April 8, 2009


Knowing Your Chances: What Health Stats Really Mean: Learn how to put aside unjustified fears and hopes and how to weigh your real risk of illness--or likelihood of recovery (Gerd Gigerenzer, Wolfgang Gaissmaier, Elke Kurz-Milcke, Lisa M. Schwartz and Steven Woloshin, April 2009, Scientific American)

Medicine has held a long-standing antagonism toward statistics. For centuries, treatment was based on an ethic of personal trust as opposed to quantitative facts, which were dismissed as impersonal or irrelevant to the individual. Even today many doctors think of themselves as artists, relying more on intuition and faith in their own judgment than on numbers. For their part, many patients prefer to trust their doctors rather than even asking for data to analyze. For example, in a 2008 unpublished survey by one of us (Gigerenzer) and his colleagues, two thirds of more than 100 American economists said they had not weighed any pros and cons of getting a prostate cancer screening test but simply followed their doctor’s recommendation.

Moreover, individuals often shy away from statistics because they have an emotional need for certainty—a concept at odds with statistical literacy, which prepares us to make decisions in the face of uncertainty. Much of the public harbors illusory certainty about the reliability of tests such as those for cancers and HIV, suggests a survey Gigerenzer conducted in 2006.

Furthermore, statistically unsophisticated patients and their doctors tend to wildly overestimate the benefits of screening tests and are blind to their harms. For example, mammography reduces the risk of a woman in her 50s dying from breast cancer from about five to four in 1,000 over some 13 years, but 60 percent of a random sample of U.S. women believed the benefit to be 80 times as high. Americans are similarly overenthusiastic about total-body computed tomographic scans: in a random sample of 500 Americans, nearly three quarters said they would prefer a free total-body CT scan to $1,000 in cash. Yet no professional medical organization endorses such scans, and several discourage them because screening tests such as this one can result in important harm from a cascade of medical quandries and invasive treatments triggered by ambiguous findings. [...]

While running for president, Giuliani claimed that health care in the U.S. was superior to that in England. He apparently used data from the year 2000, when 49 British men in every 100,000 were diagnosed with prostate cancer, of whom 28 died within five years—about 44 percent. Using a similar approach, he cited a corresponding 82 percent five-year survival rate in the U.S., suggesting that Americans with prostate cancer were twice as likely to survive as their British counterparts were. That implication, however, is false because these survival statistics largely reflect diagnostic differences between the two countries rather than better treatment and prolonged survival in the U.S.

To understand why, imagine a group of prostate cancer patients diagnosed (by their symptoms) at age 67 in the U.K., all of whom die at 70. Each survived only three years, so the five-year survival of this group is 0 percent. Now imagine that the same group is diagnosed in the U.S., where doctors detect most prostate cancer by screening for prostate-specific antigens (PSA). (The PSA test is not routinely used in Britain.) These U.S. patients are diagnosed earlier, at age 60, but they all still die at age 70. All have now survived 10 years, and thus their five-year survival rate is 100 percent. Even though the survival rate has changed dramatically, nothing has changed about the time of death. This example shows how setting the time of diagnosis earlier can boost survival rates (lead-time bias), even if no life is prolonged or saved.

Spuriously high survival rates can also result from overdiagnosis, the detection of abnormalities that are technically cancer but will never progress to cause symptoms in the patient’s lifetime. Say 1,000 men with progressive cancer do not undergo screening. After five years 440 are still alive, which results in a survival rate of 44 percent. Meanwhile in another population of men, PSA screening detects 1,000 people with progressive cancer and 2,000 people with nonprogressive cancer (who by definition will not die of cancer in five years). These nonprogressive cases are now added to the 440 who survived progressive cancer, which inflates the survival rate to 81 percent. Although the survival rate changed dramatically, the number of people who die has not changed at all.

In the U.S., screening for prostate cancer using the PSA test in the late 1980s led to an explosion in the number of new prostate cancer diagnoses. In Britain, the effect has been much smaller because of far less use of the PSA test. This diagnostic disparity largely explains why five-year survival for prostate cancer is higher in the U.S. (The most recent figures are 98 percent five-year survival in the U.S. versus 71 percent in Britain.)

Despite the difference in survival rates, mortality rates in the two countries are close to the same: about 26 prostate cancer deaths per 100,000 American men versus 27 per 100,000 in Britain. That fact suggests the PSA test has needlessly flagged prostate cancer in many American men, resulting in a lot of unnecessary surgery and radiation treatment, which often leads to impotence or incontinence.

In essence, the only reason to have your prostate checked is sexual experimentation.

Posted by Orrin Judd at April 8, 2009 11:09 AM
blog comments powered by Disqus