Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

Variation of Benefits and Harms of Breast Cancer Screening With Age

Variation of Benefits and Harms of Breast Cancer Screening With Age Abstract The critical issue in deciding whether to recommend breast cancer screening for women in their forties is to determine whether potential benefits are substantially greater than potential harms. Recent evidence from randomized clinical trials makes it likely that, after 10-12 years of follow-up, there is a real benefit from screening women ages 40-49, on the order of a 15-20% reduction in the relative risk of breast cancer death. This relative risk reduction translates into an absolute risk reduction of 1-2 women whose lives are extended from screening 1,000 women in their forties annually for 10 years (i.e., about one life extended per 5,000 mammograms). The absolute benefit of screening increases with age. Evidence about potential harms is less well established, but it is compelling that there are 15-40 times as many false positive as true positive mammograms (depending on the patient's age), and that at least some of the women with false positive mammograms have ongoing psychological distress as a result. Some 30% of all women who are screened annually during their forties will have at least one false positive mammogram and this probability likely decreases with advancing age. If the balance between benefits and harms is judged to be a “close call” for women in their forties, a blanket recommendation for all is inappropriate. Instead, each woman in her forties should be helped to understand the pros and cons of screening, to clarify her own values, and to consider with her primary care physician what decision would be best for her. Getting the Question Right The question I wish to address is what level of recommendation to make to women of different ages about breast cancer screening. I want to emphasize the phrase “what level of recommendation.” Some may think the answer is a simple “yes” or “no”—either we recommend or we don't. The strength of the recommendation, however, should depend on the strength of the evidence about two issues: the benefits of screening and the harms of screening. The real question, then, is not whether there is some small benefit demonstrated for screening women in their forties. The issue is larger than a “P-value.” What we need to know is where the balance lies between the magnitude of benefits and harms for different age (or other risk) groups. But what do we do in cases where the balance between benefits and harms is not clear, as I believe is the case with breast cancer screening for women in their forties? In these cases, there is a third option beyond recommending or not recommending. Physicians may also raise the issue of breast cancer screening with their patients, help them understand the benefits and harms, and encourage them to participate in making an individualized decision. My aim, then, is to provide an overview of the benefits and harms of screening for women in their forties, so that these women, with the help of their physicians, can make the most appropriate decision for themselves. Mortality Benefits of Screening Screening seeks to decrease the risk of dying of breast cancer, not the risk of getting it. The specific risk a woman is trying to reduce by being screened for the next 10 years is the risk of eventually dying of cancer diagnosed in those next 10 years. These risks for women of different ages, calculated from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) data before widespread screening, are given in the second column of Table 11 (1). Not surprisingly, the risk increases with age. To date, eight randomized trials of mammography screening among women aged 40 and older have been conducted in Sweden, the United Kingdom, Canada, and the United States. Mortality reduction in these trials is measured in terms of a “relative risk” reduction—that is, the reduction in risk of dying of breast cancer in a screened group relative to the baseline risk in an unscreened group. When the relative risk reduction from these randomized trials (Table 11, column 3) is factored in, we can calculate the absolute risk reduction (Table 11, column 4)—the number of women per 1,000 whose lives would ultimately be extended by screening over 10 years. The new evidence from the Swedish randomized trials makes it likely that there is a real benefit, and that it is on the order of a 15-20% relative risk reduction. If the relative risk reduction for women in their forties is 10-15%, then one woman would have her life extended for screening 1,000 women for 10 years. If the relative reduction for this age group is about 20%, then the lives of about two women would be extended for screening 1,000 women for 10 years (about one life extended per 5,000 mammograms). Table 11 illustrates that the benefit of screening—the number of women per 1,000 whose lives are extended—increases with age. Some have made the claim that the benefit of screening is the same for women in their forties as for those in their fifties or sixties. These claims have used the relative risk reduction as a measure of benefit. It is clear from Table 11, however, that even if the relative risk reduction is equivalent in different age groups (which is not at all certain from the trials), the absolute benefit in number of lives extended per 1,000 women screened increases with age. Effects of Screening on Nonmortality Outcomes Screening is a “double-edged sword” that can result in either benefits or harms. There is a need for more research on both nonmortality benefits and harms of screening. The potential magnitude of the effects, however, is apparent from examining the screening “cascade”—that is, the expected sequence of events following screening. This cascade is shown in Figs. 11 and 22 for a single screening of a hypothetical population of 10,000 women who are being screened regularly (i.e., “incidence” screens as opposed to “prevalence” screens). The figures assume a conservative mammogram “positivity” rate of 5%—that is, 5% of all cases will require further evaluation. This rate is indicative of many excellent mammography practices (2) and is much less than the 11% found in a recent national survey (3). Using the higher rate would double the number of false positives. Sensitivity of mammography is taken from average sensitivity in the trials (4), and incidence of cancer is taken from SEER data (5). Screen-Negative Women As seen in both figures, most women screened are negative. The great majority of these women are truly negative—they do not have breast cancer. A few, however, truly have cancer but are screen negative—that is, they are falsely negative. A research priority is to find out whether some of these women have been injured by false reassurance. It seems possible that some may ignore early symptoms of breast cancer because they have been reassured by the negative mammogram. We don't know how often this really happens. The true negatives would seem to be in a position to benefit; they could receive “peace of mind”—reassurance that they do not have cancer. But if you look carefully at the probability of having cancer before screening versus after negative screening—which for women in their forties is about 1.6 per 1,000 before screening and about 0.4 per 1,000 after a negative screen, a reduction in risk of about 1 per 1,000)—the difference doesn't seem large enough to make a truly objective woman change from worrying to relaxing. The woman was at low risk before screening and is still at low (but not zero) risk after being screen negative. Nevertheless, many women report peace of mind after a negative mammogram. This may reflect overestimation of initial risk and overinterpretation of a negative mammogram, and it suggests that we should develop ways other than mammography of reassuring women. Screen-Positive Women In many excellent mammography practices, about 5% of women are screen positive, and the great majority of these are falsely positive (i.e., they do not have cancer, despite the positive test). As shown in Figs. 11 and 22, there are 15-40 times as many false positives as true positives. And Figs. 11 and 22 are only for a single screen. The cumulative probability of having at least one false positive over 10 years of screening is unknown and should be a research priority. This probability could easily be as high as 30% (6) (or more) of all women. All screen-positive women subsequently undergo a “work-up,” which may be fine needle aspirate, ultrasound, or magnification views. Some will come to biopsy. We are only now beginning to appreciate the experience of women who face the burden of a false positive mammogram. Although more research is needed, it is clear now that many of these women will have marked anxiety in the days (sometimes weeks) between learning of their abnormal mammogram and being told that they do not have cancer. Some of these women will have continued anxiety months after being told that they do not have breast cancer (7). The experience of this large group of women should be a prime consideration in deciding whether to recommend screening. A related research priority should be to find ways to minimize the psychological trauma for false positive women. It is incorrect, however, to assume that this trauma can be erased completely by various interventions. It is entirely possible that at least some of this anxiety is inherent in the screening situation and in our current societal views of breast cancer. Women who are “true positive”—those who screen positive and are found to actually have breast cancer—are the women we usually think have been helped most by screening. Unfortunately, not all women whose cancer is detected by screening benefit from that detection. Breast cancer is a heterogeneous disease with a spectrum of natural histories (8). For our purposes here, we can simplify this spectrum into three distinct types. About 50% of true-positive women will not die from breast cancer, even if they are never screened and wait until later in life for their cancers to be detected. These cancers are slow growing and relatively treatable. Screening will not alter their natural history because their natural history is excellent. Still, the perception of many of these women, quite understandably, is that their lives have been “saved” by screening. Another type of breast cancer has an aggressive natural history and is difficult to treat. Women with this type of tumor, unfortunately, will die of breast cancer regardless of when it is found. These cancers metastasize at an early, undetectable stage. Again, the natural history of the disease is not altered by screening, and hence, there is no benefit to screening. The woman who has had an aggressive cancer found by screening has simply been made to live longer with knowledge of the diagnosis. One could argue that these women have been harmed, not helped, by screening. Finally, some cancers are more treatable when found earlier, and thus screening favorably changes their natural history. The screening trials help us estimate the number of women with this type of cancer. As shown in Table 11, the randomized screening trials indicate that somewhere between 10% and 25% of women who would have died of breast cancer have this type of cancer, i.e., that is more treatable if found early. The most recent Swedish data narrow this estimate to 15%-20%. This, then, translates into 1-2 lives extended per 10,000 women screened once (or, as noted above, 1,000 women screened annually for 10 years), or about one life extended per 5,000 mammograms. Ductal Carcinoma In Situ In addition to women who are true positive for invasive breast cancer, some will be found to have ductal carcinoma in situ (DCIS). The natural history of DCIS is unknown. Some, but likely not all, of these lesions will progress to invasive carcinoma. And when progression occurs, it may take many years (thus allowing opportunities for detection at a later age) (9). Understanding the natural history of DCIS and determining the characteristics of those lesions that will become clinically important as opposed to those that are actually “pseudodisease” (a pathologic finding that never produces clinical disease) should be a research priority. If, as we suspect, 50% or more of these lesions are clinically unimportant, then the potential for harming women by unnecessary treatment could be an important factor for women to consider in deciding about screening. Less Intensive Treatment One potential benefit of screening is the possibility that women whose cancers are found at an earlier stage will require less intensive therapy. Unfortunately, there are insufficient data to determine whether this theoretical benefit is real. Certainly many women with palpable tumors (not found by screening) are still eligible for lumpectomy rather than mastectomy. And many small, node-negative tumors (as well as DCIS) are being treated with surgery and either radiation or adjuvant chemotherapy. It is not clear whether increased screening has led to more or less intensive therapy for the population as a whole. Variation of Harms by Age As shown in Table 22, some of the potential harms of breast cancer screening vary with age. Because the sensitivity of mammography is lower for younger than older women, yet there are more total cancers among older women, the number of false negatives is similar in the different age groups. True positives are more frequent in older women, although for all women the number of true positives is small relative to false positives. The incidence of DCIS increases gradually with age, and thus we can expect that there will be slightly more women with this lesion in their fifties and sixties as compared with women in their forties. By far the largest group of women who may be harmed by screening is the false positive group. As noted earlier, this group may include as many as 30% of women in their forties screened annually for 10 years. A critical question, then, is whether the probability of a woman becoming a false positive varies by age. An important determinant of the probability of having a false positive is the initial “positivity rate” of screening—that is, the percentage of women screened who required some further work-up. There is conflicting evidence about whether this percentage varies with age. In some studies, especially those of academic practices (10), the positivity rate appears fairly constant with age. In studies of community practices (unpublished data, New Hanover Breast Cancer Screening Study, 1990; personal communication, Nancy Lee, M.D., from National Breast and Cervical Cancer Early Detection Program; personal communication, Bruce McCarthy, M.D.), younger women have higher positivity rates (and thus more false positives) than older women (2). The issue is important and should be a research priority. Even with the same positivity rate, however, the fact that the incidence of breast cancer is higher in older women means that more of the positives in younger women will be falsely positive. But there is another factor that makes it very likely that women in their forties have a larger—even a much larger—probability of a false positive than older women. This other factor is the frequency of screening. From the trials of women over 50, it appears that a large percentage of the benefit of annual screening can be obtained by screening biennially. For women in their forties, however, it is clear that if screening works at all, it must be done annually. Although there is need for research in this area, it seems likely that screening twice as frequently would produce a higher cumulative rate of false positive findings than screening biennially. The bottom line is that breast cancer screening is not the final answer to the problem of breast cancer in any age group. It certainly has benefits, however, among women ages 50 to 70 years, and, as shown by the Swedish studies, probably benefits as well for women in their forties. The benefit for women in their forties is delayed and small in terms of absolute number of lives extended per 1,000 women screened. Benefits gradually increase with age, and harms, flowing largely from the number of false positives, gradually decrease with age. Restating the Problem The problem, then, can be immediately appreciated. As they grow older, even well-informed women will naturally differ in their perceptions of the age at which the increasing probability of benefit outweighs the decreasing probability of harm. And policy makers will naturally differ in their evaluation of the age, on the population level, at which the increasing benefits of screening begin to outweigh the decreasing harms. Perhaps the disagreement should tell us something. We differ not because we disagree about what the evidence is, but rather because our values differ. There is no consensus about screening for women in their forties, nor should there be. This is a “close-call.” In such situations, women should be helped to participate in their own decisions. Some may question whether it is feasible for medical practices to help women understand the potential benefits and harms of screening, and to facilitate informed, shared decision making. Finding time for such discussions may be difficult in busy medical practices. A research priority should be to develop and evaluate “discussion aids,” such as videotapes, decision boards, and tailored brochures, as well as training of nonphysician staff, to help medical practices accomplish this task more efficiently and effectively. Some may also question whether many women will want to participate in such a decision, when their physicians do not make a strong recommendation. But the reason for not making a recommendation is not lack of information; it is rather the understanding of the issue as a “close call.” In such situations, women's values and perceptions will carry as much weight as the facts about the pros and cons of screening. Ideally, the patient herself should supply such information to the decision-making process. We need to better understand how women will react when encouraged to participate with their physicians (and other members of the medical staff) in a process of informed, shared decision making. Population Level This analysis has focused on the individual level, and the need to help individual women to participate in a process of individualized, informed decision making. One could also take a population perspective. Because of the relatively low risk of a woman dying of breast cancer diagnosed in her forties, a risk reduction of 15%-20% turns out to be a small absolute risk reduction for an individual. However, this number becomes larger if the 15%-20% is multiplied times the total number of women dying each year of breast cancer diagnosed during their forties. If, for example, about 5,000 women each year die of breast cancer diagnosed during their forties, then screening could conceivably extend the lives of 750-1,000 women each year (assuming 100% compliance with screening). Unfortunately, the harms (not to speak of the financial costs) are also multiplied. Many more than 1,000 women would face the trauma of a false positive mammogram; many would have a biopsy; many would be diagnosed with DCIS. Again, this appears to be a close call, even on the population level. Whether the decision is considered on the individual or population level, we should all be concerned by the lack of understanding of breast cancer risk and breast cancer screening by many American women. Several years ago, some colleagues and I surveyed women living in two eastern North Carolina counties, and found that worry about breast cancer was higher among women in their forties than women in their fifties and sixties. Less than 25% of women of any age understood that breast cancer risk increases with age (11). More recent surveys of nearly 4,000 women visiting primary care physicians found that over half of women in their forties overestimated their risk of breast cancer by a factor of three or more, and nearly half overestimated the benefit of screening mammography by a factor of at least 10 (unpublished data, North Carolina Prescribe for Health study, 1994). Others have found similar results (12). A blanket recommendation that all women in their forties be screened would not serve the cause of public or individual education about this issue. Furthermore, such a recommendation would be discordant with the weakness of the evidence that benefits outweigh harms. A more measured approach is needed. The recommendation for women in their forties should be that they be informed that there are pros and cons to being screened, and that reasonable women will disagree about whether to be screened. They should be encouraged to clarify their own values and then to discuss screening with their physicians, to participate in making an individualized decision. Then we should get to work on the real issue: how to efficiently and effectively reach all women with this discussion. Table 1. Benefits of screening Age   Risk per 1,000 women*   Relative risk reduction (%)   Absolute risk reduction†   40  7.8  16**  1.2      23††  1.8  50  12.9  15  1.9      30  3.9  60  19.5  30  5.9  ⩾70  25.3  30‡‡  7.6  Age   Risk per 1,000 women*   Relative risk reduction (%)   Absolute risk reduction†   40  7.8  16**  1.2      23††  1.8  50  12.9  15  1.9      30  3.9  60  19.5  30  5.9  ⩾70  25.3  30‡‡  7.6  * Rate of dying in next 15-20 years of breast cancer diagnosed in next 10 years, from SEER data, 1973-1980 and 1989-1991. † Number of lives ultimately extended per 1,000 by screening over the next 10 years. ** From Swedish meta-analysis. †† From Edinburgh trial, beginning with age 45 years. ‡ ‡Extrapolated from 60-69 age group. View Large Table 2. Harms of screening Type of finding   Harm   Relationship with age   False-negative  False reassurance  40 ≅ 50/60*  False-positive  Psychological trauma  40 > 50/60  True-positive  Living longer with knowledge of disease  50/60 > 40   No change in natural history  Pseudodisease   Labelling-psychological effects   50/60 > 40   Ductal carcinoma  Unnecessary treatment  Type of finding   Harm   Relationship with age   False-negative  False reassurance  40 ≅ 50/60*  False-positive  Psychological trauma  40 > 50/60  True-positive  Living longer with knowledge of disease  50/60 > 40   No change in natural history  Pseudodisease   Labelling-psychological effects   50/60 > 40   Ductal carcinoma  Unnecessary treatment  * Rate for women in their forties compared to rate for women in their fifties or sixties. (40 = women in their forties; 50/60 = women in their fifties and sixties). View Large Fig. 1. View largeDownload slide Extended lives: 2-6 each year (one in 1,700 to one in 5,000 screened). Reprinted with permission from (1). Fig. 1. View largeDownload slide Extended lives: 2-6 each year (one in 1,700 to one in 5,000 screened). Reprinted with permission from (1). Fig. 2. View largeDownload slide Extended lives: uncertain, but if relative risk reduction is same as women 50-70, then 1-2 lives will be extended (i.e., one in 5,000 to one in 10,000 screened). Reprinted with permission from (1). Fig. 2. View largeDownload slide Extended lives: uncertain, but if relative risk reduction is same as women 50-70, then 1-2 lives will be extended (i.e., one in 5,000 to one in 10,000 screened). Reprinted with permission from (1). References (1) Harris R, Leininger L. Clinical strategies for breast cancer screening: weighing and using the evidence. Ann Intern Med  1995; 122: 539-47. Google Scholar (2) CDC. Results from the National Breast and Cervical Cancer Early Detection Program, October 31, 1991-September 30, 1993. MMWR Morb Mortal Wkly Rep  1994; 43: 530-4. Google Scholar (3) Brown ML, Houn F, Sickles EA, Kessler LG. Screening mammography in community practice: positive predictive value of abnormal findings and yield of follow-up diagnostic procedures. AJR Am J Roentgenol  1995; 165: 1373-7. Google Scholar (4) Fletcher SW, Black W, Harris R, Rimer BK, Shapiro S. Report of the International Workshop on Screening for Breast Cancer. J Natl Cancer Inst  1993; 85: 1644-56. Google Scholar (5) Ries LA, Miller BA, Hankey BF, Kosary CL, Harras A, Edwards BK, editors. SEER Cancer Statistics Review, 1973-1991: Tables and Graphs, National Cancer Institute. NIH Pub. No. 94-2789. Bethesda (MD), 1994. Google Scholar (6) Elmore JG, Barton MB, Moceri VM, Fletcher SW. Cumulative risk of a false-positive mammogram over a 10-year period [abstract]. J Gen Intern Med  1997; 12 Suppl: 107. Google Scholar (7) Lerman C, Trock B, Rimer BK, Boyce A, Jepson C, Engstrom PF. Psychological and behavioral implications of abnormal mammograms. Ann Intern Med  1991; 114: 657-61. Google Scholar (8) Harris JR, Hellman S. Natural History of Breast Cancer. In: Diseases of the Breast. Harris JR, Lippman ME, Morrow M, Hellman S, editors. Philadelphia: Lippincott-Raven, 1996. Google Scholar (9) Ernster VL, Barclay J, Kerlikowske K, Grady D, Henderson C. Incidence of and treatment for ductal carcinoma in situ of the breast. JAMA  1996; 275: 913-8. Google Scholar (10) Kerlikowske K, Grady D, Barclay J, Sickles EA, Eaton A, Ernster V. Positive predictive value of screening mammography by age and family history of breast cancer. JAMA  1993; 270: 2444-50. Google Scholar (11) Harris RP, Fletcher SW, Gonzalez JJ, Lannin DR, Degnan D, Earp JA, et al. Mammography and age: are we targeting the wrong women? A community survey of women and physicians. Cancer  1991; 67: 2010-4. Google Scholar (12) Black WC, Nease RF Jr, Tosteson AN. Perceptions of breast cancer risk and screening effectiveness in women younger than 50 years of age. J Natl Cancer Inst  1995; 87: 720-31. Google Scholar Oxford University Press http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png JNCI Monographs Oxford University Press

Variation of Benefits and Harms of Breast Cancer Screening With Age

JNCI Monographs , Volume 1997 (22) – Jan 1, 1997

Loading next page...
 
/lp/oxford-university-press/variation-of-benefits-and-harms-of-breast-cancer-screening-with-age-iaQNOgYuIs

References (13)

Publisher
Oxford University Press
Copyright
Oxford University Press
ISSN
1052-6773
eISSN
1745-6614
DOI
10.1093/jncimono/1997.22.139
Publisher site
See Article on Publisher Site

Abstract

Abstract The critical issue in deciding whether to recommend breast cancer screening for women in their forties is to determine whether potential benefits are substantially greater than potential harms. Recent evidence from randomized clinical trials makes it likely that, after 10-12 years of follow-up, there is a real benefit from screening women ages 40-49, on the order of a 15-20% reduction in the relative risk of breast cancer death. This relative risk reduction translates into an absolute risk reduction of 1-2 women whose lives are extended from screening 1,000 women in their forties annually for 10 years (i.e., about one life extended per 5,000 mammograms). The absolute benefit of screening increases with age. Evidence about potential harms is less well established, but it is compelling that there are 15-40 times as many false positive as true positive mammograms (depending on the patient's age), and that at least some of the women with false positive mammograms have ongoing psychological distress as a result. Some 30% of all women who are screened annually during their forties will have at least one false positive mammogram and this probability likely decreases with advancing age. If the balance between benefits and harms is judged to be a “close call” for women in their forties, a blanket recommendation for all is inappropriate. Instead, each woman in her forties should be helped to understand the pros and cons of screening, to clarify her own values, and to consider with her primary care physician what decision would be best for her. Getting the Question Right The question I wish to address is what level of recommendation to make to women of different ages about breast cancer screening. I want to emphasize the phrase “what level of recommendation.” Some may think the answer is a simple “yes” or “no”—either we recommend or we don't. The strength of the recommendation, however, should depend on the strength of the evidence about two issues: the benefits of screening and the harms of screening. The real question, then, is not whether there is some small benefit demonstrated for screening women in their forties. The issue is larger than a “P-value.” What we need to know is where the balance lies between the magnitude of benefits and harms for different age (or other risk) groups. But what do we do in cases where the balance between benefits and harms is not clear, as I believe is the case with breast cancer screening for women in their forties? In these cases, there is a third option beyond recommending or not recommending. Physicians may also raise the issue of breast cancer screening with their patients, help them understand the benefits and harms, and encourage them to participate in making an individualized decision. My aim, then, is to provide an overview of the benefits and harms of screening for women in their forties, so that these women, with the help of their physicians, can make the most appropriate decision for themselves. Mortality Benefits of Screening Screening seeks to decrease the risk of dying of breast cancer, not the risk of getting it. The specific risk a woman is trying to reduce by being screened for the next 10 years is the risk of eventually dying of cancer diagnosed in those next 10 years. These risks for women of different ages, calculated from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) data before widespread screening, are given in the second column of Table 11 (1). Not surprisingly, the risk increases with age. To date, eight randomized trials of mammography screening among women aged 40 and older have been conducted in Sweden, the United Kingdom, Canada, and the United States. Mortality reduction in these trials is measured in terms of a “relative risk” reduction—that is, the reduction in risk of dying of breast cancer in a screened group relative to the baseline risk in an unscreened group. When the relative risk reduction from these randomized trials (Table 11, column 3) is factored in, we can calculate the absolute risk reduction (Table 11, column 4)—the number of women per 1,000 whose lives would ultimately be extended by screening over 10 years. The new evidence from the Swedish randomized trials makes it likely that there is a real benefit, and that it is on the order of a 15-20% relative risk reduction. If the relative risk reduction for women in their forties is 10-15%, then one woman would have her life extended for screening 1,000 women for 10 years. If the relative reduction for this age group is about 20%, then the lives of about two women would be extended for screening 1,000 women for 10 years (about one life extended per 5,000 mammograms). Table 11 illustrates that the benefit of screening—the number of women per 1,000 whose lives are extended—increases with age. Some have made the claim that the benefit of screening is the same for women in their forties as for those in their fifties or sixties. These claims have used the relative risk reduction as a measure of benefit. It is clear from Table 11, however, that even if the relative risk reduction is equivalent in different age groups (which is not at all certain from the trials), the absolute benefit in number of lives extended per 1,000 women screened increases with age. Effects of Screening on Nonmortality Outcomes Screening is a “double-edged sword” that can result in either benefits or harms. There is a need for more research on both nonmortality benefits and harms of screening. The potential magnitude of the effects, however, is apparent from examining the screening “cascade”—that is, the expected sequence of events following screening. This cascade is shown in Figs. 11 and 22 for a single screening of a hypothetical population of 10,000 women who are being screened regularly (i.e., “incidence” screens as opposed to “prevalence” screens). The figures assume a conservative mammogram “positivity” rate of 5%—that is, 5% of all cases will require further evaluation. This rate is indicative of many excellent mammography practices (2) and is much less than the 11% found in a recent national survey (3). Using the higher rate would double the number of false positives. Sensitivity of mammography is taken from average sensitivity in the trials (4), and incidence of cancer is taken from SEER data (5). Screen-Negative Women As seen in both figures, most women screened are negative. The great majority of these women are truly negative—they do not have breast cancer. A few, however, truly have cancer but are screen negative—that is, they are falsely negative. A research priority is to find out whether some of these women have been injured by false reassurance. It seems possible that some may ignore early symptoms of breast cancer because they have been reassured by the negative mammogram. We don't know how often this really happens. The true negatives would seem to be in a position to benefit; they could receive “peace of mind”—reassurance that they do not have cancer. But if you look carefully at the probability of having cancer before screening versus after negative screening—which for women in their forties is about 1.6 per 1,000 before screening and about 0.4 per 1,000 after a negative screen, a reduction in risk of about 1 per 1,000)—the difference doesn't seem large enough to make a truly objective woman change from worrying to relaxing. The woman was at low risk before screening and is still at low (but not zero) risk after being screen negative. Nevertheless, many women report peace of mind after a negative mammogram. This may reflect overestimation of initial risk and overinterpretation of a negative mammogram, and it suggests that we should develop ways other than mammography of reassuring women. Screen-Positive Women In many excellent mammography practices, about 5% of women are screen positive, and the great majority of these are falsely positive (i.e., they do not have cancer, despite the positive test). As shown in Figs. 11 and 22, there are 15-40 times as many false positives as true positives. And Figs. 11 and 22 are only for a single screen. The cumulative probability of having at least one false positive over 10 years of screening is unknown and should be a research priority. This probability could easily be as high as 30% (6) (or more) of all women. All screen-positive women subsequently undergo a “work-up,” which may be fine needle aspirate, ultrasound, or magnification views. Some will come to biopsy. We are only now beginning to appreciate the experience of women who face the burden of a false positive mammogram. Although more research is needed, it is clear now that many of these women will have marked anxiety in the days (sometimes weeks) between learning of their abnormal mammogram and being told that they do not have cancer. Some of these women will have continued anxiety months after being told that they do not have breast cancer (7). The experience of this large group of women should be a prime consideration in deciding whether to recommend screening. A related research priority should be to find ways to minimize the psychological trauma for false positive women. It is incorrect, however, to assume that this trauma can be erased completely by various interventions. It is entirely possible that at least some of this anxiety is inherent in the screening situation and in our current societal views of breast cancer. Women who are “true positive”—those who screen positive and are found to actually have breast cancer—are the women we usually think have been helped most by screening. Unfortunately, not all women whose cancer is detected by screening benefit from that detection. Breast cancer is a heterogeneous disease with a spectrum of natural histories (8). For our purposes here, we can simplify this spectrum into three distinct types. About 50% of true-positive women will not die from breast cancer, even if they are never screened and wait until later in life for their cancers to be detected. These cancers are slow growing and relatively treatable. Screening will not alter their natural history because their natural history is excellent. Still, the perception of many of these women, quite understandably, is that their lives have been “saved” by screening. Another type of breast cancer has an aggressive natural history and is difficult to treat. Women with this type of tumor, unfortunately, will die of breast cancer regardless of when it is found. These cancers metastasize at an early, undetectable stage. Again, the natural history of the disease is not altered by screening, and hence, there is no benefit to screening. The woman who has had an aggressive cancer found by screening has simply been made to live longer with knowledge of the diagnosis. One could argue that these women have been harmed, not helped, by screening. Finally, some cancers are more treatable when found earlier, and thus screening favorably changes their natural history. The screening trials help us estimate the number of women with this type of cancer. As shown in Table 11, the randomized screening trials indicate that somewhere between 10% and 25% of women who would have died of breast cancer have this type of cancer, i.e., that is more treatable if found early. The most recent Swedish data narrow this estimate to 15%-20%. This, then, translates into 1-2 lives extended per 10,000 women screened once (or, as noted above, 1,000 women screened annually for 10 years), or about one life extended per 5,000 mammograms. Ductal Carcinoma In Situ In addition to women who are true positive for invasive breast cancer, some will be found to have ductal carcinoma in situ (DCIS). The natural history of DCIS is unknown. Some, but likely not all, of these lesions will progress to invasive carcinoma. And when progression occurs, it may take many years (thus allowing opportunities for detection at a later age) (9). Understanding the natural history of DCIS and determining the characteristics of those lesions that will become clinically important as opposed to those that are actually “pseudodisease” (a pathologic finding that never produces clinical disease) should be a research priority. If, as we suspect, 50% or more of these lesions are clinically unimportant, then the potential for harming women by unnecessary treatment could be an important factor for women to consider in deciding about screening. Less Intensive Treatment One potential benefit of screening is the possibility that women whose cancers are found at an earlier stage will require less intensive therapy. Unfortunately, there are insufficient data to determine whether this theoretical benefit is real. Certainly many women with palpable tumors (not found by screening) are still eligible for lumpectomy rather than mastectomy. And many small, node-negative tumors (as well as DCIS) are being treated with surgery and either radiation or adjuvant chemotherapy. It is not clear whether increased screening has led to more or less intensive therapy for the population as a whole. Variation of Harms by Age As shown in Table 22, some of the potential harms of breast cancer screening vary with age. Because the sensitivity of mammography is lower for younger than older women, yet there are more total cancers among older women, the number of false negatives is similar in the different age groups. True positives are more frequent in older women, although for all women the number of true positives is small relative to false positives. The incidence of DCIS increases gradually with age, and thus we can expect that there will be slightly more women with this lesion in their fifties and sixties as compared with women in their forties. By far the largest group of women who may be harmed by screening is the false positive group. As noted earlier, this group may include as many as 30% of women in their forties screened annually for 10 years. A critical question, then, is whether the probability of a woman becoming a false positive varies by age. An important determinant of the probability of having a false positive is the initial “positivity rate” of screening—that is, the percentage of women screened who required some further work-up. There is conflicting evidence about whether this percentage varies with age. In some studies, especially those of academic practices (10), the positivity rate appears fairly constant with age. In studies of community practices (unpublished data, New Hanover Breast Cancer Screening Study, 1990; personal communication, Nancy Lee, M.D., from National Breast and Cervical Cancer Early Detection Program; personal communication, Bruce McCarthy, M.D.), younger women have higher positivity rates (and thus more false positives) than older women (2). The issue is important and should be a research priority. Even with the same positivity rate, however, the fact that the incidence of breast cancer is higher in older women means that more of the positives in younger women will be falsely positive. But there is another factor that makes it very likely that women in their forties have a larger—even a much larger—probability of a false positive than older women. This other factor is the frequency of screening. From the trials of women over 50, it appears that a large percentage of the benefit of annual screening can be obtained by screening biennially. For women in their forties, however, it is clear that if screening works at all, it must be done annually. Although there is need for research in this area, it seems likely that screening twice as frequently would produce a higher cumulative rate of false positive findings than screening biennially. The bottom line is that breast cancer screening is not the final answer to the problem of breast cancer in any age group. It certainly has benefits, however, among women ages 50 to 70 years, and, as shown by the Swedish studies, probably benefits as well for women in their forties. The benefit for women in their forties is delayed and small in terms of absolute number of lives extended per 1,000 women screened. Benefits gradually increase with age, and harms, flowing largely from the number of false positives, gradually decrease with age. Restating the Problem The problem, then, can be immediately appreciated. As they grow older, even well-informed women will naturally differ in their perceptions of the age at which the increasing probability of benefit outweighs the decreasing probability of harm. And policy makers will naturally differ in their evaluation of the age, on the population level, at which the increasing benefits of screening begin to outweigh the decreasing harms. Perhaps the disagreement should tell us something. We differ not because we disagree about what the evidence is, but rather because our values differ. There is no consensus about screening for women in their forties, nor should there be. This is a “close-call.” In such situations, women should be helped to participate in their own decisions. Some may question whether it is feasible for medical practices to help women understand the potential benefits and harms of screening, and to facilitate informed, shared decision making. Finding time for such discussions may be difficult in busy medical practices. A research priority should be to develop and evaluate “discussion aids,” such as videotapes, decision boards, and tailored brochures, as well as training of nonphysician staff, to help medical practices accomplish this task more efficiently and effectively. Some may also question whether many women will want to participate in such a decision, when their physicians do not make a strong recommendation. But the reason for not making a recommendation is not lack of information; it is rather the understanding of the issue as a “close call.” In such situations, women's values and perceptions will carry as much weight as the facts about the pros and cons of screening. Ideally, the patient herself should supply such information to the decision-making process. We need to better understand how women will react when encouraged to participate with their physicians (and other members of the medical staff) in a process of informed, shared decision making. Population Level This analysis has focused on the individual level, and the need to help individual women to participate in a process of individualized, informed decision making. One could also take a population perspective. Because of the relatively low risk of a woman dying of breast cancer diagnosed in her forties, a risk reduction of 15%-20% turns out to be a small absolute risk reduction for an individual. However, this number becomes larger if the 15%-20% is multiplied times the total number of women dying each year of breast cancer diagnosed during their forties. If, for example, about 5,000 women each year die of breast cancer diagnosed during their forties, then screening could conceivably extend the lives of 750-1,000 women each year (assuming 100% compliance with screening). Unfortunately, the harms (not to speak of the financial costs) are also multiplied. Many more than 1,000 women would face the trauma of a false positive mammogram; many would have a biopsy; many would be diagnosed with DCIS. Again, this appears to be a close call, even on the population level. Whether the decision is considered on the individual or population level, we should all be concerned by the lack of understanding of breast cancer risk and breast cancer screening by many American women. Several years ago, some colleagues and I surveyed women living in two eastern North Carolina counties, and found that worry about breast cancer was higher among women in their forties than women in their fifties and sixties. Less than 25% of women of any age understood that breast cancer risk increases with age (11). More recent surveys of nearly 4,000 women visiting primary care physicians found that over half of women in their forties overestimated their risk of breast cancer by a factor of three or more, and nearly half overestimated the benefit of screening mammography by a factor of at least 10 (unpublished data, North Carolina Prescribe for Health study, 1994). Others have found similar results (12). A blanket recommendation that all women in their forties be screened would not serve the cause of public or individual education about this issue. Furthermore, such a recommendation would be discordant with the weakness of the evidence that benefits outweigh harms. A more measured approach is needed. The recommendation for women in their forties should be that they be informed that there are pros and cons to being screened, and that reasonable women will disagree about whether to be screened. They should be encouraged to clarify their own values and then to discuss screening with their physicians, to participate in making an individualized decision. Then we should get to work on the real issue: how to efficiently and effectively reach all women with this discussion. Table 1. Benefits of screening Age   Risk per 1,000 women*   Relative risk reduction (%)   Absolute risk reduction†   40  7.8  16**  1.2      23††  1.8  50  12.9  15  1.9      30  3.9  60  19.5  30  5.9  ⩾70  25.3  30‡‡  7.6  Age   Risk per 1,000 women*   Relative risk reduction (%)   Absolute risk reduction†   40  7.8  16**  1.2      23††  1.8  50  12.9  15  1.9      30  3.9  60  19.5  30  5.9  ⩾70  25.3  30‡‡  7.6  * Rate of dying in next 15-20 years of breast cancer diagnosed in next 10 years, from SEER data, 1973-1980 and 1989-1991. † Number of lives ultimately extended per 1,000 by screening over the next 10 years. ** From Swedish meta-analysis. †† From Edinburgh trial, beginning with age 45 years. ‡ ‡Extrapolated from 60-69 age group. View Large Table 2. Harms of screening Type of finding   Harm   Relationship with age   False-negative  False reassurance  40 ≅ 50/60*  False-positive  Psychological trauma  40 > 50/60  True-positive  Living longer with knowledge of disease  50/60 > 40   No change in natural history  Pseudodisease   Labelling-psychological effects   50/60 > 40   Ductal carcinoma  Unnecessary treatment  Type of finding   Harm   Relationship with age   False-negative  False reassurance  40 ≅ 50/60*  False-positive  Psychological trauma  40 > 50/60  True-positive  Living longer with knowledge of disease  50/60 > 40   No change in natural history  Pseudodisease   Labelling-psychological effects   50/60 > 40   Ductal carcinoma  Unnecessary treatment  * Rate for women in their forties compared to rate for women in their fifties or sixties. (40 = women in their forties; 50/60 = women in their fifties and sixties). View Large Fig. 1. View largeDownload slide Extended lives: 2-6 each year (one in 1,700 to one in 5,000 screened). Reprinted with permission from (1). Fig. 1. View largeDownload slide Extended lives: 2-6 each year (one in 1,700 to one in 5,000 screened). Reprinted with permission from (1). Fig. 2. View largeDownload slide Extended lives: uncertain, but if relative risk reduction is same as women 50-70, then 1-2 lives will be extended (i.e., one in 5,000 to one in 10,000 screened). Reprinted with permission from (1). Fig. 2. View largeDownload slide Extended lives: uncertain, but if relative risk reduction is same as women 50-70, then 1-2 lives will be extended (i.e., one in 5,000 to one in 10,000 screened). Reprinted with permission from (1). References (1) Harris R, Leininger L. Clinical strategies for breast cancer screening: weighing and using the evidence. Ann Intern Med  1995; 122: 539-47. Google Scholar (2) CDC. Results from the National Breast and Cervical Cancer Early Detection Program, October 31, 1991-September 30, 1993. MMWR Morb Mortal Wkly Rep  1994; 43: 530-4. Google Scholar (3) Brown ML, Houn F, Sickles EA, Kessler LG. Screening mammography in community practice: positive predictive value of abnormal findings and yield of follow-up diagnostic procedures. AJR Am J Roentgenol  1995; 165: 1373-7. Google Scholar (4) Fletcher SW, Black W, Harris R, Rimer BK, Shapiro S. Report of the International Workshop on Screening for Breast Cancer. J Natl Cancer Inst  1993; 85: 1644-56. Google Scholar (5) Ries LA, Miller BA, Hankey BF, Kosary CL, Harras A, Edwards BK, editors. SEER Cancer Statistics Review, 1973-1991: Tables and Graphs, National Cancer Institute. NIH Pub. No. 94-2789. Bethesda (MD), 1994. Google Scholar (6) Elmore JG, Barton MB, Moceri VM, Fletcher SW. Cumulative risk of a false-positive mammogram over a 10-year period [abstract]. J Gen Intern Med  1997; 12 Suppl: 107. Google Scholar (7) Lerman C, Trock B, Rimer BK, Boyce A, Jepson C, Engstrom PF. Psychological and behavioral implications of abnormal mammograms. Ann Intern Med  1991; 114: 657-61. Google Scholar (8) Harris JR, Hellman S. Natural History of Breast Cancer. In: Diseases of the Breast. Harris JR, Lippman ME, Morrow M, Hellman S, editors. Philadelphia: Lippincott-Raven, 1996. Google Scholar (9) Ernster VL, Barclay J, Kerlikowske K, Grady D, Henderson C. Incidence of and treatment for ductal carcinoma in situ of the breast. JAMA  1996; 275: 913-8. Google Scholar (10) Kerlikowske K, Grady D, Barclay J, Sickles EA, Eaton A, Ernster V. Positive predictive value of screening mammography by age and family history of breast cancer. JAMA  1993; 270: 2444-50. Google Scholar (11) Harris RP, Fletcher SW, Gonzalez JJ, Lannin DR, Degnan D, Earp JA, et al. Mammography and age: are we targeting the wrong women? A community survey of women and physicians. Cancer  1991; 67: 2010-4. Google Scholar (12) Black WC, Nease RF Jr, Tosteson AN. Perceptions of breast cancer risk and screening effectiveness in women younger than 50 years of age. J Natl Cancer Inst  1995; 87: 720-31. Google Scholar Oxford University Press

Journal

JNCI MonographsOxford University Press

Published: Jan 1, 1997

There are no references for this article.