Statins are effective for men, but it is not clear they are effective for women. More >
What's Good Health Information?
Determining Cause — Experimental StudiesStudies showing causality are usually experiments. Laboratory experiments are what we all think of when we think of experiments, but in medicine, clinical trials (discussed in Part 1 of this series) are a common form of experiment. In order to definitively say that variable A causes variable B, a study has to have been carefully designed so as to avoid the interference of other, unmeasured, variables. Studies of disease-causing bacteria are an example. After perhaps first seeing a correlation — mice with a certain disease all seem to have the same bacteria in their blood — researchers would likely next set up a controlled study — in which they structured, or controlled, the variables so they could be sure that it was the bacteria that caused the disease and not some other factor.
In this kind of experimental study, they would separate the people participating in the study, called subjects, into two groups, a control group and an experimental group. The experimental group is the group of subjects receiving the treatment that is being tested in the study. The control group is a group of subjects who are as similar as possible (in age and health and socioeconomic status, for example) to the experimental group, but who are not given the treatment and so serve as a good basis for comparison as a way to determine if the treatment works.
Particularly in medical studies, members of the control group are often given a placebo. This is an apparent treatment, usually in the form of a sugar pill or other harmless substance that is given to a subject who believes that it is, or may be, real. Placebos are used to make sure that any effects of the substance being studied are entirely the result of the substance's own action and are not caused by any psychological effects, such as, say, wanting to believe that taking a pill will work. This corrects for the placebo effect and allows researchers to tell whether the real treatment is the true cause of any effect. It also increases the likelihood that any changes in the experimental group are caused by the treatment and not by external factors.
Double-Blind studies are a type of study in which neither the subjects nor the researchers conducting the study know who is getting the real treatment and who is getting the placebo. Double-blind studies are useful because the subjects in the experimental and control groups are less likely to be affected by psychological factors, such as their expectations about the treatment they are receiving, that may make the treatment seem more or less effective. And the researchers are protected from their own pre-conceptions and biases which can influence a) how they treat subjects (such as taking the experimental group more seriously); or b) a subconscious tendency to look for "expected" findings; or c) intentionally or unintentionally suggesting "correct" responses to subjects. All these might affect how subjects respond to the experiment. Controlled, double-blind studies are considered to result in the most reliable findings.
Uncovering Relationships — Correlational StudiesStudies showing correlation look for an association between two variables. The correlation can be positive, with both variables going up or down together, as in the finding that the more nutritious a breakfast one eats, the better one will score on a test of mental functioning.
The correlation can also be negative, meaning that when one variable goes up the other goes down. For example, the more often a person eats breakfast, the less likely they are to become obese.
Pay attention the next time your local TV news anchor describes a scientific finding. Often, they will confuse correlation with cause, saying, for example, that eating breakfast actually causes weight loss, when that is not really what the study showed. Even if that were true, it would take a controlled experiment to prove it.
Other Studies Frequently Seen in the NewsVery often, the studies behind the splashy pronouncements we hear about on the news turn out to be epidemiologic studies. As discussed in Part 1, epidemiological studies usually take place in the real world, rather than the laboratory, and involve large numbers of people. This kind of study can identify subtle trends that affect large numbers of people. For example, in the Framingham Heart Study, mentioned in Part 1, researchers asked thousands of subjects all kinds of if they smoked, how much education they had, how much meat they ate and other things over a period of decades. They then related this information to the ongoing health of the subjects' hearts.
But, as mentioned before, the relationships suggested by an epidemiologic study like the Framingham Heart Study cannot prove cause. At best, epidemiological studies can only point to a plausible cause and effect. Further controlled studies need to be done.
A relatively new kind of study is known as a meta-analysis. This is basically a study of studies. Instead of doing new research, a meta-analysis compares the results of different studies done in different ways at different times and in different places. A good example is a recent meta-analysis that looked at dozens of dietary studies and found that eating more calcium helps you lose weight. The downside of this kind of study is that the researchers are using all second-hand data that was collected in different ways by others; it can be very easy for such a study to draw false conclusions by comparing "apples and oranges." So again, testing these findings with further studies is necessary.
No comments have been made