Is cognitive behavioral therapy (CBT) especially effective in the treatment of anxiety? A recent trial - it's the one I've been dissecting here in three postings - says yes, CBT works well for children and adolescents with substantial anxiety disorders. A fuller account might say, we don't know.
The issue at stake is less the particular study - it demonstrated the efficacy of Zoloft and CBT, alone or (better) in combination - than how we interpret studies of this sort and how they affect health care policy. To gain perspective, let's pull back a few years, to the late 1970s, just as the government was getting into the business of funding large-scale psychotherapy outcome trials.
Psychologists had been conducting efficacy studies for some years, going back to the research of the early critics of psychoanalysis. By in the 1970s, the schools of treatment had multiplied. I don't recall the precise numbers, but Morris Parloff, a leader in questions of research and mental health policy, estimated that there were, say 300 named treatments and perhaps 150 indications for treatment and 4 broad age categories and 3 levels of severity, so that if you wanted to know whether strategic therapy worked for moderate levels of anorexia in adolescents, and so on, for each combination of factors, you would need to fund 540,000 outcome trials, and that is without "horserace" experiments pitting one intervention against another.
Scientists at the National Institute of Mental Health instead decided to look at a general question, along the lines of "does psychotherapy work?" The intent was to satisfy health insurers and the Federal Government. (Congress was interested in these issues, because of psychotherapy expenses in Medicare, Veterans Affairs, the Department of Defense, and so forth.) The result was the "manualization" of therapies, reducing CBT to a series of rules, along with a proxy for psychodynamic psychotherapy, called interpersonal psychotherapy (IPT), which shortly devolved into a variant of CBT. These treatments were then applied to a few indications, principally moderately severe depression in adults. For the proponents of CBT, the idea was to test their brand of therapy; but for most of the field, the chosen treatments were proxies for "psychotherapy" in general, pending further funding and research on mechanisms of change - studies that might lead to principled distinctions among approaches.
That paradisiacal age never arrived. Instead, medication took center stage. And the manualized treatments gained a central importance that was due to their availability and the interest of their proponents in outcome research. Whether CBT is better than other treatments is unknown. The best evidence, to my reading of the literature, is that for almost all indications, all therapies look equally effective. The leading critique along these lines is The Great Psychotherapy Debate, by Bruce Wampold. I will return to this book in future postings; for now, suffice it to say that Wampold demonstrates that CBT looks just like other treatments, and that when it works it is not at all clear that the mechanism has to do changes in patients' cognitions.
As for the study at hand: The psychotherapy arm offered anxious children 14 CBT sessions of 60 minutes each. The "combination" arm offered the children those 14 sessions plus 8 briefer medication-assessment sessions, with Zoloft administered. The placebo condition was the 8 medication sessions and the dispensing of a dummy pill. If we entertain the hypothesis that children do better when you spend time with them and focus on their problems, we may doubt that this research studied CBT at all. Only a comparison psychotherapy or an adequate behavioral placebo (14 hours of instruction in meditation, say) would lead us to think that the trial evaluated CBT in particular.
This study is not a bad one; in fact, its methods are excellent. But because the focus of any outcome trial is necessarily narrow, all research in this field needs to be put in context and integrated with other findings.
In a prior posting, I suggested that the Zoloft doses used in this study were so high as to limit the practical utility of its findings. Something parallel can be said about the psychotherapy component. We can be reasonably sure that psychotherapy and medication work for anxiety disorders in children. For policy purposes, that result is important: broadly speaking, this sort of treatment should be offered. But for CBT, as for Zoloft, whether this therapy should be preferred remains unknown.