| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Commentary |
1 Department of Radiology, Harvard Medical School; Department of Breast Imaging, Massachusetts General Hospital; Department of Radiology, Wang Ambulatory Care Center, Avon Foundation Comprehensive Breast Evaluation Center, 15 Parkman St., Ste. 240, Boston, MA 02114.
Received August 3, 2004; accepted after revision October 26, 2004.
Address correspondence to D. B. Kopans.
Most readers are not aware of the fact that medical journals may have an unspoken bias in the papers that they select for publication. These biases can have far-reaching consequences. The controversy over mammography screening for women ages 40-49 is a case in point. Several medical journals knowingly have refused to provide women with accurate information. Consequently, women have been led to believe that the age of 50 has some real significance with regard to screening when the fact is that it has none.
Background
Despite the fact that most major groups in the United States now support mammographic screening for women ages 40-49, there is still controversy about this policy, and many European countries as well as Canada do not encourage screening before the age of 50. What is not commonly understood is that the age of 50 is meaningless. It has no biologic or scientific importance. It is an arbitrary threshold that is the result of flawed data analyses. The age of 50 was chosen years ago as a surrogate for menopause. As a result, data began to be analyzed dichotomously using the age of 50 as the reference age. This resulted in making data that actually changed gradually with increasing age appear to change abruptly at the age of 50. These types of analyses were repeated, and the age of 50 was imbued with undeserved importance. Despite the fact that the data clearly show a benefit for screening beginning by the age of 40, because the spread of misinformation has gone on for years, writers still suggest that the benefit from screening before the age of 50 remains controversial [1]. I believe that, unfortunately, one of the main reasons for the continued controversy lies in the fact that a number of medical journals have had an undeclared biased publication policy that has prevented an accurate presentation of the data. These journals (New England Journal of Medicine, the Journal of the American Medical Association, the Journal of the National Cancer Institute, and the Annals of Internal Medicine), whose publications are monitored closely by the media and, consequently, are quickly disseminated to the public, appear to have a publication policy that will not permit a detailed explanation as to why screening for women ages 40-49 is beneficial. These journals at best have provided what they claim are "balanced reviews" that continue to suggest that the age of 50 represents a real threshold. At the height of the controversy (between 1993 and 1997), these journals published 19 articles that argued against screening for women ages 40-49 while they only published four articles in support, and two of the latter were limited by the editors [2, 3], while those writing in opposition to screening did not have the same limitations [4, 5].
Compounding the imbalance is the fact that one of these journals has perpetuated a misrepresentation that, apparently, cannot be corrected. Several years ago, the National Cancer Institute sold the rights and the title of the Journal of the National Cancer Institute (JNCI) to the Oxford University Press. Despite the fact that the JNCI no longer has any relationship to the National Cancer Institute, the reporters and the public with whom I have spoken are unaware of this and the articles in this journal continue to be reported and are believed, incorrectly, to have the imprimatur of the National Cancer Institute.
The undeclared journal biases have continued. Readers only have access to the articles that are accepted for publication. It is rare that the reader knows when an article has been refused for publication by a journal. Consequently, it is almost impossible to prove journal bias. The article following this commentary was submitted to the NEJM, JAMA, and the Annals of Internal Medicine. All three journals turned it down without any scientific criticism. When I challenged the New England Journal of Medicine and raised the concern that they were not providing the reader with an accurate representation of the issue, they cited a paper that they felt objectively addressed the issue [6]. A review of this article shows that it too perpetuates the false impression that the age of 50 has some real importance. Not only are the ethics of these undeclared publication biases questionable, but the fact that these journals correctly champion "informed decision making" makes the rejection of this information somewhat hypocritical. I appreciate the fact that the AJR has agreed to publish this commentary as well as the article following it. One might expect a specialty journal to publish articles that tend to favor that specialty, but the general medical journals have a different obligation, particularly those whose articles go quickly to the public. It may be impossible to prevent publication bias, but scientifically responsible journals should acknowledge that they might have a bias, just as authors are required to disclose potential conflicts of interest, so that the reader can be informed. My goal is not only to stimulate an open discussion of the basic issue (the age of 50 and breast cancer screening), but to permit an open discussion on the issue of bias and ethics in journal publications.
Informed Decision Making: Age of 50 Is Arbitrary and Has No Demonstrated Influence on Breast Cancer Screening in Women
OBJECTIVE. The article discusses the fact that for more than 20 years, there has been controversy concerning whether there is a benefit from mammographic screening for women ages 40-49. Based on a decision made many years ago to try to determine how menopause might influence the value of screening, the age of 50 was chosen as a surrogate for menopause. Despite the fact that there are no data to support this age as having any biologic relevance, it continues to be used as if the parameters and benefits of screening change abruptly at the age of 50.
CONCLUSION. The data clearly show that the age of 50 is nothing more than an arbitrary age, and it is far from clear whether menopause has any important effect on screening. The "harms" of screening have been cited as reasons for not recommending screening before the age of 50, but since the ratio of harm to benefit changes steadily with increasing age, there is no legitimate rationale for this determination. Women and their physicians should be informed that the age of 50 is an arbitrary threshold in terms of mammography screening so that they can make informed decisions about their own health care.
It is obvious that individuals should be provided with as much information about their health as is possible so that they may participate in and make informed decisions about their health care. This certainly is true for breast cancer screening. Even as more and more women are participating in breast cancer screening, concerns continue to be raised that they may not be aware of the possible "harms" caused by that participation [7, 8]. These include the anxiety that is associated with having a mammogram, the risk of having a false-positive examination that could lead to having additional mammographic evaluation, sonography, and even a breast biopsy for what turns out to be a benign lesion. It also is true that the potential deforming and toxic treatment for the cancers that are detected by screening may be of no benefit, the possibility that the cancers that are detected may be nonlethal [9], and the possibility that the deforming and toxic treatments may cause unnecessary morbidity. These are all legitimate concerns and women should be apprised of the possibilities so that they can make informed decisions as to whether to participate in screening.
One of the most debated issues that has been raised over the past decade is the question as to whether women ages 40-49 should be advised to be screened for breast cancer. In article after article in a number of the major medical journals, analysts repeatedly argued that there was no benefit from mammographic screening for women ages 40-49 [10-29]. Once it became clear that there was a benefit, it was argued that the data were so weak [6] that the benefit did not outweigh the harms. Women have been advised that they should defer their participation in screening until the age of 50. In some countries, screening before the age of 50 is not even offered [30]. In all of the articles that have argued against screening before the age of 50, there has been little if any mention that the age of 50 was chosen as an arbitrary threshold with no scientific support for that selection. In fact, there are no data showing that the age of 50 has any biologic significance. Furthermore, none of the ungrouped screening results change abruptly at age 50, or at any other age. The misapprehension is based on the fact that the data are continually grouped to analyze women younger than the age of 50 as compared with all women age 50 and over. The dichotomous analysis makes any changes that occur with increasing age appear to change suddenly at the age of 50. I have been unable to find any article in the major medical journals that counseled against screening for women ages 40-49, that also clearly alerted women and their physicians to this fact. Many authors who have repeatedly argued against screening before the age of 50 have correctly expressed the importance of informing women regarding the risks of mammography. Unfortunately, the same authors have neglected to inform women and their physicians that the age 50 is an arbitrary threshold.
Even a review of the Surveillance, Epidemiology, and End Results (SEER) database of the National Cancer Institute (NCI) shows the pervasiveness of the misunderstanding. The database provides breast cancer data for women in general, but the data also are broken down for women ages 49 and younger, and age 50 and older [31] as if the age of 50 actually had some demonstrable importance. A recent landmark study that evaluated the death rate from breast cancer in the populations of two counties in Sweden demonstrated a 44% decrease that occurred as a result of the institution of wide-spread mammography screening in the two counties [32]. The investigators showed that this benefit was not just for the total group ages 40-69, but that there was a 48% decrease in the breast cancer death rate when they evaluated women ages 40-49 separately. In reality, there was no good scientific reason for the investigators to stratify the data by age in this fashion, but this was done as a result of the continuing debate as to whether there was a benefit from screening women before the age of 50.
Obviously, thresholds often are necessary, but when a threshold is used that is not based on science, it should be made clear that it is either arbitrary or based on other considerations, for example, cost versus benefit. There may be well-defined reasons for a threshold. Few argue, for example, that women should begin mammography screening before the age of 40. I believe that the reason for the general acceptance of this threshold is that most agree that the only valid proof of a benefit from screening mammography must come from randomized, controlled trials. Except the Gothenberg trial in which women age 39 participated, none of the other randomized, controlled trials of screening included women under the age of 40. Consequently, it would be difficult to urge screening before the age of 40 since there are no properly performed, scientific studies to support it. In fact, the age of 40 is also a somewhat arbitrary choice made by the trialists. Randomized, controlled trials are very expensive to perform. The cost increases with every person who is enrolled. Breast cancer screening trials require that there be sufficient numbers of women in the trial so that there will be a sufficient number of women who are diagnosed with cancer during the trial and sufficient numbers of women who die from breast cancer so that a decrease in deaths among the screened group will be statistically significant. Women under the age of 40 were not included in the trials because the incidence of breast cancer is so low among these women that it was not considered to be a large enough problem to justify screening, and their inclusion would have necessitated much larger trials, making the cost prohibitive.
The age of 50 became a threshold for far different reasons. It was, initially, chosen as a surrogate for the age of menopause. Based on the obvious influences that hormones have on the development of breast cancers, investigators assumed that menopause would influence the results from screening. Since it is difficult exactly to define the age of menopause, the age of 50 was chosen, years ago, as the average age of menopause. Unfortunately, although many researchers now recognize that the age of 50 is arbitrary, and that the influence of menopause on screening is, at best, very small and unclear, women and their physicians have been led to believe that the age of 50 and menopause have some true relevance for screening, when none has been shown. This has been reinforced by the fact that paper after paper [16-34] evaluated screening data by artificially dividing the data and combining the results for women ages 49 and younger, as if they were a uniform group, and comparing them to the combined data for women ages 50 and over, as if they too were a uniform group. In fact, neither group is uniform, but this type of dichotomous comparison has made it appear as if there are changes that take place at the age of 50, when there are none. By using a similar dichotomous age grouping, it could be shown that the hairs on our heads suddenly turn gray at the age of 50 (or any other age around which the data are grouped).
The age of 50 continues to be used as the age at which some health planners and physicians continue to urge women to begin mammographic screening. This is surprising in that there have never been any data pertaining to screening that show any factors which change abruptly at the age of 50, or for that matter, any other age from 40 and up [33]. Some would argue that many screening guidelines now support beginning by the age of 40, so that the point is moot. This does not mean the truth should remain hidden. Certainly those who argue in favor of the importance of providing women with all the information about mammographic screening should state clearly for women and their physicians that the age of 50 is arbitrary and that the arguments suggesting it has relevance are artifacts of data grouping.
Randomized, Controlled Trials Were Not Designed to Evaluate Women Ages 40-49 Separately
The most reliable data on the benefits of mammographic screening come from the randomized, controlled trials (RCT). However, except for the National Breast Screening Study of Canada (NBSS), none of the RCTs was, prospectively, designed to permit retrospective stratification by age. Although promoted as a trial to evaluate women ages 40-49, the NBSS actually was not well designed to evaluate these women. In the first place, it was underpowered [34, 35]. Had the Canadian trial gone according to plan, it could only have shown a statistically significant mortality reduction of 40% or larger [36] when a 25-30% mortality reduction was more realistic. In addition to being underpowered, the execution of the trial was significantly flawed. In an effort to meet its hoped-for power, women with advanced breast cancer, who could not benefit from screening, were allowed to participate in this screening study. This bad decision was compounded by the fact that the NBSS violated one of the very basic rules for performing an RCT. The allocation process was not blinded. The randomization was on open lists. This made it possible for specific participants to be placed, selectively, in the screened group or the control group simply by skipping a line. No RCT today would permit such an open randomization process for the obvious reason that truly random allocation too easily could be compromised and the study biased. Adding to these two major errors, all the women in the trial had a clinical breast examination before allocation. Because women who had palpable abnormalities consistent with advanced cancer were allowed to participate in the trial, because the study design identified them before allocation, and because the allocation was on open lists, it was possible to place them, nonrandomly, in the screened group. A review of the allocation process appeared to show no evidence of compromise by looking for erasures on the allocation lists [37], but because the lists were open and the women were not allocated until after the clinical breast examination, erasures were unnecessary. One only had to skip a line to ensure that someone with a worrisome lesion was allocated to screening [38]. There certainly was the opportunity to subvert the randomization process and this has been suggested by the data [10, 39]. Not only was there a statistically significant excess of women with advanced cancers who were allocated to screening, and an excess of women with lymph-node-positive cancers who were assigned to the screened group, but the actual breast cancer survival in the control group was so much better than expected that an allocation imbalance was suggested. These factors have raised concern about the validity of the NBSS results. In addition, control group contamination by women who were supposed to be unscreened controls, but went outside the program and had screening mammograms, further weakens the power of the trial, not to mention the issue of the poor quality mammography that was demonstrated in a blinded review [40, 41], and corroborated by the study's own reference physicist [42]. Consequently, the data are unreliable from the only trial that was purported to be able to analyze the benefit of screening for women ages 40-49.
None of the remaining trials was designed to permit legitimate, retrospective subgroup analysis of women ages 40-49. I have used the term "legitimate" because unplanned subgroup analysis of data that lack statistical power is commonly performed to determine what the next research goal should be (for example, subgroup analysis of data on the drug raloxifene, which was being studied for the prevention of osteoporosis, suggested that it might reduce the risk of breast cancer). This led to the STAR Trial, but this kind of analysis cannot be used, legitimately, to make health care recommendations. If unplanned subgroup analyses using data that lacked statistical power were legitimate for making recommendations, then, reductio ad absurdum, trials would only need two participants. When the RCTs have been analyzed, as they were designed, from the early analyses to the most recent, they have demonstrated statistically significant mortality reduction for women ages 40-74 [43, 44]. Those who have used retrospective stratification by age never have justified this on a scientific basis. Clearly, if statistical power was not important, then trials could be much smaller and less expensive. Experts have warned repeatedly that the use of data lacking statistical power can be very misleading [45, 46], yet this basic factor has been ignored in the analyses of the RCTs of breast cancer screening.
Confusion arose when investigators evaluated women ages 40-49, separately, and required a statistically significant decrease in deaths for these women in the first 5 years of follow-up after the start of the trials [4]. What some analysts failed to appreciate, despite the fact that we presented it in 1993 [47] and published it in 1994 [48], was that all of the trials put together lacked the statistical power to show a statistically significant 25% mortality reduction for screening women ages 40-49 because they were not designed to be able to show this in the early years of follow-up. Analysts had required the retrospectively stratified data to provide a result that was mathematically impossible to achieve. The trials involved fewer than half of the nearly 500,000 women ages 40-49 that would be needed to show a 25% mortality reduction with statistical significance within 5 years of the start of the trials. Although the trials actually showed a benefit, the benefit in the early years of follow-up could not achieve significance because of this lack of power. The controversy arose because some analysts, ignoring the science, used the lack of statistical significance to advise against screening for women ages 40-49 [5]. Despite repeated warnings, ignoring these basic scientific facts led to major misinterpretations that persist today.
Despite the fact that the trials were never designed to permit separate legitimate subgroup analysis of women ages 40-49 in the early years of follow-up, with longer follow-up, and with the larger number of women/years, the benefit that had been evident for screening women ages 40-49 became statistically significant [49]. Unfortunately, instead of recognizing the fallacy of the original analyses, analysts went on to argue that because the benefit was "delayed," it must be weaker for women ages 40-49 than the results for women ages 50 and over [35]. The benefit for older women seemed to appear almost as soon as screening began. It is not surprising that the absolute benefit (number of women per 1,000) might be lower for women below any age that is chosen than it is for women above a chosen age, since the incidence of breast cancer increases steadily with increasing age This argument, however, is not unique to the age of 50. It applies to any age that one might chose. The absolute benefit from screening for women under the age of 62 is less than the absolute benefit for women ages 62 and over. The relative benefit (percent reduction in deaths), however, is at least as great if not greater for women in their 40s as for those age 50 and over. What most opponents of screening before the age of 50 fail to tell women and their physicians is that despite the fact the trials were not powered to permit retrospective subgroup analyses, in the Gothenberg trial there was a 44% statistically significant mortality reduction for women under the age of 50 [50], and in Malmö there has been a statistically significant mortality reduction of 35% for women screened under the age of 50 [51]. These are greater relative reductions than have been reported for women ages 50 and over. The Swedish trials combined have a 29% statistically significant mortality reduction for women ages 40-49 [52], and combining all of the population-based trials, there is a 26% statistically significant benefit for women ages 40-49 [15]. Adding the Canadian study, with all of its problems, there still is a significant 16% benefit [15].
Those who have argued that the benefit for women ages 40-49 is weaker because it is "delayed" and did not appear in the first 5 years after the start of screening have overlooked the Malmö II trial [53] in which the benefit began almost immediately for younger women. They also have failed to explain how an immediate benefit could occur. It is clear that one of the reasons why randomized, controlled trials are necessary is due to the phenomenon of length bias sampling. Trials of periodic screening are more likely to interrupt moderate and slower-growing cancers (that do not kill for many years) than they are to interrupt the fast-growing tumors that kill quickly [52]. Consequently, it would be difficult to explain how an immediate benefit could occur from periodic screening since women have to die in the control group before a benefit can be seen in the screened group. Nevertheless, analysts have repeatedly ignored this and, unsupported by science, used the "delayed" benefit seen among younger women as an argument against screening them.
In 1995, de Koning, based on breast cancer modeling and a published mortality reduction of 10%, wrote that the benefit in younger women could be explained by the possibility that women who were under the age of 50, at the time of allocation, reached the age of 50 during the trials, and their lives were saved as a result [29]. When he was provided with updated information at the Consensus Development Conference held by the National Institutes of Health (NIH) in January 1997, showing that the mortality reduction in the Swedish data was closer to 20%, he agreed that much of the benefit might actually be due to screening before the age of 50. Unfortunately, although he has reaffirmed this to me personally he has just recently received the actual data and, consequently, has not yet published a revised analysis [personal communication from Harry de Koning, February 19, 2004]. Consequently, the original analysis continues to be cited [35, 54].
Data Grouping Can Be Misleading
Not only have recommendations that were based on inappropriate retrospective subgroup analyses of data lacking statistical power led to a major misunderstanding, but this also has been reinforced by the repeated and misleading grouping of data in other analyses. The data for women ages 49 and younger constantly have been grouped together as if they are a uniform population and the same has been done for women ages 50 and over. There is little question that it frequently is necessary to group data for purposes of statistical analysis since there may be insufficient numbers of individuals to permit analysis when smaller increments are used. Grouping by decade or 5-year increments might have caused less confusion, but in the analysis of breast cancer screening data, many researchers have, until recently, compared just two groups: those ages 49 and younger to those ages 50 and over. This has made the data, which actually change gradually with increasing age [39], appear as if they change abruptly at the age of 50. This was first apparent in evaluations of the tissue density of the breast based on mammograms. Investigators were misled into thinking that the breast tissues were radiographically dense until the age of 50 (or menopause) and then they turned to fat and became radiolucent, making cancer easier to detect. This is, in fact, far from reality. The majority of women in their 30s do, indeed, have dense breast tissue, but the percentage changes gradually and steadily with increasing age with no abrupt change at the age of 50 or at menopause [53, 55]. We have confirmed this in a review of more than 150,000 mammographic studies (unpublished data). There are many young women with radiolucent breasts and many older women who have dense breast tissue and nothing happens abruptly to the radiographic density of the breast at the age of 50, or at menopause for most women.
Inappropriate age grouping has been used to suggest that the detection rate of breast cancer changes at the age of 50. In a study that has been cited repeatedly, the authors compared the breast cancer detection rate for women ages 30-49 to all women ages 50 to 79 [15]. It is surprising that the authors even included women in their 30s since these women were not being considered for routine screening. The inclusion of these younger women, however, further skewed the analysis. Because of the very low incidence of breast cancer among women in their 30s, when the data were combined, their inclusion lowered the overall incidence for the younger women just as the higher incidence among women in their 60s and 70s elevated the number for ages 50 and above. The authors argued that since the cancer detection rate was only two per 1,000 for women ages 30-49, while it was 10 per 1,000 for women ages 50 and over, screening should be concentrated on women ages 50 and over. This conclusion was repeated by others [56], reinforcing the misinformation. In fact, the actual data contained in the paper showed a gradual increase in cancer detection rates with no abrupt change at the age of 50. If women ages 40-49 had been compared with those ages 50-59, the cancer detection rates were much closer, and not significantly different. The detection rates, as one might expect, paralleled the prior probability of cancer in the population, which increases steadily with increasing age. It was only the dichotomous analysis of the grouped data that made the results appear to shift suddenly at the age of 50. This artificial stratification reinforced the idea that the age of 50 had some relevance when the data actually showed that it had none.
Our own review of recall rates from screening, recommendations for biopsy, and cancer detection rates showed that, if these percentages are evaluated by individual age without grouping, there is no abrupt change at age 50, or any other age, in any of the factors [57]. Our results showed that the recall rate decreased gradually from approximately 8% at the age of 40 to approximately 6% by the age of 79. The recommendation for a biopsy based on an abnormal mammogram was the same, regardless of age, and the positive predictive value and cancer detection rates increased steadily with increasing age, reflecting the prior probability of breast cancer in the population. I am unaware of any data from any of the screening trials that, if analyzed without grouping the data, would show any change that occurred suddenly at the age of 50. The use of the age of 50 is arbitrary and meaningless when it comes to mammography screening.
As noted earlier, the anxiety engendered by screening, the false-positive mammograms, biopsies with benign results that have resulted from false-positive mammograms, and treatment for possible "pseudocancers" that might never have been lethal if left undiscovered, are considered the "harms" of screening. If these women had not been screened, they would not have had these events happen. Because none of these parameters changes abruptly at any age, and the ones that do change with age change steadily and gradually, the harm-benefit ratio is a continuum and there is no more reason to pick one age as a threshold, than there is another.
Age grouping also has been used to trivialize the amount of breast cancer among women ages 40-49. In 1993, it was argued that since only 16% of breast cancers were diagnosed among women in their 40s, screening should concentrate on women ages 50 and over who accounted for 84% of the cancers. Authors who used this argument failed to point out that the same could be said for women ages 50-59 since they "only" accounted for 17% of the cancers in that year. Clearly, the incidence of cancer increases steadily with age, but the absolute number of cancers is related to the number of women at each age. In fact, in 1995 there were more total breast cancers diagnosed among women in their 40s than among women in their 50s due to the large number of women in their 40s.
Conclusion
Unfortunately, despite the legitimate calls for informing women of the facts, one fundamental fact has been ignored, and women and their physicians continue to be told that something happens at the age of 50 with regard to breast cancer screening. This is simply false. It is incumbent on those who choose to use the age of 50 as a threshold to provide ungrouped data to justify its use as anything but arbitrary. An age threshold may still be justified for other reasons, but if it is an arbitrary threshold, it should be clearly stated that there are no data to support its use other than the belief of the analyst, and women and their physicians should be alerted to this fact.
The controversy began by selecting the age of 50 to test a hypothesis. It was perpetuated by the scientifically inappropriate use of retrospective, unplanned subgroup analyses of data that lacked statistical power, and received additional support from the misleading, dichotomous comparison of women ages 40-49 (and younger) with women ages 50 and over. This led to the false belief that the age of 50 had some real significance. The very basis for not recommending screening for women ages 40-49 is scientifically unsupportable. Some may feel that they do not think the risk-benefit ratio is sufficient to urge screening for these women. This certainly is their prerogative, but women should be given accurate data for specific ages (not grouped) so that they can make this decision for themselves. Furthermore, I believe that important observations about breast cancer also have been overlooked because the data have been grouped and analyzed dichotomously. Whenever possible, analysts should use ungrouped data to avoid masking important observations.
If it is decided for economic reasons, for example, that it is too costly to screen before a certain age, then women should be so informed. This would provide them with an opportunity to discuss the choice of the threshold. Suggesting that a threshold is based on science when the fact remains that it is not, however, is an illegitimate method of avoiding a controversial discussion. In the future, authors should be required to avoid the use of misleading data grouping. Those who must group their data should alert the reader as to how this might influence the interpretation of their results. Anyone advising women and physicians to begin screening at the age of 50 should alert them that this is a purely arbitrary threshold. If screening is not recommended for women in their 40s, it should be explained that this is primarily based on economic considerations or someone's decision that the harm-benefit ratio does not, in their mind, become favorable until that age. In this case, the ratios should be provided by no more than 5-year groupings so that women can see if they change dramatically, or are merely a steady change. Women and their physicians should not be told that the analyses that suggest that the results of screening are weaker for women ages 40-49 are based on science since the true scientific evidence shows that the relative screening benefit for women in their 40s is the same or greater than for those age 50 and above. Women should be provided with accurate information so that they can make informed decisions.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  D. B. Kopans Screening Mammography for Women Age 40 to 49 Years Ann Intern Med, November 20, 2007; 147(10): 740 - 741. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
