August 11, 1998


A Study Guide to Scientific Studies


To paraphrase P. T. Barnum, there's a study born every minute, and there are countless suckers out there ready to believe anything they hear or read. Many people make significant changes in what they eat, drink, take or do based on research that is far from conclusive. Some become unduly alarmed about suspected but unproven risks to health, especially ones in the air, water or food supply, that they cannot control. Others decide that, because what is "fact" one week is sometimes contradicted by a new study finding the next, they won't change any habit until "the scientists make up their minds."

This kind of confusion is understandable given two facts of modern life. First, fed in part by media attention to health issues, the public's concern about health and demand for health information has increased significantly in recent decades. Second, few people, including many who report health findings in print or on the air, know how to interpret scientific research or understand the certainties -- and uncertainties -- inherent in different kinds of studies.

Thus, what medical researchers may report as an "association" often becomes a cause-and-effect relationship in the minds of readers and listeners. Or what takes place in a test tube or laboratory mouse is automatically assumed to apply to humans as well.

"People should view health news skeptically and critically," said Dr. Scott Emerson, a biostatistician at the University of Washington in Seattle. "It's important to realize how many variables there are in medical research and look for details that help illustrate the significance of the findings."

So before you switch back to butter, start gobbling supplements or give up coffee, look beyond the scary headlines and broadcast bulletins and try to find out just how solid the new "facts" are. To know when to act on research findings and when to sit tight, you must first determine what kind of study is involved and recognize its potential pitfalls.

What Kind of Study Is It?

The following types of research investigations are described in the order of how certain you can be that their findings can be generally applied.

These studies are done to suggest a possible hazard or benefit of a substance or circumstance or to elaborate on observations that have been made in people. They are carried out either "in vitro" -- that is, on cells or tissue samples in an actual or proverbial test tube, or "in vivo" -- in laboratory animals like mice or guinea pigs.

The advantage of such studies is that they can be tightly controlled. The scientist can be sure, for example, that comparison groups are genetically identical and the conditions to which they are exposed (except for the factor being studied) are identical.

But it is a long way from the test tube or laboratory mouse to man or woman. The fact that all people do not have the same genes, hormones, metabolism or life circumstances may greatly modify the effects. Or human biology may be such that whatever happens in a test tube or lower animal would not apply to us. Thus, for example, laboratory studies indicated that saccharin may cause cancer in mice, but no reliable study in people has yet borne that out.

These are observational studies, and they are the origin of most of the health reports you are likely to hear. They are also fraught with pitfalls. They cannot establish cause and effect; they can only suggest a relationship between two factors or events. These kinds of studies are commonly used to ferret out possible risks or preventives for diseases, for example, high cholesterol as a risk factor for heart disease or vitamin E supplements to prevent Alzheimer's disease.

Epidemiological studies are best at identifying very powerful associations, like the one between smoking and lung cancer. When the relationship between factors is weaker, like the link between alcohol and breast cancer, epidemiological studies are less helpful and often produce contradictory results. Also, the larger the study and the longer it has been carried out, the more certain you can be of the findings, because researchers can better account for factors that could lead to a spurious conclusion.

Observational studies are of two main types: case-control studies and cohort studies. In a case-control study, researchers compare factors found among people with a certain disease to factors among a comparable group of people without that disease. Thus, a study of people with ulcers revealed that infection with the bacterium, Helicobacter pylori, was far more common among them than in people without ulcers. But doctors could not be certain the bacterium was a cause until they showed that curing the infection could cure ulcers.

In a cohort study, large groups of people are followed for a long time. Researchers try to identify factors -- possible causes and preventives -- associated with illnesses that develop over time. Among the most helpful of these studies has been the 45-year-old Framingham Heart Study and the huge Nurses Health Study. By adjusting their data statistically to account for the influence of extraneous factors that might confound the findings, researchers can often zero in on important associations, like a much lower death rate from heart disease and a small increase in breast cancer risk among women who take postmenopausal estrogen.

A study that randomly assigns people to two treatment groups, with neither the researchers nor the participants knowing which group is which until the study is completed, is the gold standard of scientific research. To be sure that estrogen replacement -- not some other characteristic of women who choose to take hormones -- is responsible for their lower risk of heart disease, women must be randomly assigned to take hormones or a look-alike placebo, then followed for years to determine their fate.

Of course, not every suspected association can be subjected to a clinical trial. Given the overwhelming evidence for the risks of smoking, it would be unethical to assign one group to smoke and another to never smoke just to prove that smoking causes cancer.

Finally, regardless of the kind of study you are examining, it is important to appreciate the most fundamental fact of medical research: A single study does not a fact make. No matter how reputable the scientist or the institution where the research was done, no matter how compelling the findings appear to be, they must be confirmed by one or more independent studies before scientists will accept them as gospel. In fact, it often takes many studies that examine an issue from different angles with increasing degrees of certainty to convince well-informed researchers and public health advisers that it is time to act on the conclusions.

As Dr. Emerson put it, "Each study a researcher performs is just a small step. It takes many small steps to build a significant understanding."

Next week: Famous health scares.

PERSONAL HEALTH is published weekly, on Tuesdays. Click here for a list of links to other columns in the series.

Home | Site Index | Site Search | Forums | Archives | Marketplace

Quick News | Page One Plus | International | National/N.Y. | Business | Technology | Science | Sports | Weather | Editorial | Op-Ed | Arts | Automobiles | Books | Diversions | Job Market | Real Estate | Travel

Help/Feedback | Classifieds | Services | New York Today

Copyright 1998 The New York Times Company