© 1999 by Oxford University Press
Journal of the National Cancer Institute Monographs, No. 25, 124-133,
1999
© 1999 Oxford University Press
Risk Communication in Clinical Practice: Putting Cancer in Context
Affiliations of authors: Department of Veterans Affairs Medical Center, White River Junction, VT; Center for Evaluative Clinical Sciences, Dartmouth Medical School, Hanover, NH; Norris Cotton Cancer Center, Lebanon, NH.
Correspondence to: Lisa M. Schwartz, M.D., M.S., and Steven Woloshin, M.D., M.S., VA Outcomes Group (111B), VA Medical & Regional Office Center, 215 N. Main St., White River Junction, VT 05009.
 |
ABSTRACT |
|---|
 Top Abstract IntroductionOffice-Based Tools: Helping...Patient Education: Teaching...Guidance for Communicators:...ConclusionsNotesReferences |
|---|
CONTEXT: Clinicians are increasingly urged—even mandated—to help patients make informed medical decisions by paying more attention to risk counseling. For many, the role of risk counseling is new and unfamiliar. This effort is made more difficult given the practical constraints created by 15-minute visits and competing demands (e.g., patient's chief complaint and institutional needs). OBJECTIVE: We detail a three-part approach for improving risk communication, acknowledging the role of clinicians, patients, and other communicators (i.e., media or public health agencies). PROPOSED APPROACH: Office-based tools to help clinicians do more. We suggest two ways to help make up-to-date estimates of disease risk and treatment benefit easily available during office visits. We propose the development of a comprehensive population database about disease risk and treatment benefit to be created and maintained by the federal government. Educating patients. We propose "Understanding Numbers in Health" a tutorial that reviews basic concepts of probability and their application to medical studies to help people become better critical readers of health information. Guidance for communicators. Finally, we propose a writer's guide to risk communication: a set of principles to help health communicators present data to the public clearly and objectively. CONCLUSION: In addition to tools to help clinicians better communicate risk information, serious efforts to improve risk communication must go beyond the clinic. Efforts that help the public to better interpret health risk information and guide communicators to better present such information are a place to start.
|
INTRODUCTION |
|---|
|
TopAbstractIntroductionOffice-Based Tools: Helping...Patient Education: Teaching...Guidance for Communicators:...ConclusionsNotesReferences |
|---|
Clinicians are increasingly urged [even mandated (1)] to do better risk counseling to help patients make informed medical decisions. The central counseling role that clinicians are expected to play is typified by the recent National Institutes of Health (NIH) Consensus Panel on breast cancer screening for women ages 40–49 that stated that ". . . a woman should have access to the best possible information in an understandable and usable form. Her health care provider must be equipped with sufficient information to facilitate her decision-making process" (2). Unfortunately, the high expectations articulated in this and similar recommendations rarely are accompanied by practical advice.
There are several reasons why efforts focused on the clinician may have limited effect in improving risk communication. First, for many clinicians, risk communication is an unfamiliar discipline. The emphasis on the patient's role in medical decisions is a relatively recent phenomenon (i.e., shared decision making) (3). Few clinicians receive training in methods to promote effective communication with patients (about risk or any subject), and, in fact, little is known about the best ways to communicate such information. Moreover, the relevant data to be communicated have only recently become available and are not easily accessible at the time of office visits.
Second, the competing demands of clinical practice limit what clinicians can be expected to do within the real-world constraint of the standard 15-minute visit (4). The clinician first must address the patient's chief complaint, the concerns and symptoms that brought the patient to the office (which is usually not "I need help to make sense of the health risks I face"). In addition, the growing institutional demands aimed at measuring and at improving the quality of health care have already left many clinicians feeling burdened. These demands invariably result in increased paperwork for data collection and for monitoring adherence to practice guidelines. Given the foregoing, it is not surprising that the limited data available suggest risk is rarely discussed in typical clinical encounters (5).
Risk communication is particularly important in discussions about cancer. Because cancer is
an especially dreaded diagnosis, information about the chance of developing cancer or the effect
of various preventive strategies in reducing cancer risk or the chances of dying of cancer may be
extremely welcome. However, information about cancer alone (or a particular cancer) may
overemphasize the risk compared with other health issues. In this paper, we focus on ways to
improve the presentation and interpretation of quantitative data about risk in general (Table 1
). First, we make suggestions for simple office-based tools to help
clinicians communicate about prevention. Next, we discuss a strategy for educating patients to be
better consumers of data. We conclude with guidance for communicators to improve the quality
of data disseminated to the public by news media and public health agencies.
|
|
OFFICE-BASED TOOLS: HELPING CLINICIANS COMMUNICATE THE VALUE OF PREVENTION |
|---|
|
TopAbstractIntroductionOffice-Based Tools: Helping...Patient Education: Teaching...Guidance for Communicators:...ConclusionsNotesReferences |
|---|
Rationale
A fundamental goal of health risk communication is to help people better understand the important health risks they face. This goal, a basic concept of contemporary medical ethics (3), also has practical implications. Patients who received more information from their physician were more satisfied and had higher compliance with medical regimens (6). At a minimum, understanding the magnitude of a risk (i.e., how big of a threat is breast cancer to me?) entails having some idea of what the risk is (what does it mean to have breast cancer?) and the chances of developing or dying of the condition. Although it is often assumed that physicians spend much time communicating with their patients about risk, remarkably little is known about if and how such communication actually occurs. In the one study (5) documenting doctor-patient risk communication (defined as discussion about behavior change, compliance with screening tests, or preventive treatments), risk was discussed in only 26% of primary care visits and was described numerically in only 3%. One reason why physicians may not engage in risk communication with patients is that they lack easy access to the relevant data. Simple office-based tools may help overcome this barrier.
Office-based tools may be of value in stimulating and in facilitating discussions about disease risks. Patients may want to know the answer to questions such as, what is the chance that a person my age will die of heart disease or breast cancer in the next 10 years? Similarly, patients may also find information about the benefit of various risk-reducing strategies valuable: for example, how does my chance of dying of breast cancer change if I have annual mammograms? To be useful, such office-based tools need to be up to date, immediately available, and easy for both clinicians and patients to use and to understand.
Disease-Specific Tools
A number of tools that generate disease-specific risk estimates for an individual patient are now available. For example, the American Heart Association (7) has a web site where an individual's risk of myocardial infarction can be calculated with the use of a model generated from the Framingham data. The Northern New England Cardiovascular Group (8) uses a preprogrammed hand-held computer to provide patients considering coronary artery bypass graft surgery with an estimate of the mortality risk they face from surgery. Recently, the National Cancer Institute (NCI) (9) issued the Breast Cancer Risk Assessment Tool that provides women with their risk of developing breast cancer to help women contemplating tamoxifen for the primary prevention of breast cancer.
Implementing these tools in clinical practice entails collecting the necessary risk factor information from patients [e.g., the breast cancer risk factors required for Gail model (10) could be collected before a clinic visit] and generating a risk report. Such risk reports could then be attached to each patient's chart at the time of a scheduled clinic appointment with their clinician to maximize the chance of discussion. Some evidence (11-13) suggests that such personalized messages may be more effective than generic messages. Whether the extra time, cost, and technical difficulty of these personalized reports outweigh this potential advantage is unknown.
Although such tools are appealing because the disease-specific estimate is personalized, the inherent focus on a single disease taken out of context may overweigh its importance. When making a decision, a patient may find it helpful to understand where this particular disease fits into the important health threats he or she faces. Patients may find it particularly helpful to know: How does my chance of dying of this particular disease compare with other diseases? What is my overall chance of dying? How does the overall mortality benefit of one intervention (e.g., mammography) compare with the benefit of another (e.g., giving up cigarettes)?
Comprehensive Tools: Disease Risk and Benefit Wall Charts
To provide this context, we propose the creation of charts with age-
and sex-specific data about disease risks and treatment benefits.
Tables 2 and 3 present examples of
such simple office tools. Such low-tech tools, although lacking the
glamour of interactive computer applications, have several distinct
advantages. Simple tools are inexpensive and could be used anywhere
(e.g., posted in any clinic office). Furthermore, simple tools require
no special hardware and no additional personnel or maintenance.
|
|
Disease Risk Chart The disease risk chart shown in Table 2
displays 10-year
disease-specific mortality data for five major diseases—in this case,
coronary artery disease, breast cancer, lung cancer, colorectal cancer,
and ovarian cancer—for women within 5-year age categories
(14,15). Moving across the table allows the user to compare
the magnitude of each disease risk. Because many people may be even
more concerned about their overall chance of dying, the final column
displays all-cause mortality to provide context about how much each
disease contributes to the overall chance of dying. Whereas mortality
data can be represented as counts, proportions, or rates, we use counts
with a stable denominator (e.g., 100 000 women) in our example,
because there is some evidence (16,17) suggesting that people
find counts easiest to understand.
Benefit Chart
Table 3 presents an example of a benefit chart. The goal of this
chart is to help patients compare the relative effect of a change in
behavior or specific intervention on all-cause mortality. Our example
displays age- and sex-specific 10-year all-cause mortality with or
without a given intervention. The numbers shown in the chart are crude
estimates that are accurate in terms of order of magnitude. The first
scenario in the chart considers 100 000 smokers and displays
their chance of dying in the next 10 years if they all continue to
smoke or if they all quit smoking and the net effect—about 6500 deaths
prevented among 55-59 year-old smokers. Another scenario considers
100 000 women who do not have an annual screening mammography
and those who do and shows the net effect of 200 deaths prevented for
55-59 year-old women. These examples show that, for a 55-year-old
female smoker, giving up cigarettes has a substantially greater effect
on all-cause mortality than annual mammography.
Ideally, we would create a benefit chart for an intervention only if the efficacy of screening or behavioral changes has been demonstrated in randomized trials (e.g., mammography for women in their 50s) or when observational analytic studies have convincingly demonstrated benefit and the interventions are routinely recommended (e.g., Pap screening for cervical cancer, smoking cessation). Because age and comorbidity (i.e., competing risks of disability or death that patients face in addition to the risk under consideration) will importantly influence the benefit of any intervention (behavioral changes or screening tests), the benefit charts may encourage explicit discussion between the patient and clinician about these issues. Because interventions can also have harms, an important challenge remains in how to convey data about side effects, bad outcomes, and so forth. Studies comparing the effectiveness of our proposed comprehensive tools, disease-specific tools, and usual care are needed to learn which better helps patients make important medical decisions.
Data Source for Charts
The data required to construct such charts are currently available from a variety of sources [e.g., statistical abstracts, the National Center for Health Statistics, and the Surveillance, Epidemiology, and End Results (SEER) Program1] but would be difficult to consolidate and update. A health risk database could be developed, maintained, and made publicly available by the federal government. Such a central repository of risk information would serve the public good in much the same way as Statistical Abstracts of the U.S. or other federally maintained databases. A distinct advantage of a federal agency taking on this responsibility would be to minimize the incentives to advocate for a specific disease. The National Center for Health Statistics already collects disease data and would be an ideal candidate for the disease risk chart. Because benefit data would require more critical interpretation of the literature, the Agency for Health Care Policy and Research, with its interest and expertise in evidence-based medicine, would be a natural choice for this responsibility.
|
PATIENT EDUCATION: TEACHING PATIENTS TO BE BETTER CONSUMERS OF DATA |
|---|
|
TopAbstractIntroductionOffice-Based Tools: Helping...Patient Education: Teaching...Guidance for Communicators:...ConclusionsNotesReferences |
|---|
Rationale
Efforts to promote informed patient decision making have become increasingly common. In general, these efforts have focused on providing disease-specific facts. The rationale underlying this approach is straightforward: to make informed decisions requires information. If people lack key facts, their decisions cannot be informed. The solution, then, is to provide the facts.
Unfortunately, there are reasons to question the likely effect of this commonsense approach to patient education. First, patients may not be ready for the data. That is, problems with numeracy (i.e., low quantitative literacy) are common. For example, in the National Adult Literacy Survey (18), 47% of adult Americans could not calculate the difference between regular and a sale price from an advertisement. Low levels of numeracy strongly relate to difficulty in making use of quantitative data about the risk reduction of screening mammography (19). Second, patients may not know how to interpret the information they are given. Educators have long understood that presenting facts without first preparing the audience to receive them (i.e., integrate them into some organizing structure) is ineffective and probably counterproductive. In such a case, the members of the audience will absorb little information (which will be quickly forgotten), will not understand how the information fits into their own experience, and may misinterpret what it means. With little experience in using data, for example, patients may be especially susceptible to the framing effects frequently discussed in the cognitive psychology literature (i.e., how simple changes in the format of otherwise identical numerical information can profoundly influence its interpretation) (20-26).
To see how well patient educational materials convey quantitative data to patients under the
best of circumstances, we performed a structured literature review (27) to
identify randomized trials of interventions designed to communicate quantitative data about
disease risk or treatment benefit. Of 70 trials studying patient education interventions, only four
attempted to provide patients with some sort of framework for approaching a medical encounter
(although none dealt with the interpretation of quantitative data). The rest presented facts without
any interpretative framework. Whereas the majority of these trials sought to convey quantitative
data (n = 47 articles), we found only seven randomized trials evaluating patient
comprehension of these data (19,28-33). The table in the Appendix
summarizes the results of five of the seven trials that tested a patient education tool (28-32). Although it is difficult to compare across studies because the interventions
and metrics of efficacy are quite diverse, the interventions had variable effects and, in general,
left substantial room for improvement.
|
Tutorial: Understanding Numbers in Health
Rather than relying on clinicians and communicators to interpret information for the public, we propose to develop the public's capacity to be critical consumers of health information—to prepare patients to receive data. Our proposal consists of creating a generic patient's user guide to health information that deals with the following five subjects.
1) What is risk? Attempts to discuss medical risk are easily undermined by
confusing and imprecise use of language. The tutorial begins by addressing common sources of
confusion (multiple meanings of the word "risk"), how to use words (and the
limits of words) in describing risk, and ways to quantify risk (probability, percents, proportions,
and rates). We will also introduce the reader to a scale that we have developed to facilitate
quantification and communication of risk—particularly small risks less than 1%
(Fig. 1).
|
2) What to look for in a statement about risk. This section teaches the reader to look for various essential elements in statements about risk. Readers will be sensitized to ask questions, such as: What is the risk under discussion (e.g., is it the risk of being diagnosed or of dying?)? What is the time frame under consideration (e.g., next 5 years or lifetime—and what does "lifetime" mean?)? Who is at risk (i.e., does the statement refer to all women? women of a certain age? women with specific characteristics such as a family history of breast cancer?)?
3) Putting risk in context. A salient but rare outcome, such as a celebrity dying of a rare cancer, may give undue weight to certain health risks. The tutorial emphasizes the need to put risks into perspective. Readers will be encouraged to ask questions, such as: How does the chance of this disease compare with other diseases or other familiar events? How dangerous is the disease (i.e., appreciating the difference between developing a condition and dying of it)? To illustrate competing risks, we will make use of disease risk charts discussed previously.
4) Changing risk. This section focuses on how to interpret statements that measure changes in risk given some exposure or intervention (e.g., relative and absolute risk reduction or number needed to treat) and introduce the concept of framing (e.g., dying versus not dying). Benefit charts could be used to highlight that not all risk factors and interventions are equally important.
5) Evidence. The final section points out that there is uncertainty in what we know and introduces the idea of grading evidence by highlighting basic concepts of study design (e.g., observational study versus randomized clinical trial). We encourage readers to have a healthy skepticism and ask themselves, "Can I believe what I am being told? Could it be wrong?"
Limitations
Our approach has several potential problems that should be acknowledged. First, some patients say they do not want information. Many of these patients would therefore have no interest in our tutorial. It is possible that for some people, however, an expressed lack of interest in information may really mean they are afraid they will not be able to understand what they are told. Our tutorial may make quantitative information accessible to people who might otherwise have given up. Next, patients' interest in the tutorial may change substantially under different circumstances. For example, it is possible that patients facing new and serious decisions (e.g., at the time of learning a new cancer diagnosis) may feel too emotionally overwhelmed to make use of the materials, whereas patients making decisions under less intense circumstances (e.g., an older man considering prostate-specific antigen screening) may find the materials especially useful. Finally, understanding whether the tutorial succeeds in teaching (i.e., what do patients learn?) and whether the materials help people make decisions will require careful study and will be the subject of future research.
Application
Assuming that we are able to demonstrate that the tutorial is usable, acceptable, and effective in a randomized trial, we could envision the tutorial being used in a variety of settings. The tutorial could be available for use in school curricula (i.e., modeled on "Chance," an Internet-based, quantitative literacy course that is based on current chance events in the news and is available at: http://www.dartmouth.edu/%7Echance/Chance.html). In clinical settings, the tutorial could be distributed as part of general patient orientation to a practice or could be distributed at the time that important decisions are being made (e.g., men newly diagnosed with prostate cancer or women newly diagnosed with breast cancer deciding on a treatment course) and interpretation of data becomes critical.
|
GUIDANCE FOR COMMUNICATORS: IMPROVING THE QUALITY OFDATA THAT PATIENTS SEE |
|---|
|
TopAbstractIntroductionOffice-Based Tools: Helping...Patient Education: Teaching...Guidance for Communicators:...ConclusionsNotesReferences |
|---|
Rationale
Communicators face the difficult task of translating—often under short deadlines—complex, probabilistic information into a format accessible to a general public with limited grounding in science and with limited ability to make use of probabilistic information (18,19). Communicators themselves may have only superficial training in the critical evaluation of medical literature. They may be unprepared to recognize potential biases, methodologic weaknesses, or questionable statistical manipulations that ought to raise caution about the validity or generalizability of a study's results. For example, the case for cancer screening is often made with a statement that the 5-year survival of patients diagnosed with early stage cancers is much greater than that of patients diagnosed with later stage cancers. These inherently biased statements do not demonstrate that screening is beneficial. Rather, these statement simply say that patients diagnosed earlier live with a cancer diagnosis longer (34,35). It is only from the results of randomized trials that demonstrate that those who are screened have lower cancer death rates than those who were not screened that we can know the true effect of screening.
Evaluation of the accuracy of communications targeting the general public are limited, but frequent problems with news reports have been demonstrated (36,37), and a recent review (38) of Australian public health brochures about screening mammography documented unbalanced and incomplete presentation of data, suggesting an underlying attempt to persuade rather than inform.
Guidance for Communicators
Table 4 presents a set of principles that we hope
will guide communicators in how to present the data
completely, objectively, and understandably. To illustrate some
of these principles, we will use examples from the NCI's "Breast Cancer
Risk Tool: An Interactive Patient Education Tool" (9).
|
Delineate the Main Message Clearly Breast cancer risk. Fig. 2
shows the NCI
screen that displays the main message about breast cancer risk. A
woman's Gail risk factors are entered, and then her calculated breast
cancer risk is displayed graphically and described in text as follows:
|
- "Estimated risk for invasive cancer over the next 5 years is 0.6%"
"Estimated risk for invasive cancer over her lifetime is 11.1%"
The NCI tool has done well in clearly defining this main message: the outcome under consideration is clearly stated as the 5-year and lifetime (to age 90) risk of a diagnosis of invasive breast cancer.
Risk is expressed as a percentage in text and on a linear percentage scale (i.e.,
0%-50%, marked with 5% increments). This dual presentation is a
particularly good idea because many people have trouble working percentages alone (18,19), especially percentages less than 1%. For example, only 20% of
female veterans—almost all of whom had graduated from high school—were able
to correctly convert 0.1% to 1 in 1000 (19). There is surprisingly
little guidance available on how best to present such quantitative information. Some prior studies
(16,17) suggest that counts (e.g., imagine 1000 women, 10 die) may be
easier to understand than percentages, and, in a recent study (39), we
demonstrated that people have great difficulty with expressions of the form "1 in
____." Unfortunately, the design of the graphic in Fig. 1 is not
ideal. It is practically impossible to indicate probabilities below 1% (a relevant range for
many likely users). One approach to this problem is to use a scale, like the one shown in Fig. 1, designed to facilitate expression of small probabilities.
Benefit of tamoxifen. The NCI tool has a second main message that is to inform women about the benefit of tamoxifen in the primary prevention of breast cancer. This message is less well done.
"Women [taking tamoxifen] had about 49% fewer diagnoses of invasive breast cancer"
The benefit of tamoxifen is only expressed as a relative risk reduction without an explicit statement about baseline risk. Several studies (20-23) have shown that physicians and patients find the benefit of an intervention more compelling when it is expressed as a relative risk reduction rather than the corresponding absolute risk reduction. Whereas most typical risk reduction expressions may be difficult to understand, the relative risk reduction without the baseline risk—the format used in the NCI tool—is particularly difficult (19). In the Breast Cancer Prevention Trial (40), the baseline risk (the chance per year of developing invasive breast cancer for women in the placebo arm) was 68 cases per 10 000 women per year. Applying the 49% relative risk reduction yields a risk of 34 cases per 10 000 women per year in the tamoxifen arm.
Curiously, one of the more salient potential harms of tamoxifen—an increased chance of developing uterine cancer— is presented using absolute event rates for each group:
". . . annual rate of uterine cancer in the tamoxifen arm was 30 per 10 000 compared to 8 per 10 000 in the placebo arm"
This asymmetric framing tends to emphasize the benefit of
tamoxifen while minimizing the harm (Fig. 3). If the
increased uterine cancer was expressed using the relative risk format,
the statement would read ". . . 275% more uterine cancer" and
would likely elicit a very different feeling. On the basis of this
framing phenomenon, we believe it is important to present both the
benefits and harms of a treatment using the same frame. To enhance the
effectiveness of such messages, we suggest that communicators present
changes in risk using absolute event rates (19).
|
Provide Context The purpose of the NCI tool is "to measure a woman's risk of invasive breast cancer." The risk provided is the risk of getting breast cancer. For many women, however, the more relevant risk is her chance of dying of breast cancer. Presenting incidence data without mortality data fails to provide important context about how often breast cancer results in death. A related issue involves competing risks for death—for example, how a woman's chance of dying of breast cancer compares with her chance of dying of heart disease.
Another important aspect of context relates to calibrating users to the magnitude of the probabilities presented. It has been demonstrated that even experts are often poorly calibrated to the magnitude of various risks (41). Providing comparisons with the chance of familiar events—such as having a minor car accident—may help make the numbers more meaningful. Such comparisons might help users put their breast cancer risk (i.e., numbers like 0.6% given above) into perspective.
In addition, when discussing factors that change risk, it is important to emphasize that all risk factors do not increase risk to the same degree—something that is not done in the tool. For example, a woman may be able to better judge her breast cancer risk by knowing that family history and age raise the chance of breast cancer to a far greater extent that having the onset of menarche at an early age.
Acknowledge Uncertainty In both the presentation of disease risk and treatment benefit described above, only point estimates (e.g., 49% risk reduction) are provided. These single numbers without 95% confidence intervals imply a false sense of certainty in the expected outcome. This concern is mostly relevant to discussion of treatment benefit rather than disease risk. Whereas the formal statistical definition of 95% confidence intervals may be difficult to communicate, we suggest a simple statement that uses the lower and upper bound of the confidence interval in the following way:
"If 1000 woman do not take tamoxifen, six will be diagnosed with invasive breast cancer in the next year. If these 1000 women all take tamoxifen, our best guess is that three of these six women will not get breast cancer. It is possible that tamoxifen actually prevents as few as two women or as many as four women from getting breast cancer."
In addition to the uncertainty of statistical estimates, there is uncertainty extrapolating from populations to individuals (42). An approach suggested to convey this kind of uncertainty (43) is to use qualifying statements, such as:
"There is no way of knowing whether you will be one of the women who gets breast cancer. In addition, if you take tamoxifen, there is no way of knowing whether you will be one of the women who benefited from it."
An explicit acknowledgment of uncertainty should also accompany messages based on the results of a single study, intermediate end point, or extrapolations across populations. It is rare that a single study provides a definitive answer about a particular question. Consequently, it is critical to put the results of a single study into the context of similar studies and to grade the quality of the evidence (i.e., give less weight to the results of observational studies or subgroup analyses and more weight to randomized trials). Communicators should be particularly cautious about drawing strong inferences when small differences are reported by studies with weaker designs. Intermediate end points should also raise caution because changes in such end points (e.g., tumor shrinkage) may not translate into clinically important improvements in patient outcomes (e.g., improved length or quality of life). An additional concern is whether the study results are really generalizable to the patients the clinic sees: Would this population have met entry criteria for the trial? Is this disease a relatively minor competing risk for this population?
Remember Health One of the main objectives of medical care is to improve the health of the population. Recently, health communicators have begun to increasingly focus on increasing individual's awareness of the disease risks they face and in identifying strategies to modify these risks. In many cases, compliance with recommended risk reduction strategies (e.g., screening for colorectal cancer) has been considered suboptimal. To improve compliance, a number of public health campaigns now actively seek to persuade the public to adopt specific preventive strategies. Many campaigns use scare tactics to promote a particular behavior (e.g., "feeling well is sometimes the first sign of colon cancer—get screened today!" or "you can't see it or feel it, but you may have cancer").
Whereas such persuasive tactics may elicit intended behavioral changes, they have other consequences that may paradoxically worsen a population's sense of health (44-47). First, rather than promoting a sense of health, such tactics may simply increase everyone's sense of vulnerability and anxiety about disease. These messages make it clear that no one is really healthy. For example, telling the story of a 30-year-old woman with breast cancer may garner a lot of attention and motivate some older women to undergo screening mammography, but it will probably also frighten many young women who stand to gain little if anything from mammography. Second, these aggressive tactics may convey a false sense of the magnitude and certainty of the benefits of interventions, engendering unrealistic expectations. Finally, the heightened emphasis on taking personal responsibility for reducing one's risk may lead people diagnosed with disease to blame themselves.
Ironically, the increasing prevalence of persuasive messages coincides with a shift in contemporary medical ethics to a shared decision-making model in which the clinician's role is not to persuade patients to adopt a particular behavior (i.e., use any means necessary to get them to eat that fifth daily fruit) but, where possible, to help patients understand the risks and benefits of the options they face so they can make informed choices between them (e.g., "I understand the pros and cons, and I choose to eat this ice cream"). This model places increasing emphasis on the role of patient preferences and values in medical decision making—because physicians and patients may have different interpretations of well-being.
Whether the evidence of benefit for an intervention is questionable or certain, we believe it is important to consider the likely net effect of such tactics on well-being. We argue that the fundamental purpose of risk communication is to provide individuals with the facts they need to make informed decisions. Increasing the public's sense of vulnerability to inspire a healthy behavior undermines well-being and may result in net harm. We encourage communicators to be sensitive about the potential side effects of their messages.
|
CONCLUSIONS |
|---|
|
TopAbstractIntroductionOffice-Based Tools: Helping...Patient Education: Teaching...Guidance for Communicators:...ConclusionsNotesReferences |
|---|
Although clinicians clearly need to be part of the solution, competing demands and inadequate training in how to talk to patients about risk limit what clinicians can contribute toward improving the state of medical risk communication. Moreover, use of criteria such as Health Plan Employer Data and Information Set (HEDIS) report cards to measure "the quality of care" by the degree with which practice complies with guidelines rather than on some measure of the quality of decision making creates a perverse incentive to prescribe rather than to discuss treatment options.
By acknowledging the realities of clinical practice, we advocate a three-part plan to improve clinical risk communication.
1) Help clinicians to do more by providing clinicians with simple and efficient office-based tools to generate and display up-to-date risk and benefit estimates
2) Educate patients with a reader's guide for patients to help the public more critically evaluate the ubiquitous health risk data to which they are exposed
3) Disseminate Guidance for Communicators, a writer's guide to risk communication to help journalists and public health agencies express risk information in a clear, balanced, and understandable way
Like any intervention, ours will need careful study to evaluate whether it is effective and acceptable to clinicians, communicators, and patients.
|
NOTES |
|---|
1 SEER is a set of geographically defined, population-based, central cancerregistries in the United States, operated by local nonprofit organizations under contract to theNational Cancer Institute (NCI). Registry data are submitted electronically without personalidentifiers to the NCI on a biannual basis, and the NCI makes the data available to the public forscientific research.
Supported by U.S. Army Medical Research and Materiel Command Breast Cancer Research Program New Investigator Award DAMD17-96-MM-6712 as well as by VA Career Development Awards in Health Services Research and Development.
The views expressed herein do not necessarily represent the views of the Department of Veterans Affairs or the United States Government.
Drs. Schwartz and Woloshin are joint first authors of this manuscript. The order of their names is entirely arbitrary.
|
REFERENCES |
|---|
|
TopAbstractIntroductionOffice-Based Tools: Helping...Patient Education: Teaching...Guidance for Communicators:...ConclusionsNotesReferences |
|---|
1 Veterans Health Administration. Journey of change. Washington (DC): Department of Veterans Affairs; 1997
2 National Institutes of Health Consensus Development Conference. Breast cancer screening for women ages 40-49: program and abstracts. Bethesda (MD): National Cancer Institute; 1997.
3
Emanuel EJ, Emanuel LL. Four models of the physician-patient
relationship. JAMA 1992;267:2221-6.
4 National Center for Health Statistics. National ambulatory medical care survey: 1996. Hyattsville (MD): U.S. Department of Health and Human Services; 1997. Available online from: http://www.cdc.gov/nchswwv/data/ad295.pdf
5 Kalet A, Roberts JC, Fletcher R. How do physicians talk with their patients about risks? J Gen Intern Med 1994;9:402-4.[Web of Science][Medline]
6 Hall JA, Roter DL, Katz NR. Meta-analysis of correlates of provider behavior in medical encounters. Med Care 1988;26:657-75.[Web of Science][Medline]
7
Wilson PW, D'Agostino RB, Levy D, Belanger AM,
Silbershatz H, Kannel WB, et al. Prediction of coronary heart disease using risk factor categories.Circulation
1998;97:1837-47.
8
O'Connor GT, Plume SK, Olmstead EM, Coffin LH,
Morton JR, Maloney CT, et al. Multivariate prediction of in-hospital mortality associated with
coronary artery bypass graft surgery. Northern New England Cardiovascular Disease Study
Group. Circulation 1992;85:2110-8.
9 National Cancer Institute. Breast cancer risk tool: an interactive patient education tool [software]. Bethesda (MD): National Cancer Institute; 1998.
10 Gail MH, Brinton LA, Byar DP, Corle DK, Green SB, Schairer C, et al. Projecting individualized probabilities of developing breast cancer for white females who are being examined annually. J Natl Cancer Inst 1989;24:1879-86.cancerlit;90079979
11 Kalichman SC, Coley B. Context framing to enhance HIV-antibody-testing messages targeted to African American women. Health Psychol 1995;14:247-54.[CrossRef][Web of Science][Medline]
12
Skinner CS, Strecher VJ, Hospers H. Physicians'
recommendations for mammography: do tailored messages make a difference? Am J
Public Health 1994;84:43-9.
13 Rimer BK, Glassman B. Tailoring communications for primary care settings. Methods Inf Med 1998;37:171-7.[Web of Science][Medline]cancerlit;98320816
14 Ries LAG, Kosary CL, Hankey BF, Miller BA, Edwards BK, editors. SEER cancer statistics review, 1973-1996. Bethesda (MD): National Cancer Institute; 1999. Available online from: http://www.seer.ims.nci.nih.gov/
15 National Center for Health Statistics. Mortality data [online] 1999. Available from: http://www.cdc.gov/nchswww/datawh/statab/upubd/mortabs.htm
16
Gigerenzer G. The psychology of good judgment: frequency
formats and simple algorithms. Med Decis Making 1996;16:273-80.
17 Hoffrage U, Gigerenzer G. Using natural frequencies to improve diagnostic inferences. Acad Med 1998;73:538-40.[Web of Science][Medline]cancerlit;98272789
18 Kirsch IS, Jungeblut A, Jenkins L, Kolstad A. Adult literacy in America. Washington (DC): Office of Education Research and Improvement, U.S. Department of Education; 1993.
19
Schwartz LM, Woloshin S, Black W, Welch HG. The role of numeracy in understanding the benefit of screening mammography. Ann Intern Med 1997;127:966-72.
20 Malenka DJ, Baron JA, Johansen S, Wahrenberger JW, Ross JM. The framing effect of relative and absolute risk. J Gen Intern Med 1993;8:543-8.[Web of Science][Medline]
21 Forrow L, Taylor WC, Arnold RM. Absolutely relative: how research results are summarized can affect treatment decisions. Am J Med 1992;92:121-4.[CrossRef][Web of Science][Medline]
22 Naylor CD, Chen E, Strauss B. Measured enthusiasm: does the method of reporting trial results alter perceptions of therapeutic effectiveness? Ann Intern Med 1992;117:916-21.
23
Hux JE, Naylor CD. Communicating the benefits of chronic
preventive therapy: does the format of efficacy data determine patients' acceptance of
treatment? Med Decis Making 1995;15:152-7.
24 O'Connor AM. Effects of framing and level of probability on patients' preferences for cancer chemotherapy. J Clin Epidemiol 1989;42:119-26.[CrossRef][Web of Science][Medline]cancerlit;89140907
25
Tversky A, Kahneman D. The framing of decisions and the
psychology of choice. Science 1981;211:453-8.
26 McNeil BJ, Pauker SG, Sox HC Jr, Tversky A. On the elicitation of preferences for alternative therapies. N Engl J Med 1982;306:1259-62.[Abstract]cancerlit;82173068
27 Woloshin S, Schwartz LM. How can we help people make sense of medical data? Eff Clin Pract 1999;2:176-83.[Medline]
28
O'Connor AM, Tugwell P, Wells GA, Elmslie T, Jolly E,
Hollingworth G, et al. Randomized trial of a portable, self-administered decision aid for
postmenopausal women considering long-term preventive hormone therapy. Med Decis
Making 1998;18:295-303.
29
Lerman C, Lustbader E, Rimer B, Daly M, Miller S, Sands C, et
al. Effects of individualized breast cancer risk counseling: a randomized trial. J Natl
Cancer Inst 1995;87:286-92.
30 Flood AB, Wennberg JE, Nease RF Jr, Fowler FJ Jr, Ding J, Hynes LM. The importance of patient preference in the decision to screen for prostate cancer. J Gen Intern Med 1996;11:342-9.[Web of Science][Medline]cancerlit;96396620
31 Inglis S, Farnill D. The effects of providing preoperative statistical anaesthetic-risk information. Anaesth Intensive Care 1993;21:799-805.[Web of Science][Medline]
32 Krueter MW, Strecher VJ. Changing inaccurate perceptions of health risk: results from a randomized trial. Health Psychol 1995;14:56-63.[CrossRef][Web of Science][Medline]cancerlit;95255174
33 Halpern DF, Blackman S, Salzman B. Using statistical risk information to assess oral contraceptive safety. App Cogn Psychol 1989;3:251-60.
34
Black WC, Welch HG. Advances in diagnostic imaging and
overestimations of disease prevalence and the benefits of therapy. N Engl J Med 1993;328:1237-43.
35 Black W. Should this patient be screened for cancer? Eff Clin Pract 1999;2:86-95.[Medline]
36 Freimuth VS, Greenberg RH, DeWitt J, Romano R. Covering cancer: newspapers and the public interest. J Communication 1983;34:62-73.[CrossRef][Web of Science]
37 Oxman AD, Guyatt GH, Cook DH, Jaeschke R, Heddle N, Keller J. An index of scientific quality for health reports in the lay press. J Clin Epidemiol 1993;46:987-1001.[CrossRef][Web of Science][Medline]
38
Slaytor EK, Ward JE. How risks of breast cancer and benefits of
screening are communicated to women: analysis of 58 pamphlets. BMJ 1998;317:263-4.
39 Woloshin S, Schwartz LM, Byram SJ, Fischhoff B, Welch HG. Scales for assessing perceptions of event probability: a validation study. Med Decis Making 1998;14:490.
40
Fisher B, Costantino JP, Wickerham DL, Redmond CK,
Kavahah M, Cronin WM et al. Tamoxifen for prevention of breast cancer: Report of the National
Surgical Adjuvant Breast and Bowel Project P-1 Study. J Natl Cancer Inst 1998;90:1371-88.
41 Lichtenstein S, Fischhoff B, Phillips LD. Calibration of probabilities: the state of the art to 1980. In: Kahneman D, Slovic P, Tversky A, editors. Judgment under uncertainty: heuristics and biases. Cambridge (UK): Cambridge University Press; 1982. p. 306-34.
42
Steiner JF. Talking about treatment: the language of populations
and the language of individuals. Ann Intern Med 1999;130:618-22.
43
Skolbekken JA. Communicating the risk reduction achieved by
cholesterol reducing drugs. BMJ 1998;316:1956-8.
44 Feldman W. How serious are the adverse effects of screening? J Gen Intern Med 1990;5(5 Suppl):S50-3.[Medline]
45
Welch HG, Burke W. Uncertainties in genetic testing for chronic disease. JAMA 1998;280:1525-7.
46
Fitzgerald FT. The tyranny of health. N Engl J Med 1994;331:196-8.
47
Meador CK. The last well person. N Engl J Med 1994;330:440-1.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. Ayer, J. Chhatwal, O. Alagoz, C. E. Kahn Jr, R. W. Woods, and E. S. Burnside Comparison of Logistic Regression and Artificial Neural Network Models in Breast Cancer Risk Estimation RadioGraphics, November 18, 2009; (2009) rg.301095057v2. [Abstract] [Full Text]
|
|
|
|
 |
|
 A. E. Barnato, H. A. Llewellyn-Thomas, E. M. Peters, L. Siminoff, E. D. Collins, and M. J. Barry Communication and Decision Making in Cancer Care: Setting Research Priorities for Decision Support/Patients' Decision Aids Med Decis Making, October 1, 2007; 27(5): 626 - 634. [Abstract] [PDF]
|
|
|
|
 |
|
 R. E. Gramling and J. I. Vidrine Risk Communication During Screening for Genomic Breast Cancer Susceptibility American Journal of Lifestyle Medicine, January 1, 2007; 1(1): 54 - 58. [Abstract] [PDF]
|
|
|
|
 |
|
 L. M. Schwartz, S. Woloshin, F. J. Fowler Jr, and H. G. Welch Enthusiasm for Cancer Screening in the United States JAMA, January 7, 2004; 291(1): 71 - 78. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. G. Mulley Jr. and K. Sepucha Making Good Decisions about Breast Cancer Chemoprevention Ann Intern Med, July 2, 2002; 137(1): 52 - 54. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||