Randomized controlled trials are the gold standard of scientific testing for new medical interventions. They have become the standard that must be met by pharmaceutical companies in the process of working out what level of efficacy and safety can be achieved by an experimental drug.
The three words for this method of clinical testing - randomized controlled trial (RCT) - represent important elements of the scientific design:
- Randomized - the decision about whether a patient in the trial receives the new treatment or the control treatment (or placebo) is made randomly
- Controlled - the trial uses a control group for comparison or reference. In the control group, the patients do not receive the new treatment being tested, but receive a reference treatment or placebo instead
- Trial - the drug or treatment is on trial during an RCT; it will be approved for wider use only if the results of the testing program indicate that there is a worthwhile level of efficacy, which must be balanced against an acceptable level of adverse effects (safety).
Why are clinical trials 'randomized'?
Randomization is important in a clinical trial in order to prevent the results from being skewed. If the decision about who receives the new treatment in a clinical trial is not decided by random, the patients and scientists can influence the results.
The selection of patients into one arm of a clinical trial or the other (into the treatment arm or the control group) is left to the play of chance.
When patients are randomly assigned to the treatment or control arm of a clinical trial, the results are free from a specific type of selection bias.
For example, without randomization, scientists may consciously or subconsciously assign patients to the treatment arm if they look more likely to benefit from the treatment under investigation, thus making a treatment look more beneficial overall than it may in fact be.
Conversely, if scientists are looking to demonstrate that a certain treatment is ineffective or unsafe, they may assign patients with a higher risk of complications or a lower chance of success to receive that treatment.
The chances of allocation bias may be considered higher in trials run by researchers either directly or indirectly funded by a pharmaceutical company looking to prove efficacy and safety of a new drug. This is why researchers must disclose any potential conflict of interest when conducting a clinical trial, as pharmaceutical manufacturers have a clear (and usually very large) financial interest in achieving positive results.
Knowing which patients are getting the experimental drug can result in sound motivations as well as questionable ones. Doctors could have a well-intentioned influence over the results.
If the active treatment seems to be producing bad side-effects, for example, doctors may protect certain types of patient from the drug. Treating subjects in the trial differently in this way reduces the integrity of like-for-like comparisons and gives false results.
In summary, randomization in a clinical trial helps to remove bias and to ensure that the two groups being compared are truly similar, and that - as far as possible - the only difference that could influence the outcome in their disease is whether or not they are receiving the new drug.
Double- and single-blind trials
In addition to randomization, a high-quality clinical trial is typically double-blind, or single-blind where double-blinding is not possible. This means that the patients do not know if they are receiving the active treatment or a placebo (single-blind), or that patients and the physicians treating and assessing those patients are unaware of who is in the active treatment group (double-blind).
This kind of "blinding" helps to prevent results being skewed by patients who feel better simply because they know they are receiving active treatment (or, conversely, who do not report feeling better because they know they are receiving a placebo or sham treatment). Similarly, a double-blind trial reduces the potential for a treatment effect in patients and reduces the risk of the physicians or researchers reporting greater effects in the treatment group or lesser effects in the control group.
In some cases, it is difficult to blind patients and/or researchers to who is in the treatment and control groups. For example, research on acupuncture can be difficult to carry out in a double-blind fashion as practitioners are inherently aware that they are administering a sham or active treatment. In such cases, a trial may be single-blind (i.e. patients do not know if they are receiving active or sham treatment), but may use an independent third party practitioner to assess patients without knowing if they received the active or sham treatment.
Fast facts on randomized controlled trials
Here are some key points about RCTs. More detail and supporting information is in the body of this article.
- Randomized controlled trials are the gold standard of clinical testing applied to new medical interventions.
- RCTs are usually required in pharmaceutical testing programs before regulators will allow new drugs to be sold.
- Randomization means pure chance is used to decide who gets the new treatment and who goes into the comparison group of the trial.
- Randomization is done so that 'selection bias' is removed when treatment is allocated, making the results of the comparison more reliable.
- RCTs are "controlled" so that investigators can reasonably expect any health effects to be the result of the drug treatment, by observing effects in the reference group not given the new drug.
- Placebos may be used for the control group - dummy treatments that look the same as the experimental treatment but contain nothing.
- It is not always ethical or practical to give a placebo, such as when this would mean denying treatment to people who have a life-threatening or serious illness.
- A standard treatment that is already established against the disease can be used in the control group for comparison.
Why are clinical trials 'controlled'?
The purpose of a control group in a randomized controlled trial is to help reduce the likelihood that any benefits (and risks) identified during the process can be attributed to factors other than the drug itself.
The trial could use a placebo control - a dummy pill that looks exactly like the treatment, but lacks the active ingredient.
If there is no control - no reference or comparison group - any improvement in health or other outcome cannot be attributed to the drug or treatment.
There may be others factors about being in the clinical trial that could explain the results, and without comparing what happens in similar patients taking part under similar conditions - but not getting the new drug - there is no proper measurement of any health changes that may be observed.
This value of employing a control group is produced only if the trial is sufficiently large. It must have enough people taking part to ensure that chance differences and unusual cases do not have a large effect on the results.
A control group is typically made up of people who are matched for age, sex and ethnicity, along with any other factors that may influence the effect of a drug or treatment, such as weight, smoking status or comorbidities.
The control group may receive a placebo - a dummy treatment that looks exactly the same as the drug being tested, but that does not actually contain the active agent - or may receive standard treatment without the additional treatment under investigation.
In some cases, typically those investigating the benefits of an intervention in healthy individuals, the control group may receive no treatment and simply be made up of individuals similar to those receiving a supplement or therapy.
The quality of the control group is important, in terms of how well its participants match those of the active group. For example, randomization (explained above) helps to ensure that there is no bias influencing the selection of people into the control group.
Good quality clinical trials will publish baseline measurements for both the treatment and control arms of the trial, allowing for direct comparison.
Comparison with standard treatment
Many trials investigating new drugs or treatments for a disease are designed so that the control group receives an established (standard) treatment for that disease. The purpose of this type of control is to find out if there is any comparative benefit from the new drug versus the standard treatment. Even if the new drug does appear to have a beneficial effect, the established treatment may still be safer and more efficacious.
Comparative drug trials are important beyond the pharmaceutical development process as they can help guide decisions about allocation of healthcare resources.
For example, healthcare policymakers around the world are particularly interested in how a new drug fares against existing treatment options, taking into account cost-effectiveness, effects on quality of life, and other factors that add to the picture of overall benefit and cost to society, as well as to individuals.
Policy makers also have to account for the lack of diversity in clinical trials when making decisions about health care guidelines and funding. Historically, clinical trials have typically been carried out using white male patients, resulting in the approval of a range of drugs and interventions that have subsequently been seen to be less effective or riskier in a different demographic.
Research in non-human animals or a limited section of the population is, in most cases, insufficient to recommend widespread use of a drug or treatment for the general population. Indeed, the approval of some drugs based on animal research has led to significant harm to human health as non-human animals are generally a poor model for the human response to a drug or treatment.
Impractical or unethical use of dummy treatments in RCTs
Rigorous scientific trial design is not always practical. It can be difficult to achieve a true placebo, to disguise the dummy completely. And in some cases, it is unethical to give a dummy treatment.
Desire for a randomized controlled trial may be defeated by the following examples of practical limitation:
- Treatments that are more invasive - involving devices or surgery, for example - may be impossible to mock-up realistically in the comparison group
- There may be too few people with a certain disease available for investigation in both treatment and non-treatment groups
- The recruitment of patients to a particular trial may be too difficult.
A placebo-controlled trial is often simply an unfair prospect for participants. For example, a placebo is not allowed to be given for comparison in serious or life-limiting disease if it would mean denying usual care.
Here, the comparison group would instead be given a treatment that is already available, so that the patients would not be sacrificing their standard care for the sake of a dummy treatment.
If no existing treatment is available, another trial design may be used. Decisions about whether trial designs are fair on the participants are made by independent ethical review boards. A clinical trial cannot go ahead without this ethical approval.