In November 2009, the US Preventive Services Task Force (USPSTF) released new guidelines for breast cancer screening, which recommended against mammography screening for women aged 40-49 and advised against teaching breast self-examination. However, it avoided condemning clinical breast examination (CBE) by primary care physicians, stating that the data were insufficient to recommend for or against. What does this mean for women in that age bracket?
The case against the recommendation: Dr Daniel Kopans
The USPSTF guidelines have condemned a scan that has been proven to save lives and perpetuated CBE, a test that has no value, except when the background detection is of cancers larger than 2cm.
This is despite decades of data showing that, of the three tests, CBE has the least chance of detecting an early cancer. Without mammography most women will still detect their own cancers. If they are not screened using mammography, and are not to examine themselves, then this recommendation takes women back to before 1960 when many waited until a lump was impossible to ignore and then sought help, often too late.
USPSTF representatives, appearing before Congress, stated that their intent was not to deny women access to screening, but to urge that they consult with their physicians to decide whether the benefits of screening outweighed the ‘harms’: the anxiety of having a test, breast compression, false positives and needle biopsies. This is revisionism at its worst. The written guidelines – the only ones that carry any weight – recommend against. They do not say “we recommend that you discuss this with your doctor”, and ignore the goal of informed decision making.
Consider the evidence
The major problem with the USPSTF guidelines is that they ignore or discount the scientific evidence. This goes back to 1993 when the National Cancer Institute (NCI) set an unsupportable requirement that a statistically significant decrease in deaths had to be shown for women aged 40-49 in the randomised controlled trials (RCT) of screening when analysed as a separate subgroup within five years of screening commencing.
Since the trials were not designed to evaluate women aged 40-49 separately, it was a mathematically impossible requirement to fulfil. This retrospective and unplanned subgroup analysis lacked any statistical power to permit using it to make medical recommendations. Nevertheless, it was used by the NCI to drop support for screening women aged 40-49 and to support screening for women aged 50-plus every two years. This review, and others, ignored the fact that when RCTs are analysed as intended they always showed a statistically significant benefit from screening women beginning at the age of 40.
In 1997, the NCI convened a conference to review its guidelines and new data from Swedish trials, which showed that screening had statistically significant benefits for women in their forties with a longer follow-up. Breast cancer deaths among women aged 39-49 decreased by a statistically significant 44% in Gothenberg. Significant decreases were also seen in Malmo (35%) and the five Swedish trials had a statistically significant decrease of 29%. When the seven population-based trials were evaluated the decrease in deaths was a significant 26%.
Even adding the Canadian National Breast Screening Study, which used an unblinded allocation process that biased the trial against screening from the start, there was a 15% decrease in deaths.
Donald Berry, co-chairman of the panel, told the media that there was insufficient evidence to recommend screening for women ages 40-49. It was falsely reported that the panel had unanimously agreed this point. One panel member resigned in disagreement, and two others wrote a dissenting opinion that recognised the benefits.
The USPSTF suggested that there was new data that justified its updated review. In fact, the only important new data that has become available since 1997 are national statistics that show that, as more women participate in mammography screening, the breast cancer death rate continues to decrease.
The task force also appears unaware of the fact that there is no scientific support for using the age of 50 as a threshold for mammography screening. None of the parameters of screening, such as recall rates, recommendations for biopsy or cancer detection rates, change abruptly at any age; they occur gradually with increasing age.
The age of 50 has been imbued with unsupportable importance by grouping the data dichotomously, comparing women aged 49 and younger with those aged 50 and over. This makes the gradual changes appear to suddenly jump at the age of 50. In some cases this has been amplified by grouping women in their thirties with women in their forties, which pulls down the breast cancer detection rate. Ungrouped data showed no sudden jump.
Although the USPSTF evaluated women in three groups (aged 40-49, 50-59 and 60-69) the effect of grouping is the same. The data appear to change abruptly at age 50, and again at age 60. In fact, there is no significant difference between a woman aged 49 and one who is 51. The USPSTF has no scientific support for using the age of 50 as anything but an arbitrary threshold with regard to mammography screening, yet it has not informed women or their physicians of this.
Population studies in Sweden and the Netherlands have shown that the majority of the decrease in deaths is due to mammography screening, with a much smaller component due to improved therapies. The USPSTF ignored these directly measured data, instead accepting computer modelling to determine what proportion each contributed to the decrease.
Computer models provide results that are based on the assumptions programmed into them, and the ones used by the task force did not agree with each other; estimates for benefits attributable to mammography screening ranged from 23% to 65%.
These were the same models used by the task force to arrive at their conclusions. The USPSTF has refused to explain why it ignored the data from Sweden and the Netherlands.
Counting the cost
Although the USPSTF charter prohibits cost / benefit analyses, the USPSTF circumvented this prescription by using the number needed to screen (NNTS) to save one life. It compared women in three groups – 40-49, 50-59 and 60-69 – using estimates of the NNTS for each group, which in reality gradually increases with age, and the 15% decrease in deaths. Its computer models estimated that approximately 1,900 women in their forties needed to be screened to save one life.
The USPSTF decided that this was too many, but an NNTS of 1,300 for women in their fifties was acceptable. Had they used a conservative decrease in deaths of 30%, the number derived from actual data, the NNTS for women in their forties would be reduced to 950, well within the range that included women in their fifties.True cost/benefit analyses must take into account other factors such as the years of life saved. More than 40% of the years of life lost due to breast cancer are among women diagnosed in their forties.
Mammography is one of the major medical advances of the past 50 years. Inexplicably, it continues to be attacked despite the fact that virtually all critiques have been addressed scientifically. No one has ever suggested that it is the ultimate answer to breast cancer; it does not find all cancers and even when it does it may not be early enough to effect a cure.
While we await a cure, or a safe method of preventing breast cancer, we should not deny women access to mammography screening. The USPSTF guidelines are not based on the scientific evidence and, as USPSTF admits, will lead to thousands of unnecessary deaths if adopted. They should be revised or withdrawn.
Defending the recommendation: Dr Ned Calonge
In light of such stinging criticism, Dr Ned Calonge, USPSTF chair, directly responds to Dr Kopans’ concerns.
The USPSTF suggested that there was new data that justified its updated review. The only important new data that has become available since 1997 are national statistics that show that, as more women participate in screening, the breast cancer death rate continues to decrease.
This statement is incorrect. The AGE trial was designed to specifically evaluate the benefit of mammographic screening in women aged 40-49 and was published in 2006, well after the 2002 USPSTF recommendation update.
Also, the 2009 Nelson evidence review included the first-ever analysis of the performance of mammography – including detection rates and rates of false positives – from the Breast Cancer Surveillance Consortium database, which includes data from seven million mammograms in 2.5 million US women. The 2009 CISNET modelling study also provided important new information used by the USPSTF in its recommendation.
There is no scientific support for using the age of 50 as a threshold for mammography screening.
The USPSTF did not use the age of 50 as a threshold for mammographic screening. What the USPSTF actually recommends is that women under 50 should understand the risks and benefits of screening and make their own decision, in consultation with their clinician.
has never used cost in considering a recommendation. NNTS is an estimate of the benefit of the test.”
The risk of breast cancer is less in women under 50 and increases incrementally each year in the 40-49 age interval and beyond. Similarly, the rate of false positives is highest for women aged 40 and decreases incrementally as they grow older. The USPSTF concluded that the small (but real) mortality benefit in women under 50, compared with the dramatic increased risk of false positives and all the associated health harms, should prompt a more deliberative and informed discussion and decision for women in this age interval.
The patient’s choice should be supported and the USPSTF members believe that women choosing to start mammography earlier should have these studies covered by insurance. However, the USPSTF is not charged with making coverage decisions, only for creating evidence-based recommendations for use by primary care clinicians.
The USPSTF ignored population studies in Sweden and the Netherlands showing that the decreased death rate is due to screening.
There are eight RCTs of mammography in women aged under 50. None of these studies have, individually, found a statistically significant mortality reduction associated with mammography. This is because death from breast cancer in this age group is relatively low, and the reduction associated with mammography screening is likewise low. However, when the study results are combined in meta-analysis, the estimated 15% mortality reduction becomes statistically significant.
There have been four Scandinavian observational studies published since the systematic evidence review that the USPSTF used was completed. Two found mortality reductions much lower than 15% (including no benefit) and two found mortality reductions higher than 15%. These findings indicate the problem with using observational studies, with their well-known and well-documented sources of bias, to estimate efficacy.
Although the USPSTF charter prohibits cost/benefit analyses, it circumvented this prescription by using the NNTS to save one life.
As stated in testimony in front of Congress, the USPSTF has never used cost in considering a recommendation. NNTS is an estimate of the benefit of the test.
The USPSTF process evaluates this benefit against the harms of screening. Extra radiation (though small), anxiety, additional studies and biopsies for false positive results that provide no health benefit are harms that increase with the number of mammograms performed.
Due to the dramatic increase in the rate of false positive results in women under 50 and the small benefit, the USPSTF concluded that screening in this age group should not be automatic, but should prompt a discussion of risks and benefits to the individual.
The USPSTF guidelines will lead to thousands of unnecessary deaths if adopted.
There is no evidence to support this assertion. The recommendation is for women to make their own informed decision. Shouldn’t that be their right? Shouldn’t it be a clinician’s responsibility to inform a patient of the potential benefits and the potential harms,
then support her decision?