening outside of the subspecialty medical societies and little public advocacy or interest in the intervention.

We wonder whether decision making about breast cancer screening, by both policymakers and private individuals, is being driven and perhaps distorted by concerns related to gender. Understanding this aspect of the debate may be useful in defusing it.

Can the Debate Be Defused?

We think that there are ways to reduce the contentiousness of this debate. One possible benefit of a "cooling off" is an increased chance that good decisions will be made, both at policy and individual levels.

The debate may be defused in part simply by understanding the anatomy of a decision, so that specific disagreements may be identified and explored. Within such a framework, common goals presumably may be identified, even if deliberating parties have different perspectives. Lessons from other domains about how to handle disagreement [38] may be useful here.

As noted above, common ground should be relatively easy to find for step 1. Thus, there is currently agreement about the magnitude of the mortality reduction; the uncertainty about the nature and frequency of harm associated with screening could be addressed by further research and analysis, which are urgently needed. It might help substantially if we first make sure that we have all of the evidence and then explicitly acknowledge that we agree about the facts.

Defusing step 2 is harder, but not impossible. It would help to agree at the outset that informed judgments are an essential part of decision making, whether by individuals or by society, and that differences in judgments should be expected and respected. Shared-and informed-decision making can be seen as a common middle ground for debates of this type, with both "sides" working toward fully informing the public and individual persons about the pros and cons of each option.

Getting both sides to focus on fully informing individual persons and the public is, we suggest, a key to defusing the debate about breast cancer screening for women in their 40s. The task of informing the public about this issue should not be left to advocacy groups alone; physicians and physician groups should use their positions to lead this educational effort. They need to learn how to work with the media to inform the public. This kind of involvement should improve the "environment" in which decisions are made, especially if health professionals can maintain their objectivity and resist becoming advocates for a particular point of view rather than advocates for health.

The aim of this educational effort should be to help persons understand their actual risks as well as the true benefits, harms, and costs of screening and to understand why daily experience might lead to biased perceptions. The educational program might provide positive information on other ways to reduce the risk for breast cancer, such as decreasing alcohol use [39] and increasing exercise [40]. We think that such a deliberate effort is warranted to help ensure that decisions at both the individual and policy levels are truly "informed" and not substantially distorted by misperceptions.

Finally, we think that defusing the debate requires both sides to agree that cost-effectiveness needs to be factored into decisions about health policy.

Conclusions

The debate over mammography concerns a policy problem in which there is, at a substantial economic cost, a large benefit for a small number of persons and a small amount of harm for a much larger number. The problem is similar to others that we increasingly face, such as those associated with prostate cancer screening and some kinds of genetic screening. As our science becomes ever more sophisticated, we discover ways to make small improvements in health, but large breakthroughs are rare. Cost and harm are often associated with widespread implementation of new interventions "at the margin." The harm may sometimes be subtle and may involve the diversion of resources that could be better directed toward other important health problems. If we are always ready to pay "any price" to reduce mortality from our favorite disease by even a small amount, we deserve the expensive, relatively ineffective, and sometimes harmful health care system that we will ultimately produce.

Rational decision making, whether for individual persons or for the public, requires a more thoughtful approach. We need to learn how to better understand and manage problems that have small degrees of risk and for which persons disagree about what to do. We need to learn how to conduct debate in a rational and productive manner and especially how to avoid distortions that lead to decisions we might later regret. And we need to work together to find ways to help the public better understand decisions of this type and the reasons why disagreement exists.

Screening mammography for women in their 40s is effective, but only to a limited degree, and it brings with it real costs and potential harms. There is a choice to be made; reasonable persons may disagree. Our job is to inform women in their 40s, to work with them to clarify their values, and to respect their individual decisions.

We can hope that in 10 to 20 years, when we can evaluate whatever policies result from this debate, we will not judge those policies to have caused more harm than good and wish that we had known better.

Dr. Harris: CB #7508, Building 52, University of North Carolina School of Medicine, Mason Farm Road, Chapel Hill, NC 27599-7508.