You wouldn’t know it from the pink billboards but questions about the benefits and risks of screening mammography have been ongoing in the medical scientific community for decades. No screening test has been studied more extensively, and study after study confirms that the vast majority of women (70 to 90%) do not have their lives lengthened as a result of routine screening. What’s more there’s a good chance they will be overdiagnosed (5 to 50 percent) and over-treated (20 to 30 percent), sometimes for conditions that are not life-threatening. None of this is news in the research community. Yet it wreaks havoc within the realms of advocacy, mass media, and public health.
An article in the Annals of Internal Medicine back in 1976 reviewed existing clinical and experimental data to discern whether screening mammography for entire populations made sense. At the time there was little data on the long-term effects of mammography or the inherent risks associated with radiation. It concluded that “promotion of mammography as a general public health measure was premature, that the possible benefits of mammography received more emphasis than its defects, and that the evidence that mammography alone played a significant role in the reduction of breast cancer mortality was weak and indirect.” That was 35 years ago. Since then there have been numerous studies of screening mammography, including randomized clinical trials and statistical modeling, all of which are designed to analyze the impact of routine screening on the primary end in mind–mortality reduction. What does the bulk of the data find? Mortality reduction as a result of routine screening is weak and indirect.
Seven major clinical trials were conducted between 1963 and 1982 that involved over half a million women. In 2000 the Lancet published the first systematic review of these trials, which assessed their quality in terms of agreed-upon standards for well-conducted and reliable research and also evaluated the findings. Researchers identified no trials with high-quality data and and two trials that were highly flawed (Edinburgh and New York). The remaining trials had either medium-quality (Canada and Malmö) or poor-quality data (Göteborg, Stockholm, and Two-County). When taking the methodology into account they concluded that there was no reliable evidence to justify mass screening of entire populations of women. These studies were then evaluated in greater detail and reported in the Lancet in 2001. The best conclusion was that mortality reduction due to routine screening was about 15 percent for women ages 50 to 69. The remaining 85 percent died at the same rate as unscreened women who developed breast cancer. Two additional meta-analyses of these seven clinical trials (2002 Humphrey review and the 2006 Armstrong review) also concluded that reductions in mortality resulting from mammography screening were quite variable, and fairly modest.
The Institute of Medicine reviewed breast cancer detection technologies too, releasing a report in 2001 called Mammography and Beyond. These reviewers found that screening outcomes had not been assessed adequately, which confused the potential value of routine mammograms in reducing deaths. They also identified several problems with the technology, such as the fact that mammograms are very difficult to interpret in women with dense breast tissue (something especially common in pre-menopausal women); that dense tissue interferes with the identification of abnormalities thereby leading to higher rates of false-positive results (i.e., those that do not, upon biopsy, show the presence of cancer) and also higher false negative results (i.e., those that miss tumors that actually are cancerous). They also found that routine screening frequently gives inconclusive results and that the ability of current screening tools “to determine the lethal potential of breast abnormalities is crude at best.” A followup symposium in 2005 reported similar concerns and reiterated that the reduction in the relative risk of dying from breast cancer was somewhere between 20 and 30 percent depending on age at screening.
The National Cancer Institute took up the mammography question in 2005, not through the testing of clinical effectiveness but through the use of high-level statistical modeling. The Cancer Intervention and Surveillance Modeling Network (CISNET) used retrospective surveillance data to create statistical models to illustrate past trends and project future trends related to screening and mortality. Reports based on CISNET estimated that routine screening reduced mortality between 16 and 22 percent. Based on the confluence of increased screening and improved treatment the researchers concluded that the overall decrease in breast cancer mortality from 1990 to 2000 (i.e., averaging about 20%) must have been related to mammography. There is no way to know for sure. However, statistical modeling for assessing risks and benefits is quite different from using clinical trials that involve real people in comparison groups who are observed over a period of time. Without contextual information about disease profiles, patient demographics, treatment regimens, and quality of care the association between routine screening and the overall decrease in mortality is, unfortunately, only a correlation.
In general there has been a consensus across studies that routine screening reduces the total breast cancer death rate somewhere between 15 and 25 percent. However, a Norwegian study reported in the New England Journal of Medicine (2010) found that mammograms combined with modern treatment reduced the death rate by only 10 percent, and that the effect of mammograms alone could be as low as 2 percent or even zero. This study only included a two-year followup, and must be viewed within its national context. Norwegians are insured under a universal, tax-funded, single-payer health system that is notable in terms of providing an important safety net for the population. The risk of dying of breast cancer is declining in Norway and the current rates are at the same level or lower than European averages. However, this study illuminates a critical insight: improvements in treatments may be responsible for improved outcomes for some groups of women regardless of when, or how, their tumors were found. Women over age 70 who were exposed to the program’s multidisciplinary medical teams but did not undergo mammography, for example, had an 8 percent reduction in breast-cancer mortality. This would explain why more recent clinical trials in advanced countries with improved treatment regimes are likely to find lesser or no effect on mortality due to screening.
None of this data has had an effect on U.S. society’s overzealous commitment to mammograms. Quarrels that focus on whether to start having them at age 35, 40, or 50 (though important in their own right) serve to detract attention from the fact that the overall reduction in mortality from screening is quite low. Compare the benefits of routine mammography with that of screening for cervical cancer. The Papanicolaou (Pap) smear has been called “the best screening tool ever introduced for any cancer.” It not only reduces the incidence of cervical cancer by 90 percent by detecting pre-cancerous conditions but has led to a 70% reduction in mortality in developed countries. An article in the American Journal of Clinical Oncology reports that among women who are screened regularly, the Pap smear may have reduced cervical cancer mortality by as much as 99 percent. Now that’s a screening tool that works!
The success of the Pap smear sets the bar high for breast cancer screening. So far, routine mammography screening hasn’t measured up. Yet studies abound to find out whether the existing technology benefits a small, or smaller, percentage of women. A study of two counties in Sweden found a mortality benefit of a whopping 30 percent. Early detection campaigners jumped for joy with statements that the study reconfirms that mammography saves lives. We already know that screening saves some lives, but that it doesn’t save most. What is being celebrated here?
There is much to consider about the Swedish study in the context of the mountains of data that already exist about screening. First, the findings from the Swedish study were not stratified by age so nothing can be said about the benefits of screening for different age groups of women. Second, the study did not systematically evaluate outcomes related to false-positive and false-negative results. Third, the methodology has been questioned previously in terms of the comparison groups and it is not clear that breast cancer deaths were counted correctly. Fourth, as Dr. H. Gilbert Welch has pointed out the study did not take into account the advances in breast cancer therapy over the last two decades. Finally, would the Swedish findings be generalizable to a population such as our own that does not have universal healthcare? If all of these issues could be rectified, we would still have a situation in which the vast majority of women routinely screened for breast cancer (in the Swedish study, 70 percent) would see no survival benefit due to screening.
Yes, routine screening does save some lives: About 1 in every 2000 women who are screened. That’s an individual stroke of luck, not a public health strategy. Instead of researching the same screening tool over and over again to look for minute differences in mortality benefits that continue to tell us what we already know, why not invest those research dollars into learning how to prevent cancer in the first place or keep those who are diagnosed from dying from the disease? [See also Kathi Kolb's essay on Betty Ford and the Status of Mammography.]
Added Note: The print advertisement below is a typical example within the breast cancer “awareness” realm. Featured in an October special issue, it leaves no doubt that routine screening is unequivocally good for women.
- It claims that mammograms are successful in finding 90 percent of breast cancers but includes no information about how many results are inconclusive, false-positives, or false-negatives.
- It suggests that that routine screening may “prevent and diagnose” though screening is clearly not a form of primary prevention.
- The ad even puts the onus on women themselves as being personally responsible for doing the “one thing” that might help them to beat breast cancer.
- It also relies on peer pressure to coerce women into doing what their friends are doing.
- This common advertisement embellishes and even fabricates information in support of its product (mammograms) while omitting real information that could help women to understand the risks, benefits, and limitations of screening technology.
Such strategies may be expected in advertising. But how different are they from everyday awareness campaigns and educational materials given to women about breast cancer???