There are a variety of human intervention study designs in nutrition research, but the most common, especially in pharmaceutical/medical research, is the clinical trial. A clinical trial is a scientifically controlled study using consenting people to find the safety and effectiveness of different items/regimens. Clinical trials are the "gold standard" research method. Their findings carry the most weight when making decisions about a certain research area because they are the most rigorous scientific studies. Every pharmaceutical must go through a series of clinical trials (specifically randomized, double-blind, placebo-controlled trials) before being approved in the US by the Food and Drug Administration (FDA). As shown in the figure below, human intervention studies/clinical trials are normally prospective. By the end of this section you should have an understanding of what randomized, double-blind, and placebo-controlled means.
Placebo and The Placebo Effect
A placebo is a inert pill or treatment that serves as the control comparison to an active intervention. Some example placebo pills are shown below.
The use of a placebo is necessary in research because of a phenomenon known as the "placebo effect". The placebo effect is the postive effect that can occur from a subject's belief that the intervention that they are receiving will improve their condition, even if they are not receiving an active intervention. An example would be an athlete who consumes a sports drink and runs the 100 meter dash in 11.00 seconds. The athlete then, under the exact same conditions, drinks what he is told is "Super Duper Sports Drink" and runs the 100 meter dash in 10.50 seconds. But what the athlete didn't know was that Super Duper Sports Drink was the Sports Drink + Food Coloring. There was nothing different between the drinks, but the athlete believed that the "Super Duper Sports Drink" was going to help him run faster, so he did. This improvement is due to the placebo effect.
Ironically, a study similar to the example given above has been conducted and its results support that there is a “placebo effect.”
Randomization is the process of randomly assigning subjects to groups to decrease bias. Bias is a systematic error that may influence results. Bias can occur in assigning subjects to groups in a way that will influence the results. An example of bias in a study of an antidepressant drug is shown below. In this nonrandomized antidepressant drug example, researchers (who know what the subjects are receiving) put depressed subjects into the placebo group, while "less depressed" subjects are put into the antidepressant drug group. As a result, even if the drug isn't effective, the group assignment may make the drug appear effective, thus biasing the results as shown below.
This is a bit of an extreme example, but even if the researchers are trying to prevent bias, sometimes bias can still occur. However, if the subjects are randomized, the sick and the healthy people will ideally be equally distributed between the groups. Thus, the trial will be unbiased and a true test of whether or not the drug is effective.
Blinding is a technique to prevent bias in human intervention studies. A study without blinding is referred to as "open label" because both the subject and the researchers know what intervention the subject is receiving (i.e. placebo or drug). This can lead to bias, so these types of trials are used less frequently.
In a single-blind study, the researcher knows what intervention the subject is receiving, but the subject does not. If the subjects are randomized, these types of trials should produce robust results, but it is still possible that the researcher can bias the results.
Finally there is the double-blind study, where neither the researcher nor the subject know what intervention the subject is receiving. A separate board reviews the collected results and decides the fate of the trial. This is the "gold standard" because it prevents observer bias from occurring.
The following video does a nice job explaining and illustrating how double-blind randomized trials are performed.
*The above video does not contain captioning, and is posted in a way that we cannot add it to it. Below is the transcript for what is said in the video.
To conduct a randomized controlled trial a statistician will select a sample size large enough to produce a significant result.
Care must be taken that test subjects are properly representative of the target population and not tainted by selection biases that might skew the results.
The subjects are blinded knowing as little as possible about what is being tested.
They are randomly and blindly assigned to one of several groups.
There may be a group that will receive the treatment being studied, a group receiving an established treatment and always at least one control group receiving a control or placebo treatment.
Test administrators are also blinded.
This is called double blinding.
Such as that they don’t know what group each subject is assigned to and whenever possible they don’t know what the treatment is they are administering.
Everything is coded to avoid experimenter bias and to cancel out any effects like patients trying to respond the way experimenters want them to.
The trial lasts long enough to satisfy the statisticians and the scientist.
Finally, when the results are tabulated by a blinded statistician, this is called triple blinding.
We get the results.
The cloaks of anonymity are whisked aside and we finally learn with statistical certainty which treatments are effective and which are not.
When this process shows significant benefits for a new treatment and the trial can be repeated by other experimenters and yields similar results, then and only then do scientists say that this is a product that works and is supported by evidence.
- Levin K. (2006) Study design III: Cross-sectional studies. Evidence - Based Dentistry 7(1): 24.