NEW MAMMOGRAM GUIDELINES: MORE NOISE THAN SCIENCE

Last week, the U.S. Preventive Services Task Force issued new mammogram screening guidelines that have stirred considerable controversy in political and scientific circles. If the taskforce report were credible and convincing, there would be no need to suspect a hidden motive. However, these findings are deeply flawed and highly suspect. Here is a synopsis of findings released by the taskforce:

Credit: (O)CT(O)PUS

According to the taskforce: If you screen 1,904 women in the 40 to 49 year-old age group, there is a high probability of preventing one cancer death. Similarly, to prevent one cancer death in the 50s age group, you would have to screen 1,339 women. It goes without saying that cancer risks increase with age, meaning that women in older age groups are more likely to test positive for cancer than women in younger age groups.

With simple arithmetic, you can estimate how many lives per million can be saved. For women in the 40s age group: If you screen 1,904 women to save one life, it is reasonable to assume that you can save 525 lives if a million women were screened. What does this mean when applied to the U.S. population? According to the U.S. Census Bureau (Table 1), there are 22,346,000 women in the 40s age group (14.7%). When you divide this population by 1,904, the probabilistic number of lives that can be saved is 11,736.

If you accept this assumption, then you cannot accept these findings of the U.S. Preventive Services Task Force, which state (source):

The USPSTF recommends against routine screening mammography in women aged 40 to 49 years (...) This is a C recommendation.

The USPSTF recommends against teaching breast self-examination (BSE). This is a D recommendation.

In other words, the taskforce recommends no mammogram screenings for women in the 40s age group and no breast self-examinations for women of any age. Why? According to the study panel, the relative risks outweigh the benefits, i.e., risk factors such as false-positive test results, anxiety, and pain from increased biopsy rates resulting from over-diagnoses. Yet, there are contradictory statements within the taskforce report, such as:

Breast cancer mortality has been decreasing since 1990 by 2.3% per year overall and by 3.3% for women aged 40 to 50 years. This decrease is largely attributed to the combination of mammography screening with improved treatment.

On one hand, the study panel recommends a reduced screening regimen for women in the 40s age group yet attributes reductions in mortality to a “combination of mammography screening with improved treatment.” How contradictory!

If you think these conclusions are strange, the measurements of relative risks are even more contradictory. According to the taskforce report, the relative risks are virtually identical for both groups: 15% for the 40s age group, and 14% for the 50s group. In other words, the tradeoff is more peace of mind within the context of “ignorance is bliss.” Either the data are internally inconsistent, or the conclusions contradict the data, or Aesop wrote the final study report.

A few words about methods and statistics: It seems different branches of the federal government employ different research protocols. If this were the FDA, for instance, no report would pass muster if it did not include at least one domestic study covering these age distributions. The USPSTF report includes only one domestic study for the 40s age group, but none covering the 50s age group. Other meta-analyses are based on non-domestic data sets that are less representative of U.S. population demographics.

Furthermore, the USPSTF report applies less than rigorous statistical methods. For example, the study panel uses the term “credible interval” instead of the more conventional term “confidence interval” as a measure of statistical validity. By definition, the term “credible interval” is a subjective subset of the confidence interval … implying considerable wiggle room to employ a fudge factor when one wants to force data into a preconceived conclusion.

Why are substandard study methods and statistics being used to justify new mammogram screening guidelines? Consider this comment by one of our Swash Zone writers under my last post on this subject (link):

maleeper (@11:02 AM, November 20, 2009): “Thanks for raising the issue of the USPSTF ill-advised report on breast cancer. The study was commissioned during the Bush administration, which may be why no oncologists or radio-therapists served on the study panel.

According to the American College of Radiology, the USPSTF recommendations ignored direct scientific evidence from large clinical trials and "also ignored peer reviewed journal articles that critqued studies on which their recommendations rely."

One trial that was used for the study was translated from Russian to English so that it could be used in the study, while many reputable articles were ignored.

Such selectivity in sources cited leads me to believe that the taskforce may have decided the results it wanted ahead of time, since they clearly will save money for insurance companies for a few years. The USPSTF panel then found studies to back up their assumptions, regardless of the cost of human life.

What do you get when junk science conspires with junk journalism? You get editorials like this one from the New York Times:

There is nothing wrong with a healthy public debate about mammography within the medical community and among women who must decide when and how often to get screened. It should not be injected into the partisan debate over health care reform.

Unless, of course, the findings of the taskforce are specious and suspicious.