is it time to rethink clinical trial guidelines?

May 9, 2009

There’s much to be sad for the idea that when you’re feeling sad, taking a pill for a pick-me up, most of which is in the somewhat fuzzy realm of psychology and psychiatry. On a more scientific note however, we can say that antidepressants are probably only going to work 40% to 55% of the time. Why are these drugs having a relatively low rate of effectiveness? The problem may not be with the antidepressants themselves, but with the guidelines for their Stage III clinical trials, the results of which are used for regulatory approval and marketing.

pill target

A group of researchers took a very good representation of patients being treated for depression known as the STAR*D project, and tried to apply the criteria for enrolling into Stage III antidepressant trials to a sample of 2,855 of them. The result? Just 22.2% of these patients qualified to be in the trials while almost 8 in 10 had a number of additional health and mental problems which barred them from entering. Since the data on those sample patients was collected from 2001 to 2004, the study could also take a look how well those who were qualified for enrollment ultimately fared.

As you might expect, the response rate to the drug was good with 51.6% of the qualified groups saying it was having an effect and their depressions subsided faster, but they also suffered from a 34.4% rate of remission compared to a 24.7% rate for the group that would be rejected from clinical trials. Basically, they were easier cases that relapsed more often than patients with compounding factors like a longer family history of mental disorders, drug addiction or alcoholism. But the big problem is the fact that they weren’t a good representation of an average patient with a case of depression and the results would be rosier than they should’ve been and higher relapse rates which could’ve affected the treatment they would’ve received from a psychiatrist.

Before you start saying that drug companies were manipulating the population for marketing purposes, you’ll need to consider that all trials use homogenous groups of patients to minimize adverse side-effects in those with a multitude of conditions and what they’re really testing is how well the drug treats what it was supposed to treat. However, as we’ve seen above, the results might not work for a typical patient who has multiple health conditions that could alter the response to the drug being tested. And that’s why years after a drug was on the market, we might look back and say: “hey, this wasn’t nearly as good as advertised! what gives?” even though there was noting wrong or fraudulent about the trielson which the promises of efficacy were based.

So maybe Stage III trials should be based not on a homogenous population which has only a certain number of complicating factors, but on groups that accurately represent the patients with the condition the drugs were intended to treat for much more realistic efficacy data? Yes, the number of adverse side effects will rise and so will the complexity of the trials, but in the end, we’ll have better information about our medication. The authors of the study on antidepressant trials certainly seem to think so in their conclusions.

See: Wisniewski, S., et al (2009). Can Phase III Trial Results of Antidepressant Medications Be Generalized to Clinical Practice? American Journal of Psychiatry, 166 (5), 599-607 DOI: 10.1176/appi.ajp.2008.08071027

Share on FacebookTweet about this on TwitterShare on RedditShare on LinkedInShare on Google+Share on StumbleUpon