Rats! Animal studies poor predictor of human effect

By Phil Galewitz
The Palm Beach Post
From the Summer 2007 issue of HealthBeat

The front-page headline in the South Florida Sun-Sentinel: "Red wine could reduce the risk of developing Alzheimer's, new study says."

The lede: "Researchers have uncovered yet another reason to toast red wine: A glass a day could keep Alzheimer's disease at bay. A new study, to be published next month, suggests moderate consumption of ruby-hued cabernet sauvignon can reduce the risk of developing Alzheimer's, a devastating neurological disorder that strikes as many as half of those age 85 or older."

It was a catchy headline and lede.

There was just one problem: The study was done on mice. That little fact wasn't noted until two-thirds of the way through the 900-word story published last October. Nowhere did the author touch on the tenuous link between studies on animals and what the results mean for human beings.

Almost everyday, newspapers and television stations tout the result of animal studies published in major and minor medical and science journals or presented at medical meetings.

But just how useful are animal studies to human health? Not very, according to two studies published late last year in the British Medical Journal and the Journal of the American Medical Association.

Animal studies are of "limited usefulness to human health because they are of poor quality and their results often conflict with human trials," concluded researchers in the British Medical Journal study.

The researchers, British scientists and an investigator from Yale University, compared treatment effects in animal models with human clinical trials. They looked at the outcomes of six drugs for conditions such as head injury, stroke and osteoporosis. Three of the six studies on human and animals showed no link.

For example, corticosteroids showed a benefit in treating animals with head injuries, but not humans. Similarly, giving the drug tirilizad to animals suggested a benefit in animals, but showed no help in humans.

On the positive side, the study did find corticosteroids reduced neonatal respiratory distress syndrome in animal studies and in clinical trials, though the data were sparse and found no evidence of agreement comparing mortality rates.

"This report is certainly not an argument against animal experimentation," says Michael Bracken, a study author and professor of epidemiology at Yale University. But he adds: "The animal studies are often too small or designed in ways that don't shed much light on human disease."

There's another reason the results in animal studies are not found in clinical trials. Animals are biologically different from people.

Test animals are often young, rarely have other illnesses, and are not exposed to other medical interventions that humans often receive, the British Medical Journal study said.

The JAMA study also found major problems in translating the results of animal studies into human benefits. The JAMA study, conducted by two University of Toronto researchers, analyzed animal studies published in major science journals to later clinical trials.

Just 37 percent of the 76 animal studies analyzed were ever replicated in human randomized trials – the gold standard in clinical testing. About 45 percent of the tests in animals went untested in people, 37 percent were tested in humans but not through randomized trials. Of the limited number of animal studies that were replicated, 18 percent of the human tests contradicted the animal findings.

The JAMA study found the average time it took to replicate an animal study to clinical trials was seven years, though it could take as long as 15 years.

"Patients and physicians should remain cautious about extrapolating the findings of prominent animal research to the care of human disease," the JAMA study concluded. "Poor replication of even high-quality animal studies should be expected by those who conduct clinical research."

Given the tenuous link between animal studies and human trials, why should newspapers and television stations even bother reporting the results of animal studies?

"Animal studies are often given more space and bigger headlines – in short, more weight – than they deserve simply because the topics are so provocative," says Tami Dennis, health editor at the Los Angeles Times. "The amount of work, time and money necessary before humans are actually affected is much less compelling."

Clearly, there is an allure to these animal tests. Consider these headlines.

Los Angeles Times: "Drug may help fight mental impairment; Fed PTZ daily, mice with the retardation of Down syndrome score as well as normal mice. Next: human tests"

The Washington Post: "A Compound in Red Wine Makes Fat Mice Healthy"

The Oakland Tribune: "Potlike chemicals found useful"

The Salt Lake Tribune: "U. study may aid development of diabetes drug"

While some stories note the shortcomings of animal studies, most do not.

Glenn O'Neal, health editor for USA Today, says generally mainstream newspapers should be hesitant about publishing major stories based solely on the results of animal studies.

"Animal studies are typically too preliminary for major coverage, but in rare cases these studies do provide some insight into a potential breakthrough and are worth pursuing. The key is including the study's impact – the proper context and the ‘take home message' – up high so that the reader can quickly assess the merit of the study."

Newsrooms can do better at helping readers assess the value of animal studies, Dennis says.

"If an animal study is truly worth a news story (and this question should be asked more frequently than it is), the relevance of those findings should be made clear," Dennis says. "The story should put the research in the proper context – and explain just how much further scientists need to go before they truly find a cure for cancer or obesity or Parkinson's."

"If findings are worth publicizing, they're worth putting in the appropriate context,"

Phil Galewitz is editor of HealthBeat and a health writer for The Palm Beach Post.