In case you've been under a rock, an expert panel recommended that most women in their 40s don't need a mammogram screen for breast cancer. The group, the U.S. Preventive Services Task Force, created a firestorm: Many (not all) oncologists, primary care physicians, and other groups disagreed with the recommendation. Details on the panel's process are summarized at the end of this post. (BTW, the U.S. government is now trying to distance itself from the USPSTF action.)
Evidence is easy. Decision-making is hard. This illustrates how tricky it is to translate scientific evidence into real-world advice. It also illustrates why we need to formally recognize more forms of evidence -- going beyond systems that weigh the strength of medical evidence (A, B, etc.). And finally, this shows how desperately we need to present evidence differently: USPSTF's 'Evidence Synthesis' document is 95 pages of paragraphs, bulleted lists, and tables [pdf here]. There are better ways to present evidence so the intended audience can make more effective use of the information.
How do you save one-third of a life? Today, NPR's Talk of the Nation hosted a discussion of the guidelines, including comments from several patients -- some had experienced excruciating stress over false positives, while some had cancers that likely would have killed them were it not for mammography. The feelings they displayed underscore why it can sound inhumane to express possible outcomes in statistical terms. Alas, one doc -- when explaining how many 'unneeded' mammograms are being performed -- said 3,000 women would need to undergo mammograms to 'save 1/3 of a life' ...huh? (No doubt he was referring to the 'number needed to treat' - a common term in evidence-based medicine, representing how many people with a particular ailment must receive a particular treatment to save one life, statistically speaking. But he missed the mark with this attempt to explain statistics to the gen-pop.)
Values vs. evidence? Whether or not this specific recommendation is flawed, I think it's refreshing that USPSTF considered evidence other than statistical significance and other traditional quantitative findings. They didn't simply do a cost-benefit analysis, weighing excess screening/treatment against the possibility of undiagnosed/untreated disease. USPSTF also considered the stress caused by false positives and unnecessary treatments: "The harms resulting from screening for breast cancer include psychological harms, unnecessary imaging tests and biopsies in women without cancer, and inconvenience due to false-positive screening results."
In a 2007 post, Oh dear. Are we going to need a new word for evidence?, I wrote about the book Evidence-Based To Value-Based Medicine, by three MDs (Melissa Brown, Gary Brown, & Sanjay Sharma). "They claim that medical practice is evolving from evidence-based medicine to an even higher quality of patient care, what they describe as value-based medicine: An approach that measures patient-perceived value and integrates relevant costs provided by health care interventions, allowing a more accurate measure of the overall worth to stakeholders. The authors present ways to evaluate medical treatments in the context of the patient's quality of life." (Of course, weighing all this stuff is easier to do at an abstract level -- or, if you're an actual patient, in hindsight.)
We need more categories of evidence. Question: Aren't "patient-perceived values" and "relevant costs" both forms of evidence that should be factored into the overall equation? This is why we need to broaden our definition of 'evidence' and do more - much more - to distinguish between various flavors of evidence, recognizing the validity of each, and acknowledging their strengths & weaknesses.
No gold stars for mainstrean media. Media have been all over this story. But few identified who made the recommendation (USPSTF), or described how the panel arrived at conclusions. Almost no one linked to the actual recommendations, evidence, and discussion (here they are). But there were some insightful articles on the subject: In Mammogram guidelines spark heated debate, the LA Times did a good job of explaining the tradeoffs and the difficulty of issuing a blanket health recommendation like this.
From the USPSTF web site, here's their approach to synthesizing evidence, and the specifics of their mammogram recommendation.
"The Task Force makes its recommendations on the basis of explicit criteria. Recommendations issued by the USPSTF are intended for use in the primary care setting. The USPSTF recommendation statements present health care providers with information about the evidence behind each recommendation, allowing clinicians to make informed decisions about implementation. The USPSTF is supported by an Evidence-based Practice Center (EPC). Under contract to AHRQ, the EPC conducts systematic reviews of the evidence on specific topics in clinical prevention that serve as the scientific basis for USPSTF recommendations.
"The USPSTF reviews the evidence, estimates the magnitude of benefits and harms for each preventive service, reaches consensus about the net benefit for each preventive service, and issues a recommendation. The Task Force grades the strength of the evidence from "A" (strongly recommends), "B" (recommends), "C" (no recommendation for or against), "D" (recommends against), or "I" (insufficient evidence to recommend for or against)."
"Benefits of Detection and Early Intervention: There is convincing evidence that screening with film mammography reduces breast cancer mortality, with a greater absolute reduction for women aged 50 to 74 years than for women aged 40 to 49 years. The strongest evidence for the greatest benefit is among women aged 60 to 69 years. Among women 75 years or older, evidence of benefits of mammography is lacking.... The evidence for benefits of digital mammography and MRI of the breast, as a substitute for film mammography, is also lacking."
"Harms of Detection and Early Intervention: The harms resulting from screening for breast cancer include psychological harms, unnecessary imaging tests and biopsies in women without cancer, and inconvenience due to false-positive screening results. Furthermore, one must also consider the harms associated with treatment of cancer that would not become clinically apparent during a woman's lifetime (overdiagnosis), as well as the harms of unnecessary earlier treatment of breast cancer that would have become clinically apparent but would not have shortened a woman's life.
"Adequate evidence suggests that the overall harms associated with mammography are moderate for every age group considered, although the main components of the harms shift over time. Although false-positive test results, overdiagnosis, and unnecessary earlier treatment are problems for all age groups, false-positive results are more common for women aged 40 to 49 years, whereas overdiagnosis is a greater concern for women in the older age groups.
"USPSTF Assessment: The USPSTF has reached the following conclusions: For biennial screening mammography in women aged 40 to 49 years, there is moderate certainty that the net benefit is small. Although the USPSTF recognizes that the benefit of screening seems equivalent for women aged 40 to 49 years and 50 to 59 years, the incidence of breast cancer and the consequences differ. The USPSTF emphasizes the adverse consequences for most women—who will not develop breast cancer — and therefore use the number needed to screen to save 1 life as its metric. By this metric, the USPSTF concludes that there is moderate evidence that the net benefit is small for women aged 40 to 49 years."
I agree, they could have done better presenting their evidence, but it's also about their judgement and it's really hard to make judgements when lives hang in the balance. Another way to put the evidence is: how much money must be spent and how many women must endure a biopsy from a false positive in order to save one life. MSNBC recently reported on MRI studies (http://www.msnbc.msn.com/id/18160689). MRIs cost much more, and have a higher false positive rate (12.5% about double the false positive rate of mammograms), but in this study, they also identified a 3.1% false negative rate in the mammograms (identifying cancer the mammograms missed). Among those 30 (false negative) women, how many would have died without the MRI. I guess I'm just saying that context is important.
Posted by: Howard | Thursday, 19 November 2009 at 08:08 AM