Benign Breast Disease — The Risks of Communicating Risk
http://www.100md.com
《新英格兰医药杂志》
The term "risk" appears in the title of more than 10,000 medical articles published in 2004 (2 percent of the total) — nine times as many as appeared in 1975. In this issue of the Journal, the article by Hartmann et al. on benign breast disease and the risk of breast cancer1 continues the trend, as does this editorial. Hartmann et al. studied a cohort of women who had a benign breast lesion and found that the histologic appearance of the initial biopsy specimen was associated with the risk of breast cancer. As compared with women in the general population, women with nonproliferative findings on breast biopsy had a relative risk of breast cancer of 1.27, those with proliferative changes but no atypia had a relative risk of 1.88, and those with atypical hyperplasia had a relative risk of 4.24. The effect of atypia on the risk of cancer seemed to be independent of a family history of breast cancer.
These data solidify what has long been known about the risk of breast cancer among women with benign breast disease2,3 and help stratify women with a benign lesion into high-risk and low-risk groups. The information will be useful for a surprising number of women: within a decade of starting annual screening, approximately 20 percent of women in the United States will have undergone a breast biopsy4; most of these biopsies show no evidence of cancer.
Hartmann et al. studied a large cohort of women and used current definitions to review all cases, but their results are limited by the retrospectively gathered information on family history (presented for only 53 percent of the women) and the lack of data on breast density and other risk factors. Menopausal status, moreover, was derived mainly from the women's ages. Other variables that underlie both the likelihood of a biopsy and the increase in the rate of detection of breast cancer, such as a high level of concern about breast cancer and consequent frequent examinations when atypia is noted, will require consideration in future studies.
It is unclear whether an atypical histologic appearance is a precursor lesion or a marker of a general tendency to develop breast cancer. Only half of invasive breast cancers arise in the same breast in which atypical hyperplasia was previously diagnosed, suggesting that this lesion is a marker of generalized risk.1,5
Additional refinement of risk may come with the identification of molecular markers; in the meantime, reproducibility of findings among pathologists must be improved if we plan to base risk estimates on histologic findings.5,6 Hartmann et al. provide no data on reproducibility, despite prior studies that have shown major disagreements in the assessment of atypia.5,6
How should clinicians communicate the risk of breast cancer and the implications of a benign breast lesion to women? Most of us, who cannot interpret numbers nearly as well as words, have difficulty understanding numerical expressions of risk.7 In medical schools, courses in statistics usually do not go far enough in teaching statistical or probabilistic thinking, and few teach strategies for effective communication. Hence, most physicians are poorly equipped to discuss risk factors in a way that is readily comprehensible to their patients. This deficiency puts the ideal of informed consent in jeopardy.
Three simple techniques can be helpful.7,8,9,10,11 First, have numerical risk data on hand while seeing patients; second, communicate risk in a clear way; and third, pay attention to positive and negative framing. Consider a woman who asks about her breast-cancer risk and, like most women, has had no prior breast biopsy. She is white and 45 years old, had her first menstrual period at the age of 12 and delivered her first child after the age of 30, and has no first-degree relative with breast cancer. According to the Gail risk model, easily obtained on a Web site,12 her risk of a diagnosis of breast cancer within the next five years is 1.1 percent. Her risk of not receiving a diagnosis of breast cancer within the same period is 98.9 percent. A more transparent formulation is that among 1000 women with these characteristics, 11 would receive a diagnosis of breast cancer within the next five years, whereas 989 would not. The woman should understand that this is a risk of diagnosis, not death, and that treatment has markedly improved over time.
How can the results of the study by Hartmann et al. be explained to a woman with a benign breast lesion? They found that among 6061 women with nonproliferative disease, breast cancer developed in 379, as compared with an expected number of 297.7. This difference is reported as a relative risk of 1.27. The result can be communicated in terms of relative risks, which are misunderstood by many physicians and most patients, or absolute risks, which foster insight. Women with nonproliferative hyperplasia had a 27 percent increase in the risk of breast cancer in the ensuing 15 years. This is a relative risk and will most likely be misunderstood. Absolute risks are clearer (for simplicity, numbers are rounded): in the study by Hartmann et al., among 100 women in the general population, breast cancer developed in 5 within an average of 15 years of follow-up (Figure 1A). Among 100 women with nonproliferative histologic findings, this number increased to approximately 6 (Figure 1B). Thus, the increase in absolute risk is about 1 in 100. This is a simple way to describe the 27 percent increase in risk reported by Hartmann et al.
Figure 1. Examples of Outcomes among 100 Women Followed for an Average of 15 Years.
Each box represents a woman in the general population, and each circle represents a woman who has undergone a breast biopsy. The black circles and boxes represent women with a diagnosis of breast cancer after an average of 15 years of follow-up. In Panel A, 100 women in the general population are followed for an average of 15 years, and 5 subsequently receive a diagnosis of breast cancer. In Panel B, 100 women have nonproliferative findings on breast biopsy, and 6 subsequently receive a diagnosis of breast cancer. In Panel C, 100 women have atypical hyperplasia on breast biopsy, and 19 subsequently receive a diagnosis of breast cancer.
Women with proliferative disease but without atypia have an increase in the relative risk of breast cancer of 88 percent. Some women will falsely conclude that breast cancer will develop in 88 percent of such women. A more comprehensible way of communicating the same information is to say that among 100 women with this condition, the number in whom breast cancer will develop increases from 5 to about 10. Women with atypical hyperplasia have an increase in relative risk of 324 percent, equivalent to an increase in absolute risk from about 5 among 100 women in the general population to 19 among 100 women with atypical hyperplasia (compare Figure 1A and Figure 1C). The use of relative risks suggests greater effects than truly exist, whereas the use of absolute risks (or equivalent clear forms, such as the number needed to treat or the number needed to screen) prevents this misunderstanding. The use of relative risks should be avoided or employed in combination with more comprehensible forms of communicating risk.
Framing is the presentation of logically equivalent information in different forms. Positive framing emphasizes the absence of disease; negative framing emphasizes the presence of disease. Expressing the absolute risk in a positive frame would lead us to say that among 100 women in the general population, breast cancer will not develop in 95 of them within the next 15 years (Figure 1A); among 100 women with a biopsy revealing nonproliferative disease, 94 will not receive a diagnosis of breast cancer (Figure 1B). People are sensitive to framing. Negative framing evokes a willingness to participate in a treatment or a screening, whereas positive framing may not.
Once information about risk is communicated, options for follow-up should be discussed (Table 1).13,14 The recommended course of action is — and will remain for some time — annual mammographic screening with or without a clinical breast examination. If the woman wants to do more, she can perform breast self-examination, although this is no longer recommended by most expert groups. Annual screening with magnetic resonance imaging is not recommended for women whose only risk factor is benign breast disease. Genetic testing is recommended only for women with risk factors for BRCA mutations; it is unlikely to provide useful information for others. The risk of breast cancer among high-risk women can be decreased by chemoprevention and prophylactic surgery, though the potential harms need to be considered.
View this table:
[in this window]
[in a new window]
Table 1. Options for Women at Increased Risk for Breast Cancer.
Informed decisions require that physicians know what the numbers mean and communicate them in ways that patients understand. Improving communication about risk is often treated as a "soft" topic, less important than improving forms of technology. But the best technology offers optimal results only when consumers understand its risks and benefits.15
Source Information
From the Department of Medicine, University of Washington School of Medicine, Seattle (J.G.E.); and the Max Planck Institute for Human Development, Berlin (G.G.).
References
Hartmann LC, Sellers TA, Frost MH, et al. Benign breast disease and the risk of breast cancer. N Engl J Med 2005;353:229-237.
Armstrong K, Eisen A, Weber B. Assessing the risk of breast cancer. N Engl J Med 2000;342:564-571.
Fletcher SW, Elmore JG. Mammographic screening for breast cancer. N Engl J Med 2003;348:1672-1680.
Elmore JG, Barton MB, Moceri VM, Polk S, Arena PJ, Fletcher SW. Ten-year risk of false positive screening mammograms and clinical breast examinations. N Engl J Med 1998;338:1089-1096.
Schnitt SJ. Benign breast disease and breast cancer risk: morphology and beyond. Am J Surg Pathol 2003;27:836-841.
Wells WA, Carney PA, Eliassen MS, Tostenson AN, Greenberg ER. A statewide study of diagnostic agreement in breast pathology. J Natl Cancer Inst 1998;90:142-145.
Gigerenzer G. Calculated risks: how to know when numbers deceive you. New York: Simon & Schuster, 2002.
Edwards A, Elwyn G, Mulley A. Explaining risks: turning numerical data into meaningful pictures. BMJ 2002;324:827-830. [Free Full Text]
Paling J. Strategies to help patients understand risks. BMJ 2003;327:745-748.
Alaszewski A, Horlick-Jones T. How can doctors communicate information about risk more effectively? BMJ 2003;327:728-731.
Jekel J, Elmore J, Katz D. Epidemiology, biostatistics and preventive medicine. Philadelphia: Harcourt Health Sciences, 2001.
National Cancer Institute. Models for prediction of breast cancer risk: Gail model. (Accessed June 30, 2005, at http://www.cancer.gov/bcrisktool/.)
Preventive Services Task Force. Chemoprevention of breast cancer: recommendations and rationale. Ann Intern Med 2002;137:56-58.
Elmore JG, Armstrong K, Lehman CD, Fletcher SW. Screening for breast cancer. JAMA 2005;293:1245-1256.
Deyo RA, Patrick DL. Hope or hype: the obsession with medical advances and the high cost of false promises. New York: AMACOM, 2005:335.(Joann G. Elmore, M.D., M.)
These data solidify what has long been known about the risk of breast cancer among women with benign breast disease2,3 and help stratify women with a benign lesion into high-risk and low-risk groups. The information will be useful for a surprising number of women: within a decade of starting annual screening, approximately 20 percent of women in the United States will have undergone a breast biopsy4; most of these biopsies show no evidence of cancer.
Hartmann et al. studied a large cohort of women and used current definitions to review all cases, but their results are limited by the retrospectively gathered information on family history (presented for only 53 percent of the women) and the lack of data on breast density and other risk factors. Menopausal status, moreover, was derived mainly from the women's ages. Other variables that underlie both the likelihood of a biopsy and the increase in the rate of detection of breast cancer, such as a high level of concern about breast cancer and consequent frequent examinations when atypia is noted, will require consideration in future studies.
It is unclear whether an atypical histologic appearance is a precursor lesion or a marker of a general tendency to develop breast cancer. Only half of invasive breast cancers arise in the same breast in which atypical hyperplasia was previously diagnosed, suggesting that this lesion is a marker of generalized risk.1,5
Additional refinement of risk may come with the identification of molecular markers; in the meantime, reproducibility of findings among pathologists must be improved if we plan to base risk estimates on histologic findings.5,6 Hartmann et al. provide no data on reproducibility, despite prior studies that have shown major disagreements in the assessment of atypia.5,6
How should clinicians communicate the risk of breast cancer and the implications of a benign breast lesion to women? Most of us, who cannot interpret numbers nearly as well as words, have difficulty understanding numerical expressions of risk.7 In medical schools, courses in statistics usually do not go far enough in teaching statistical or probabilistic thinking, and few teach strategies for effective communication. Hence, most physicians are poorly equipped to discuss risk factors in a way that is readily comprehensible to their patients. This deficiency puts the ideal of informed consent in jeopardy.
Three simple techniques can be helpful.7,8,9,10,11 First, have numerical risk data on hand while seeing patients; second, communicate risk in a clear way; and third, pay attention to positive and negative framing. Consider a woman who asks about her breast-cancer risk and, like most women, has had no prior breast biopsy. She is white and 45 years old, had her first menstrual period at the age of 12 and delivered her first child after the age of 30, and has no first-degree relative with breast cancer. According to the Gail risk model, easily obtained on a Web site,12 her risk of a diagnosis of breast cancer within the next five years is 1.1 percent. Her risk of not receiving a diagnosis of breast cancer within the same period is 98.9 percent. A more transparent formulation is that among 1000 women with these characteristics, 11 would receive a diagnosis of breast cancer within the next five years, whereas 989 would not. The woman should understand that this is a risk of diagnosis, not death, and that treatment has markedly improved over time.
How can the results of the study by Hartmann et al. be explained to a woman with a benign breast lesion? They found that among 6061 women with nonproliferative disease, breast cancer developed in 379, as compared with an expected number of 297.7. This difference is reported as a relative risk of 1.27. The result can be communicated in terms of relative risks, which are misunderstood by many physicians and most patients, or absolute risks, which foster insight. Women with nonproliferative hyperplasia had a 27 percent increase in the risk of breast cancer in the ensuing 15 years. This is a relative risk and will most likely be misunderstood. Absolute risks are clearer (for simplicity, numbers are rounded): in the study by Hartmann et al., among 100 women in the general population, breast cancer developed in 5 within an average of 15 years of follow-up (Figure 1A). Among 100 women with nonproliferative histologic findings, this number increased to approximately 6 (Figure 1B). Thus, the increase in absolute risk is about 1 in 100. This is a simple way to describe the 27 percent increase in risk reported by Hartmann et al.
Figure 1. Examples of Outcomes among 100 Women Followed for an Average of 15 Years.
Each box represents a woman in the general population, and each circle represents a woman who has undergone a breast biopsy. The black circles and boxes represent women with a diagnosis of breast cancer after an average of 15 years of follow-up. In Panel A, 100 women in the general population are followed for an average of 15 years, and 5 subsequently receive a diagnosis of breast cancer. In Panel B, 100 women have nonproliferative findings on breast biopsy, and 6 subsequently receive a diagnosis of breast cancer. In Panel C, 100 women have atypical hyperplasia on breast biopsy, and 19 subsequently receive a diagnosis of breast cancer.
Women with proliferative disease but without atypia have an increase in the relative risk of breast cancer of 88 percent. Some women will falsely conclude that breast cancer will develop in 88 percent of such women. A more comprehensible way of communicating the same information is to say that among 100 women with this condition, the number in whom breast cancer will develop increases from 5 to about 10. Women with atypical hyperplasia have an increase in relative risk of 324 percent, equivalent to an increase in absolute risk from about 5 among 100 women in the general population to 19 among 100 women with atypical hyperplasia (compare Figure 1A and Figure 1C). The use of relative risks suggests greater effects than truly exist, whereas the use of absolute risks (or equivalent clear forms, such as the number needed to treat or the number needed to screen) prevents this misunderstanding. The use of relative risks should be avoided or employed in combination with more comprehensible forms of communicating risk.
Framing is the presentation of logically equivalent information in different forms. Positive framing emphasizes the absence of disease; negative framing emphasizes the presence of disease. Expressing the absolute risk in a positive frame would lead us to say that among 100 women in the general population, breast cancer will not develop in 95 of them within the next 15 years (Figure 1A); among 100 women with a biopsy revealing nonproliferative disease, 94 will not receive a diagnosis of breast cancer (Figure 1B). People are sensitive to framing. Negative framing evokes a willingness to participate in a treatment or a screening, whereas positive framing may not.
Once information about risk is communicated, options for follow-up should be discussed (Table 1).13,14 The recommended course of action is — and will remain for some time — annual mammographic screening with or without a clinical breast examination. If the woman wants to do more, she can perform breast self-examination, although this is no longer recommended by most expert groups. Annual screening with magnetic resonance imaging is not recommended for women whose only risk factor is benign breast disease. Genetic testing is recommended only for women with risk factors for BRCA mutations; it is unlikely to provide useful information for others. The risk of breast cancer among high-risk women can be decreased by chemoprevention and prophylactic surgery, though the potential harms need to be considered.
View this table:
[in this window]
[in a new window]
Table 1. Options for Women at Increased Risk for Breast Cancer.
Informed decisions require that physicians know what the numbers mean and communicate them in ways that patients understand. Improving communication about risk is often treated as a "soft" topic, less important than improving forms of technology. But the best technology offers optimal results only when consumers understand its risks and benefits.15
Source Information
From the Department of Medicine, University of Washington School of Medicine, Seattle (J.G.E.); and the Max Planck Institute for Human Development, Berlin (G.G.).
References
Hartmann LC, Sellers TA, Frost MH, et al. Benign breast disease and the risk of breast cancer. N Engl J Med 2005;353:229-237.
Armstrong K, Eisen A, Weber B. Assessing the risk of breast cancer. N Engl J Med 2000;342:564-571.
Fletcher SW, Elmore JG. Mammographic screening for breast cancer. N Engl J Med 2003;348:1672-1680.
Elmore JG, Barton MB, Moceri VM, Polk S, Arena PJ, Fletcher SW. Ten-year risk of false positive screening mammograms and clinical breast examinations. N Engl J Med 1998;338:1089-1096.
Schnitt SJ. Benign breast disease and breast cancer risk: morphology and beyond. Am J Surg Pathol 2003;27:836-841.
Wells WA, Carney PA, Eliassen MS, Tostenson AN, Greenberg ER. A statewide study of diagnostic agreement in breast pathology. J Natl Cancer Inst 1998;90:142-145.
Gigerenzer G. Calculated risks: how to know when numbers deceive you. New York: Simon & Schuster, 2002.
Edwards A, Elwyn G, Mulley A. Explaining risks: turning numerical data into meaningful pictures. BMJ 2002;324:827-830. [Free Full Text]
Paling J. Strategies to help patients understand risks. BMJ 2003;327:745-748.
Alaszewski A, Horlick-Jones T. How can doctors communicate information about risk more effectively? BMJ 2003;327:728-731.
Jekel J, Elmore J, Katz D. Epidemiology, biostatistics and preventive medicine. Philadelphia: Harcourt Health Sciences, 2001.
National Cancer Institute. Models for prediction of breast cancer risk: Gail model. (Accessed June 30, 2005, at http://www.cancer.gov/bcrisktool/.)
Preventive Services Task Force. Chemoprevention of breast cancer: recommendations and rationale. Ann Intern Med 2002;137:56-58.
Elmore JG, Armstrong K, Lehman CD, Fletcher SW. Screening for breast cancer. JAMA 2005;293:1245-1256.
Deyo RA, Patrick DL. Hope or hype: the obsession with medical advances and the high cost of false promises. New York: AMACOM, 2005:335.(Joann G. Elmore, M.D., M.)