Archive for the ‘Absolute vs. Relative Risk’ Category

Absolute and Relative Risk: Shell Games

Monday, March 8th, 2010

When researchers, reporters, and others use data to compare two or more different groups, they may present their results in two very different and often confusing ways to emphasize a point of view. These relations may be expressed as either absolute or relative differences. An absolute difference is a subtraction; a relative difference is a ratio.To emphasize how easily people and even most physicians can be fooled, consider the following: Which drug would you rather take, one that reduces your risk of cancer by 50 percent, or another drug that only reduces your risk of cancer from two to one out of 100? Most people would choose the drug that reduces their risk of cancer by 50 percent, but in fact both these numbers refer to the same outcome. They’re just two different ways of looking at the same numbers. Without any qualification, both statements “reduced the risk by 50%” and “reduced the risk by 1 in a hundred” (1%) could be construed as representing either an absolute or relative difference. But note the difference in “feel” between 50% and 1%. Which figure sticks in your mind?

The headlines read, “Tamoxifen Cuts Breast Cancer Risk by 33% in Healthy Women!,” yet it turns out, among all the women in a study who took Tamoxifen, less than 2% got breast cancer, and among those that took the placebo, less than 3% got breast cancer. The real difference was 1%. {“How to Lie With Statistics,” Real Health Breakthroughs, Dr. William Campbell Douglass, 2004}

One of the main studies being cited in support of a drug for advanced breast cancer, Herceptin©, saw 34 deaths in the control group (2.0% of the participants) and 23 deaths (1.4%) in the group treated with Herceptin. According to the authors, this translates into a 46% Relative Reduction in cancer deaths, (wrong calculation; should have been 2.0-1.4 divided by 2.0 or 30%) But the true absolute reduction in deaths is only 0.6% (2.0%-1.4%), almost certainly not statistically significant in this series. Is this a miracle drug? The number, of course, is pure marketing and statistical spin. As reported some time ago in the New Scientist magazine, one of the main cheerleaders for Herceptin is none other than Hortobagyi, a paid consultant of Genentech, who received somewhere between $10,000 and $100,000 from the drug company. He’s one of the proponents who calls Herceptin a “cure.”

Keep in mind that headlines promoting a drug will almost always refer to relative risk, “A breathtaking 40% reduction in risk!” -and this numerical shell game will be copied in the mainstream media, press, medical journals, even the FDA... Pharmaceutical companies, marketing reps, even some physicians anxious to publish and usually supported by commercial drug interests are constantly pushing and exaggerating the supposed benefits of their drugs while minimizing their risks.