Map of life expectancy at birth from Global Education Project.

Tuesday, March 04, 2008

Cross of Gold

The Randomized Controlled Trial (RCT) is the so-called gold standard for determining what is and is not evidenced based therapy. When we talk about the clinical trials submitted to the FDA in new drug applications, that's what we're talking about. When people call for Evidence Based Medicine, they mostly want doctors to use interventions that have been "proven" through RCTs.

So it was refreshing to day to hear Malcolm Gordon, who is an official of the Center for Mental Health Services, tell us about many of the reasons why RCTs don't necessarily tell us what does and does not work when it comes to mental health services. For example, they are very expensive to do, so they seldom go on long enough or have large enough sample sizes to support adequate subgroup analysis. An intervention that works for some people might not work for others, but we just don't get the information we need from RCTs to understand that. Attrition is a big problem in trials, particularly of mental health treatments, and it can seriously mess up the statistics. A truly major problem is that interventions are developed, implemented and tested in highly specialized settings and circumstances that just don't correspond to anything that is likely to happen in the real world of the mental health system.

I'd like to offer another major problem with RCTs in the context of mental health, which expect Dr. Gordon wanted to get to but didn't have time. They depend on very highly specified, standardized interventions. But counseling interventions just aren't like that. You can't read people a script, you have to respond to specific needs, capacities, and wishes of the individual client. Some people need more of this and less of that, more or less time in therapy. Above all, a lot depends on the talent of the counselor and the fit between counselor and client. A counselor might help some people and not others -- even hurt some people and help some people -- and the same people, if they went to different counselors, would end up with different patterns of outcomes, even though everybody is implementing the "same" intervention.

What is more, these interventions are given in settings that have varying capacities for case management, flexibility with scheduling, providing supportive services; different language and cultural capabilities; different physical and social settings. All of this matters as much or more than the intervention manual.

That's why, in my view, the so-called evidence base for mental health consists mostly of cognitive behavioral therapy, and pills. Pills for the most part don't work all that well, but they're easy to standardize and test in RCTs. Of course, they're also very profitable so a lot of money is invested in getting them approved. CBT, among all the counseling interventions, is probably easiest to standardized, and it's also relatively short term, so it's cheaper to test and you have less trouble with attrition.

In other words, we're looking under the lamp post because that's where the light is. The first part of the morning consisted of listening to stories from survivors of trauma -- women who had been sexually abused as children, or victims of violent crime. They were convinced that they had benefited enormously from therapy -- that it had enabled them to put their lives back together. But I'm quite sure that none of the counseling they received has been subjected to any form of RCT.

No comments: