Giuliani Is Still Standing By Questionable Figures

Wednesday, November 7, 2007


The former New York mayor would have us believe that he was off by one percentage point at most in calculating his chances of surviving prostate cancer in Britain. In fact, he was spectacularly wrong the first time and equally wrong the second time. Epidemiologists say that his claim rests on a faulty statistical methodology that would not earn a passing grade at top medical schools in the United States.

Let's begin by deconstructing the original Giuliani claim, featured in a campaign ad in New Hampshire. It rests on a crude statistical calculation by his medical adviser, David Gratzer, on the basis of a 2000 study by a pair of health experts from Johns Hopkins University. According to Gratzer, "49 Britons per 100,000 were diagnosed with prostate cancer, and 28 per 100,000 died of it. This means that 57 percent of Britons diagnosed with prostate cancer died of it; and consequently, that just 43 percent survived."

There are several problems with this line of reasoning, according to health experts.

To make statistically valid comparisons in epidemiology, it is necessary to track the same population. Because prostate cancer is a slow-developing tumor, it is probable that the Britons who died of prostate cancer in 2000 developed the disease 15 to 20 years earlier. They represent an entirely different cohort of cancer sufferers than those who were diagnosed with the disease in 2000. The number of Britons diagnosed with the disease is itself a subset of the number of Britons with the disease.

"You would get an 'F' in epidemiology at Johns Hopkins if you did that calculation," said Johns Hopkins professor Gerard Anderson, whose 2000 study "Multinational Comparisons of Health Systems Data" has been cited by Gratzer as a source for his statistics. "Numerators and denominators have to be the same population."

Five-year survival rates for those with prostate cancer are higher in the United States than in Britain but, according to Howard Parnes of the National Cancer Institute, this is largely a statistical illusion. Americans are screened for the disease earlier and more systematically than Britons.

If you are detected with prostate cancer at age 58 in year one of a disease that takes 15 years to kill you, your chances of surviving another five years (until the age of 63) are obviously much higher than if your cancer is detected in year 11, at the age of 68. Both Anderson and Parnes say that it is impossible, on the basis of the available data, to conclude that Americans have a significantly better chance of surviving prostate cancer than Britons.

Instead of acknowledging his error, Giuliani chose to repeat it on several occasions, including a campaign event here in Washington last Friday. We invited the members of the Giuliani campaign to name a reputable prostate cancer research expert or epidemiologist who would publicly support the candidate's claim. It has not so far responded. If it can produce the scientific evidence, we will reconsider our verdict.

In the meantime, four Pinocchios for recidivism.

© 2007 The Washington Post Company