This is the online version of the NIH Word on Health, Consumer Health Information Based on Research from the National Institutes of Health

August 2004


Different Trials, Different Results
How to Explain Them?

A Word to the Wise...
Different Types of Research Studies

To better understand news reports of medical findings, you need to learn about the different types of studies. Each has its own strengths and weaknesses. Which type of study researchers will use depends on many factors.

Case Control Study — A type of observational study. Researchers identify a group with a particular disease or condition, then look at another group without the condition for comparison to figure out what differences between the two groups might be causing the disease.

Cohort Study — Another type of observational study. Researchers select participants based on their exposure status (to a particular pollutant, for example) and then try to find a similar group of people who haven't been exposed. They compare the two groups to see whether the factor in question is causing any health problems.

Intervention Study — In intervention studies, researchers decide who gets a particular treatment or exposure to something. The best-known type of intervention study, the randomized clinical trial, is considered the gold standard of clinical research studies. In randomized clinical trials, people are assigned to two or more study groups by chance (randomly). One of the groups, the control group, receives a preparation such as a pill that looks just like the treatment or drug being tested, but actually does nothing (called a placebo). Comparing treatment groups to control groups is the best way to see if a treatment is really effective.

News reports about medical studies can get confusing. Have you ever thought that something was supposed to be good for your health, but then suddenly heard it was bad for you? How in the world can medical studies come to such different conclusions about the same thing?

Dr. Julie Buring, an epidemiologist at Brigham and Women's Hospital in Boston, recently gave some insight when she spoke at a cancer research symposium at the National Institutes of Health. Her keynote address explained why there are often discrepancies between the two major types of medical studies you often hear about, observational studies and randomized trials.

To understand why these two types of research studies can come up with different answers, you first need to understand what they are. In observational studies, researchers simply observe people, grouping them according to whether or not they have chosen to take a particular drug, live a particular lifestyle, or have been exposed to something of interest. The researchers then compare these groups of people to see their subsequent risk of disease.

In contrast, exposures in intervention studies are controlled by investigators, who assign people to two or more study groups and then control what treatment or exposure to something of interest each group receives. The best-known type of intervention study is the randomized clinical trial, considered to be the gold standard of clinical research studies. (See the side box for more details.)

Dr. Buring gave some examples that have been widely reported in the news to illustrate how observational studies and randomized clinical trials can have different outcomes. One example involves the discovery of a surprising relationship between beta-carotene and heightened lung cancer risk.

"In animal laboratory studies," she explained, "beta-carotene could block the carcinogenic process and inhibit specific tumor growth." There was also, she said, a large body of evidence associating the consumption of fruits and vegetables with a lower risk of cancer. Many researchers believed that these pieces of evidence, taken together, showed that beta-carotene helped people ward off cancer.

But in randomized clinical trials, beta-carotene showed no benefit in preventing cancer and actually increased the risk of lung cancer among heavy smokers.

There are many possible explanations for this discrepancy, Dr. Buring said. People may have made an incorrect leap to one specific factor. It could have been something besides beta-carotene, or there might have been other nutrients in the diet that worked with beta-carotene to prevent cancer.

"Many people prefer taking a pill to changing their diet," Dr. Buring said. But those consuming high levels of beta-carotene in the observational studies had also been eating lots of fruits and vegetables, and were likely unique in other ways — eating more fiber, exercising more, and perhaps differing in other habits researchers haven't even thought of.

Maybe the clinical trial simply lasted for too short a time to see an effect, Dr. Buring offered as another explanation. Or it might have been that the actual blood levels of beta-carotene in the clinical trial patients differed significantly from those in the observational studies.

As for the smokers, researchers have since been able to explain why their risk of lung cancer increased. It turns out that, when combined with changes inside the body brought about by smoking, high doses of beta-carotene actually create harmful compounds.

Dr. Buring also recounted the recent findings from the Women's Health Initiative studies on hormone replacement therapy (HRT). Many doctors had been giving their patients HRT based on their own observations, she explained, and some thought it was actually unethical to conduct a randomized trial, which would require denying the benefits of these hormones to control groups.

But it turned out that randomized clinical trials had to be stopped early because the risks of using HRT for extended periods outweighed their benefits.

There were many possible reasons for these unexpected results. One likely explanation is that women taking HRT in the observational studies may have been living healthier lifestyles and seeing their doctors more often.

This explanation highlights an inherent limitation in observational studies, Dr. Buring said: "People who select one way to live also select others." Unforeseen or "confounding factors" in an observational study can lead to a clinical trial that returns unexpected results.

Asking Different Questions

Many people get frustrated with these discrepancies, Dr. Buring acknowledged, particularly when the media hypes contradictory conclusions. But she argued that these discrepancies aren't necessarily a sign of bad studies. "Discrepancies may mean the system is working," she said. "This is exactly why we do trials. Apparent discrepancies are not necessarily contradictory or wrong; they may be looking at different questions."

So why do observational studies at all, if randomized clinical trials are the best way to answer a question? "Sometimes, an observational study is the only feasible or ethical approach," Dr. Buring explained. For instance, researchers cannot ethically expose people to a suspected carcinogen to find out whether or not they develop cancer.

Dr. Buring concluded that "Neither a reliance on randomized clinical trials nor observational studies is appropriate." Both are needed to answer a medical question.

So if you find yourself wondering how seriously to take the latest research report, ask yourself what kind of study it was. Is this really the final word on the topic, or do researchers need to explore the topic further? Those "contradictory" research results, once you put them in perspective, may suddenly not be as baffling as they seemed.u

— Written by Harrison Wein, Ph.D.

For more information on understanding news reports on medical risks, see Understanding Risk from the April 2004 issue of The NIH Word on Health at

You can also view or order a free copy of the National Institute on Aging publication Understanding Risk: What Do Those Headlines Really Mean? at Or, call 1-800-222-2225 (TTY: 1-800-222-4225).

Up to Top | Contents | Previous Story | Next Story

[ Q&A About NIH | Career Opportunities | Visitor Information | FOIA ]
[ Site Map | Employee Information | Información en español | Search ]

[ Contact Us | Privacy Notice | Disclaimers | Accessibility ]

N I H logo - link to the National Institutes of Health

National Institutes of Health (NIH)
9000 Rockville Pike
Bethesda, Maryland 20892

    H H S logo - link to U. S. Department of Health and Human Services

Department of Health
and Human Services