Latest News

Statistics on disease prevalence often inflated

WASHINGTON—Relax. Grim estimates of how many millions of Americans have various diseases are often as high and wild as a baseball pitcher in spring.

And they're all over the place.

A Baltimore Sun report says 20 million Americans suffer from depression. A patient-care newsletter says 10 million Americans older than 50 have the bone-wasting disease osteoporosis. Other published reports say 13 million Americans have hypothyroidism, 7.9 million are alcoholics, 40 million have the hearing defect known as tinnitus, 62 million have digestive diseases and 70 million have some form of arthritis.

Add up the published claims about disease prevalence and the average American has at least two ailments at a time.

Who's pushing the high numbers? Skeptical bio-statisticians blame drug companies and reporters for much of the hype. They also blame research institutes and disease foundations seeking more public spending on particular diseases.

"They always take the high-end numbers," said Mary Grace Kovar, a senior health statistician at the University of Chicago's National Opinion Research Center in Washington. "They want the money, power and prestige" that flow when a disease looks like a major problem.

Former National Institutes of Health director Dr. Harold Varmus, who's now the president of Memorial Sloan-Kettering Cancer Center in New York, fought hard against using such estimates to justify research spending. He called it "body-based budgeting" and argued that NIH's billions should be targeted instead to areas that promise the greatest scientific and therapeutic advances.

These days, prevalence statistics are often part of a larger effort to persuade people that what they consider a human condition is really a disease. Your grandfather snored. But you—and 18 million other Americans, according to the Sleep Foundation—have a breathing disorder called sleep apnea. Your mother blushed and perspired. But you—and 7 million other Americans—have an excessive sweating disorder called hyperhidrosis, according to the International Hyperhidrosis Society.

Kovar and other bio-statisticians fault reporters for thumping the tub—or worse—to make new diseases newsy.

A case in point: stories about shopping addiction, a vaguely defined compulsion that some drug companies would like to treat with antidepressants.

According to the Atlanta Journal-Constitution, 17 million Americans are compulsive shoppers. A doctor on the British Broadcasting Corp.'s popular health Web site ( says it's 15 million. Ronald Faber, a University of Minnesota Twin Cities professor whose 1992 study provided the high-end figures for both articles, begs to differ. Faber said reporters almost always ignore his report's conclusion that the low-end estimate of 2 million to 4 million was the better one.

"Everybody wants the topic they're talking about to sound important," Faber said in an interview. "To get the story read, you need to grab people's attention, and big numbers grab attention."

Daniel Zelterman, a biostatistics professor at Yale University Medical School in New Haven, Conn., said all the ominous health numbers obscured the real bottom line: "People are living longer. We're living better. Things are looking up, looking good, as far as health is concerned."

Zelterman's theory is that prevalence statistics routinely are skewed by the researcher's motive: "Do you want to say we're winning the war on cancer or that we need more research?"

There's plenty of room for subjectivity, given the way most of these numbers are generated: A researcher selects a random population of several thousand or less and asks a series of carefully worded questions. To the answers, the researcher applies one of several official definitions of a disease and determines approximately how many people in the test group have it. This estimate is expressed as a range and projected onto a much larger population.

Reporters and patients' lobbies often seize the range's high end, drop the low end and hype the disease's prevalence with statements such as "Researchers estimate that as many as 30 million. ..."

Even from authoritative, disease-specialist institutes such as NIH, "you're much more likely to get an overestimate than an underestimate" of a disease's prevalence, said Susan Ellenberg, a professor of biostatistics at the University of Pennsylvania School of Medicine in Philadelphia.

The reason: "The more people who have something, the more likely they'll get funding for it," Ellenberg said.

"Most people at NIH institutes have more of a sense of moral responsibility than to do that," said Ronald Manderscheid, a top health-data expert at the federal Substance Abuse and Mental Health Services Administration.

While some disease foundations use high-end estimates, not all of them do. The American Cancer Society, for example, omits from its figures on cancer's prevalence the minor skin cancers and slow-growing tumors that pose no health threats. Were it raising alarms, it would count them.

According to bio-statistician John Bailar, an adviser to the National Academy of Sciences, prevalence numbers are so untrustworthy that he and his colleagues usually ignore them. Estimates based on doctors' reports or mortality statistics derived from death certificates are more reliable, he said.

Deceptions involving prevalence, Bailar said, generally entail "accurate but misleading" numbers that are "presented to the public with the understanding that a lot of people won't understand what they're being told."

Drug makers, research institutes and reporters sometimes end up working together in these efforts.

Consider the male sexual malady now known as erectile dysfunction.

A recent health newsletter from the Gale Group Inc. said erectile dysfunction "affects up to 30 million Americans, according to the National Institute of Diabetes and Digestive and Kidney Diseases."

The institute's Web site supports the published statement, but says the number is a high-end figure and hints that it's unreliable: "Estimates range from 15 million to 30 million, depending on the definition used."

What "depending on the definition" means is that the reported prevalence varies if men are asked whether they currently have the problem, or have had it in the past, or have ever had it.

In addition, the highest numbers ensue when men are asked questions to which it's easy to say yes, said Dr. Ira Sharlip, a San Francisco urologist and spokesman for the American Urological Association.

"Ask a man, `Do you ever have soft erections?' and he's more likely to say yes than when you ask, `Are you able to penetrate your partner?'" Sharlip said.

The easy affirmatives drive up the numbers of men who say they've had erectile dysfunction, Sharlip added, but include millions who don't consider it a problem. If the condition is called by the stark traditional term "impotence"—meaning unable to engage in intercourse because of an inability to attain an erection—the numbers plummet.

In 1997, a year before Viagra was introduced, 10,000 men older than 45 who were living in Goteborg, Sweden, received postcard questionnaires asking if they were impotent. Of the 75 percent of cards returned, 7.6 percent were checked yes. Projected to U.S. men over 45, that comes to a prevalence of just 3.7 million. (That's the lowest of any prevalence figures, Sharlip noted.)

What does the Web site for the anti-impotence drug Levitra say? "Over half of all men over 40 experience some degree of decreased erectile function." That comes to about 27 million, based on the U.S. Census.

The high-end number, Yale's Zelterman said, helps persuade men to admit a problem and seek treatment.

"If you know lots of people have it," he said, "you're more likely to declare yourself among them, especially if there's something to do about it."

In other words, high-end figures for erectile dysfunction solicit for treatment people who don't think they're sick.

With that same general purpose, Viagra, Levitra and Cialis spent $425 million on advertising last year, according to Nielsen Monitor-Plus, an independent ad-auditing agency.


(c) 2005, Knight Ridder/Tribune Information Services.

GRAPHIC (from KRT Graphics, 202-383-6064): 20050413 BADNUMBERS

Need to map

Related stories from McClatchy DC