Medical Studies: Don't Believe Everything the Media Says
JoAnn E. Manson, MD, DrPH
Harvard Medical School
hen the media reports on medical news, complex research gets reduced to sound bites -- which may be misleading. Find a full account of the study (try www.nlm.nih.gov/medlineplus) and ask...
Are this study's results consistent with other evidence? Accuracy usually involves consistent results from different researchers... using different types of studies... and involving different people. Example: The link between smoking and lung cancer that has been seen in so many studies.
Is it an observational study or a randomized clinical trial? An observational study tracks behavior and health outcomes without intervening in participants' lives. This can uncover "associations" but cannot prove a cause-and-effect link. Example: The apparent health benefits of vitamin supplements seen in observational studies simply may reflect that people who choose to use such supplements tend to have more healthful habits overall.
In a randomized clinical trial, researchers actively intervene by assigning participants at random to receive treatment or a placebo -- making this the "gold standard" of research.
Is it an animal study? Animal studies allow far greater control than human studies -- but results from other species may not apply directly to people.
How many participants were there? The larger the study, the less likely its findings are due to chance.
How long did the study last? A long-term study may detect risks or benefits that go unnoticed in shorter studies. Example: Hormone therapy using estrogen plus progestin increases risk for breast cancer -- but only after four to five years. A two-year study would not uncover this relationship.
Did the study look at actual disease outcomes? Because it takes years for certain diseases to develop, many studies examine "markers" of disease. Example: Lower cholesterol levels suggest a reduced risk for heart disease -- though cholesterol reductions do not always lead to actual decreases in heart disease risk. Research looking at concrete outcomes, such as the occurrence of heart attacks, is more reliable.
Who were the participants? A study is less valid if participants are not typical of the people who use the therapy. Example: The first clinical trial of estrogen therapy to reduce heart disease was done on men, not women!
What does increased risk really mean? It is scary to hear that a risk factor (such as exposure to a toxin) increases risk for a certain disease by, say, 50%. But suppose that two cases of disease normally occur per 10,000 women who haven't been exposed to the toxin. In that case, a 50% increase would mean that toxic exposure leads to three cases per 10,000 women -- which isn't so scary.
Bottom Line/Women's Health interviewed JoAnn E. Manson, MD, DrPH, a professor of medicine and women's health at Harvard Medical School and chief of the division of preventive medicine at Brigham and Women's Hospital, both in Boston. She is one of the lead investigators for two highly influential studies on women's health -- the Harvard Nurses' Health Study and the Women's Health Initiative. Dr. Manson is the author, with Shari Bassuk, ScD, of Hot Flashes, Hormones & Your Health (McGraw-Hill).