Some general principles may help the practitioner decide whether to screen average-risk young women using mammography.

Benefits and Harms of Screening Mammography: A Brief Summary

A meta-analysis of five Swedish randomized trials of screening mammography provides information about the effect of screening mammography on breast cancer mortality [3]. The combined cohort comprised 282 777 women aged 40 to 74 years. The five studies were similar except for the interval between mammograms, which ranged from 18 months to 33 months. The 12-year cumulative mortality from breast cancer was 3.9 deaths per 1000 enrollees in the screened group and 5.1 deaths per 1000 enrollees in the control group. The difference in these two mortality rates is the risk for death from breast cancer that is attributable to not being screened periodically for 12 years (1.2 deaths from breast cancer per 1000 women screened). The inverse of the attributable risk is the number of women aged 40 to 74 years who must be screened for 12 years to prevent one death from breast cancer: 883.

Because of the large size of the combined cohort, the investigators were able to analyze the effect of screening in different age groups. In screened women aged 50 to 59 years, the relative risk for dying of breast cancer was 0.71 (95% CI, 0.57 to 0.89); in screened women aged 60 to 69 years, this risk was also 0.71 (CI, 0.56 to 0.91). Mammographic screening reduced the risk for dying of breast cancer by approximately 30% in women aged 50 to 69 years, a result that is highly statistically significant. The situation is different in women aged 40 to 49 years. In this age group, screened women have a relative risk of 0.87 (CI, 0.63 to 1.20) for dying of breast cancer. Because the CI includes 1.0, the apparent 13% risk reduction is not statistically significant.

The meta-analysis is informative about the benefits of screening but not about its potential harms. A cross-sectional study of 30 814 women in northern California described the consequences of a first mammogram [4]. The proportion of abnormal mammograms was similar (6% to 8%) in women younger than 50 years of age and in women aged 50 to 69 years. The yield of the first mammogram was five times higher in women 50 years of age and older (10 cancers per 1000 studies compared with 2 cancers per 1000 studies). The probability of cancer after a positive mammogram was 4% in women aged 40 to 49 years, 9% in women aged 50 to 59 years, and 17% in women aged 60 to 69 years. The number of tests done per cancer detected after a positive mammogram was 48.3 in women 50 years of age or younger and 14.8 in women older than 50 years of age. Clearly, mammography is much more efficient at detecting breast cancer in older women.

Some general principles for using ambiguous data may help in deciding whether to screen; these principles can be applied to the decision to do mammography in average-risk young women.

Physiological Plausibility

General principles: If you are uncertain about the effectiveness of an intervention and cannot wait for definitive clinical evidence, biological thinking may help. If the effect of the intervention is unequivocal in one population but uncertain in another population, ask whether the two populations differ in a physiologic characteristic that is known to affect the efficacy of the intervention.

Application to screening mammography: Why would mammography in younger women produce more false-positive results and have little or perhaps no effect on breast cancer mortality? The breasts of younger women have less fat and are therefore more radiodense than the breasts of older women. This greater radiodensity means that small tumors are more difficult to detect before they have metastasized. In fact, the proportion of cancers that are detected by screening is lower in younger women than in older ones. Furthermore, it is harder to be sure that a mammographic image is normal and that there is no need for a follow-up mammogram or a biopsy. Finally, breast cancers are more likely to grow rapidly in younger women, so that early detection and treatment may be less effective in preventing metastasis.

Be Sure that Current Thinking is Based on Valid Evidence

General principles: The quest to make screening as cost-effective as possible means searching for subgroups of patients in whom screening is ineffective. A recent editorial provides a thoughtful approach to interpreting the results of studying subgroups of patients in a randomized trial [5]. The authors suggest a guideline for subgroup analysis: Trust differences in subgroup results only when the intervention works unambiguously in one group and fails utterly in another. Mistrust differences of degree in which the intervention is effective in all groups but more effective in some groups than in others.

Application to screening mammography: The effect of mammography on deaths from breast cancer in women younger than 50 years of age is derived from a subgroup analysis that invites caution in interpretation. Mammography was clearly effective in women aged 50 to 69 years, but it did not fail utterly in younger women. According to the suggested guideline for subgroup analysis, we should mistrust this difference. In fact, our best guess is that mammography in younger women has a small effect on breast cancer mortality. If this effect were statistically significant, which it is not, we should mistrust it because it represents a difference of degree rather than a qualitative difference. Clearly, we should resist the temptation to draw a strong conclusion about whether mammography is less effective in women younger than 50 years of age.

Costs Do Matter

General principles: An individual screening test has low cost and low yield. The annual incidence of breast cancer in women 40 to 44 years of age is 111 cases per year in 100 000 women [6]. On the other hand, the lifetime probability of breast cancer is 9.3% [6]. However, one must screen repeatedly to detect breast cancer at an early stage and, thus, a program of periodic screening has a relatively high cost in return for its high lifetime yield. Periodic screening for an entire population incurs very large costs.

Application to screening mammography: The probability of cancer after a positive mammogram is much lower in younger women, which means that many more follow-up tests, including biopsies, are required to diagnose cancer [4]. These figures translate into more costs for the health care system and more mental anguish for younger women who have screening mammography.

An analysis of the absolute reduction in the risk for death from breast cancer leads to the same conclusion. One must screen 883 women 40 to 74 years of age for 12 years to prevent one death from breast cancer, a Figure derived from the absolute reduction in the risk for dying of breast cancer (1.2 deaths from breast cancer per 1000 women screened).

The absolute risk reduction is the product of the relative risk reduction and the baseline risk for death from breast cancer, both of which are smaller in younger women than in the entire population of women 40 to 74 years of age. Therefore, mammography done in young women should have a much smaller effect on the absolute risk for death from breast cancer, and many more than 883 young women (approximately 2 1/2 times as many) must be screened to prevent a death from breast cancer.

Primum No Nocere

General principles: The Hippocratic corpus states that "As to diseases, make a habit of two things—to help, or at least to do no harm." In other words, when you are certain, act. Otherwise, move cautiously. Many believe that this principle has particular force when applied to healthy persons, who are by definition the persons being screened. It's easy to endorse the principle of avoiding policies that are known to cause net harm. But what about the situation in which the balance of harms and benefits is unknown? The Hippocratic corpus tells us that when we are in doubt we should take special care to avoid actions that might cause harm. Therefore, if a clinical policy has known harms and unknown benefits, we should be cautious. We should proceed with even greater caution if the benefits are not known exactly but are likely to be small.

Application to screening mammography: Mammography in younger women is costly, produces a high proportion of false-positive mammograms, and leads to mental anguish. Whether it confers benefit is uncertain, and the best estimate is that the benefit is small. The principle of primum no nocere argues strongly for caution in advocating routine mammography for average-risk women younger than 50 years of age.

Talk to the Patient

General principles: Persons differ in that which matters most to them. The only way to find out what matters to a person is to talk with her. Shared decision making, in which a well-informed patient participates in decision making as a full partner with the physician, is a natural consequence of asking the patient what matters most. Eddy [6] has suggested that the physician should actively solicit the patient's preferences when the decision to screen is a "close call." These are the circumstances in which the patient's preferences should regularly tip the balance.

Application to screening mammography: The efficacy of mammographic screening in younger women is clearly in doubt, and there is good evidence about its harms. The physician should bring up the topic of breast cancer screening, explain what is known of the harms and benefits of screening, and be sure that the patient understands. The issue for most women is not whether to have breast cancer screening but when to begin. The physician may wish to contrast the proven efficacy of mammography in older women with the uncertainty about its efficacy in younger women. Contrasting the consequences of screening in the two age groups Table 1 may also be useful.

Plan for the Usual; Adapt for the Unusual

General principles: All physicians have "clinical policies" that represent their approach to the "usual" patient. Guidelines, best practices, and algorithms are clinical policies that we apply to the usual patient and modify for the unusual patient. Choosing a clinical policy means thinking about cost and balancing harms and benefits. We should choose policies that, despite the possibility of harm in an individual patient, will lead to net benefit when applied to many patients. Despite the important role of clinical policies in patient care, patient care decisions should be made on an individual basis. When the needs of an individual patient suggest departing from a clinical policy, it is time to listen to those needs and act accordingly.

Application to screening mammography: I am uncertain about the benefits of mammography in women younger than 50 years of age, and I am concerned about the costs and the harms. Therefore, my clinical policy is to recommend screening mammography to average-risk women only when they reach 50 years of age. Three organizations that adhere to a high standard of evidence have issued guidelines that make the same recommendation [6-8]. I discuss screening mammography with female patients under the age of 50 years, and particular circumstances, such as a family history of breast cancer, early menopause, or the fears of the patient, may alter the care that I provide.