This article is more than 5 years old. Information may no longer be current.
How do you rank and use medical evidence?
Click Here to Manage Email Alerts
Randomized clinical trials are the gold standard
From the point of view of testing clinical questions that relate to therapeutic interventions, the randomized clinical trial is still the gold standard. The reason for this is that if researchers ask a solid question; the trial is adequately designed and powered; and the randomization works, you will get an answer, whether or not it is the outcome that was hypothesized. Randomized clinical trials thus provide the opportunity to answer important clinical questions.
|
|
The randomized clinical trial contrasts with other approaches, such as meta-analyses and epidemiology studies. These might provide very important and compelling information, but, in my mind, they more frequently suggest hypotheses to be tested than the true answer, in contrast to a well-designed and executed randomized clinical trial. Dependent on the analytical approach used, it may be possible to get very different results. This is very much the case with meta-analyses where the analytical approach being applied can result in the exclusion of important studies, some of which may be quite compelling.
At this stage, for providing clinical information regarding one therapy vs. another or even the safety of one therapy vs. another, the results of randomized clinical trials must still be considered the best evidence we have.
Steven E. Kahn, MD, is a professor of medicine at VA Puget Sound Health Care System and University of Washington, Seattle.
Mixed bag of evidence to determine drug safety
The scientific community is in agreement that good clinical trials give good answers. To determine efficacy, we need to rely on large, well-done clinical trials and meta-analyses. To determine safety, we need to use all the information we can get our hands on; clinical trials are helpful but do not provide all the answers.
|
|
There are three reasons why clinical trials are not helpful. First, there is often a lag time to harm. Second, there is a statistical power issue in that very few trials are large enough to detect something like a rare adverse event. Third, the researchers may not be looking for a particular event; it may be something new and unexpected. With retrospective observational studies, people can be studied for extended periods of time; a large number of patients are usually included; and researchers know what they are looking for.
Traditionally, the FDA has not relied much on retrospective observational studies and meta-analyses; it prefers individual trials. People often behave as if all clinical trials are the same — and they are not. There are good trials and there are bad trials. Just because it is a large trial does not mean it is a very useful source of information.
Curt D. Furberg, MD, PhD, is professor of public health sciences at Wake Forest University School of Medicine.